Carbohydrate Mimics: Concepts and Methods

Beat Ernst, Gerald W. Hart, Pierre Sinay

Carbohydrates in Chemistry and Biology

Related Titles from WILEY-VCH A. E. Stiitz (ed.) Iminosugars as Glycosidase Inhibitors XIV, 397 pages with 152 figures, 24 in color, and 48 tables plus a comprehensive 123-table of glycosidase inhibitors 1999. ISBN 3-527-29544-5. Cloth

Y. Chapleur (ed.) Carbohydrate Mimics - Concepts and Methods XXVIII, 604 pages with 375 figures and 52 tables 1998. ISBN 3-527-29526-7. Cloth H. van Bekkum, H. Roper, A. G. Vorhaben (eds.) Carbohydrates and Organic Raw Materials I11 X, 315 pages with 156 figures and 68 tables 1996. ISBN 3-527-30079-1. Cloth

T. K. Lindhorst Essentials of Carbohydrate Chemistry and Biochemistry XIV, 323 pages with 244 figures and 11 tables 2000. ISBN 3-527-29543-7. Paperback

Beat Ernst, Gerald W. Hart, Pierre Sinay

Carbohydrates in Chemistry and Biology Part I

Chemistry of Saccharides

VOl. 1

Chemical Synthesis of Glycosides and Glycomimetics

@WILEY-VCH Weinheim - New York * Chichester Brisbane - Singapore * Toronto

Beat Emst, Gerald W. Hart, Pierre Sinay

Carbohydrates in Chemistry and Biology Part I

Chemistry of Saccharides

VOl. 2

Enzymatic Synthesis of Glycosides and Carbohydrate-Receptor Interaction

Weinheim - New York - Chichester Brisbane - Singapore - Toronto

Beat Ernst, Gerald W. Hart, Pierre Sinay

Carbohydrates in Chemistry and Biology Part I1

Biology of Saccharides

VOl. 3

Biosynthesis and Degradation of Glycoconjugates

Weinheim - New York * Chichester Brisbane - Singapore * Toronto

Beat Ernst, Gerald W. Hart, Pierre Sinay

Carbohydrates in Chemistry and Biology Part I1

Biology of Saccharides

VOl. 4

Lectins and Saccharide Biology

@WILEY-VCH Weinheim - New York * Chichester Brisbane * Singapore * Toronto

Prof. Dr. B. Ernst Institut fur Molekulare Pharmazie Universitat Basel Klingenbergstrasse SO 4051 Basel Switzerland

T~ -~~

Prof. Dr. G. W. Hart Dept. of Biological Chemistry Johns Hopkins University School of Medicine 725 N. Wolf St. Rm. 401 Hunterian Baltimore, MD 21205-2185 USA ___p

Prof. Dr. P. Sinay Dept. de Chimie, URA 1686 Ecole Normale SupCrieure 24 rue Lhomond 75231ParisCedex05 France

~~~p~~

~

~~~~

.

p

~

~

.

-1

~

~

~

This book was carefully produced. Nevertheless, editors, authors and publisher do not warrant the information contained therein to be free of errors. Readers are advised to keep in mind that statements, data, illustrations, procedural details or other items may inadvertently be inaccurate.

!-.~p-p.-p-ppp

~p

p~

-~

Cover picture: 3,6-dideoxyhexose binding to antibody se155.4 Courtesy of David Bundle, University of Alberta

Library of Congress Card No.: applied for A catalogue record for this book is available from the British Library. Die Deutsche Bibliothek - CIP Cataloguing-in-Publication-Data A catalogue record for this publication is available from Die Deutsche Bibliothek ISBN 3-527-295 11-9

0 WILEY-VCH Verlag GmbH, D-69469 Weinheim (Federal Republic of Germany). 2000 Printed on acid-free and chlorine-free paper. All rights reserved (including those of translation in other languages). No part of this book may be reproduced in any form - by photoprinting, microfilm, or any other means - nor transmitted or translated into machine language without written permission from the publishers. Registered names, trademarks, etc. used in this book, even when not specifically marked as such, are not to be considered unprotected by law. Composition: Asco Typesetters, Hongkong. Printing: betz-druck gmbH, D-6429 1Darmstadt. Bookbinding: Wilhelm Osswald & Co., D-67433 Neustadt. Printed in the Federal Republic of Germany.

About the Editors

Beat Ernst studied chemistry at the ETH in Zurich, Switzerland, where he completed his PhD-thesis about novel tricyclic hydrocarbons under the guidance of Professors Oskar Jeger and Camille Ganter. After his post-doctoral research on tirandamycinic acid with Professor R. E. Ireland at caltech in Pasadena, he joined the Central Research Laboratories of Ciba-Geigy (now Novartis) in Basel where he headed the Carbohydrate Research Section. In 1998 he became Professor of Molecular Pharmacy at the University of Basel, Switzerland. His major research interests are Carbohydrate Chemistry and Glycobiologymainly the chemical and enzymatic syntheses of oligosaccharides, investigation of carbohydrate/lectin interactions and carbohydrate mimics. In 1991 he was awarded the Werner-Price of the Swiss Chemical Society, and in 1993 he became CIBA-fellow for his research contributions.

Gerald Warren Hart is Director and DeLamar Professor of Biological Chemistry at Johns Hopkins University in Baltimore, MD. He studied biology, and chemistry at Washburn University and completed his PhD-thesis in developmental biology at Kansas State University before he was appointed assistant professor of Biological Chemistry at Johns Hopkins University School of Medicine. From 1993 to 1997 he held a chair at the department of Biochemistry and Molecular Genetics at the University of Alabama before he returned to Johns Hopkins University. In 1989 he founded the journal Glycobiology which became the leading journal in the field. His research results and contributions to the field of glycobiology are honored with many prices and awards-President of the Society of Glycobiology, Member of the Council of the ASBMB, First International Glycoconjugate Organization Award, just to name a few.

VI

About the Editors Pierre Sinay studied chemistry at Ecole Nationale Superieure des Industries Chimiques, Nancy, France, and received his PhD from the University of Nancy. After his post-doctoral research at Harvard University he became Professor at the University of Orleans before he was appointed professor of organic chemistry at Pierre et Marie Curie University and Ecole Normale Superieure in Paris. His main research focuses on the chemical synthesis of oligosaccharides of biological relevance and on the development of new synthetic methods in the carbohydrate field. He is Editor-in-Chief of the journal Carbohydrate Letters and he has published numerous scientific papers and patents. In 1987 he was elected president of the organic chemistry division of the French Chemical Society. In 1978 he was awarded the Le Be1 Price of the French Chemical Society, in 1996 he was triple-honoured with the Desnuelles Price, the Bethellot Medal and with the election as corresponding member of the French Academy of Sciences.

Contents

Part I

.

Vol 1

Chemistry of Saccharides Chemical Synthesis of Glycosides and Glycomimetics List of Contributors ...................................................

LV

Abbreviations Used in Volumes 1 and 2 ...........................

LXIII

I

Chemical Synthesis of Glycosides ..................................

1

1

Introduction to Volumes 1 and 2 .......................................

3

2

Trichloroacetimidates .................................................. Richard R. Schmidt and Karl-Heinz Jung Introduction ............................................................ Methods ................................................................ 0-Glycosides ............... ......................................... Synthesis of Oligosaccharides ......................................... p-Glucosides, p.Galactosides, a.Rhanmosides. etc.................... Aminosugar Trichloroacetimidates.................................... p-Mannosides ......................................................... 2-Deoxyglycosides .................................................... Miscellaneous Compounds ........................................... Complex Oligosaccharides............................................ Inositol Glycosides.................................................... Glycosylation of Sphingosine Derivatives and Mimics .............. Glycosylation of Amino Acids ....................................... Polycyclic and Macrocyclic Glycosides ..............................

5

2.1 2.2 2.3 2.3.1

2.3.2 2.3.3 2.3.4 2.3.5

5 6 7 7 7 8 13 13 14 14 36 38 40 42

VIII

Contents

2.3.6 2.3.7 2.4 2.5 2.6 2.7

Glycosides of Phosphoric and Carboxylic Acids .................... Solid-Phase Synthesis................................................. S-Glycosides .......................................................... N- and P-Glycosides.................................................. C-Glycosides.......................................................... Conclusion and Outlook ............................................. References ............................................................

3

Iterative Assembly of Glycals and Glycal Derivatives: The Synthesis of Glycosylated Natural Products and Complex Oligosaccharides... Lawrence J . Williams. Robert M . Garbaccio. and Samuel J . Danishefsky Introduction .......................................................... Ciclamycin 0 .......................................................... Allosamidin ........................................................... KS-502 and Rebeccamycin ........................................... Extension to Thioethyl Donors ...................................... LewisY ................................................................. Globo H .............................................................. KH-1 .................................................................. Concluding Remarks ................................................. Acknowledgments .................................................... References ............................................................

3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9

4

4.1 4.2 4.2.1 4.2.2 4.2.3 4.3 4.3.1 4.3.2

4.4 4.4.1 4.4.2 4.4.3

44 45 49 51 51 53 53

61

61 64 66 69 74 76 82 86 90 90 90

Thioglycosides ........................................................ 93 Stefan Oscarson Introduction .......................................................... 93 Synthesis of Thioglycosides .......................................... 94 From Anomeric Acetates............................................. 94 From Glycosyl Halides ............................................... 95 Protecting Group Manipulations in Thioglycosides ................. 96 Glycosylations with Thioglycoside Donors .......................... 97 A Two-step Procedure: Transformation of Thioglycosides into 97 Other Types of Glycosyl Donors ..................................... 99 Direct Activation of Thioglycoside Donors ......................... 99 Heavy Metal Salt Promoters ......................................... Halonium, sulfonium and carbonium type promoters ............... 100 Single-ElectronActivation ........................................... 106 Other Types of Donors With an Anomeric Sulfur .................. 108 Applications of Thioglycosides....................................... 110 Block Syntheses, Orthogonal Glycosylations ........................ 110 Thioglycosides as Acceptors.......................................... 110 Thioglycosides as Both Donors and Acceptors ...................... 111 Intramolecular Glycosidations ....................................... 112 Solid Phase Synthesis ................................................. 113 Reference6 ............................................................ 113

Contents 5

5.1 5.2 5.3 5.3.1 5.3.2 5.3.3 5.3.4 5.3.5

5.3.6 5.4 5.4.1 5.4.2 5.4.3 5.4.4

6

6.1 6.2 6.3 6.4 6.5 6.6 6.7 7

7.1 7.2 7.3 7.4 7.4.1

IX

Glycosylation Methods: Use of Phosphites ........................... 117 Zhiyuan Zhang and Chi-Huey Wong Introduction ........................................................... 117 118 Preparation of Glycosyl Phosphites .................................. Glycosylation using Glycosyl Phosphites ............................ 119 119 Mechanism ............................................................ Low Temperature-Dependent Stereoselectivity ...................... 121 Glycosylation of Sialyl Phosphites ................................... 122 Glycosylation of C-2-Acylated Glycosyl Phosphites ................. 123 Glycosylation with C-2-O-Benzylated Glycosyl Phosphites ......... 124 Glycosylation using Glucosyl Phosphites with a Benzyl Group at C-2 .................................................................... 124 Glycosylation using Galactosyl and Fucosyl Phosphites with a 125 Benzyl Group at C-2.................................................. Glycosylation using other Glycosyl Phosphites with a Benzyl Group at C-2 .......................................................... 126 Glycosylation with 2-Deoxy Glycosyl Phosphites ................... 127 128 Other Applications of Glycosyl Phosphites .......................... Synthesis of CMP-NeuAc ..... .................................... 129 Synthesis of GDP-Fucose ............................................ 129 Formation of Glycosyl Phosphonate. ................................ 131 Transformation to other Types of Glycosyl Donor ..................131 131 Phosphate ............................................................. Phosphorimidate ...................................................... 131 132 References............................................................. Glycosylation Methods: Use of n-Pentenyl Glycosides ............... 135 Bert Fraser.Reid. G. Anilkumar. Mark R. Gilbert. Subodh Joshi. and Ralf Kraehmer Introduction ........................................................... 135 Fundamental Reactions .............................................. 135 Determination of Relative Reactivities ............................... 138 n-Pentenyl Orthoesters as Glycosyl Donors .......................... 141 144 n-Pentenyl Orthoesters as Latent C2 Esters .......................... 146 Protecting Groups .................................................... Solid-Phase Iterative Couple-Deprotect-Couple Strategy .......... 146 153 References .............................................................. Glycosylidene Diazirines .............................................. Andrea Vasella. Bruno Bernet. Martin Weber. and Wolfgang Wenger Introduction ........................................................... Synthesis of Glycosylidene Diazirines ................................ Stability of the Glycosylidene Diazirines ............................. Glycosidation by Glycosylidene Diazirines .......................... General Aspects .......................................................

155 155 155 158 158 158

X 7.4.2 7.4.3 7.5 7.6 7.7

8

8.1 8.2 8.2.1 8.2.2 8.2.3 8.2.4 8.2.5 8.2.6 8.2.7 8.2.8 8.2.9 8.3 8.3.1 8.3.2 8.3.3 8.3.4 8.3.5

Contents

Glycosidation of Strongly Acidic Hydroxy Compounds ............ 162 Glycosidation of Phenols ............................................. 162 163 Glycosidation of Fluorinated Alcohols .............................. Glycosylation of Weakly Acidic Hydroxy Compounds ............. 163 163 Glycosidation of Monovalent Alcohols .............................. Glycosidation of Diols and Triols ................................... 164 Synthesis of Spirocyclopropanes ..................................... 168 Addition to Aldehydes and Ketones ................................. 170 Exploratory Use of Diazirines: Formation of Glycosyl Phosphines, Stannanes. N.Sulfonylamines. Esters. Boranes. and Alanes. and of 1.l-Difluorides ........................................................ 171 Acknowledgments .................................................... 174 174 References ............................................................ Glycosylation Methods: Alkylations of Reducing Sugars ............ 177 Jun-ichi Tamura Introduction .......................................................... 177 Anomeric O-Alkylation .............................................. 177 Anomeric O-Alkylation of Ribofuranose with Primary Triflates: Effect of the Protecting Group at 0 - 5 of Ribofuranose ............. 178 Anomeric O-Alkylation of Mannofuranose with Primary Triflates: 179 The Crown Ether Effect .............................................. Anomeric O-Alkylation of Gluco- and Galactopyranoses with Primary Triflates: High P-Selectivity as a Result of the Reactive Anomeric p-Anion .................................................... 180 Anomeric O-Alkylation of Acyl-Protected Nucleophiles with Primary Triflates ...................................................... 181 Anomeric O-Alkylation of Mannopyranose with Primary Triflates: Possibility of Intramolecular Complexation of the Nucleophile .... 184 Anomeric O-Alkylation of KDO with Primary Triflates ............ 185 Anomeric O-Alkylation of Some Protected Aldoses with Primary 186 Triflate ................................................................ Anomeric O-Alkylation of Unprotected Aldoses with Primary 187 Triflate, Bromides, and Cyclic Sulfates .............................. Anomeric O-Alkylation with Secondary Triflates and Nonaflate ... 188 Glycosylation via the Locked Anomeric Configuration ............. 189 Synthesis of Methyl. Allyl. and Benzyl Glycosides via Stannylene Acetals ................................................................ 189 Epimerization at C-2 by the Locked Anomeric Configuration 189 Method ............................................................... The Locked Anomeric Configuration Method for Rhamnosyl Stannylene Acetal ..................................................... 190 The Locked Anomeric Configuration Method for Mannosyl Stannylene Acetal: Isomerization of Acetal [25. 261 ................. 190 The Locked Anomeric Configuration Method for Stannylene Acetal with the Glucose Configuration [25. 261 ..................... 191

Contents

XI

8.4

Conclusion ............................................................ References .............................................................

9

Other Methods of Glycosylation ...................................... 195 Luigi Panza and Luigi Lay 195 Introduction and Summary ........................................... Highlights ............................................................. 195 Enol Ethers ........................................................... 197 Endo-En01 Ethers ..................................................... 198 Exo-Enol Ethers ...................................................... 201 Endo-Glycals ......................................................... 202 Exo-Glycals .......................................................... 204 Vinyl Glycosides ...................................................... 206 1-Hydroxy Sugars ..................................................... 209 Acidic Activation ..................................................... 210 Acidic Activation With Additional Reagents ........................ 211 Dehydrative Glycosylation ........................................... 212 In the Presence of the Acceptor From the Beginning ................213 214 Mitsunobu Glycosylation ............................................. I-0-Silyl Glycosides .................................................. 215 Esters and Related Derivatives ....................................... 216 216 Esters .................................................................. Sugar Carbonates and Derivatives ................................... 221 Orthoesters and Oxazolines........................................... 223 Phosphorus and Sulfur Derivatives .................................. 229 References ............................................................. 233

9.1 9.1.1 9.1.2 9.1.3 9.1.4 9.1.5 9.2 9.2.1 9.2.2 9.2.3 9.2.4 9.3 9.3.1 9.3.2 9.3.3 9.3.4

10 10.1 10.2 10.3 10.4 10.4.1 10.4.2 10.5

10.5.1

192 193

Polymer-Supported Synthesis of Oligosaccharides.................... 239 Jiri J. Krepinsky and Stephen P . Douglas Introduction ........................................................... 239 General Reflections ................................................... 240 Polymer Supports ..................................................... 246 One-Phase Systems (Syntheses in Solution) .......................... 247 248 Polyethyleneglycol,. monomethylether (MPEG) ..................... 249 Linear Polystyrene .................................................... Linkers ................................................................ 250 Succinoyl Diester ..................................................... 250 Dioxyxylyl Diether (DOX) ........................................... 252 254 Chemistry Investigations ............................................. Two-Phase Systems (Syntheses on Solid Supports) ..................255 Controlled Pore Glass ................................................ 256 256 Cross-Linked Polystyrene ............................................. Polyethylene Grafts on Cross-Linked Polystyrene. ..................256 Linkers ................................................................ 259 259 Dialkyl- or Diaryl-Silyl ............................................... Thioglycoside Linkers ................................................ 259

XI1

10.6 10.7 10.8 10.9 11 11.1 11.1.1 11.1.2

11.1.3 11.2 11.2.1

11.2.2 11.3 11.3.1

11.3.2

Contents

Linkers Cleavable by Photolysis ..................................... Examples of Syntheses ............................................... Combinatorial Libraries .............................................. Capping ............................................................... Concluding Remarks ................................................. References ............................................................

260 260 261 262 262 262

Glycopeptide Synthesis in Solution and on the Solid Phase .......... 267 Horst Kunz and Michael Schultz Introduction .......................................................... 267 Which Protecting Groups are Suitable for Carbohydrates (Table 2)? ............................................................. 269 Which Glycosylation Methods are Useful for the Formation of Glycopeptides?........................................................ 271 Formation of Asparagine N-Glycosides ............................. 271 a-Fucosylation ........................................................ 271 272 Formation of the P-Lactosamine Linkage ........................... a-Sialylation .......................................................... 272 Glycopeptides Containing Particularly Sensitive Linkages .......... 272 Acid Sensitivity ....................................................... 273 273 Base Sensitivity ....................................................... Synthesis of Glycopeptides in Solution .............................. 274 0-Glycopeptides ...................................................... 274 Glycopeptides Carrying N-Acetylgalactosamine (Tn-Antigen)...... 274 Glycopeptides Carrying the T-Antigen (Gal-GalNAc) ............. 276 Glycopeptides Carrying the Sialyl T Antigen (NeuAca2,6[Galpl,31GalNAc) ....................................... 278 279 Glycopeptides Carrying 0-GlcNAc .................................. N-Glycopeptides ...................................................... 280 N-Glycopeptides Carrying Natural Saccharide Side-Chains ........ 280 N-Glycopeptides with Lewis-Type Saccharide Side-Chains ......... 285 Glycopeptide Synthesis on the Solid Phase .......................... 286 0-Glycopeptides ...................................................... 287 Glycopeptides Carrying N-Acetylgalactosamine (Tn-Antigen)...... 287 0-Glycopeptides Carrying the T Antigen (Gal-GalNAc) .......... 290 0-Glycopeptides Carrying the Sialyl Tn Antigen (NeuNAc~a2, 6. GalNAc) .............................................................. 291 0-Glycopeptides Carrying the 2, 3.Sialyl T Antigen ................. 293 0-Glycopeptides Carrying 0-GlcNAc Side-Chains .................294 295 0-Glycopeptides Carrying 0-Linked Fucose ........................ 0-Glycopeptides Carrying a Sialyl Lewis Antigen Structure ........ 296 N-Glycopeptides ...................................................... 297 The Construction of N-Glycopeptide Libraries on the Solid 298 Phase .................................................................. Sequential N-Glycopeptide Synthesis on the Solid Phase with Oligosaccharides from Natural Sources.............................. 299

Contents

XI11

11.4

Conclusion ............................................................ References .............................................................

12

Glycolipid Synthesis ................................................... 305 Hidehuru Ishida Introduction ........................................................... 305 Synthesis of Ganglio-Series Gangliosides ............................ 305 305 Retrosynthetic Analysis of Ganglioside G D l a ...................... Preparation of Sialylgalactose Donor as Building Block ............ 306 Construction of Oligosaccharide ..................................... 308 Transformation of Oligosaccharide into Glycolipid .................310 Synthesis of Polysialo Ganglio-Series Gangliosides.................. 311 Retrosynthetic Analysis of GQl b .................................... 313 313 Preparation of Building Block ........................................ Construction of Oligosaccharide ..................................... 314 315 Conclusion ............................................................ References ............................................................. 316

12.1 12.2 12.2.1 12.2.2 12.2.3 12.2.4 12.3 12.3.1 12.3.2 12.3.3 12.4 13

13.1 13.2 13.2.1

13.2.2 13.2.3 13.2.4 13.2.5

13.2.6 13.2.7 13.2.8 13.2.9 13.3 13.4

Stereoselective Synthesis of P-Mannosides............................ Vince Pozsguy Introduction ........................................................... Chemical Methods .................................................... Glycosylation with Mannosyl Donors ............................... Mannosylation using Insoluble Promoters ........................... The Sulfonate Approach .............................................. Intramolecular Mannosylation ....................................... Other Mannosyl Donor-Based Methods ............................. Epimerization of p-Glucopyranosides at C-2 ........................ The Oxidation-Reduction Approach ................................ Direct Inversion ....................................................... The 2-Ulosyl Donor Method ......................................... Anomeric O-alkylation ............................................... Alkylation of l-O-Metal Complexes ................................. The Stannylene Acetal Method ...................................... Miscellaneous Methods ............................................... Radical Inversion of the Anomeric Chirality of a-DMannopyranosides .................................................... Reductive Cleavage of Cyclic Orthoesters ........................... De nouo Syntheses .................................................... 2-Acetamido-2-deoxy-~-~-mannopyranosides ....................... Aryl p-D-mannopyranosides.......................................... l-Thio-P-D-mannopyranosides ....................................... P-D-Mannopyranosylamines .......................................... Enzymatic Synthesis .................................................. Conclusions ........................................................... References.............................................................

300 300

319 319 320 320 320 322 324 327 329 329 329 331 332 332 332 333 333 334 334 335 336 336 337 337 338 338

XIV 14

14.1 14.2 14.3 14.4 14.4.1 14.4.2 14.4.3 14.5 14.5.1 14.5.2 14.5.3

15

15.1 15.2 15.2.1 15.2.2 15.2.3 15.2.4 15.2.5 15.2.6 15.2.7 15.3 15.4 15.5 15.5.1 15.5.2 15.5.3 15.5.4 15.6 15.6.1 15.6.2 15.6.3 15.6.4 15.6.5

Contents

Special Problems in Glycosylation Reactions: Sialidations ........... 345 Mukoto Kiso. Hidehuru Ishida and Hirorni It0 Introduction .......................................................... 345 Sialidation by the Koenigs-Knorr Method .......................... 345 347 Sialidation Using an Auxiliary Group at C-3 ....................... Sialidation Using 2.Thioglycosides. Xanthates. or Phosphites of 349 Sialic Acids in Acetonitrile ........................................... Thioglycosides ........................................................ 349 Xanthates and Phosphites ............................................ 356 359 Reaction Mechanism ................................................. Further Solutions to the Problem .................................... 359 Combination of C-3 Auxiliary and Sterically less Hindered Sugar Acceptors ............................................................. 359 Combination of C-3 Auxiliary and Specific Activation of the 360 Anomeric Center C-2 ................................................. Thioglycoside of N ,N.Diacetylneuraminic Acid and Combination 363 with C-3 Auxiliary .................................................... References ............................................................ 364 Special Problems in Glycosylation Reactions: 2-Deoxy Sugars ...... 367 Aluin Veyri2res Introduction .......................................................... 367 Electrophilic Additions to Glycals: Mechanistic Aspects and Applications to the Synthesis of 2-Deoxyglycosides . ................ 368 Protonation of Glycals ............................................... 369 370 Enzyme-Catalyzed Additions to Glycals............................. Halogenation of Glycals ............................................. 370 372 Bromo- and Iodoalkoxylation of Glycals ............................ 377 Epoxidation of Glycals ............................................... Addition of Sulfur Based Electrophiles to Glycals .................. 379 Addition of Selenium Based Electrophiles to Glycals ............... 382 The Cycloaddition Way to Glycosyl Transfer ....................... 384 385 Fluoroglycosylation of Glycals ....................................... 386 Glycosyl Donors with a C-2 Heteroatom ............................ 2-Bromo-2-deoxyglycosyl bromides .................................. 386 2-Deoxy-2-(thiophenyl)-glycosylfluorides ........................... 387 388 2,6.Anhydro-2-Thi o-Glycosyl Donors ............................... 1,2-Di-O-Acetyl-P-Hexopyranoses and N-Formylglucosamine Derivatives ............................................................ 392 2-Deoxyglycosyl Donors ............................................. 393 394 2-Deoxy-Hexopyranoses ............................................. Tert-Butyldimethylsilyl 2-Deoxyglycosides .......................... 394 1-0-Acyl- and Acetimidyl-2-Deoxy.Hexopyranoses ................. 394 395 2-Deoxyglycosyl Bromides and Fluorides ........................... S-(2-Deoxyglycosyl)phosphorodithioates............................ 396

Contents 15.6.6 15.6.7 15.6.8 5.7 5.7.1 5.7.2

6

16.1 16.2 16.2.1 16.2.2 16.3 16.3.1 16.3.2 16.3.3 16.3.4 16.4 16.4.1 16.4.2 16.4.3 16.5

17

17.1 17.2 17.3 17.4 17.5 17.5.1 17.5.2 17.6 17.7 17.8

XV

2-Deoxyglycosyl Phosphates. Phosphoramidites and Phosphites .... 397 2-Deoxy Thioglycosides .............................................. 398 399 2-Deoxyglycosyl Sulfoxides........................................... Other Approaches to 2-Deoxyglycosides............................. 400 Cyclization of Acyclic Sugars ........................................ 401 Use of Alkoxy-Substituted Anomeric Radicals ...................... 402 References............................................................. 403 Orthogonal Strategy in Oligosaccharide Synthesis ...................407 Osurnu Kun ie Introduction ........................................................... 407 Analysis of the Strategic Aspects of Oligosaccharide Synthesis ..... 408 General Aspects ....................................................... 408 The Pursuit of Efficiency in Oligosaccharide Synthesis ..............408 The Introduction of the Orthogonal Glycosylation Strategy ........ 410 Limitation of Current Concepts ...................................... 410 The Orthogonal Coupling Concept .................................. 412 413 What is Orthogonality Anyway? ..................................... Orthogonal Glycosylation and Solid-Phase Oligosaccharide 414 Synthesis .............................................................. The Orthogonal Glycosylation Strategy ............................. 414 Orthogonal Chain Elongation of Homo-Oligosaccharides: 414 Synthesis of Chito-Oligosaccharides [ 191 ............................. Orthogonal Coupling for Hetero-Oligomer Synthesis [22]...........418 Application to Polymer-Supported Synthesis [26] ...................420 421 Conclusions and Prospects ........................................... Acknowledgments .................................................... 424 References ............................................................. 424 Protecting Groups: Effects on Reactivity. Glycosylation Stereoselectivity. and Coupling Efficiency ............................ 427 Luke G. Green and Steven V. Ley Introduction ........................................................... 427 Glycosidic Mechanism ................................................ 428 Electronic and Torsional Effects ..................................... 430 Influence of Protecting Group on Donor Reactivity ................. 431 Stereoselectivity ....................................................... 436 Neighboring-Group Participation .................................... 436 437 Reactivity Control .................................................... influence of the Protecting Group on the Acceptor .................441 Steric Effects on Glycosylation ....................................... 443 444 Conclusions ........................................................... Acknowledgments .................................................... 445 References ............................................................. 446

XVI 18

18.1 18.2 18.2.1 18.2.2 18.3 18.4 19 19.1 19.2 19.2.1 19.2.2 19.2.3 19.2.4 19.2.5 19.2.6 19.3 19.3.1 19.3.2 19.3.3 19.3.4 19.3.5 19.3.6 19.4 19.4.1 19.4.2 19.4.3 19.5

Contents

Intramolecular Glycosidation Reactions .............................. 449 Jacob Madsen and Mikael Bols Introduction .......................................................... 449 Reactions in which the Tether Participates in the Reaction ......... 450 Tethering to the Glycosyl Donor .................................... 450 Carbon Tethers ....................................................... 450 Silicon Tethers ........................................................ 454 Tethering to the Leaving Group ..................................... 459 Reactions in which the Tether does not Participate in the Reaction 459 Conclusion ............................................................ 464 References ............................................................ 465 Classics In Total Synthesis of Oligosaccharides and Glycoconjugates....................................................... 467 Jean-Maurice Mallet and Pierre Sinay Introduction .......................................................... 467 Syntheses of Nod factors ............................................. 467 Introduction .......................................................... 467 The K . C . Nicolaou Synthesis (1992) [3]............................. 468 The J.-M. Beau Synthesis (1994) [ 121 ................................ 471 The T . Ogawa Synthesis (1994) [ 161 ................................. 475 The Y . Z . Hui Synthesis (1992) [ 181 ................................. 477 Conclusion ............................................................ 480 Synthesis of the Antithrombin-Binding Pentasaccharide Sequence in Heparin (1984) [ 19, 201 ............................................... 480 Introduction .......................................................... 480 An Overview of the Synthesis of the Protected Pentasaccharide 73 ..................................................................... 481 Synthesis of the Disaccharidic Bromide Donor 68 .................. 483 Synthesis of the Disaccharidic Acceptor 69 .......................... 484 Synthesis of the Protected Pentasaccharide 73 ....................... 484 Synthesis of the Active Site of Heparin .............................. 485 Total Synthesis of VIM-2 Ganglioside [31] .......................... 485 Introduction .......................................................... 485 The Total Synthesis of VIM-2-a General Strategy ................ 486 Preparation of the Key Protected Octasaccharide 87 ................ 487 Epilogue .............................................................. 490 References ............................................................ 491

I1

Synthesis of Oligosaccharide Mimics .................................

493

20

Synthesis of C-Oligosaccharides...................................... Troels Skrydstrup. Boris Vauzeilles. and Jean-Marie Beau Introduction .......................................................... The Anionic Approach ............................................... C5-Alkynyl Anions ...................................................

495

20.1 20.2 20.2.1

495 496 496

Contents

XVII

20.2.2 20.2.3 20.2.4 20.2.5 20.3 20.3.1 20.3.3 20.4 20.5

C1-Glycal Carbanions ................................................ 500 Anomeric Samarium Species ......................................... 502 C-Branched Carbanions .............................................. 506 C6-Phosphoranes ..................................................... 508 The Radical Approach ............................................... 511 Intermolecular Anomeric Radical Addition ......................... 511 513 Intramolecular Anomeric Radical Addition ......................... The Partial de Nuuo Approach ....................................... 518 The Cycloaddition and Rearrangement Approach ..................527 528 References .............................................................

21

531 Synthesis of Oligosaccharide Mimics: S-Analogs..................... Jon K. Fairweather and Hugues Driguez Introduction ........................................................... 531 General Synthesis ..................................................... 532 Preparation of Thioglycoses .......................................... 532 l-Thioglycoses ........................................................ 532 2-, 3-, 4-, 5.. or 6-Thioglycoses ....................................... 532 Selective S-Deprotection of Thioglycoses ............................ 533 Glycosylation Methods ............................................... 534 Establishment of 1,6-Thio Linkages .................................. 534 6-Thiodisaccharides ................................................... 534 6-Thiooligosaccharides ............................................... 538 538 Branched Thiocyclodextrins .......................................... Establishment of I , 4-Thio Linkages .................................. 541 1,4-Thiodisaccharides ................................................. 541 General Approaches .................................................. 541 SN~-Displacement on Triflates ....................................... 541 546 1,4-Thiooligosaccharides ............................................. Conventional Approaches ............................................ 546 548 Chemoenzymatic Approaches ........................................ 549 Michael Addition to Unsaturated Acceptors ........................ Solid-Support Synthesis........ ................................... 550 551 Establishment of 1, 3.Thio Linkages .................................. 1,3-Thiodisaccharides ................................................. 551 Conventional Methods ............................................... 551 Cyclic Sulfamidate and Aziridine .................................... 551 1,3-Thiooligosaccharides ............................................. 552 Establishment of 1,2.Thio Linkages .................................. 553 1,2-Thiodisaccharides................................................. 553 Conventional Methods ............................................... 554 Other Approaches .................................................... 555 Establishment of I , l-Thio Linkages .................................. 557 558 Establishment of Mixed Thio linkages ............................... Thiooligosaccharides and Proteins ................................... 558 The Conformation of Thiooligosaccharides in Solution ............. 558

21.1 21.2 21.2.1

21.3 21.3.1 21.3.2 21.3.3 21.4 21.4.1 21.4.2

21.5 21.5.1 21.5.2 21.6 21.6.1 21.7 21.8 21.9 21.9.1

XVIII

Contents

21.9.2

Enzyme-Substrate Interactions ...................................... a-Glucan- Active Enzymes ............................................

21.9.3 21.10

P-Glucan-Active Enzymes ............................................ Lectin-Ligand Interactions .......................................... Conclusion ............................................................ Acknowledgments .................................................... References ............................................................

22 22.1 22.2 22.2.1 22.2.2 22.3 22.4 22.4.1 22.4.2 22.4.3 22.4.4

Saccharide-Peptide Hybrids .......................................... Huns Peter Wessel Introduction .......................................................... Carbohydrate Amino Acids .......................................... Natural Carbohydrate Amino Acids ................................. Synthetic Carbohydrate Amino Acids ............................... Amide-Linked Carbohydrate Polymers .............................. Amide-Linked Carbohydrate Oligomers ............................. Solution Synthesis .................................................... Solid-Phase Synthesis................................................. Biological Activity .................................................... Conformational Properties ........................................... References ............................................................

565

Index ...................................................................

11

Part I

Chemistry of Saccharides

Vol. 2

Enzymatic Synthesis of Glycosides and CarbohydrateReceptor Interaction

111

Enzymatic Synthesis of Glycosides ..................................

23

On the Origin of Oligosaccharide Species-Glycosyltransferases in Action ................................................................ Dirk H . van den Eijnden Introduction ......................................................... Protein N-Glycosylation: Pre-assembly of Oligosaccharide-PPDolichol and en bloc Transfer ...................................... Trimming of the Polypeptide-Bound Oligosaccharide .............. Folding and Quality Control ........................................ Committed Steps in the Formation of Complex-Type Oligosaccharide Chains and Branching ............................. Topology of the Reaction Catalyzed by a Typical GlcNAcT ...... Elongation and Termination Reactions in the truns-Golgi ......... Activity with Branched Substrates ..................................

23.1 23.2 23.3 23.4 23.5 23.6 23.7 23.8

560 560 561 562 562 562 562

565 566 566 567 572 574 574 578 579 582 583

587 589 589 591 592 593 594 596 596 598

Contents

23.9 23.10 23.11 23.12 23.13 23.14 23.15 23.16 23.17 23.18 23.19 23.20 23.21 23.22 23.23 23.24 23.25 23.26 23.27 23.28 23.29 24 24.1 24.2 24.2.1

24.2.2

24.3 24.4

XIX

600 Branch Specificity .................................................... Essential Requirements for Activity with LacNAc ................. 601 Further Terminal Reactions in Complex-Type Oligosaccharide 602 Synthesis ............................................................. Specific Modifications of Polylactosaminoglycans .................. 603 The Invariable Core of N-linked Oligosaccharide Chains, and 606 Site- and Protein-Specific Processing ................................ Comparison of the Synthesis of Type 1 (Gal(P1-3)GlcNAcP-R) 607 and Type 2 (Gal(P1-4)GlcNAcP-R) Chains ........................ The LacdiN Ac Pathway of Complex-Type Oligosaccharide Synthesis ................... ....................................... 607 608 Protein O-Glycosylation ............................................. Glycosyltransferase Families ........................................ 608 610 Sialyltransferase Family ............................................. 611 a2-Fucosyltransferase Family ....................................... 612 a3/4-Fucosyltransferase Family ..................................... a3-Galactosyl/N-Acetylgalactosaminyltransferase(Histo-Blood Group ABO) Family ................................................ 613 P6-N- Acetylglucosaminyltransferase Family ........................ 613 Polypeptide N-Acetylgalactosaminyltransferase Family ............ 614 P4-N-AcetylgalactosaminyltransferaseFamily ...................... 615 P4-Galactosyltransferase Family .................................... 615 P3-Galactosyltransferase Family .................................... 617 P3-Glucuronyltransferase Family ................................... 617 Glycosyltransferases Standing Alone ................................ 617 618 Concluding Remarks ................................................ References............................................................ 618 Synthesis of Sugar Nucleotides ...................................... Reinhold Ohrlein Introduction .......................................................... Synthesis of Sugar Nucleotides ...................................... Chemical Synthesis .................................................. UDP-Activated Donors ............................................. CMP-Activated Sugars.............................................. GDP-Activated Donors ............................................. Comments ............................................................ Chemo-Enzymatic Synthesis ........................................ Uridine Diphosphate-Activated Donor Sugars ..................... CMP-Activated Sugars.............................................. GDP-activated sugars ............................................... Comments ............................................................ In situ Generation of Sugar Nucleotides ............................ Comments ............................................................ Outlook .............................................................. References............................................................

62.5 62.5 626 626 626 629 632 634 635 635 637 639 640 641 641 644 644

XX 25

25.1 25.2 25.2.1 25.2.2 25.2.3 25.2.4 25.2.5 25.2.6 25.2.7 25.3 25.3.1 25.3.2 25.3.3 25.3.4 25.4 25.5

26 26.1 26.2 26.3 26.4 26.4.1 26.4.2 26.4.3 26.4.4 26.4.5 26.5

27

27.1 27.2 27.2.1

Contents

Enzymatic Glycosylations with Glycosyltransferases ................ Ossi Renkonen Introduction ......................................................... In vitro Synthesis of the Core Region of 0.Glycans ................ Initialization of O-Glycan Biosynthesis ............................. Synthesis of Core 1 .................................................. Synthesis of Core 2 .................................................. Synthesis of Core 3 and Core 4 ..................................... In vitro Extension of Core 1 Glycans ............................... In vitro Extension of Core 2 Glycans ............................... Extension of Core 3 and Core 4 Glycans ........................... Enzymatic in vitro Synthesis of Polylactosamine Backbones ....... Enzymatic Synthesis of the Primary Chains of Blood Group i-Type ................................................................ Distal Branching of i-Type Polylactosamine Backbones ........... Central Branching of i-Type Polylactosamine Backbones.......... P4-Galactosylation in Polylactosamine Backbones ................. a3-Sialylation of N-Acetyllactosaminoglycansat the Terminal Gal a3-Fucosylation of Lactosamine Saccharides....................... References ........................................................... Recycling of Sugar Nucleotides in Enzymatic Glycosylation........ Kathryn M . Koeller and Chi-Huey Wong Introduction ......................................................... Glycosyltransferases of the Leloir Pathway and their Sugar Nucleotide Substrates ............................................... Design of Regeneration Systems .................................... Practical Regeneration Systems ..................................... UDP-Galactose ..................................................... Other UDP-Sugars .................................................. CMP-NeuAc ........................................................ GDP-Sugars ......................................................... Other Carbohydrate-Based Regeneration Systems ................. Conclusion ........................................................... References ........................................................... Enzymatic Glycosylations with Non-Natural Donors and Acceptors............................................................. Xiangping Qian. Keiko Sujino. and Monica M . Palcic Introduction ......................................................... Enzymatic Glycosylations ........................................... Galactosylations ..................................................... P 1.4-Galactosyltransferase .......................................... al.3-Galactosyltransferase ..........................................

647 647 648 648 648 649 650 650 651 651 651 652 653 654 656 656 657 659 663 663 663 665 666 666 669 671 676 680 682 683

685 685 686 686 686 688

Contents

27.2.2

27.2.3 27.2.4 27.3

28

28.1 28.2 28.3 28.3.1

28.3.2 28.4 28.4.1

28.4.2

29

29.1 29.2 29.3 29.4 29.5 29.6

XXI

Fucosylations ........................................................ Human Milk a173/4.Fucosyltransferase ............................. FucT I11 and VI ..................................................... FucT V ............................................................... Sialylations ........................................................... a2,3-Sialyltransferase and a2,6.Sialyltransferase .................... N.Acetylglucosaminylation .......................................... N-Acetylglucosaminyltransferase I, 11. and I11...................... N-Acetylglucosaminyltransferase V ................................. Summary ............................................................. Acknowledgments ................................................... References ............................................................

690 690 692 692 692 692 696 696 698 700 700 700

Solid-Phase Synthesis with Glycosyltransferases .................... Claudine AugC. Christine Le Narvor. and Andri Lubineau Introduction .......................................................... General Aspects ...................................................... Enzymatic Synthesis on Insoluble Supports ......................... Enzymatic Synthesis of Oligosaccharides ........................... Use of an Amino-Functionalized Water-Compatible Polyacrylamide Gel .................................................. Use of a Sepharose Matrix .......................................... Use of Controlled-Pore Glass ....................................... Enzymatic Synthesis of Glycopeptides .............................. Use of Controlled-Pore Glass ....................................... Use of Polyethylene Glycol Polyacrylamide (PEGA)............... Enzymatic Synthesis of Oligosaccharides and Glycoconjugates on Soluble Supports ..................................................... Enzymatic Synthesis of Oligosaccharides ........................... Use of Water-Soluble Amino-Substituted Poly(viny1 alcohol) ..... Use of Water-Soluble Glycopolymer Synthesized by Polymerization ....................................................... Enzymatic Synthesis of Glycolipids on Water-Soluble Polyacrylamide-Poly (N-acryloxysuccinimide) (PAN) .............. References ............................................................

705

Glycosidase-Catalysed Oligosaccharide Synthesis ................... David J. Vocadlo and Stephen G. Withers Introduction .......................................................... Background on Glycosidases ........................................ Basic Mechanisms ................................................... Synthesis by the ‘Thermodynamic’ Approach ...................... The Kinetic Approach ............................................... Recent Developments and New Directions ......................... References ............................................................

705 705 707 707 707 708 711 712 712 715 715 715 715 717 718 722 723 723 723 724 724 728 732 838

XXII 30

30.1 30.2 30.2.1 30.2.2 30.2.3 30.2.4 30.3 30.3.1 30.3.2 30.3.3 30.4 30.4.1 30.4.2 30.4.3

Contents

Production of Heterologous Oligosaccharides by Recombinant Bacteria (Recombinant Oligosaccharides) ........................... Roberto A . Geremia and Eric Samain Introduction ......................................................... Concept and Methodology of Heterologous (‘Recombinant’) Oligosaccharide Production in E. coli............................... Biosynthesis of Nod Factors ........................................ Expression Systems and Cloning Strategy .......................... High Cell-Density Cultivation ...................................... Purification of Recombinant Oligosaccharides ..................... Examples of Recombinant Oligosaccharides ....................... Production of Chitin Oligosaccharides in E . coli Expressing NodC ................................................................ Production of Nod Factor Precursors .............................. Production of Derivatives of N-Acetyllactosamine ................. Conclusions and Future Perspectives ............................... Production of Labeled Chitin Oligosaccharides to Study Their Interactions with Proteins ........................................... Improvement of Oligosaccharide Production, and Metabolic Engineering .......................................................... Production of More Complex Oligosaccharides .................... Acknowledgments ................................................... References ...........................................................

845 845 847 847 849 851 852 852 852 853 855 856 856 858 858 859 859

IV

Carbohydrate-Protein Interactions..................................

861

31

Protein-Carbohydrate Interaction: Fundamental Considerations ... Nikki F. Burkhalter. Sarah M . Dimick. and Eric J . Toone Introduction ......................................................... Association in Aqueous Solution ................................... Gas Phase Non Covalent Interactions .............................. Dipole-Dipole Interactions ......................................... Dipole-Induced Dipole ............................................. Dispersive Interactions .............................................. Specific Forces: Hydrogen Bonding and n - ( ~Bonding .............. The Effect of Water on Intermolecular Interactions ................ Coulombic Stabilization ............................................. Hydrogen Bonding .................................................. Dispersive Interactions .............................................. ‘Hydrophobic’ Interactions .......................................... The Evaluation of Protein-Carbohydrate Binding ................. Precipitin Assays ..................................................... Enzyme-Linked Lectin Assay (ELLA).............................. Isothermal Titration Microcalorimetry ............................. The Interpretation of Calorimetric Data ............................

863

31.1 31.2 31.2.1

31.2.2

3 1.2.3 31.3 31.3.1 31.3.2 31.3.3 31.4

863 864 864 864 866 867 868 869 870 871 872 872 876 877 878 878 882

Contents XXIII 31.4.1 3 1.4.2 31.4.3 31.5 31.6 31.6.1 31.6.2 31.6.3

32

32.1 32.2 32.3 32.4 32.5 32.6 32.7 32.8 32.9 32.9.1 32.9.2 32.9.3 32.9.4 32.10 32.10.1 32.10.2

Solvation/Desolvation ............................................... Solvation Entropy ................................................... Translational/Rotational Entropy ................................... Other Contributions to Thermodynamics of Association .......... Proton Transfer ...................................................... Salt Effects/Binding Site Reorganization ........................... van’t Hoff versus Calorimetric Enthalpies .......................... The Thermodynamics of Protein-Carbohydrate lnteraction ....... The Role of Multivalency in Protein-Carbohydrate Interaction ... Phenomenology ...................................................... The Energetic Consequence of Ligand Linkage .................... Enthalpic Contributions to AGi ..................................... Entropic Contributions to AGi ...................................... A Molecular Basis for the Cluster Glycoside Effect ................ Acknowledgments ................................................... References............................................................ Structural Analysis of Oligosaccharides: FAB.MS. ES-MS and MALDI-MS ......................................................... Anne Dell. Howard R. Morris. Richard Easton. Stuart Haslam. Maria Panico. Mark Sutton.Smitlz. Andrew J . Reason. and Kay-Hooi Khoo Introduction .......................................................... Fast Atom Bombardment-Mass Spectrometry (FAB-MS) ......... Matrix Assisted Laser Desorption Ionization-Time of Flight-Mass Spectrometry (MALDI-TOF-MS) .................................. Electrospray-Mass Spectrometry (ES-MS).......................... Appearance of Mass Spectra Obtained in FAB.MS. MALDI-MS and ES-MS Experiments ............................................ Assignment of Mass Values ......................................... Derivatisation ........................................................ Fragmentation Pathways ............................................ Protocols for MS Analysis........................................... Protocol l-Sample Loading for FAB-MS Analysis ............... Protocol 2-Sample Loading for NanoES-MS and MS-MS Analysis on the Q.TOF .............................................. Protocol 3-Sample Loading for LC-ES-MS and LC-ES-MS-MS on the Q.TOF ........................................................ Protocol 4-Sample Loading for MALDI-MS Analysis ........... Applications of FAB.MS. MALDI-MS and ES-MS in Glycobiology ......................................................... Case Study 1-Molecular Weight Profiling of Polysaccharides by MALDI-MS ......................................................... Case Study 2-Analysis of Glycoproteins by LC-ES-MS and FAB-MS .....................................................

882 883 884 885 885 885 886 887 901 901 905 906 907 910 911 911

915

915 915 917 918 919 921 921 922 924 924 925 925 925 926 926

XXIV 32.10.3 32.10.4 32.10.5 32.10.6 32.11 33 33.1 33.2 33.2.1 33.2.2 33.2.3 33.2.4 33.2.5

33.3 33.3.1 33.3.2 33.4 33.4.1 33.4.2 33.5 33.5.1 33.5.2 33.5.3 33.5.4 33.5.5 33.6 34

34.1 34.2 34.3 34.3.1

Contents Case Study 3-Characterization of a Novel N-Glycan by FAB-MS and FAB.MS.MS ......................................... Case Study 4-High Sensitivity Sequencing of a Novel Glycopeptide by Q-TOF ES-MS-MS and MALDI.MS ............ Case Study 5-FAB-MS Screening of Biological Samples for Glycan Content ...................................................... Case Study 6-MS Analysis of Mycobacterial Glycoconjugates . . Concluding Remarks ................................................ References ........................................................... Conformational Analysis in Solution by NMR ...................... S. W. Homans Introduction ......................................................... Solution Conformations of Oligosaccharides ....................... The NMR Technique ................................................ Conformational Parameters in Oligosaccharides ................... Conformational Restraints .......................................... I3C Isotopic Enrichment ............................................ Additional Conformational Restraints .............................. Exchangeable Protons ............................................... Heteronuclear Overhauser Effects................................... 13C- c Coupling-Constants ....................................... Dipolar Couplings ................................................... Experimental Restraints in Conformational Analysis .............. Restraining Protocol ................................................. Biharmonic Restraints ............................................... Time-Dependent Restraints ......................................... Dynamical Simulated Annealing .................................... Analysis of Oligosaccharide Dynamics ............................. Monte-Carlo Simulations ........................................... Molecular Dynamics Simulations ................................... A Case Study on Neu5Aca2-3GalP1-4Glc ......................... Resonance Assignments in Neu5Aca2-3Galfll-4Glc ............... ROE Connectivities ................................................. ‘Global Minimum’ Conformation of Neu5Aca2-3GalP1-4Glc .... Conformational Dynamics of Neu5Aca2-3Gal~1-4Glc............ Short-range vs Long-range Restraints .............................. Conclusions .......................................................... References ........................................................... Oligosaccharide Conformations by Diffraction Methods ............ Serge Pdrez. Catherine Gautier. and Anne Imberty Introduction ......................................................... General Analysis .................................................... Crystalline Conformations of Disaccharide Moieties .............. The Disaccharides ...................................................

930 933 935 942 944 945 947 947 947 947 948 949 949 950 950 952 953 954 955 955 955 957 957 958 959 959 959 960 960 961 962 963 966 966 969 969 970 973 973

Contents 34.3.2 34.4 34.5 34.6 34.7 34.8

35 35.1 35.2 35.3 35.4 35.5 35.6 35.7 35.8 36

36.1 36.2 36.3 36.4 36.5 36.6 36.7 36.8 36.9

37 37.1 37.1.1 37.1.2 37.2

XXV

The Analogs (S. C. N. ....). ......................................... Hydrogen Bonding in Crystalline Oligosaccharides ................ Packing Features ..................................................... Selected Examples ................................................... Crystalline Conformations of Oligosaccharides Complexed with Lectins................................................................ Concluding Remarks ................................................ References ............................................................

985 987 988 990 992 996 998

Transfer NOE Experiments for the Study of Carbohydrate-Protein Interactions ........................................................... 1003 Thomas Peters Introduction .......................................................... 1003 1004 The Transfer NOE Experiment ..................................... 1006 Measurement of trNOEs .......................................... Bioactive Conformations of Carbohydrate Ligands From trNOE Experiments .......................................................... 1008 Spin Diffusion may Generate Misleading Distance Constraints ... 1009 The Conformation of Sialyl Lewis' Bound to E-selectin ...........1011 Interaction of Bacterial Lipopolysaccharide Fragments with Monoclonal Antibodies .............................................. 1016 Conclusions and Future Directions ................................. 1019 References ............................................................ 1021 Carbohydrate-Protein Interactions: Use of the Laser Photo Chemically Induced Dynamic Nuclear Polarization(C1DNP)-NMR Technique ............................................................ 1025 Hans-Christian Siebert and Johannes F. G. Vliegenthart Introduction .......................................................... 1025 The CIDNP Method ................................................. 1026 CIDNP-related Molecular Modelling ............................... 1027 Applications .......................................................... 1027 Hevein-like Lectins .................................................. 1029 Galactoside-binding Lectins from Plant and Animal Origin ....... 1032 Sialidase from Clostridium Perfringens (Wild Type and Mutants) . 1037 CIDNP Analysis of Glycoproteins .................................. 1039 Conclusions .......................................................... 1040 Acknowledgments ................................................... 1041 1042 References ............................................................ Biacore ............................................................... 1045 Wolfgang Jager Introduction .......................................................... 1045 Real-time Analysis by Surface Plasmon Resonance ................1045 Information in a Sensorgram ........................................ 1047 Experimental Procedures ............................................ 1048

XXVI

Contents

37.2.1 31.2.2 37.2.3 37.2.4 37.2.5 37.3 37.3.1 37.3.2 37.3.3

Immobilization of Biomolecules at the Sensor Surface ............. 1048 Surface Regeneration ................................................ 1050 Interaction Analysis and Controls .................................. 1051 1052 Determination of Kinetic Rate Constants .......................... Affinity Determination .............................................. 1053 1054 Application Areas ................................................... Selectin Binding to a Glycoprotein Ligand ......................... 1054 Oligosaccharide Characterization ................................... 1055 In situ Modification of Immobilized Carbohydrates ................1056 References ........................................................... 1056

V

Carbohydrate-Carbohydrate Interactions ...........................

38

Carbohydrate-Carbohydrate Interactions ........................... 061 Dorothe Spillmann and Max M . Burger Introduction ......................................................... 1061 From Structural Components to Cell Recognition ................. 1063 Carbohydrate-Carbohydrate Interactions as Part of Structural Components ......................................................... 1063 Extracellular Matrix of Seaweeds-Agarose, Carrageenan and 1063 Alginate .............................................................. Cell Walls ............................................................ 1064 Mammalian Extracellular Matrix Components .................... 1066 Carbohydrate-Carbohydrate Interactions as Part of Recognition 1068 Keys? ................................................................. Carbohydrate Interactions in Invertebrates-The Marine Sponge Microciona prolifera as a Model System ............................ 1069 Carbohydrate Interactions in Vertebrates-Embryonal and Tumor 1071 Cells .................................................................. Repulsive Carbohydrate-Carbohydrate Interactions ............... 1072 Molecular Aspects of Carbohydrate Interactions ...................1074 1074 Polyvalence to Inforce Weak Interactions .......................... Arrangement of Motifs and the Possibility to Control Specificity . 1075 Molecular Basis of Carbohydrate-Carbohydrate Interactions ..... 1076 Experimental Approaches ........................................... 1078 General Considerations ............................................. 1078 Affinity Interactions ................................................. 1079 1079 Cell Binding Studies ................................................. Aggregation of de n o w Complexes ................................. 1081 Affinity Chromatography ........................................... 1082 Distribution between Compartments ............................... 1082 Microscopy .......................................................... 1083 Electron Microscopy ................................................ 1083 Atomic Force Microscopy .......................................... 1083 1084 Crystallography ......................................................

38.1 38.2 38.2.1

38.2.2

38.3 38.3.1 38.3.2 38.3.3 38.4 38.4.1 38.4.2

38.4.3 38.4.4

059

Contents XXVII

38.4.5 38.4.6 38.4.7 38.4.8

Mass Spectrometry .................................................. Nuclear Magnetic Resonance ....................................... Molecular Modelling ................................................ Tools ................................................................. Synthetic Oligosaccharides .......................................... Antibodies against Carbohydrate Motifs ........................... Cells .................................................................. References ............................................................

1085 1085 1086 1086 1086 1087 1088 1088

V1

Carbohydrate-Nucleic Acid Interactions ............................

1093

39

Carbohydrate-Nucleic Acid Interactions ............................ Heinz E . Moser Introduction .......................................................... Carbohydrates Binding to DNA .................................... Ene-Diyne Antibiotics and Antitumor Agents ...................... Esperamicins ......................................................... Calicheamicins ....................................................... Anthracyclins ........................................................ Pluramycins and Aureolic Acids .................................... Carbohydrates Binding to RNA .................................... Aminoglycosides ..................................................... References ............................................................

1095

39.1 39.2 39.2.1 39.2.2 39.2.3 39.3 39.3.1

Index ...................................................................

Part I1

Biology of Saccharides

Vol . 3

Biosynthesis and Degradation of Glycoconjugates Introduction to Volumes 3 and 4 ....................................... Abbreviations Used in Volumes 3 and 4 .............................

1095 1096 1096 1096 1100 1106 1111 1112 1113 1120 11

V LV

I

Biosynthesis of Glycoconjugates .......................................

1

1

Metabolism of Sugars and Sugar Nucleotides ......................... Hudson H . Freeze Introduction ............................................................ Basic Principles ........................................................ Transporters Deliver Monosaccharides to Cells ...................... Intracellular Sources of Sugars ........................................ Salvage .................................................................

3

1.1 1.2 1.3 1.4 1.4.1

3 3 4 5 5

XXVIII Contents 1.4.2

1.5 1.6 1.7 2

2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 2.10

3

3.1 3.2 3.3 3.4 3.5

Activation and Interconversion of Monosaccharides................. Glycogen ............................................................... Glucose ................................................................ Glucuronic acid ........................................................ Iduronic acid ........................................................... Xylose .................................................................. Mannose ............................................................... Fucose .................i ............................................... Galactose ............................................................. N-Acetylglucosamine................................................. N-Acetylgalactosamine............................................... Sialic acids ............................................................ Sugar Nucleotide Transporters ....................................... Control of Sugar Nucleotide Levels ................................. Possible Future Directions ........................................... References ............................................................

6 6 7 8 8 8 8 9 10 10 10 11 11 13 13 14

Nucleotide Sugar Transporters ....................................... Rita Gerardy-Schahn and Matthias Eckhardt Introduction .......................................................... General Considerations .............................................. The Requirement for Nucleotide Sugar Transporters and Their Mechanism of Function: A Comprehensive Overview of the Last 20 Years .............................................................. Molecular Cloning of Nucleotide Sugar Transporters .............. The Structure of Nucleotide Sugar Transporters .................... The Subcelluar Distribution of Nucleotide Sugar Transporters ..... Molecular Defects that Cause Inactive UDP-Galactose and CMP-Sialic Acid Transporters ....................................... Association Between Defects in Nucleotide Sugar Transporters and Diseases ............................................................... Involvement of Nucleotide Sugar Transporters in the Regulation of Glycosylation ......................................................... Future Perspectives ................................................... Acknowledgements ................................................... References ............................................................

19

Biosynthesis of Oligosaccharyl Dolichol .............................. Sharon S. Krag General Overview .................................................... Oligosaccharyl Dolichol .............................................. Key Enzymatic Steps in the Assembly Process ...................... Topology of the Assembly Process................................... Utilization of Oligosaccharyl Dolichol .............................. Acknowledgment ..................................................... References ............................................................

19 20

20 22 25 27 28 29 29 30 31 32 37 37 38 39 42 42 43 43

Contents XXIX 4

4.1 4.2 4.2.1 4.2.2 4.2.3 4.2.4 4.3 4.4 4.4.1 4.4.2 4.4.3 4.4.4 4.4.5 4.4.6 4.4.7 4.4.8 4.5

5

5.1 5.2 5.3 5.4 5.5 5.6 5.7 5.8

6

6.1 6.2

Biochemistry and Molecular Biology of the N-Oligosaccharyltransferase Complex................................................... Roland Knauer and Ludwig Lehle Introduction ........................................................... Biochemistry of OST ................................................. Lipid-Saccharide Donor ............................................. Acceptor Specificity of OST .......................................... Catalytic Mechanism of OST ........................................ Regulation of OST Activity .......................................... Isolation of OST Complexes from Different Sources ................ Molecular Biology of OST ........................................... WBPl/OST48 ......................................................... SWPl/Ribophorin I1 ................................................. OSTl/Ribophorin I ................................................... OST3/OST6 ........................................................... OST5 .................................................................. OST4 .................................................................. OST2/DAD1 .......................................................... STT3 .................................................................. Structural Organization of the OST Complex ....................... Acknowledgments .................................................... References ............................................................. Processing Enzymes Involved in the Deglucosylation of N-Linked Oligosaccharides of Glycoproteins: Glucosidases I and 11 and Endomannosidase ..................................................... Robert G. Spiro Introduction ........................................................... Glucosidase I .......................................................... Glucosidase 11......................................................... Endo-a-mannosidase .................................................. Concerted Action of Deglucosylation Enzymes ...................... Mutants ............................................................... Role of Monoglucosylated N-Linked Oligosaccharides and Glucose Trimming Enzymes in Regulating Quality Control of Glycoproteins ......................................................... Effect of Glucosidase Inhibitors on Viral Proliferation .............. Acknowledgments .................................................... References............................................................. a-Mannosidases in Asparagine-linked Oligosaccharide Processing and Catabolism ........................................................ Kelley W. Moremen Overview .............................................................. Introduction ...........................................................

45 45 46 47 48 49 51 51 52 54 54 55 55 56 56 57 58 59 60 60

65 65 66 68 70 72 74

75 77 78 78

81 81 82

XXX 6.2.1 6.2.2 6.2.3 6.2.4 6.2.5 6.3 6.3.1 6.3.2

6.3.3

6.4

7 7.1 7.2 7.3 7.4 7.5

8 8.1 8.2

Contents

Roles of N - and O-Linked Glycans and Compartmentalization of Biosynthetic and Catabolic Reactions ............................... 82 82 Processing of Asn-Linked Oligosaccharides ......................... Early Trimming Events: importance for quality control glycoprotein degradation and anteriograde transport ..................... 85 Glycoprotein Catabolism: multiple routes for glycoprotein 87 breakdown ............................................................ Consequences of Genetic Defects in Oligosaccharide Biosynthesis and Catabolism ....................................................... 88 Mannosidases in Glycoprotein Processing and Catabolism ......... 89 89 Classification of Mannosidases ....................................... Class 1 Mannosidases: enzymes of the ER and Golgi ............... 93 ER mannosidase I subfamily ......................................... 93 95 Golgi mannosidase I sub-family ..................................... Fungal secreted mannosidases ....................................... 97 New genes with unknown functions ................................. 98 Class 2 Mannosidases: enzymes of the cytosol, ER, Golgi, and Lysosomes ............................................................ 98 99 Golgi mannosidase I1................................................. Lysosomal mannosidase ............................................. 101 Epididymal/sperm mannosidase .................................... 103 Heterogeneous cluster of mannosidase homologs among eukary a, 104 eubacteria, and archaea ............................................. Conclusions and Future Prospects .................................. 106 107 Acknowledgments ................................................... 107 References ........................................................... The Role of UDP-Glcyglycoprotein Glucosyltransferase as a Sensor of Glycoprotein Conformations ...................................... Armando J. Parodi Introduction ......................................................... General Properties ................................................... GT Recognizes Glycoprotein Conformations ...................... The Primary Structure of the UDP-Glcyglycoprotein Glucosyltransferase .................................................. The Role of Monoglucosylated Oligosaccharides in Glycoprotein Folding ............................................................... Acknowledgments ................................................... References ........................................................... Mannosyltransferases................................................ Peter Orlean Introduction ......................................................... Occurrence of Covalently-linked Mannose .........................

119 119 120 121 122 123 126 127 129 129 130

Contents XXXI

8.2.1 8.2.2 8.2.3 8.2.4 8.3 8.3.1 8.3.2 8.3.3 8.3.4 8.4 8.4.1

8.4.2

8.4.3 8.4.4 8.5 8.6

9

9.1 9.2 9.3 9.3.1 9.3.2 9.4 9.5 9.6

Eukaryotic Secretory Glycoproteins ................................ Glycophospholipids .................................................. Eubacterial and Archaeal Mannose-containing Molecules ......... C-linked Mannose ................................................... Biochemistry of Mannosyl Transfer ................................. Many Linkages, Two Donors ....................................... Donor Specificity .................................................... Acceptor Specificity.................................................. Structural Features of Man-T ....................................... Man-T Families and the Pathways They Participate in ............ Man-Ts of the ER [l-51 ............................................. Alglp ................................................................. Alg2p/Algl l p ........................................................ Dpmlp ............................................................... Alg3p ................................................................. Alg9p/PIG-Bp family ................................................ Pmtlp family ......................................................... Golgi Man-Ts and Fungal Mannan Synthesis ...................... Ochlp family ......................................................... ................................. Mnn9p family ........... MnnlOp/Mnnl l p family ............................................ Mnnlp family ........................................................ Ktrlp family ......................................................... “Missing” Eukaryotic Man-T ....................................... Eubacterial and Archaeal Man-T ................................... Coordinating Man Transfer with the Cell Cycle and Morphogenesis ....................................................... Concluding Remarks ................................................ Acknowledgments ................................................... References............................................................

130 130 130 130 131 131 131 132 132 133 134 134 134 135 135 135 136 136 137 137 138 138 138 138 139

Branching of N-Glycans: N-Acetylglucosaminyltransferases........ Hurry Schuchter Introduction .......................................................... Processing of N-Glycans within the Endomembrane Assembly Line .................................................................. General Properties of the N-Acetylglucosaminyltransferases....... Domain Structure .................................................... Targeting to the Golgi Apparatus ................................... UDP-GlcNAc:Manal-3R [GlcNAc to Manwl-31 p.1,2. N . Acetylglucosaminyltransferase I (GnT I, EC 2.4.1.101) ............ UDP-GlcNAc:Manal-6R [GlcNAc to Manal-61 p.1,2. N . Acetylglucosaminyltransferase 11 (GnT 11, E.C. 2.4.1.143). ........ The Role of GnT I and I1 in Mammalian Development ...........

145

139 140 140 140

145 146 148 148 150 150 152 153

XXXII 9.7 9.7.1 9.7.2 9.8 9.9 9.9.1 9.10 9.11

10 10.1 10.2 10.2.1 10.3 10.3.1 10.3.2 10.3.3 10.3.4 10.3.5 10.4 10.4.1 10.4.2 10.4.3 10.4.4 10.5 10.5.1 10.5.2 10.5.3 10.5.4

Contents UDP-G1cNAc:RI-Manal-6[GlcNAc~l-2Manul-3]Man~l-4R~ [GlcNAc to ManP 1-41 P- 1,4. N.Acetylglucosaminyltransferase 111 155 (GnT 111. E.C. 2.4.1.144). ........................................... 156 Overexpression of GnT 111 Activity ................................. 157 GnT 111 Activity and Cancer ........................................ UDP-GlcNAc:R1Manal-3R2 [GlcNAc to Manal-31 (3.1,4. N . Acetylglucosaminyltransferase IV (GnT IV, E.C. 2.4.1.145). ...... 157 UDP-GlcNAc:Rl Manal-6R2 [GlcNAc to Manal-61 P-1,6-NAcetylglucosaminyltransferase V (GnT V, E.C.2.4.1.155) ......... 158 159 GnT V Activity and Cancer ......................................... UDP-GlcNAc:Rl (R2)Manal-6R3 [GlcNAc to Manul-61 P-1, 4-Nl -Acetylglucosaminyltransferase VI (GnT VI) ............. 161 161 GnT VII and GnT VIII ............................................. 162 References ........................................................... The Galactosyltransferases .......................................... Nancy L . Shaper. Martin Charron. Neng- Wen Lo. Jane R. Scocca. and Joel H . Shuper Introduction ......................................................... Using the Databanks to Obtain Information on the Galactosyltransferases ............................................... Nomenclature ........................................................ The Dual Role of P4-Galactosyltransferase-I (P4GalT-I) in Oligosaccharide and Lactose Biosynthesis: The Early Days ....... p4GalT-I: Isolation and Characterization of cDNA Clones ....... The Murine NGalT-I Gene: Genomic Organization and Structure of the 5'-End ......................................................... P4GalT-I and Lactose Biosynthesis ................................. P4GalT-I and the Vertebrate P4GalT Gene Family ................ Evolution of the 04-Galactosyltransferase Gene Family ........... The Vertebrate P3Galactosyltransferase (P3GalT) Gene Family . . General Characteristics of the P3-Galactosyltransferase Gene Family Members .................................................... P3GalT-IV: UDP-ga1actose:GM 1 P3-galactosyltransferase (GM1 Synthase; GalT-3) ................................................... Other Vertebrate P-Galactosyltransferase Activities................ UDP-Ga1actose:Ceramide P-Galactosyltransferase (CGalT; EC 2.4.1.45). ............................................................. The Vertebrate u3-Galactosyltransferase Gene Family ............ u3-Galactosyltransferase (a3GalT: UDP-Gal:Galp4GlcNAca3Galactosyltransferase; EC 2.4.1.87) ................................. The Blood Group B a3-Galactosyltransferase (EC 2.4.1.37). ...... The Forssman Glycolipid Synthetase (EC 2.4.1.88). ............... Evolution of the u3GalT Gene Family .............................

175 175 177 177 178 181 181 182 182 184 185 186 187 187 187 188 188 190 191 191

Contents XXXIII

10.6

A UDP-Gal:Gal@3GalNAca4Galactosyltransferase Activity ..... 192 192 Acknowledgments ................................................... References............................................................ 192

11

Fucosyltransferases ................................................... 197 Ernest0 T. A . Marques. Jr. Introduction .......................................................... 197 198 General Characteristics .............................................. Nomenclature ........................................................ 198 199 Gene Structure ....................................................... Sequence Peptide Motifs ............................................. 199 Specificity ............................................................ 199 200 Protein Structure and Topology ..................................... Enzymatic Reaction Mechanism .................................... 201 203 Inhibitors ............................................................. 203 Specific Fucosyltransferases ......................................... GDP-Fucose: Fucal (Fucal,2Fuc)a2-fucosyltransferase ............ 204 GDP-Fucose: GalPl (Fucal,2Gal)a2-fucosyltransferase ............ 204 GDP-Fucose: Gal~l,4/3GlcNAc(Fucal,3/4GlcNAc)a3/ 4204 fucosyltransferases ................................................... 204 Blood group Lewis: FucT 111, V and VI ............................ Myeloid enzyme: FucT IV ........................................... 205 Leukocyte enzyme: FucT VII ....................................... 206 206 Neuronal enzyme: FucT IX ......................................... GDP-Fucose: Gal~l,3GlcNAc(Fuca1, 3GlcNAc) bacterial (Helicobacter pylori) a3-fucosyltransferase ......................... 207 GDP-Fucose: GlcNAc-N(Fuca1,6GlcNAc)a6fucosyltransferases . 207 GDP-Fucose: O-Ser(Fuca1-O-Ser)GlcNAc polypeptide 207 fucosyltransferases ................................................... Unconventional Types of Fucosylation: FucPl -P-Ser and cytoplasmic Fucal,2-GalP1,3-GlcNAc-Pro (Dictyostelium discoideum) 207 FucP 1-P-Ser .......................................................... 207 208 Fucal,2-GalP1,3-GlcNAc-Pro ...................................... Acknowledgments ................................................... 208 208 References............................................................

11.1 11.2 11.2.1 11.2.2 11.2.3 11.2.4 11.2.5 11.2.6 11.2.7 11.3 11.3.1 11.3.2 11.3.3

11.3.4 11.3.5 11.3.6 11.3.7

12

12.1 12.2 12.3 12.4 12.5 12.6 12.7

Sialyltransferases..................................................... Joseph T. Y. Lau and Sherry A . Wuensch Introduction .......................................................... General Features of Sialyltransferases .............................. Cloning and ldentification Strategies for Sialyltransferases ........ Sialyltransferase Classification and Nomenclature .................. The a2, 3-ST Family ................................................. The 1x2,6-ST Family ................................................. The 1x2.8.ST Family .................................................

213 213 213 215 216 216 217 218

XXXIV Contents 12.8

Regulation and Functionality of Sialyltransferases................. 219 221 References ...........................................................

13

Biochemistry of Sialic Acid Diversity................................ Roland Schauer Introduction ......................................................... Occurrence and Biosynthesis ........................................ General Biological Functions ....................................... N-Glycolylneuraminic Acid ......................................... 0-Acetylated Sialic Acids ........................................... 0-Methylated and 0-Sulfated Sialic Acids ......................... Acknowledgments ................................................... References ...........................................................

13.1 13.2 13.3 13.4 13.5 13.6

14 14.1 14.2 14.3 14.4 14.5 14.6 14.7

15

15.1 15.2 15.3 15.4 15.5 15.6 15.7 15.8 15.9 15.10

Carbohydrate Sulfotransferases...................................... Steven D. Rosen. Annette Bistrup. and Stefan Hemmerich Introduction ......................................................... Basic Features of Sulfotransferase Reactions ....................... Tyrosine Sulfation ................................................... Diversity of Carbohydrate Sulfation ................................ Biochemical Demonstration of Carbohydrate Sulfotransferases ... Molecular Cloning of Carbohydrate Sulfotransferases ............. Primary Structures of Carbohydrate Sulfotransferases ............. Acknowledgments ................................................... References ........................................................... Novel Variant Pathways in Complex-type Oligosaccharide Synthesis ............................................................. Dirk H . van den Eijnden Introduction ......................................................... The lacNAc Pathway of Complex-type Oligosaccharide Synthesis Occurrence and Biology of 1acdiNAc-based Complex-type Oligosaccharides ..................................................... Biosynthesis of lacdiNAc Backbone Units .......................... The lacdiNAc Pathway of Complex-type Oligosaccharide Synthesis ............................................................. Other Shared Properties of 04-GalT and 04-GalNAcT ............ Cloning of a snail UDP-G1cNAc:GlcNAcP-R 04-Nacetylglucosaminyltransferase ....................................... The Chitobio Pathway of Complex-type Oligosaccharide Synthesis ............................................................. Competition Between Pathways ..................................... Concluding Remarks ................................................ References ...........................................................

227 227 227 229 231 234 238 239 239 245 245 245 246 246 249 250 252 256 256

261 261 261 262 263 264 266 266 267 267 269 269

Contents XXXV 16

16.1 16.2 16.3 16.4 16.4.1 16.4.2

16.5 16.6 16.7

17

17.1 17.2 17.2.1 17.2.2 17.2.3 17.2.4 17.2.5 17.3

Control of Mucin-Type 0-Glycosylation: 0-Glycan Occupancy is Directed by Substrate Specificities of Polypeptide GalNAcTransferases .......................................................... 273 Helle Hassan. Eric P. Bennett. Ulla Mandel. Michael A . Hollingsworth. and Henvik Cluusen Introduction .......................................................... 273 The Mammalian UDP-GalNAc: Polypeptide GalNAc-Transferase Gene Family ......................................................... 274 The GalNAc-Transferase Gene Family is Evolutionarily Old ..... 276 The Kinetic Properties of GalNAc-Transferase Isoforms are Different.............................................................. 278 Lessons from in vivo Analysis of GalNAc-transferase Substrate Specificities ........................................................... 219 Lessons from in vitro Analysis of the Acceptor Substrate Specificities of GalNAc-transferase Isoforms ....................... 280 Isoforms may have distinct acceptor substrate specificities ......... 281 Isoforms may have overlapping substrate specificities .............. 283 Isoforms may act in different order on substrates with multiple 283 acceptor sites ......................................................... Isoforms may require prior (GalNAc) glycosylation ............... 283 Expression of the GalNAc-Transferase Genes are Differentially Regulated ............................................................ 285 288 Predictive Value of in vitro 0-glycosylation? ....................... Conclusions and Future Perspectives................................ 288 References ............................................................ 289 Glycosyltransferase Inhibitors ........................................ Xianyping Qian and Monica M . Palcic Introduction .......................................................... Inhibitors of Glycosyltransferases................................... Inhibitors of Galactosyltransferases ................................. Inhibitors of pl, 4-galactosyltransferase ............................. Inhibitors of cr.l,3-galactosyltransferase............................. Inhibitors of Fucosyltransferases .................................... Inhibitors of 1x1,2-fucosyltransferases ............................... Inhibitors of a1,3-fucosyltransferases ............................... Inhibitors of Sialyltransferases ...................................... Inhibitors of a2, 6-sialyltransferase .................................. Inhibitors of a2, 3.sialyltransferase .................................. Inhibitors of N-Acetylglucosaminyltransferases .................... Inhibitors of Human Blood Group A and B Glycosyltransferases. Summary ............................................................. Acknowledgments ................................................... References............................................................

293 293 296 296 296 297 298 300 300 301 302 304 305 306 309 309 309

XXXVI Contents 18

18.1 18.2 18.3 18.4 18.5 18.6 18.7 18.8 18.9 18.10 18.11 18.12

19

19.1 19.2 19.3 19.4 19.5 19.6 19.7

20

20.1 20.2 20.3 20.3.1 20.3.2 20.3.3 20.3.4 20.3.5 20.4 20.4.1 20.4.2 20.4.3 20.5

Biosynthesis of the O-Glycan Chains of Mucins and Mucin Type Glycoproteins ........................................................ Inka Brockhausen Summary ............................................................. Introduction ......................................................... Structures of O-Glycans ............................................. Functions of Much Type O-Glycans ............................... Primary O-Glycosylation ............................................ Synthesis of O-Glycan Core 1....................................... Synthesis of O-Glycan Core 2 ....................................... Synthesis of O-Glycan Core 3 ....................................... Synthesis of O-Glycan Core 4 ....................................... Synthesis of O-Glycan Cores 5-8 ................................... Elongation and Branching Reactions ............................... Synthesis of Terminal Structures .................................... Acknowledgments ................................................... References ........................................................... Glycosyltransferases in Glycosphingolipid Biosynthesis ............. Subhash Basu. Kamal Das. and Manju Basu Introduction ......................................................... Fucosyltransferases in Glycolipid Biosynthesis ..................... Galactosyltransferases in Glycolipid Biosynthesis .................. N-Acetylgalactosaminyltransferases in Glycolipid Biosynthesis.... N-Acetylglucosaminyltransferases in Glycolipid Biosynthesis...... Sialyltransferases in Glycolipid Biosynthesis ....................... Glucuronyltransferases in Glycolipid Biosynthesis ................. Acknowledgments ................................................... References ........................................................... Biosynthesis of Glycogen ............................................ Peter J . Roach Summary ............................................................. Introduction ......................................................... Glycogenin and the Initiation of Glycogen Synthesis .............. History ............................................................... Properties ............................................................ Reaction Mechanism ................................................ Domain Structure ................................................... Function ............................................................. Glycogen Synthase and the Bulk Synthesis of Glycogen ........... Properties ............................................................ Structure of Glycogen Synthase ..................................... Branching Enzyme .................................................. Intermediates in the Biosynthesis of Glycogen .....................

313 313 313 314 314 315 315 317 319 319 319 320 321 324 324 329 329 329 332 334 336 337 340 342 342 349 349 350 351 351 351 352 352 354 354 354 355 356 357

Contents XXXVII 20.6

Conclusion ........................................................... Acknowledgments ................................................... References ............................................................

358 359 359

21

Biosynthesis of Hyaluronan .......................................... Paraskevi Heldin and Torvard C. Laurent Introduction .......................................................... Site of Biosynthesis .................................................. Biosynthetic Precursors .............................................. Hyaluronan Synthases ............................................... Microbial Enzymes .................................................. Vertebrate Synthases ................................................. Mechanism of Synthesis ............................................. Chain Elongation .................................................... Translocation ........................................................ Shedding ............................................................. Regulation of HA Synthesis ......................................... Concluding Remarks ................................................ Acknowledgments ................................................... References ............................................................

363

21.1 21.2 21.3 21.4 21.4.1 21.4.2 21.5 21.5.1 21.5.2 21.5.3 21.6 21.7

22

22.1 22.2 22.2.1 22.2.2 22.2.3 22.2.4 22.3 22.3.1 22.3.2 22.3.3

22.3.4

Biosynthesis of Chondroitin Sulfate and Dermatan Sulfate Proteoglycans ........................................................ Geetha Sugumaran and Barbara A4. Vertel Introduction .......................................................... Proteoglycan Structure .............................................. Proteoglycans and Their Core Proteins ............................. What Initiates GAG Chain Addition? .............................. The Linkage Region ................................................. .................................... CS and DS Chains ........... Biosynthesis of CS and DS Proteoglycans .......................... Biosynthesis of the Core Protein .................................... Origin of Sugar and Sulfate Precursors ............................. Addition of the Linkage Oligosaccharides .......................... Xylosylation .......................................................... Galactosylation ...................................................... Addition of GlcA and completion of the common tetrasaccharide linkage region ........................................................ Initiation of CS/DS chains by addition of the first GalNAc ....... Formation of the CS/DS Chains .................................... Addition of the repeating disaccharides ............................. Epimerization of GlcA to IdoA to form DS ........................ Sulfation of GalNAc................................................. Sulfation of uronic acid ..............................................

363 364 364 365 365 366 367 368 369 369 370 371 372 372

375 375 379 379 381 381 382 383 383 384 385 385 386 387 388 388 388 389 390 391

XXXVIII Contents 22.4

Concluding Remarks/Perspectives .................................. Acknowledgments ................................................... References ...........................................................

391 392 392

23

Biosynthesis of Heparin and Heparan Sulfate Proteoglycans ....... Lena Kjellin and U r Lindahl Introduction ......................................................... The Proteoglycans: Structure. Location and Functions ............ Biosynthesis of the Polysaccharide Backbone ...................... Outline of Polymer-Modification Reactions ........................ The N.Deacetylase/N.Sulfotransferases ............................. The C5-Epimerase ................................................... The 2-O-Sulfotransferase............................................ The 6-O-Sulfotransferases........................................... The 3 - 0 Sulfotransferases........................................... The Products. Heparin and Heparan Sulfate ....................... Interactions with Proteins ........................................... Regulation of HS Biosynthesis ...................................... References ...........................................................

395

23.1 23.2 23.3 23.4 23.4.1 23.4.2 23.4.3 23.4.4 23.4.5 23.5 23.6 23.7

24 24.1 24.2 24.2.1 24.2.2 24.2.3 24.2.4 24.3 24.3.1 24.3.2 24.3.3 24.3.4 24.4 24.5

25

25.1 25.2 25.2.1

395 396 396 397 399 399 399 400 400 400 401 402 403

Biosynthesis of Proteoglycans with Keratan Sulfates................ James L . Funderburgh Introduction: Keratan Sulfate Renaissance ......................... Keratan Sulfate Structure and Distribution ........................ Corneal KS .......................................................... Non-corneal KSI .................................................... KSII .................................................................. KSIII ................................................................. Keratan Sulfate Proteoglycans ...................................... SLRPs ................................................................ Aggrecan ............................................................. Cell-Associated KS .................................................. Brain ................................................................. Enzymatic Reactions of KS Biosynthesis ........................... Metabolic Control of KS Synthesis ................................. Acknowledgments ................................................... References ...........................................................

407

The Biosynthesis of GPI Anchors .................................... Yasu S. Morita, Alvaro Acosta.Serrano. and Paul T. Englund Introduction ......................................................... Structure of GPI Anchors ........................................... Glycan Core Modifications .........................................

417

407 407 408 409 409 410 410 410 411 411 412 412 413 414 414

417 417 417

Contents XXXIX 25.2.2 25.3 25.3.1 25.3.2 25.3.3 25.3.4 25.3.5 25.3.6 25.3.7 25.3.8 25.4 25.4.1 25.4.2 25.4.3 25.5 25.6

Variations in Anchor Lipid Structure ............................... GPl Precursor Synthesis ............................................. GlcNAc-PI Synthesis ................................................ GlcNAc-PI Deacetylation ........................................... Inositol Acylation .................................................... GPI Mannosylation .................................................. Transfer of EtN-P ................................................... Lipid Remodeling.................................................... Addition of Carbohydrate Side Chains ............................. Topology of Biosynthetic Pathways ................................. Attachment of the GPI Precursor to a Protein ..................... Basic Features ....................................................... Protein Machinery for GPI Addition ............................... Signal Sequence for GPI Addition. ................................. Evolution of GPI Biosynthesis ...................................... Future Studies ....................................................... Acknowledgments ................................................... References............................................................

419 419 420 421 421 422 423 423 424 424 425 425 426 426 426 427 427 427

26

435

26.1 26.2 26.3 26.3.1 26.3.2 26.3.3 26.3.4 26.3.5 26.3.6 26.4 26.5

Escherichiu coli Lipid A: A Potent Activator of Innate Immunity ... Teresa A . Garrett and Christian R . H . Raetz Introduction .......................................................... Structure of Lipopolysaccharide..................................... Lipid A Biosynthesis in E. coli ...................................... Acylation of UDP-GlcNAc ......................................... Disaccharide Formation ............................................. Phosphorylation by the Lipid A 4' Kinase .......................... Kdo Addition and the Late Acyltransferases ....................... Other Acyltransferases............................................... Transport of Lipid A and the Role of MsbA ....................... Lipid A Activation of Signal Transduction in Animal Cells ....... Summary ............................................................. Acknowledgments ................................................... References............................................................

435 435 437 439 440 440 441 442 442 444 447 447 447

11

Glycosidases ...........................................................

453

27

Lysosomal Degradation of Glycolipids............................... 455 Thomas Kolter and Konrad Sandhof Summary ............................................................. 455 Introduction .......................................................... 455 Mechanisms of Lysosonial Glycolipid Degradation ................ 456 Glycosidases ......................................................... 456

27.1 27.2 27.3 27.3.1

XL 27.3.2 27.3.3 27.3.4 27.3.5 27.3.6 27.4 27.4.1 27.4.2 27.4.3 27.4.4 27.4.5 27.4.6 27.4.7 27.5 27.5.1 27.5.2 27.6

28 28.1 28.2 28.3 28.4 28.4.1 28.4.2 28.4.3 28.5

28.5.1 28.5.2 28.5.3 28.6

29 29.1 29.2

Contents Topology of Endocytosis and Lysosomal Glycolipid Degradation Sphingolipid Activator Proteins ..................................... The GM2-activator and its role in lysosomal digestion ............ SAP-A to SAP-D .................................................... Lateral Pressure ...................................................... Lipid Composition .................................................. Membrane Curvature ............................................... Degradation of Selected Lipids ....... ........................... Ganglioside GM2 ................................................... Lactosylceramide .................................................... Glucosylceramide .................................................... Ceramide ............................................................. Sphingomyelin ....................................................... Sulfatide ............................................................. Galactosylceramide .................................................. Pathobiochemistry ................................................... Animal Models for Sphingolipidoses ............................... Therapy .............................................................. Future Directions .................................................... References ...........................................................

457 458 459 460 460 461 462 462 462 464 464 465 465 465 466 466 467 469 470 470

Lysosomal Degradation of Glycoproteins ........................... Nathan N . Aronson. Jr. Summary ............................................................. Introduction ......................................................... Roles of Lysosomes ................................................. Lysosomal Degradation of N-Linked Glycoproteins ............... General Features .................................................... Carbohydrate Digestion ............................................. Protein and Linkage Hydrolysis .................................... Formation of Thyroid Hormone via Lysosomal Degradation of Thyroglobulin ....................................................... Synthesis of Thyroid Hormone ..................................... Carbohydrate Degradation .......................................... Proteolysis ........................................................... Degradation of Free Polymannose-Type Oligosaccharides Derived from N-Linked Glycoproteins During Biosynthesis ................ References ...........................................................

473

Sialidases............................................................. Garry Taylor. Susan Crennell. Carl Thompson. and Marina Chuenkova Abstract .............................................................. Introduction .........................................................

473 473 474 475 475 476 476 477 477 478 479 479 481 485

485 485

Contents

XLI

29.3 29.4 29.5 29.6 29.7 29.8 29.9

Influenza Virus Neuraminidase ...................................... Paramyxovirus Hemagglutinin-Neuraminidase (HN) .............. Non-Viral Sialidases ................................................. Small Sialidases ...................................................... Large Sialidases ...................................................... T . cruzi Trans-Sialidase (TS) ........................................ Conclusion ........................................................... Acknowledgments ................................................... References ............................................................

486 487 487 490 491 491 493 494 494

30

Microbial Glycosidases............................................... Kenji Yamamoto. Su-Chen Li. and Yu-Teh L i Exo-Glycosidases .................................................... u-Glucosidase ........................................................ P-Glucosidase ........................................................ a-Galactosidase ...................................................... P-Galactosidase ...................................................... u.Mannosidase ....................................................... 0.Mannosidase ....................................................... P-N-Acetylhexosaminidase .......................................... u.N.Acetylga1actosaminidase ........................................ u-L-Fucosidase ....................................................... P.D.Fucosidase ....................................................... Sialidase .............................................................. KDNase .............................................................. u-L-Rhamnosidase ................................................... p-Xylosidase ......................................................... Endo.Glycosidases ...................................................

497

30.1 30.1.1 30.1.2 30.1.3 30.1.4 30.1.5 30.1.6 30.1.7 30.1.8 30.1.9 30.1.10 30.1.11 30.1.12 30.1.13 30.1.14 30.2 30.2.1 30.2.2 30.2.3 30.2.4 30.2.5 31

31.1 31.2 31.3 31.4 31.5 31.6 31.7 31.8

Endo.P.N.acetylg1ucosaminidase .................................... Peptide-N-glycanase F ............................................... Endo.u.N.acetylga1actosaminidase .................................. Endo-P-galactosidase ................................................ Endoglycoceramidase ................................................ References. ........................................................... Glycoprotein Processing Inhibitors ................................... Magid Osser and Alan D . Elbein Introduction ........................................................... Structural Classification ............................................. Distribution of Glycosidase Inhibitors in the Plant Kingdom ...... Isolation and Structural Determination ............................. Glycosidase Inhibitory Activity ..................................... Structure-Activity Relationships .................................... N-Linked Oligosaccharide Processing ............................... Inhibitors of N-Linked Oligosaccharide Processing.................

497 497 498 498 499 500 500 501 501 502 503 503 504 504 505 505 505 506 506 507 507 508 513 513 515 515 516 517 518 519 522

XLII 31.8.1 31.8.2 31.9

Contents Glucosidase Inhibitors ............................................... Mannosidase Inhibitors ............................................. Summary and Future Prospects ..................................... References ...........................................................

522 525 528 529

Index ...................................................................

11

Part I1

Biology of Saccharides

Vol. 4

Lectins and Saccharides Biology

Ill

Lectins ...............................................................

533

32

Plant Lectins ......................................................... Marilynn E. Etzler Summary ............................................................. Introduction ......................................................... Carbohydrate Specificity ............................................ Other Activities ...................................................... Structure ............................................................. Biological Roles ..................................................... Acknowledgments ................................................... References ...........................................................

535

32.1 32.2 32.3 32.4 32.5 32.6

33 33.1 33.2 33.3 33.4 33.5 33.6 33.7 33.8

34

34.1 34.2 34.3

Interactions of Oligosaccharides and Glycopeptides with Hepatic Carbohydrate Receptors ............................................. Yuan C. Lee and Reiko T. Lee Summary ............................................................. Introduction ......................................................... Molecular Characteristics of Hepatic Lectins ...................... Cellular Aspects of HL .............................................. Binding Specificity................................................... Photoaffinity Labeling ............................................... Subunit Organization on Rat Hepatocyte Surface ................. Applications ......................................................... References ........................................................... P-Type Lectins and Lysosomal Enzyme Trafficking ................ Patricia G. Marron-Teradu and Nancy M . Dahms Introduction ......................................................... Intracellular Trafficking of the MPRs .............................. Primary Structure and Biosynthesis of the MPRs ..................

535 535 536 539 540 543 546 547

549 549 550 551 552 553 557 558 559 560 563 563 564 566

Contents

XLIII

34.3.1 34.3.2 34.4 34.5 34.5.1 34.5.2 34.6

CI-MPR .............................................................. CD-MPR ............................................................. Lysosomal Enzyme Recognition by the MPRs ..................... Structural Determinants of Man-6-P Recognition .................. Expression of Mutant Forms of the MPRs ......................... Crystal Structure of the CD-MPR .................................. Conclusions .......................................................... Acknowledgments ................................................... References............................................................

35

The Siglec Family of I-Type Lectins ................................. 579 Paul R. Crocker and Soerye Kelm Introduction .......................................................... 579 The Immunoglobulin Superfamily and Carbohydrate Recognition 579 Siglecs as a Family of Sialic Acid Binding Proteins ................ 580 Biology of Siglecs .................................................... 581 Sialic Acids in Cellular Recognition ................................ 583 Mode of Carbohydrate Recognition by Siglecs ..................... 584 Importance of Multivalent Binding ................................. 588 Sialic Acid Recognition by the Immunoglobulin FoldEvolutionary Considerations ........................................ 588 Role of cis Interactions in Modulating Adhesion to Other Cells in trans ............................................................... 589 Sialoadhesin as a Macrophage Adhesion Molecule................. 590 Signalling Versus Adhesion Mediated by Siglecs ................... 591 Conclusions .......................................................... 592 Acknowledgments ................................................... 592 References............................................................ 592

35.1 35.2 35.3 35.4 35.5 35.6 35.7 35.8 35.9 35.10 35.11 35.12

36

36.1 36.2 36.2.1 36.2.2 36.3 36.3.1 36.3.2 36.3.3 36.3.4 36.3.5 36.3.6 36.3.7 36.4

C-Type Lectins and Collectins ....................................... Russell Wallis Summary ............................................................. Structure and Function of C-Type Animal Lectins ................. The Carbohydrate-Recognition Domain ............................ Ligand Binding ...................................................... Mannose-Binding Protein and Collectins ........................... Domain Organization ............................................... MBPs as Prototype Collectins ....................................... Ligand Binding by Serum MBP ..................................... MBP and Innate Immunity .......................................... Liver MBP ........................................................... Pulmonary Surfactant Proteins ...................................... Conglutinin and CL-43 .............................................. Conclusions ..........................................................

566 567 569 571 571 572 574 574 575

597 597 598 599 600 601 601 603 603 604 607 608 608 609

XLIV

Contents

Acknowledgments ................................................... References ...........................................................

609 609

Selectins .............................................................. Rodger P. McEver Introduction ......................................................... Structure of Selectins ................................................ Selectin Ligands ..................................................... Requirements for Selectins to Mediate Tethering and Rolling of Leukocytes under Hydrodynamic Flow ............................ Functions of Selectins and their Ligands in vivo ................... Conclusions .......................................................... References ...........................................................

613

Galectins ............................................................. Douglas N . W. Cooper and Samuel H . Barondes Introduction ......................................................... Galectin Structure ................................................... Novel Candidate Galectins .......................................... Unorthodox Subcellular Targeting .................................. Regulation of Galectin Expression .................................. Galectin Binding Specificity and Identified Ligands ................ Physiological Functions ............................................. Summary ............................................................. References ...........................................................

625 625 626 631 635 637 639 640 642 642

IV

Saccharide Biology ...................................................

649

39

Structures and Functions of Nuclear and Cytoplasmic Glycoproteins Robert S. Haltiwanger Introduction ......................................................... O-Linked N-Acetylglucosamine (0.GlcNAc) ....................... O-GlcNAc Appears to be a Regulatory Modification much like Phosphorylation ..................................................... Modulation of Protein Stability and Function by O-GlcNAc ..... Other Forms of Nuclear and Cytoplasmic Glycosylation .......... Unique Cytoplasmic Forms of Glycosylation ...................... Conventional Forms of Glycosylation in the Nucleus and Cytoplasm ........................................................... Nuclear and Cytoplasmic Lectins ................................... Conclusions .......................................................... Acknowledgments ................................................... References ...........................................................

65 1

37 37.1 37.2 37.3 37.4 37.5 37.6

38

38.1 38.2 38.3 38.4 38.5 38.6 38.7 38.8

39.1 39.2 39.2.1 39.2.2 39.3 39.3.1 39.3.2 39.3.3 39.4

613 613 614 619 621 621 622

651 652 653 655 658 658 660 661 662 662 662

Contents 40

40.1 40.2 40.3 40.3.1 40.3.2 40.3.3 40.4 40.5 40.6 40.7 40.7.1

40.7.2 40.7.3 40.7.4 40.7.5 40.8

41

41.1 41.2 41.3 41.3.1 41.3.2 41.3.3 41.3.4 41.3.5 41.4 41.4.1 41.4.2

XLV

Structure and Functions of Mucins .................................. 669 Joyce Taylor-Papadimitriou and Joy M . Burchell Classification of Mucins ............................................. 669 670 The Epithelial Mucins ............................................... Mucin Type O-Glycosylation Pathways ............................ 670 671 Initiation of O-Glycosylation ........................................ Chain Extension ..................................................... 671 Chain Termination ................................................... 671 Expression of Epithelial Mucins ..................................... 672 The Complex Gel-Forming Mucins: Processing and Function ..... 672 674 Epithelial Membrane Mucins ....................................... Studies Related to the MUCl Much ............................... 675 Changes in the Patterns of O-Glycosylation of MUCl in Breast 675 Cancer ................................................................ Differences in sites of glycosylation ................................. 675 Changes in the composition of O-glycans added to MUCl in 676 Breast Cancer ........................................................ Correlation in changes of Glycosyltransferase activities with changes in O-glycan structure in Breast Cancer .................... 676 Changes in Glycosylation of MUCl in other Cancers ............. 677 Effects of MUCl Expression on the Behavioral Properties of Cancer Cells.......................................................... 677 Effects on cell interactions and tumourogenicity .................... 677 678 MUCl Expression and Immune Responses ......................... Active Specific Immunotherapy Based on MUCl .................. 679 679 Animal models ....................................................... 680 Clinical studies ....................................................... Comments ............................................................ 681 References............................................................ 681 Biological Roles of Hyaluronan...................................... Bryan P . Toole Introduction .......................................................... Hyaluronan is a Biopolymer with Unusual Physical Properties .... Hyaluronan Binds to Several Types of Proteins (Hyaladherins) ... General Properties of Hyaladherins ................................. Structural Hyaluronan-Binding Proteins ............................ Hyaluronan Receptors ............................................... Intracellular Hyaluronan-Binding Proteins ......................... Inter-a-Trypsin Inhibitor ............................................ Hyaluronan-Dependent Pericellular Matrices Assemble Around Several Cell Types ................................................... Hyaluronan-Dependent Cellular “Coats” .......................... Assembly of Chondrocyte Pericellular Matrix ......................

685 685 685 687 687 688 688 689 689 690 690 691

XLVI 41.4.3 41.5 41.5.1 41.5.2 41.5.3 41.5.4 41.5.5 41.6

42 42.1 42.2 42.3 42.4 42.5 42.5.1 42.6 42.6.1 42.6.2 42.7 42.7.1 42.7.2 42.8 42.8.1 42.8.2 42.8.3 42.9 42.9.1 42.9.2 42.9.3 42.10

43 43.1 43.2

Contents

Tethering of Cell Surface Hyaluronan to Hyaluronan Synthase ... Hyaluronan Influences Cell Behavior During Morphogenesis and Tissue Remodeling .................................................. Migratory and Proliferating Cells are Surrounded by Hyaluronanenriched Matrices .................................................... Hydrated Pericellular Milieux Provide Cellular Pathways ......... Receptors Mediate Effects of Hyaluronan .......................... Hyaluronan-Cell Interactions in Limb Development ............... Hyaluronan-Cell Interactions in Other Physiological and Developmental Systems ............................................. Hyaluronan Plays a Crucial Role in Cancer ........................ References ...........................................................

691 693 693 693 693 694 695 696 696

Biological Roles of Heparan Sulfate Proteoglycans................. 701 Ofer Reizes. Pyong Woo Park. and Merton BernJeld Introduction ......................................................... 701 Heparan Sulfate Biosynthesis ....................................... 701 Functions of Heparan Sulfate ....................................... 702 703 Proteoglycans ........................................................ Intracellular Proteoglycans .......................................... 703 Serglycin and Heparin ............................................... 704 Cell Surface Heparan Sulfate Proteoglycans ........................ 705 Syndecans ............................................................ 705 Glypicans ............................................................ 706 Part-time Cell Surface Heparan Sulfate Proteoglycans ............. 707 Betaglycan ........................................................... 707 CD44 ................................................................. 708 Functions of Cell Surface Heparan Sulfate Proteoglycans ......... 708 Ligand Receptors .................................................... 708 709 Ligand Coreceptors .................................................. Shed Effectors ....................................................... 709 Extracellular Matrix Heparan Sulfate Proteoglycans and Their 710 Functions ............................................................ 710 Perlecan .............................................................. Agrin ................................................................. 712 Other Extracellular HSPGs ......................................... 713 713 Conclusions .......................................................... References ........................................................... 713 Biological Roles of Keratan Sulfate Proteoglycans.................. 717 Gary W. Conrad Introduction ......................................................... 717 718 Corneal Transparency ...............................................

Contents XLVII 43.3 43.4 43.5

Nerve Growth Cone Guidance ...................................... Cell Adhesion ........................................................ Other Possible Roles of KSPGs ..................................... Acknowledgment .................................................... References ............................................................

44

Developmental and Aging Changes of Chondroitin/Dermatan Sulfate Proteoglycans ........................................................ 729 J . Michael Sorrell. David A . Carrino. and Arnold I. Caplan Proteoglycans ........................................................ 729 Glycosaminoglycans ................................................. 729 Core Proteins ........................................................ 731 Hyalectans ........................................................... 731 Small Leucine-rich Proteoglycans ................................... 733 Chondrotin/Dennatan Sulfate Proteoglycans in Development and Aging ................................................................. 735 Core Proteins in Development. Aging. and Pathologies ............ 735 Chondroitin/Dermatan Sulfate Glycosaminoglycan Chains in Development. Aging. and Pathologies .............................. 736 Summary ............................................................. 740 References ............................................................ 740

44.1 44.2 44.3 44.3.1 44.3.2 44.4 44.4.1 44.4.2 44.5

45

45.1 45.2 45.3 45.4 45.5 45.6 45.7

46 46.1 46.2 46.3 46.4 46.5 46.6

719 721 722 723 723

Proteoglycans and Hyaluronan in Vascular Disease ................. Thomas N . Wight Introduction .......................................................... Proteoglycans and Hyaluronan ...................................... Versican (CSPGs) .................................................... Hyaluronan .......................................................... Decorin/Biglycan (DSPGs) .......................................... Perlecan/Syndecans (HSPGs)........................................ Summary ............................................................. Acknowledgments ................................................... References ............................................................

743

Functions of Glycosyl Phosphatidylinositols ......................... Nikola A . Baumann. Anant K. Menon. and David M . Rancour Introduction ........................................................... Parasite Coats: Extreme GPI-Anchoring ............................ Yeast GPIs and the Cell Wall ....................................... Paroxysmal Nocturnal Hemoglobinuria (PNH): Disease and Defects in GPI-Anchoring of Proteins .............................. GPIs in the Secretory and Endocytic Pathways ..................... Organization of GPI Proteins in the Plasma Membrane ...........

757

743 744 745 747 748 749 750 750 750

757 758 758 759 760 762

XLVIII

Contents

46.7 46.8 46.9 46.10 46.11

Association of GPI-Anchored Proteins with Caveolae ............. Detergent Insolubility and Signaling via GPI-Proteins ............. Membrane Release of GPI-Anchored Proteins ..................... GPIs as Second Messenger Signaling Molecules ................... Summary ............................................................. Acknowledgments ................................................... References ...........................................................

47

Glycosphingolipid Microdomains in Signal Transduction. Cancer. 771 and Development ..................................................... Sen-itiroh Hakomori and Kuzuko Handu Clustered GSLs as Functional Units ................................ 771 GSL Clusters. Associated with Signal Transducers. are Functional Units Separable from Caveolae ..................................... 772 Cell Adhesion Coupled with Signal Transduction Initiated by GSL Microdomain: Concept of Glycosignaling Domain (GSD) ........ 773 Role of GSLs in Control of Growth Factor and Hormone Receptors: Possible Relationship with GSL Microdomain ......... 774 Functional Role of Developmentally-Regulated and Tumor776 Associated GSLs ..................................................... References ........................................................... 778

47.1 47.2 47.3 47.4 47.5

48

48.1 48.2 48.3 48.3.1 48.3.2 48.4 48.5 48.6 48.7 48.8 48.9 48.10 48.10.1 48.10.2 48.10.3 48.10.4 48.11 48.12

The Primary Cell Walls of Higher Plants ........................... Jocelyn K. C. Rose, Mulcolm A . O’Neill, Peter Albersheim. and Alan Darvill Introduction (What is a Cell Wall?)................................. Purification of Cell Walls and Isolation of Wall Components ..... The Structural Components of the Primary Cell Wall ............. Cell Walls and the Diversity of Flowering Plants .................. The Structural Components of the Primary Wall .................. Biosynthesis of Wall Components .................................. Organization of the Plant Primary Cell Wall ....................... Cellulose-Xyloglucan Interactions .................................. Interactions Between Pectins and Other Cell Wall Components ... Glycoproteins in the Cell Wall ...................................... Heterogeneity in the Primary Cell Wall ............................ Function and Metabolism of Plant Primary Cell Walls ............ Mechanical Support ................................................. Regulation of Cell Expansion ....................................... Morphogenesis and Differentiation ................................. Plant Cell Wall Oligosaccharides in Defense and Cell Signalling .. Intercellular Transport and Storage ................................. Biotechnology and Future Directions in the Commercial Applications of Plant Primary Cell Walls ..........................

764 764 765 766 767 767 768

783 783 784 786 786 786 791 793 793 794 796 797 798 798 798 800 801 803 803

Contents

49

49.1 49.2 49.3 49.4 49.5 49.6

50

50.1 50.2 50.2.1 50.2.2 50.2.3 50.2.4 50.3 50.3.1 50.3.2 50.3.3

51

51.1 51.2 51.2.1 51.2.2 51.3 51.3.1 51.4

XLIX

Acknowledgment .................................................... References ............................................................

804 804

Glycolipids and Bacterial Pathogenesis .............................. ClifSord A . Lingwood Introduction .......................................................... Modulation of Glycolipid Receptor Function ...................... Stress Response and Glycolipid Receptors .......................... Subcellular Gb3 Trafficking ......................................... Model for Lipid Sorting Based on Chain Length ................... Glycosphingolipids and Signal Transduction ....................... Acknowledgments ................................................... References............................................................

809

Glycobiology of Viruses .............................................. Hildegard Geyer and Rudolf Geyer Summary ............................................................. General Aspects ...................................................... Functions of Viral Surface Glycoproteins ........................... Biosynthesis .......................................................... Function of Carbohydrate Substituents ............................. Oligosaccharide Diversity ........................................... Examples ............................................................. Friend Murine Leukemia Virus Complex ........................... Marburg Virus (MBGV) ............................................ Hepatitis B Virus (HBV) ............................................ Acknowledgments ................................................... References ............................................................

821

809 810 812 813 815 815 817 817

821 821 825 826 826 827 830 830 832 833 836 836

The Glycobiology of Influenza Viruses............................... 839 Stephen J. Stray and Gillian M . Air Introduction .......................................................... 839 Receptor Binding Proteins: Influenza A HAg and Influenza C HEF .................................................................. 840 Structure of Receptor Binding Domain and Mechanism of 843 Sialic Acid Recognition .............................................. HEF Esterase Domain and Mechanism of Cleavage ............... 844 844 Influenza NAm (types A and B) .................................... Mechanism of Sialic Acid Cleavage ................................. 845 Function of Viral Receptor Destroying Enzymes ................... 847 847 Acknowledgments ................................................... 848 References............................................................

L 52

52.1 52.2 52.3 52.4 52.5 52.6 52.7 52.7.1 52.7.2 52.7.3 52.7.4 52.7.5 52.8 52.9

53

53.1 53.2 53.3 53.3.1 53.3.2 53.3.3 53.4 53.4.1 53.4.2 53.4.3 53.4.4 53.4.5 53.4.6 53.4.7 53.5 53.6

Contents Glycobiology of Aids ................................................. Ten Feizi Abstract .............................................................. Introduction ......................................................... The Repertoire of N-Glycans on the Envelope Glycoprotein of HIV of Human Immunodeficiency Virus Produced in Different Cell Types............................................................ Evidence for the Occurrence of O-Glycans on the Envelope Glycoproteins of HIV-1 Produced in Certain Cell Lines ........... Oligosaccharides of gp 120 and gp 41 at N-Glycosylation Sites and Their Possible Influence on Viral Infectivity ................... gp 120 Glycosylation Can Influence Antigenicity and .. Immunogenicity ..................................................... Saccharides Recognized by Carbohydrate-binding Proteins and Antibodies as Potential Neutralization Epitopes on the Envelope Glycoprotein of HIV-1 .............................................. Lectins and Antibodies with Mannose-related Specificities ........ Antibodies to O-Glycan Sequences ................................. Antibodies to Blood Group A ...................................... Xeno-antibodies to Gala1-3Gal Sequence .......................... Potential Medical Relevance ........................................ Does Viral Oligosaccharide Display Influence Tissue Tropism? ... Concluding Remarks ................................................ Acknowledgment .................................................... References ...........................................................

851

Glycobiology of Protozoan and Helminthic Parasites ............... Richard D . Cummings. and A . Kwame Nyame Introduction ......................................................... General Classification of Parasites .................................. The Major Protozoan Parasites ..................................... Malaria .............................................................. . . ..................................................... Trypanosomiasis Leishmaniasis........................................................ Other Protozoan Parasites........................................... Entamoeba histolytica ............................................... Acanthamoeba ....................................................... Giardia lumblia ...................................................... Cryotosporidium parvum ............................................. Sarcocystis spp ....................................................... Toxoplusmu gondii., ................................................. Pneumocystis carinii ................................................. Helminthic Parasites ................................................. Carbohydrate-Binding Proteins in Parasitic Helminths ............

867

851 851 853 854 855 856 857 857 858 858 859 859 860 862 862 863

867 867 868 868 873 874 878 818 878 878 878 879 879 879 879 883

Contents 53.7 53.8

Unusual Glycans in Other Helminthic Parasites .................... Future Directions .................................................... Acknowledgments ................................................... References ............................................................

54

The Involvement of the Oligosaccharide Chains of Glycoproteins in Gamete Interactions at Fertilization ................................. Noritaka Hirohashi and William J . Lennarz Introduction .......................................................... Advantages of Marine Invertebrates as an Experimental System . . Induction of the Acrosome Reaction ................................ Studies in Sea Urchins ............................................... Studies in Starfish .................................................... Sperm-Egg Coat Binding ........................................... Studies in Mammals ................................................. Studies in Frog ....................................................... Studies in Ascidians .................................................. Studies in Sea Urchins ......... ................................... Carbohydrate as a Species-Specific Determinant ................... References ............................................................

54.1 54.2 54.3 54.3.1 54.3.2 54.4 54.4.1 54.4.2 54.4.3 54.4.4 54.5

55

55.1 55.2 55.3 55.4 55.4.1 55.4.2 55.5 55.6 55.6.1 55.6.2 55.6.3 55.7

56

56.1 56.2 56.3

LI 883 885 885 886

895 895 895 896 896 899 899 900 900 902 904 906 907

Glycosylation and Development ...................................... 909 Michde Aubery and Christian Derappe Summary ............................................................. 909 Introduction .......................................................... 911 Lectins as Tools to Analyze Changes in Cell-surface 911 Glycoconjugates During Development .............................. Cell-adhesion Molecules ............................................. 912 Neural Cell-adhesion Molecule ...................................... 912 The Adhesion Molecule L1 .......................................... 914 Glycosyltransferases ................................................. 914 Altered Expression of Endogenous Lectins During Development . 915 Galectins ............................................................. 915 Selectins .............................................................. 917 Other Endogenous Lectins .......................................... 917 Conclusion ........................................................... 918 References ............................................................ 918 Protein Glycosylation and Cancer ................................... James W. Dennis and Maria Granovskjj Introduction .......................................................... Protein Glycosylation Generates Molecular Diversity .............. Cancer Initiation and Progression ...................................

923 923 923 926

LII

Contents

56.4 56.5 56.6 56.7 56.8 56.9

Tumor Cell Proliferation ............................................ Cell Migration ....................................................... Sialylation and Metastasis........................................... Endogenous Lectins and Tumor Cell Adhesion .................... Carbohydrate Processing Inhibitors as Anti-Cancer Agents ....... Other Considerations ................................................ Acknowledgments ................................................... References ...........................................................

927 930 933 934 935 936 937 937

57

Lysosomal Storage Diseases ......................................... Nathan N . Aronson. Jr. Summary............................................................. Introduction ......................................................... Animal Models ...................................................... Mucopolysaccharidoses ............................................. Cathepsin K Deficiency and Pycnodysostosis ...................... Mouse Models for Tay-Sachs and Sandhoff Diseases .............. Impact of Lysosomal Diseases and Their Study .................... References ...........................................................

945

Genetic Diseases of Glycosylation ................................... Tomoya Akama and Michiko N . Fukuda Introduction ......................................................... CDGS ................................................................ CDGS Type I ........................................................ CDGS Type I1 ....................................................... HEMPAS ............................................................ References ...........................................................

959

57.1 57.2 57.3 57.4 57.5 57.6 57.7

58

58.1 58.2 58.2.1 58.2.2 58.3

59

59.1 59.2 59.3 59.3.1 59.3.2 59.3.3 59.3.4 59.3.5 59.3.6 59.4 59.5

Glycobiology of Helicobacter pylori and Gastric Disease ........... Karl-Anders Karlsson Introduction ......................................................... The Bacterial Surface and Molecular Mimicry ..................... Host Surfaces and H . pylori Recognition of Glycoconjugates: Unique Complexity.................................................. Sialic Acid ........................................................... Sulfatide ............................................................. Heparan Sulfate ..................................................... Fucose-Dependent Binding (H-1 and Lewis b) ..................... Gangliotetraosylceramide ........................................... Lactosylceramide .................................................... The Meaning of Multiple Binding Specificities ..................... Aspects for the Future ............................................... References ...........................................................

945 946 947 947 949 951 953 954

959 959 959 961 963 964 967 967 968 968 969 970 970 971 971 972 972 973 973

Contents

60 60.1 60.2 60.3 60.4 60.5 60.6 60.7 60.7.1 60.7.2 60.8 60.9 60.10 60.1 1 60.12 60.13 60.14 60.15

61

61.1 61.2 61.3 61.4 61.5

62 62.1 62.2 62.3 62.3.1

LIII

Immunoglobulin G Glycosylation and Galactosyltransferase Changes in Rheumatoid Arthritis .................................... 977 John S. Axford Introduction .......................................................... 977 977 Oligosaccharide Synthesis ........................................... Galactosyltransferase ................................................ 978 978 Immunoglobulin G .................................................. Rheumatoid Arthritis ................................................ 979 979 Quantification of IgG sugars in RAr ................................ RAr and Pregnancy .................................................. 980 980 Galactosylation of IgG .............................................. a3-Fucosylation of a1-Acid Glycoprotein .......................... 981 Agalactosyl-IgG and Rheumatoid Factor Binding ................. 982 Tissue-specific Galactosyltransferase Abnormalities in an Experimental Model of Rheumatoid Arthritis ...................... 983 Glycosylation Homeostasis within RAr Lymphocytes is Abnormal ............................................................ 985 Are the Rheumatoid Arthritis Associated Glycosylation Abnormalities Unique? .............................................. 986 Sugar Printing Rheumatic Disease is Possible ...................... 990 Rapid Profiling of IgG N-Glycans by Fluorophore-coupled Oligosaccharide Electrophoresis has the Potential of Differentiating 992 Rheumatic Diseases.................................................. In What Way could GTase Enzymatic Control be Abnormal? . . . . 992 Conclusion ........................................................... 993 References ............................................................ 994

Calnexin. Calreticulin and Glycoprotein Folding Within the Endoplasmic Reticulum .............................................. 997 Michael R. Leach and David B. Williams Structure and Properties of Calnexin and Calreticulin ............. 997 1000 Biological Functions ............................................... Mechanism of Action .............................................. 1002 Functional Relationship Between Calnexin and Calreticulin ..... 1005 Relationship with other ER Chaperones and Folding Catalysts . 1007 1008 References .......................................................... Glycobiology of The Nervous System .............................. Ronald L . Schnaar Introduction ........................................................ Nervous System Glycoconjugates-Overview .................... Nervous System Glycolipids ....................................... Galactosylceramides ...............................................

1013 1013 1013 1014 1014

LIV

Contents

62.3.2 62.4 62.4.1 62.4.2 62.5 62.6 62.6.1 62.6.2 62.7

Gangliosides and Related Anionic Glycosphingolipids ........... Nervous System Glycoproteins ..................................... Polysialic Acid ...................................................... The HNK-I Determinant .......................................... Nervous System Glycosaminoglycans ............................. Lectins in the Brain ................................................. Myelin-Associated Glycoprotein ................................... Other Nervous System Lectins ..................................... Concluding Remarks ............................................... References ..........................................................

1016 1019 1020 1020 1020 1021 1021 1022 1023 1023

63

Glycobiology of the Immune System ............................... Elizabeth F. Hounsell Infection and Pathogenesis ......................................... Control of the Immune Response .................................. Bacterial and Tumor Antigens. Mucins and Mucin-like Molecules ........................................................... Immunoglobulins and Pathology .................................. References ..........................................................

1029

63.1 63.2 63.3 63.4

64

64.1 64.2 64.2.1 64.2.2 64.2.3 64.3 64.3.1

64.3.2

Metabolic Engineering Glycosylation: Biotechnology’s Challenge to the Glycobiologist in the Next Millennium ...................... Thomas G. Warner Introduction ........................................................ Recent Developments in Carbohydrate Biosynthesis.............. Optimizing Sialylation of Recombinant Proteins by Metabolic Engineering Sialic Acid Biosynthesis............................... Optimizing Galactosylation of Recombinant Proteins by Metabolically Engineering Galactose Biosynthesis ................ Mannose Biosynthesis and Mannosylation of Recombinant Proteins ............................................................. Glycosylation Engineering Alternate Expression Hosts For Recombinant Protein Therapeutic Production .................... Engineering Glycosylation of Recombinant Proteins Expressed in Baculovirus-Insect Cells ......................................... Genes needed to supplement glycosylation of recombinant proteins in insect cells .............................................. Deleterious genes may need to be deleted or inhibited to enhance recombinant glycoprotein biosynthesis in insect cells.............. Engineering Glycosylation of Recombinant Proteins Expressed in Plants ............................................................. Genetic addition and supplementation needed to improve plant recombinant protein glycosylation ................................. Inhibition or deletion of plant glycosylation genes ................

1029 1033 1034 1036 1038

1043 1043 1044 1044 1049 1052 1053 1053 1054 1054 1056 1058 1059

Contents

64.4

Summary ........................................................... Acknowledgments ................................................. References ..........................................................

Index ...................................................................

LV

1059 1060 1060 11

List of Contributors (only main authors are listed)

Gillian Air Dept. of Biochemistry and Molecular Biology University of Oklahoma, Health Center BMSB 840, P. 0. Box 26901 Oklahoma City, OK 73190 USA Nathan N . Aronson Department of Biochemistry University of South Alabama MSB 2152 Mobile, AL 36688 USA Michele Aubery Ul80-INSERM U180 45 Rue Des Saints Peres 75006 Paris France Claudine Auge Laboratoire de Chimie Organique Multifonctionelle DR2 CNRS 462 Universite de Paris-Sud Inst. de Chimie Moleculaire (ICMO) 91405 Orsay Cedex France John S. Axford St. George’s Hospital Med. Ctr. Cranmer Terrace London SW17 ORE England

Samuel H. Barondes Langley Porter Psych. Inst. 401 Parnassus Avenue San Francisco, CA 94143-0984 USA Subhash Basu University of Notre Dame Stephan Chemistry Hall, Rm 443 Notre Dame, IN 46556 USA Jean-Marie Beau Univ. Paris-Sud Lab. de Synthese de Biomolecules URA CNRS 462, Inst. de Chimie 91405 Orsay France Merton Bernfield Harvard Medical School Enders Building, Room 961 300 Longwood Avenue Boston, MA 021 15 USA Mikael Bols Department of Organic Chemistry Aarhus University Langelandsgade 140 8000 Aarhus C Denmark

LVIII

List of Contributors

Inka Brockhausen Department of Medicine Queen’s University Etherington Hall, room 1021 Kingston, Ontario K7L 3N6 Canada Max Burger Friedrich Miescher Institut 4002 Basel Switzerland Arnold I. Caplan Case-Western Reserve University Cleveland, OH 44106 USA Henrik Clausen Department of Oral Diagnostics University of Copenhagen Norre Alle 20 DK-2200 Copenhagen N. Denmark Gary W. Conrad Kansas State University Ackert Hall Manhattan, KS 55606 USA Paul Crocker Department of Biochemistry The Welcome Trust Building University of Dundee Dundee, DDl 5HN Scotland Richard D. Cummings Department of Biochemistry and Molecular Genetics University of Oklahoma HSC 975 NE lofhSt., BRC 417 Oklahoma, OK 73104 USA Nancy Dahms Medical College of Wisconsin P. 0. Box 26509 8701 Watertown Plank Road Milwaukee, WI 53226-0509 USA Samuel J. Danishefsky Department of Chemistry Columbia University Havemeyer Hall

New York, NY 10027 and Laboratory for Bioorganic Chemistry The Sloan-Kettering Institute for Cancer Research 1275 York Avenue New York, NY 10021 USA Alan Darvill University of Georgia 220 Riverbend Road Athens, GA 30602 USA James W. Dennis Div. of Cancer and Cell Biology Mount Sinai Hosp., Res. Inst. 600 University Ave. Toronto, ON M5G 1x5 Canada Anne Dell Wolfson Laboratories Department of Biochemistry Imperial College of Science London SW7 2AY England Hugues Driguez CERMAV-CNRS B. P. 53 38041 Grenoble cedex France Dirk H. van den Eijnden Department of Medicinal Chemistry Vrije Universiteit Van der Boechorststraat 7 1081 BT Amsterdam The Netherlands Alan D. Elbein Dep. of Biochemistry and Molecular Biology University of Arkansas, Slot 516 4301 W. Markham Street Little Rock, AR 72205-7199 USA Paul Englund Johns Hopkins University School of Medicine 725 North Wolfe St. Baltimore, MD 21205 USA

List of Contributors Beat Ernst Institute of Molecular Pharmacy Pharmacenter Universitat Basel Klingenbergstrasse 50 4051 Basel Switzerland Mail: [email protected] Marilynn E. Etzler University of California Davis, CA 95616 USA Ten Feizi Clinical Research Centre Watford Road Harrow, MDX HA1 3UJ England Bert Fraser-Reid Natural Products and Glycotechnology Research Institute Inc. 41 18 Swarthmore Road Durham, North Carolina 27707 USA Hudson H. Freeze The Burnham Institute 10901 N. Torrey Pines Road La Jolla, CA 92093 USA Michiko Fukuda Burnham Institute (formerly La Jolla Cancer Res. Fndm.) 10901 North Torrey Pines Road La Jolla, CA 92037 USA James L. Funderburgh Kansas State University Ackert Hall Manhattan, KS 66506 USA Robert M. Garbaccio Laboratory for Bioorganic Chemistry The Sloan-Kettering Institute for Cancer Research 1275 York Avenue New York, NY 10021 USA

Teresa Garrett Duke University Medical Center c/o Dr. C. Raetz P. 0. Box 3711 Durham, N C 27710 England Rita Gerardy-Schahn Institut fur Medizinische Mikrobiologie Medizinische Hochschule Hannover Carl-Neuberg-Strak 1 30625 Hannover Germany Roberto A. Geremia Centre de Recherches sur les M acromolkcules Vegttales Universite Joseph Fourier Cermav-CNRS, BP 53 38041 Grenoble Cedex 9 France Rudolph Geyer Biochemisches Institut Universitat Giessen Friedrichstr. 24 35392 Giessen Germany Gary R. Gray Department of Chemistry Institute of Technology University of Minnesota 207 Pleasant Street S. E. Minneapolis, MN 55455-0431 USA Sen-Itiroh Hakomori Pacific Northwest Research Foundation 720 Broadway Seattle, WA 98122 USA Robert S. Haltiwanger State University of New York Stony Brook, NY 11794-5215 USA Gerald W. Hart Department of Biological Chemistry John Hopkins University 401 Hunterian Building 725 N. Wolfe Street Baltimore, M D 21205-2185 USA

LIX

LX

List of Contributors

Steve W. Homans School of Biochemistry and Molecular Biology University of Yeeds LSZ 9ST United Kingdom

Horst Kunz Institut fur Organische Chemie Universitat Mainz J.-Joachim-Becher-Weg 18-20 55099 Mainz Germany

Hideharu Ishida Department of Applied Bio-organic Chemistry Gifu University Gifu 501-11 Japan

Joseph T. Y. Lau Roswell Park Memorial Institute 666 Elm Street Buffalo, NY 14263 USA

Wolfgang Jager Biacore AB Jechtinger Strasse 8 79 111 Freiburg Germany Osamu Kanie Glycoscience Laboratory Mitsubishi Kasei Institute of Life Sciences (MILS) Machida-shi Tokyo 194-8511 Japan Karl-Anders Karlsson Institute of Medical Biochemistry G~teborgUniversity Medicinaraegatan 9A P. 0. Box 440 405 30 Gateborg Sweden Makoto Kiso Department of Applied Bio-organic Chemistry Gifu University Gifu, 50 1-11 Japan Sharon S. Krag Department of Biochemistry Johns Hopkins University 615 North Wolfe Street Baltimore, MD 21205 USA J. J. Krepinsky Department of Medical Genetics and Microbiology University of Toronto Toronto, ON M5S 1A8 Canada

Torvard C. Laurent Department of Medical and Physiological Chemistry University of Uppsala Avd. for Med. Kemi, Box 575 751 23 Uppsala Sweden Yuan C. Lee Department of Biology Johns Hopkins University 3400 North Charles Street Baltimore, MD 21218 USA Ludwig Lehle Lehrstuhl fur Zellbiologie und Pflanzenphysiologie Universitat Regensburg 93040 Regensburg Germany William J. Lennarz Department of Biochemistry and Cell Biology SUNY Stony Brook Room 450, Life Sciences Bldg. Stony Brook, NY 11794-5215 USA Steven V. Ley Chemical Laboratory University of Cambridge Lensfield Road Cambridge CB2 1EW England Yu-Teh Li Department of Biochemistry, SL43 Tulane University School of Medicine 1430 Tulane Avenue New Orleans, LA 701 12 USA

List of Contributors Ulf Lindahl Medical and Physiological Chemistry Uppsala Universitet P. 0. Box 575 75 12 Uppsala Sweden

Kelly L. Moremen Complex Carbohydrate Research Center University of Georgia 220 River Bend Road Athens, GA 30602-7229 USA

Clifford A. Lingwood Department of Microbiology Hospital for Sick Children 555 University Avenue Toronto, OT M5G 1x8 Canada

Heinz E. Moser GeneSoft Inc Two Corporate Drive South San Francisco, CA 9080 USA

Jean-Maurice Mallet Department de Chimie, URA 1686 Ecole Normale Superieure 24 rue Lhormond 75231 Paris Cedex 05 France Ernesto Marques Department of Pharmacology and MoleculaI Science John Hopkins University School of Medicine 725 North Wolfe Street Baltimore, M D 21205-2185 USA Jamey Marth Howard Hughes Medical Institute University of California, San Diego 9500 Gillman Drive, 0625 La Jolla, CA 92093-0625 USA Rodger McEver Oklahoma Medical Research Ed. 825 NE 13'h Street Oklahoma City, OK 73 104 USA Anant K. Menon Department of Biochemistry University of Wisconsin-Madison 420 Henry Mall Madison, WI 53706-1569 USA

LXI

Reinhold Ohrlein Ciba Speciality Chemicals Inc. K-420.2.19 4002 Base1 Switzerland Peter Orlean 309 Roger A d a m Laboratory University of Illinois at Urbana 600 South Mathews Avenue Urbana, IL 61801 USA Stefan Oscarson Department of Organic Chemistry Stockholm University Arrhenius Laboratory Svante Arrhenius vag 12 10691 Stockholm Sweden Monica M. Palcic Department of Chemistry University of Alberta Edmonton, AB, T6G 2G2 Canada Luigi Panza Dipartimento di Chimica Organica e Industriale Via Venezian 2 1 20133 Milano Italy Armando J. Parodi Biquimicas-Fundaction Campomar University of Buenos Aires Antonio Machado 151 Buenos Aires, 1405 Argentina

LXII

List of Contributors

Serge Perez Centre de Recherches sur les MacromolCcules Vegetales UniversitC Joseph Fourier Cermav-CNRS, BP 53 38041 Grenoble Cedex 9 France

Konrad Sandhoff Institut fur Organische Chemie und Biochemie Universitat Bonn Gerhard-Domagk-Str. 1 53121 Bonn Germany

Thomas Peters Medizinische Universitat Zu Lubeck Institut fur Chemie Ratzeburger Allee 160 23538 Lubeck Germany

Harry Schachter Department of Biochemistry Hospital for Sick Children 555 University Avenue Toronto, ON M5G 1x8 Canada

Vince Pozsgay National Institute of Health 6 Center Dr. MSC 2720 Bldg. 6, Rm. 2A02 Bethesda, MD 20892-2720 USA

Roland Schauer Christian-Albrechts Universitat zu Kiel Olshausenstr. 40 24098 Kiel Germany

Christian Raetz Department of Biochemistry Duke University Medical Center P.O. Box 371 1 Durham, NC 27710 USA

Ronald L. Schnaar Department of Pharmacology and Molecular Science Johns Hopkins School of Medicine 725 North Wolfe Street Baltimore, MD 21205 USA

Ossi Renkonen University of Helsinki Institute of Biotechnology P. 0. Box 56 00014 University of Helsinki Finland

Richard R. Schmidt Fakultat fur Chemie Universitat Konstanz Postf. 5560 M 125 78457 Konstanz Germany

Peter J. Roach Department of Biochemistry Indiana University School of Medicine 635 Barnhill Dr. Indianapolis, IN 46202-5122 USA

Joel Shaper Oncology Center Johns Hopkins Hospital 600 North Wolfe Street Baltimore, MD 21287 USA

Steven D. Rosen Department of Anatomy University of California, San Francisco Box 0452 San Francisco, CA 94143 USA

Pierre Sinay Departement de Chimie Ecole Normale Superieure 24 rue Lhomond 75231 Paris Cedex 50 France

List of Contributors Dorothe Spillmann Department of Medical Biochemistry and Microbiology University of Uppsala The Biomedical Centre P. 0. Box 575 751 23 Uppsala Sweden Robert C. Spiro Department of Biological Chemistry Haward Medical School J o s h Research Laboratory 1 J o s h Place Boston, MA 02215 USA Jun-ichi Tamura Faculty of Education Tottori University Tottori 680-0945 Japan Garry Taylor Department of Biology and Biochemistry University of Bath Calverton Down Bath, BA2 7AY England Joyce Taylor-Papadimitriou Epithelial Cell Biology Laboratory Imperial Cancer Research Fund 44 Lincoln's Inn Field London, WC2A 3PX England Bryan P. Toole Department of Anatomy Tufts University HSC 136 Harrison Avenue Boston, MA 021 11 USA Eric J. Toone Department of Chemistry Duke University Durham, NC 27708-0346 USA Andrea T. Vasella Laboratorium fur Organische Chemie Eidgenossische Technische Hochschule Universitatsstrasse 19 8092 Zurich Switzerland

LXIII

Alain Veyrieres Ecole Nationale Superieure de Chimie de Rennes Campus du Beaulieu Avenue du General Leclerc 35700 Rennes France Barbara Vertel Department of Cell Biology and Anatomy Chicago Medical School 3333 Green Bay Road North Chicago, IL 60064-3095 USA Johannes F. G. Vliegenhart Bijvoet Center for Biomolecular Research Utrecht University P. 0. Box 80075 3508 TB Utrecht The Netherlands David J. Vocadlo Department of Chemistry University of British Columbia Vancouver B.C. V6T 1Zl Canada Russell Wallis Department of Biochemistry Glycobiology Institute University of Oxford South Parks Road Oxford, OX1 3QU England Thomas Warner 541 Wellington Drive San Carlos, CA 94070 USA Hans Peter Wessel F. Hoffmann-La Roche AG Pharma Research Discovery Chemistry PRPV-D Building 15/110 4070 Basel Switzerland Thomas N. Wight Department of Pathology SM-30 University of Washington School of Medicine Seattle, WA 98 195 USA

LXIV

List of Contributors

David B. Williams Department of Biochemistry University of Toronto Toronto, ON M5S 1A8 Canada Lawrence J. Williams Laboratory for Bioorganic Chemistry The Sloan-Kettering Institute for Cancer Research 1275 York Avenue New York, NY 10021 USA

Stephen G. Withers Department of Chemistry University of British Columbia 2036 Main Mall Vancouver, B.C. V6T 1Zl Canada Chi-Huey Wong Department of Chemistry Scripps Research Institute 10550 N. Torrey Pines Road La Jolla, CA 92037 USA

Abbreviations Used in Volumes 1 and 2

a1,3-GalT a2,3-SialT a2,6-SialT aGalF aGlcF aMuNeuAc apNPFuc apNPGal apNPGalNAc apNPGlc apNPGlcNAc apNPMan apNPNeuAc

A aa Ac AcHmb ACP AF AFM AgOTf AGP 019.

All AllNAc Aloc APC APP APS APS

a 1,3-GaIactosyltransferase a2,3-Sialyltransferase a2,6-sialyltransferase a-Galactosyl fluoride a-Glucosyl fluoride 4-Methylumbelliferone a-D-N-acetylneuraminic acid para-Nitrophenyl a-L-fucopyranoside para-Nitrophenyl a-D-galactopyranoside para-Nitrophenyl 2-acetamido-2-deoxy-a-~-galactopyranoside para-Nitrophenyl a-D-glucopyranoside para-Nitrophenyl 2-acetamido-2-deoxy-a-~-glucopyranoside para-Nitrophenyl a-D-mannopyranoside para-Nitrophenyl-a-D-N-acetyheuraminic acid Angstrom amino acid acetyl

N-2-acetoxy-4-methoxybenzyl acyl carrier protein aggregation factor atomic force microscopy silver trifluoromethanesulfonate (silver triflate) arabinogalactan protein asparagine-linked glycosylation ally1 N-Acetylallosamine Ally loxycarbony1 antigen-presenting cell acute phase protein adenosine phosphosulfate ammonium persulfate

LXVI

Abbreviations Used in Volumes I and 2

AR ARIS ASGP-R Asialo-GM1 Asn Asn GT ASTF-GP j3 1,4-GalT PManF PMuGlc PMuGlcNAc PMuXyl PoNPGal PoNPGlc PpClPMan PPhGal PPhGlc ppNP2dGlc PpNPGal PpNPGalNAc PpNPGlc PpNPGlcNAc PpNPMan PPNPXYl PpOMePhGlc PVGal 9-BBN BDA BHK bFGF BLAST BM BMP Bn Boc BOP bP Bu BU4NSSTr Bz Bzl CAA CAM CAN

acrosome reaction acrosome reaction-inducing substance asialoglycoprotein receptor p-D-Gal( 1-3)-b-~-GalNAc(1-4)-P-D-Gal( 1-4)-P-~-Glc-O-Cer asparagine asparagine glucosyltransferase (GalG1cNAcMan)zManGlcNAcz Asn P 1,4-Galactosyltransferase P-Mannosyl fluoride 4-Methylumbelliferyl P-o-glucopyranoside 4-Methylumbelliferyl 2-acetamido-2-deoxy-~-~-glucopyranoside 4-Methylumbelliferyl P-D-xylopyranoside ortho-Nitrophenyl P-D-galactopyranoside ortho-Nitrophenyl P-D-glucopyranoside para-Chlorophenyl P-D-mannopyranoside Phenyl P-D-galactopyranoside Phenyl P-D-glucopyranoside para-Nitrophenyl2-deoxy-P-~-arabino-hexopyranoside para-Nitrophenyl P-D-galactopyranoside para-Nitrophenyl 2-acetamido-2-deoxy-P-~-galactopyranoside para-Nitrophenyl P-D-glucopyranoside para-Nitrophenyl 2-acetamido-2-deoxy-~-~-glucopyranoside para-Nitrophenyl P-D-mannopyranoside para-Nitrophenyl P-D-xylopyranoside para-Methoxyphenyl P-D-glucopyranoside Vinyl 0-D-galactopyranoside 9-Borabicyclo[3.3.llnonane butane diacetal baby hamster kidney basic fibroblast growth factor basic alignment search tool basement membrane bis(monoacylg1ycero)phosphate benzyl tert-butoxycarbonyl benzotriazol-1-yloxy-tris(dimethy1amino)phosphoniumhexafluorophosphate base pair butyl Tetrabutylammoniumsaltoltriphenylmethanethiol benzoyl benzyl carbohydrate amino acids cell adhesion molecule Cer(1V)ammoniumnitrate

Ahhreviutions Used in Volumes 1 and 2 LXVII CAT CBMIT CBS CBS Cbz CCCP CD CDA CDGS CDMT cDNA CDTA Cer CHO CgOse4 CGTase CI CI8 CIAP ClAc ClNP CMLN CMP CMP CMP-NeuAc/ CMP-NeuSAc CMP-Sia CNBr CNS CoA Coll CPG CRD

cs

CSA CSA

csc CSPG CST CTL CTP Da DAF DAG DBA DBMP

chloramphenicol acetyltransferase 1,l-carbonylbis(3-methylimidazolium triflate Gal(S03)Pl- 1Cer Cerebroside sulfate benzyloxycarbonyl carbonyl cyanide m-chlorophenylhydrazone cation dependent cyclohexane-l,2-diacetal carbohydrate-deficient glycoprotein syndrome 2-chloro-4,6-dimethoxy-l,3,5-triazine complementary DNA diaminocyclohexane tetraacetic acid ceramide Chinese hamster ovary fi-~-Gal( 1-3)-o-~-GalNAc( 1-4)-fi-~-Gal( 1-4)-p-~-Glc cyclodextrin glucosyltransferase cation independent Cycloisomalto-octaose calf intestine alkaline phosphatase [ EC.3.1.3.11 chloroacetyl 2-Chloro-4-nitrophenyl calf mesenteric lymph node cytosine monophosphate cytidine 5'-monophosphate cytidine 5'-monophosphate N-acetylneuraminic acid Cytidine 5'-monophosphate sialic acid cyanogen bromide central nervous system coenzyme A sym-collidine controlled pore glass carbohydrate-recognition domain chondroitin sulfate Campher sulfonic acid chondroitin-4-sulfate chondroitin-6-sulfate chondroitin sulfate proteoglycan castanospermine cytotoxic T cell 5 '-Cytidinetriphosphate Dalton decay accelerating factor diacylglucosamine ((strain D x strain B) x strain A) mice 2,6-di-tert-butyl-4-methylpyridine

LXVIII Abbreviations Used in Volumes 1 and 2 DBU DCC DCPhth Dde DDP DDQ DEPC DF 2dGlc DIBALH DiBz DIG DIM DIPEA DMA DMAP DME DMF DMJ DMM DMPU DMSO DMTST DMTST 2,4-DNP DNJ DNJ Do1 Dol-P Dol-PP DP DPPA DRM DS DTBMP DTBP DTBPI DTT E-CAD ECM EDCD/EDCI EDTA EGF eIF EM

1,8-diazabicyclo[5.4.01undec-7-ene N ,N’-Dicyclohexylcarbodiimide 4,s-dichlorophthaloyl N- 1- (4,4-dimethy1-2,6-dioxocyc1ohexy1idene) ethy 1 dialkyl N,N-diethylphosphoramidites 2,3-Dichloro-S,6-dicyano-1,4-benzoquinone diethylphosphoryl cyanide deactivation factor 2-Deoxy-arabino-hexose Diisobutylaluminumhydride Dibenzoyl detergent-insoluble glycolipid complex 1,4-dideoxy-1,4-imino-~-mannitol diisopropylethylamine N ,N-dimethylacetamide 4-Dimethylaminopyridine 1,Zdimethoxyethane N ,N-dimethylformamide 1-deoxymannojirimycin dimethylmaleoyl N ,N’-dimethylhexahydropyrimidin-2-one dimethyl sulfoxide dimethyl(methy1thio)sulfonium tetrafluoroborate (dimethyl (methy1thio)sulfoniumtriflate) Dimethyl(methy1thio)sulfoniumtetrafluoromethansulfonate 2,4-Dinitrophenyl deoxynojirimycin 1,5-Dideoxy-l,S-irnino-D-glucitol (deoxynojirimycin) dolichol dolichyl phosphate dolichyl pyrophosphate degree of polymerization diphenylphosphoryl azide detergent-resistant membrane dennatan sulfate 2,6-di-tert-butyl-4-methylpyridine 2,6-di-tert-butylpyridine 2,6-di-tert-butylpyridiniumiodide dithiothreitol Epithelial Cadherin I extracellular matrix 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide ethylene diamine tetraacetic acid epidermal growth factor eukaryotic initiation factor electron microscopy

Abbreviations Used in Volumes I and 2 LXIX endo H EPGase ER ER ESI EST EtN EtN-P F FAK FCOE FGF Fmoc/FMOC FSP FT FTIR FTIR Fuc Fuc FucT GA GAG Gal Gal GalNAc GalT Gal- 1-P UT Gb3 Gb;l GC GC-MS GDI, GDP GDP GDP-Fuc GDP-Man GDPMP Gg3 GI GI1 Glc Glc GlcA GlcN GlcN

endo-0-N-acetylglucosaminidase H endopolygalacturonase endoplasmic reticulum endoplasmic reticulum electrospry ionization expressed sequence tag ethanolamine ethanolamine phosphate fusion protein focal adhesion kinase fluorophore-coupled oligosaccharide electrophoresis fibroblast growth factor 9-Fluorenylmethoxylcarbonyl(fluorenyl-9-methoxycarbonyl) fucose sulfate polysaccharide a-(1-3/4)-fucosyltransferase [ EC.2.4.1.651 Fourrier transform infra red Fourier transform infra-red (spectroscopy) fucose/fucosyl fucose fucosyltransferase Golgi apparatus glycosaminoglycan galactose galactose/galactosy1 N-acetylgalactosamine galactosyltransferase Gal- 1-P-uridyltransferase Gala+4Gal~1+4Glc~lA+lCer

GalNAcPl+3Gala1+4Gal~l+4Glc~l+lCer glucosyl cerdmide gas chromatography-mass spectrometrya-DNeuAc(2-3)-P-~-Gal(1-3)-p-~-GalNAc(1-4)-[a-DNeuAc(2-3)]-P-~-Gal(1-4)-p-~-Glc-o-CerGDP guanosine diphosphate guanosine 5'-diphosphate guanosine 5I-diphosphate fucose guanosine 5'-diphosphate mannose GDP-Man pyrophosphorylase

GalNAcP1+4Galpl+4Glcpl+lCer gastro-intestinal glucosidase I1 glucose/glucosyl glucose glucuronic acid 2-Amino-2-deoxy -glucopyr anose glucosamine

LXX

Abbreviations Used in Volumes I and 2

GkN3 Glc3NAc GlcNAc GlcNAc GlcNAc, GlcNAc-Oxazoline GlcNPht GlcNTeoc Glucal GLUT GM~ GMi

GM3 GM3 GMP GnT GP GPI GRIFIN GRP GSL GT GT GTA GTB H HA HAF HAg HAS HBP HBPyU hCG HEMPAS test HeP HEPES HEV HIV HIV

2-Azido-2-deoxy-~-glucopyranose 3-Acetamido-3-deoxy-~-glucopyranose N-acetylglucosamine 3,4,6-tri-O-acetyl-N-acetylglucosamine { p-~-GlcNAc( 1-4)) ( , - l ) - ~ - G l ~ N A ~ 2-Methyl-4,5-dihydro-(1,2-dideoxy-a-~-g~ucopyranoso) [2,1-d]-1,3-oxazole 2-Deoxy-2-phthalimido-~-glucopyranose 2-Deoxy-2-aminot~ethoxylcarbonyl-~-g~ucopyranose 1,5-Anhydr0-2-deoxy-~-arabino-hex1-enitol glucose transporter GalPl+3GalNAcf31 -+[NeuAcu2+3]Gal~l+4Glc~l+1Cer p-D-Gal(1-3)-p-~-GalNAc(1-4)-[u-~-NeuAc(2-3)1-p-~Gal( 1-4)-p-D-GlC-O-Cer P-D-GalNAc(1-4)-[u-~-NeuAc(2-3)]-p-~-Gal( 1-4)-p-~Glc-0-Cer NeuAca2+3Gal~l-+4G1c~l+ lCer a-~-NeuAc(2-3)-P-D-Gal(1-4)-p-~-Glc-O-Cer guanosine monophosphate N-acetylglucosaminyltransferase glycoprotein glycosyl phosphatidylinositol galectin-related interfiber protein glycine-rich protein glycosphingolipid p-(1-4)-galactosyltransferase [ EC.2.4.1.221 (glycosyltransferase) UDP-G1c:glycoprotein glucosyltransferase human blood group A glycosyltransferase human blood group B glycosyltransferase Fucul+2Galpl+3/4GlcNAcP hyaluronic acid, hyaluronate Hulicondriu aggregation factor hemagglutinin hyaluronate synthesizing enzyme hexosamine biosynthetic pathway 0-(benzotriazole-1-yl)-N,N,N,N-bis(tetramethy1ene)uronium hexafluorophosphate human chorionic gonadotropin hereditary erythroblastic multinuclearity with a positive acidified-serum-lysis (congenital dyserythropoietic anemia type 11) heparin N-(2-Hydroxyethylpiperazine)-N'-(2-ethanesulfon~c acid) high endothelial venules Human immunodeficiency virus human immunodeficiency virus

Abbreviations Used in Volumes 1 and 2 HL HMG HN HOAll HOAT HOBt HOMO HPLC HPLC HRGP HS HSD HSPG HUS HYCRON HYP ICAM IDCP IDCT IduA IGF Ig IgSF IIDQ IL IPG IU IV3NeuAc PGalNAcGbOse4 I V NeuAcII NeuAcCgOse4 K

kb kDa Kdn

Kdn/KDN Kdo Kdo K-4 Ki KIF Krn KS KSPG

LXXI

hepatic lectin high mobility group (proteins) hemagglutinin-neuraminidase Ally1 alcohol 1-hydroxy-7-azabenztriazole 1-hydroxybenzotriazole hydrate highest occupied molecular orbital high-performance liquid chromatography high performance liquid chromatography hydroxyproline-rich glycoprotein heparan sulfate highly sulfated domain heparan sulfate proteoglycan hemolytic uremic syndrome hydroxycrotyl-oligoethylene glycol-n-alkanoyl 4-hydroxyproline intercellular adhesion molecule iodonium dicollidine perchlorate iodonium dicollidine triflate iduronic acid insulin-like growth factor immunoglobulin immunoglobulin superfamily of proteins isobutoxyisobutyldihydroquinoline interleukin inositol phosphoglycan international unit p-D-GalNAc(1-3)-p-~-GalNAc(1-3)-u-~-Gal(1-4)-p-DGal(1-4)-~-Glc a-u-NeuAc(2-3)-P-~-Gal(1-3)-p-~-GalNAc(1-4)-[a-DNeuAc(2-3)]-P-~-Gal(1-4)-Dkinase enzyme activity coefficient kilobase (thousand base pairs) kiloDalton 2-keto-3-deoxy-nononicacid, formally 5-desamino-5hydroxyneuraminic acid 3-deoxy-~-y~ycero-D-gu~ucto-2-nonulosonic acid ketodeoxoctulosonic acid (3-deoxy-~-mannooctulosonic acid) 2-keto-3-deoxy-~-manno-octulosonic acid kinase enzyme activity coefficient, at equilibrium kinase enzyme activity coefficient, for inhibition kifunensine kinase enzyme activity coefficient, for structural modification(s) keratan sulfate keratan sulfate proteoglycan

LXXII Abbreviations Used in Volumes 1 and 2 LacAm LacCer LacNAc LacNAc LAD LAMP LBP Lc3 LDA Lea Lea Leb Lea, LeX LeX LeX Ley Leu Lev LNP LPS LRP LRR LUMO LYS mAb MAd MAF MAG MALDI-TOF MALDI-TOF Man ManNAc MAP MASP Man-T MBL MBP MBz MCD MDBK MDCK ME MEM MHC MK

lactosamine GalP 1+4GlcP1+ 1Cer N-acetyllactosamine N-acetyllactosamine leucocyte adhesion deficiency lysosomal associated membrane protein lipopolysaccharide-binding protein GlcNAcPl-.3Gal~1+4GlcP1+ 1Cer Lithium diisoprypropylamide GalP 1+3[Fucal+4]GlcNAcP1+3Gal~1+GlcP P-~-Gal(l-3)-[a-~-Fuc( 1-4)]-~-GlcNAc Fuca 1+2GalP 1+3[ Fucal-+4]GlcNAc~ 1+3GalP 1-4GlcP Lewis a, Lewis x determinants p-D-Gal(1-4)-[a-~-Fuc( 1-3)]-~-GlcNAc GalP 1+4[ Fucal+3]GlcNAcP 1+3GalP 1-+4GlcP Fuca 1-2GalP 1+4[Fucal+3]GlcNAcP 1+3GalP 1-4ClcP leucine levulinoyl lectin and nucleotide phosphohydrolase lipopolysaccharaide lipoprotein receptor-related protein leucine-rich repeat (amino acid/protein) lowest unoccupied molecular orbital lysine monoclonal antibody mucosal addressin Microciona prolifera aggrigation factor myelin-associated glycoprotein mass-analyzed laser desorption ionization-time of flight matrix-assisted-laser desorption ionization time of flight mass spectrometry mannose N-acetylmannosamine microtubule-associated protein MBP-associated serine protease mannosyltransferase mannose-binding lectin mannose-binding protein p-methoxybenzoyl macular corneal dystrophy Madin-Darby bovine kidney (cells) Madin-Darby canine kidney (cells) methoxyethyl methoxyethoxymethyl major histocompatibility complex myokinase

Abbreviations Used in Volumes 1 and 2 LXXIII mol.wt. MP MPEG MPR MPS M, MS MSD MTr Mu MUC NA NAc nAChR NAm NANA N-BOC NBS NCAM NCS N DP NDP NDST NEM 3NeuAcCgOse4 Neu5Ac NeuAc NGF NIH NIS NK nLc4 Nle NMK NMP NMR NOE NPG NS NTP Nu OD (2-OH)-Lea (2-OH)-Le" (2-OH)-sLea

molecular weight p-methoxyphenyl polyethylene glycol monomethyl ether mannose 6-phosphate receptor mucopolysaccharidosis apparent molecular weight mass spectrometry multiple sulfatase deficiency, Austin disease p-methoxytrityl 4-Methylumbelliferone mucin N-acetylated disaccharide unit N-acetyl nicotinic acetylcholine receptor neuraminidase N-acetylneuraminic acid N-tert-butoxycarbonyl N-bromosuccinimide neural cell adhesion molecule N-chlorosuccinimide nucleoside diphosphates nucleoside-diphosphate N-deacetylaselN-sulfotransferase N-ethylmaleimide p-~-Gal(1-3)-o-~-GalNAc( 1-4)-[a-~-NeuAc(2-3)]-P-DGal( 1-4)-D-Glc N-acetylneuraminic acid N-acetylneuraminic acid nerve growth factor National Institutes of Health N-iodosuccinimide natur a1 killer Gal~l+4GlcNAc~1+3Gal~1-t4Glc~1-+ 1Cer Norleucin nucleoside monophosphate kinase nucleoside monophosphate nuclear magnetic resonance nuclear Overhauser effect n-pentenyl glycoside N-sulfated disaccharide unit nucleoside triphosphate nucleophile optical density P-D-Gal(1-3)-[a-~-Fuc( 1-4)I-D-Glc p-D-Gal(1-4)-[a-~-Fuc(1-3)I-D-Glc a-NeuAc(2-3)-P-~-Gal(1-3)-[a-~-Fuc(1-4)]-~-Glc

LXXIV Abbreviations Used in Volumes I and 2 (2-OH)-sLe” OR ORF OSP OST OST P PA Pac PAMP PAPS PAPS PARP PBL pClP PCR PDGF PDI PEG PEGA PEP PEP PfP

pmu

PG PGM PGM Ph PhSeOTf PhSOTf Phth PI Piv PK PMB pNP-a-Man PNS PNZ POEPOP Ppase ppGaNTases PPK PRP PSA PSGL

a-NeuAc(2-3)-P-~-Gal(1-4)-[a-~-Fuc( 1-3)]-~-Glc dimethyl(methy1thio)sulfonium trifluoromethanesulfonate open reading frame 0-specific polysaccharides oligosaccharyltransferase N-oligosaccharyltransferase phosphate 2-Aminopyridine phenacyl pathogen-associated molecular pattern) 3 ‘-phosphoadenosine 5’-phosphosulfate phosphoadenosine phosphosulfate procyclic acidic repetitive protein peripheral blood leucocytes para-Chlorophenyl polymerase chain reaction platelet-derived growth factor protein disulfide isomerase poly(ethy1ene glycol) polyethylene glycol N,N-dimethylacrylamide copolymer phospho(eno1)pyruvate phosphoenolpyruvate pentafluorophenyl N,N,N’,N’-bis(tetramethylene)-O-pentafluorophenyluronium hexafluorophosphate phosphatidylglycerol phosphoglucomutase phosphoglucomutase phenyl Phenylseleniumtriflate Phenylsulfoniumtriflate phthaloyl phosphatidylinositol pivaloyl pyruvate kinase p-methoxybenzyl p-nitropheny1-a-D-mannoside peripheral nervous system p-nitrobenzyloxycarbonyl poly oxyethylene-polyoxypropylene pyrophosphatase UDP-Ga1NAc:polypeptide N-acetylgalactosaminyltransferases polyphosphate kinase proline-rich protein polysialic acid P-Selectin glycoprotein ligand

Abbreuiutions Used in Volumes I and 2 LXXV PSGL-1 PSGL- 1 PTG PYr 9.Y. RA RAr RACE rER RHAMM ROESY RT-PCR SA SAP SAT SBA SBu SDS SDS-PAGE Se SE Ser SEt SGC SGG SGGL SGLT sia sLea sLea, sLe' sLe' 6'-SLN SLRP SMC SP SPh SPH SSEA ST ST3 ST6 STF-GP SV

sw

P-selectin glycoprotein ligand-1 P-selectin glycoprotein ligand-1 peptidoglycan pyridine quantitative yield retinoic acid rheumatoid arthritis rapic amplification of cDNA ends rough endoplasmic reticulum receptor for hyaluronan-mediated motility rotating-frame nuclear Overhauser effect reverse transcription polymerase chain reaction sialyl saposin sialyltransferase soybean agglutinin Thiobutyl sodium dodecyl sulfonate sodium dodecylsulfate-polyacrylamide gel electrophoresis secretor enzyme 2-(trimethylsily1)ethyl serine Thioethyl sulfogalactosyl ceramide sulfogalactosyl glycerolipid sulfoglucuronyl glycolipid sodium-glucose transporter sialic acid a-~-NeuAc(2-3)-P-~-Gal( 1-3)-[ a-~-Fuc( 1-4)]-~-GlcNAc(sialyl Lewisa trisaccharide) sialylated Lewis a, sialylated Lewis x determinants a-~-NeuAc(2-3)-P-~-Gal( 1-4)-[a-~-Fuc( 1-3)]-~-GlcNAc(sialyl Lewis' trisaccharide) 6'4alyl-LacNAc small leucine-rich proteoglycan sialomucin complex spacer Thiophenyl saccharide-peptide hybrids stage specific embryonic antigen sulfurotransferase recombinant rat liver a-(2-3)-sialyltransferase [EC.2.4.99.6] rat liver a-(2-6)-sialyltransferase jEC.2.4.99.11 (NeuAcGalGlcNAcMan)2ManGlcNAqAsn synaptic vesicle-associated protein swainsonine

LXXVI Abbreviations Used in Volumes 1 and 2 TATU TBAF TBAI TBDMS TBDMSOTf TBDPS TBDPSCl TBTU Bu TCA Tcoc TCP TCR TDS TEMED Teoc TES TESOTf Tf20 TFA TFDMSCl TfOH TGF TGN THF Thr tlc TM TMD TMS TMSCN TMSN3 TMSOTf to1 Tos t-PA TriBz Troc Trt Ts UDP UDP UDP-Gal UDP-Gal

0-(7-azabenzotriazol-1-yl)-3-tetramethyluronium tetrafluoroborate Tetrabutylammoniumfluoride Tetabutylammoniumiodide tert-butyldimethylsilyl tert-butyldimethylsilyl trifluoromethanesulfonate (tert-butyldimethylsilyl triflate) tert-butyldiphenylsilyl tert-butyldiphenylsilylchloride 0-(benzotriazol- 1-yl)-N,N,N',N'-tetrauroniumtetrafluoroborate tert-butyl trichloroacetyl trichloroethyloxycarbonyl tetrachlorophthaloyl T cell receptor thexyldimethylsilyl tetramethylethylenediamine 2,2,2-trichloroethoxycarbonyl triethylsilyl triethylsilyl trifluoromethanesulfonate (triethylsilyl triflate) trifluoromethane sulfonic acid anhydride trifluoroacetic acid trifluoromethyldimethylsilylchloride trifluoromethanesulfonic acid (triflic acid) transforming growth factor trans Golgi network tetrahydrofuran threonine thin layer chromatography transmembrane trans-membrane domain trimethylsilyl trimethylsilyl cyanide trimethylsilyl azide trimethylsilyl trifluoromethanesulfonate (trimethylsilyl triflate) toluene p-toluenesulfonyl (tosyl) tissue plasminogen activator factor tribenzoyl

2,2,2-trichloroethoxycarbonyl triphenylmethyl (trityl) p-toluenesulfonyl (tosyl) uridine 5'-diphosphate uranyl diphosphate UDP-a-D-galactose uridine 5'-diphosphate-galactose

Abbreviations Used in Volumes I and 2 UDPGE UDP-Glc UDPGP UMD UMP UTP

uv vc

VE VL VSG

vsv

VT vWF WBP XYl XY12 Z ZP

LXXVII

uridine 5'-diphosphate-glucose epimerase [EC.5.1.3.2.] uridine 5'-diphosphate-glucose UDP-Glc pyrophosphorylase unmodified domain uranyl monophosphate uranyl triphosphate ultraviolet vitelline coat vitelline envelope vitelline layer variant surface glycoprotein vasicular stomatis virus verotoxin von Willebrand factor wheat germ agglutinin binding protein xylose p-D-xyl(1--4)-D-xyl benzyloxycarbonyl zona pellucida

Abbreviations Used in Volumes 3 and 4

A" aa ACP AGP 4 7

APC APP APS AR ARIS ASGP-R Asn Asn GT BHK bFGF BLAST BM BMP bP CAM CAT CCCP CD CDGS CDTA Cer CHO CI CMLN CMP

Angstrom amino acid acyl carrier protein arabinogalactan protein asparagine-linked glycosylation antigen-presenting cell acute phase protein adenosine phosphosulfate acrosome reaction acrosome reaction-inducing substance asialoglycoprotein receptor asparagine asparagine glucosyltransferase baby hamster kidney basic fibroblast growth factor basic alignment search tool basement membrane bis(monoacy1glycero)phosphate base pair cell adhesion molecule chloramphenicol acetyltransferase carbonyl cyanide m-chlorophenylhydrazone cation dependent carbohydrate-deficient glycoprotein syndrome diaminocyclohexane tetraacetic acid ceramide Chinese hamster ovary cation independent calf mesenteric lymph node cytosine monophosphate

LVIII

Abbreviations Uscd in Volumes 3 and 4

CNS CoA CRD CS CSPG CST CTL Da DAF DAG DBA DIG DIM DMJ DMSO DNJ Do1 Dol-P Dol-PP DP DRM DS ECM EDTA EGF eIF EM endo H EPGase ER EST EtN EtN-P F FAK FCOE FGF FSP FTIR Fuc FucT GA GAG Gal GalNAc GalT

central nervous system coenzyme A carbohydrate-recognition domain chondroitin sulfate chondroitin sulfate proteoglycan castanospermine cytotoxic T cell Dalton decay accelerating factor diacylglucosamine ((strain D x strain B) x strain A) mice detergent-insoluble glycolipid complex 1,4-dideoxy-l,4-imino-~-mannitol 1-deoxymannojirimycin dimethyl sulfoxide deoxynojirimycin dolichol dolichyl phosphate dolichyl pyrophosphate degree of polymerization detergent-resistant membrane dermatan sulfate extracellular matrix ethylene diamine tetraacetic acid epidermal growth factor eukaryotic initiation factor electron microscopy endo-p-N-acetylglucosaminidaseH endopol y galacturonase endoplasmic reticulum expressed sequence tag ethanolamine ethanolamine phosphate fusion protein focal adhesion kinase fluorophore-coupled oligosaccharide electrophoresis fibroblast growth factor fucose sulfate polysaccharide Fourier transform infra-red (spectroscopy) fucose fucosyltransferase Golgi apparatus glycosaminoglycan galactose N-acetylgalactosamine galactosyltransferase

Ahbreniutions Used in Volumes 3 and 4

GC GC-MS GDP GI GI1 Glc GlcA GlcN GlcNAc GLUT GnT GP GPI GRIFIN GRP GSL GT GTA GTB HA H Ag HAS HBP HEMPAS hCG HeP HEV HIV HL HMG HN HPLC HRGP HS HSD HSPG HUS ICAM lduA IGF Ig IgSF IL IPG

1u

LIX

glucosyl ceramide gas chromatography-mass spectrometry guanosine diphosphate gastro-intestinal glucosidase I1 glucose glucuronic acid glucosamine N-acetylglucosamine glucose transporter N-acetylglucosaminy ltransferase glycoprotein glycosyl phosphatidylinositol galectin-related interfiber protein glycine-rich protein glycosphingolipid UDP-G1c:glycoprotein glucosyltransferase human blood group A glycosyltransferase human blood group B glycosyltransferase hyaluronic acid, hyaluronate hemagglutinin hyaluronate synthesizing enzyme hexosamine biosynthetic pathway hereditary erythroblastic multinuclearity with a positive acidifiedserum-lysis test (congenital dyserythropoietic anemia type 11) human chorionic gonadotropin heparin high endothelial venules human immunodeficiency virus hepatic lectin high mobility group (proteins) hemagglutinin-neuraminidase high performance liquid chromatography hydroxyproline-rich glycoprotein heparan sulfate highly sulfated domain heparan sulfate proteoglycan hemolytic uremic syndrome intercellular adhesion molecule iduronic acid insulin-like growth factor immunoglobulin immunoglobulin superfamily of proteins interleukin inositol phosphoglycan international unit

LX

K kb kDa Kdn

Abbreviations Used in Volumes 3 and 4

kinase enzyme activity coefficient kilobase (thousand base pairs) kiloDalton 2-keto-3-deoxy-nononic acid, formally 5-desamino-5-hydroxyneuraminic acid 2-keto-3-deoxy-~-manno-octulosonic acid Kdo kinase enzyme activity coefficient, at equilibrium Keq kinase enzyme activity coefficient, for inhibition Ki kifunensine KIF kinase enzyme activity coefficient, for structural modification(s) Km keratan sulfate KS keratan sulfate proteoglycan KSPG lactosamine LacAm N-acetyllactosamine LacNAc leucocyte adhesion deficiency LAD lysosomal associated membrane protein LAMP lipopolysaccharide-binding protein LBP Lewis a, Lewis x determinants Lea, Le" leucine Leu lectin and nucleotide phosphohydrolase LNP lipopoly saccharaide LPS lipoprotein receptor-related protein LRP leucine-rich repeat (amino acid/protein) LRR lysine LYS monoclonal antibody mAb mucosal addressin MAd myelin-associated glycoprotein MAG MALDI-TOF matrix-assisted-laser desorption ionization time of flight inass spectrometry mannose Man microtubule-associated protein MAP MBP-associated serine protease MASP mannose-binding lectin MBL mannose-binding protein MBP mannosyltransferase Man-T macular corneal dystrophy MCD Madin-Darby bovine kidney (cells) MDBK Madin-Darby canine kidney (cells) MDCK major histocompatibility complex MHC molecular weight mol.wt. mannose 6-phosphate receptor MPR mucopolysaccharidosis MPS apparent molecular weight Mr multiple sulfatase deficiency, Austin disease MSD N-acetylated disaccharide unit NA N-acetyl NAc

Ahhreviutions Used in Volumes 3 and 4

nAChR NAm NANA NCAM NDP NDST NEM NGF NIH NK NMR NS ORF OST P PAMP PAPS PARP PRL PCR PDGF PDI PEP PG PGM PI pNP-a-Man PNS PRP PSA PSGL-1 PTG RA RAr RACE rER RHAMM RT-PCR SA SAP SAT SBA SDS-PAGE Se Ser SGC

nicotinic acetylcholine receptor neuraminidase N-acetylneuraminic acid neural cell adhesion molecule nucleoside-diphosphate N-deacetylaselN-sulfotransferase N-ethylmaleimide nerve growth factor National Institutes of Health natural killer nuclear magnetic resonance N-sulfated disaccharide unit open reading frame N-oligosaccharyltransferase phosphate pathogen-associated molecular pattern phosphoadenosine phosphosulfate procyclic acidic repetitive protein peripheral blood leucocytes polymerase chain reaction platelet-derived growth factor protein disulfide isomerase phosphoenolpyruvate phosphatidylglycerol phosphoglucomutase phosphatidylinositol p-nitrophenyl-a-D-mannoside peripheral nervous system proline-rich protein polysialic acid P-selectin glycoprotein ligand-1 peptidoglycan retinoic acid rheumatoid arthritis rapic amplification of cDNA ends rough endoplasmic reticulum receptor for hyaluronan-mediated motility reverse transcription polymerase chain reaction sialyl saposin sialyltransferase soybean agglutinin sodium dodecylsulfate-polyacrylamide gel electrophoresis secretor enzyme serine sulfogalactosyl ceramide

LXI

LXII

Abbreviations Used in Volumes 3 and 4

SGG SGGL SGLT sia sLea, sleX SLRP SMC ST

sv

SW TCR TGF TGN Thr tlc TM t-PA UDP UDP-Gal UMD UMP UTP UV

vc

VE VL VSG

vsv

VT vWF WBP XYl ZP

sulfogalactosyl glycerolipid sulfoglucuronyl glycolipid sodium-glucose transporter sialic acid sialylated Lewis a, sialylated Lewis x determinants small leucine-rich proteoglycan sialomucin complex sulfurotransferase synaptic vesicle-associated protein swainsonine T cell receptor transforming growth factor trans Golgi network threonine thin layer chromatography transmembrane tissue plasminogen activator factor uranyl diphosphate UDP-a-D-galactose unmodified domain uranyl monophosphate uranyl triphosphate ultraviolet vitelline coat vitelline envelope vitelline layer variant surface glycoprotein vasicular stomatis virus verotoxin von Willebrand factor wheat germ agglutinin binding protein xylose zona pellucida

Part I

Volume 1

I Chemical Synthesis of Glycosides

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

1 Introduction to Volumes 1 and 2

As no better man advances to take this matter in hand, I hereupon offer my own poor endeavours. I promise nothing complete; because any human thing supposed to be complete must for that very reason infallibly be faulty. (from Moby Dick, Herman Melville)

Our understanding of the relationship between carbohydrate structure and biological function is still far behind that of proteins and nucleic acids. Initially, carbohydrates were only recognized as structural and energy storage molecules (e.g. cellulose, chitin and glycogen). The development of novel tools to study carbohydrate function has permitted additional biological functions to be elucidated. Their roles in protein folding, cell signaling, fertilization, pathogen binding to host tissue, leukocyte trafficking and associated inflammatory responses, tumor cell metastasis, and regulation of hormone and enzyme activities are just a few selected examples. These findings led to an immense interest in the preparation of oligosaccharides and glycoconjugates as well as the understanding of their interaction with their natural receptors. Part 1 of Carbohydrates in Chemistry and Biology is designed to be a valuable collection of chemical and enzymatic methods for the synthesis of oligosaccharides. It also covers state-of-the-art knowledge of the interactions of carbohydrates with their natural receptors. Each of the chapters is written by a leading expert in the field and covers an impressive body of references. Volume 1 of the series begins with the most challenging aspect of synthetic carbohydrate chemistry, namely the stereospecific formation of glycosidic bonds. A number of chapters describe recently developed approaches and cover numerous significant advances in glycosidation techniques. Other chapters review the attachment of sugars through linkages not occurring in nature. Volume 1 concludes by summarizing one of the most exciting and rapidly evolving area in modern carbohydrate synthesis: the use of glycosyl transferases and glycosidases in the synthesis of oligosaccharides and their analogs.

4

1 Introduction to Volumes 1 and 2

Volume 2 of the series covers the interactions of carbohydrates with their natural receptors, i.e. proteins, nucleic acids, and other carbohydrates. Whereas carbohydrate-protein interactions are well documented, the investigation of carbohydratenucleic acid and carbohydrate-carbohydrate interactions is an emerging area. The Carbohydrates in Chemistry and Biology series provides a comprehensive and up-to-date compilation of information on the fields of glycochemistry (Volumes 1 and 2) and glycobiology (Volumes 3 and 4),available today. The volumes not only provide an expert overview of the entire field, but also serve as a valuable reference tool to both, established investigators and students in many disciplines. The editors are obliged to all authors for their excellent contributions. Special thanks goes to Dr. Anette Eckerle, Wiley-VCH, who managed the realization of this series. Beat Ernst, Ph.D. Professor of Molecular Pharmacy University of Base1 Pierre Sinay, Ph.D. Professor of Organic Chemistry Ecole Normale SupCrieure, Paris

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

2 Trichloroacetimidates Richard R. Schmidt and Karl-Heinz Jung

2.1 Introduction Glycoside bond formation generally requires activation of the sugar at the anomeric center. To this end, activation through anomeric oxygen-exchange reactions (the classical Koenigs-Knorr method and its modifications, sulfur based activations, etc.) or, alternatively, activation through retention of the anomeric oxygen (trichloroacetimidate method, phosphite and phosphate activation, etc.) have been proposed. The trichloroacetimidate method is well established [ 1-1 31 and although the first paper on this method was published as recently as 1980 [ 141 it has already become a classical method. This is essentially because only catalytic amounts of a promoter are required to provide very high glycosyl donor properties whereas the anomeric oxygen-exchange based methods generally require at least equimolar amounts of a promoter system, which is obviously often associated with disadvantages of various kinds. The basic principle of the trichloroacetimidate method [ 11 consists in a simple base-catalyzed addition of trichloroacetonitrile to the anomeric hydroxy group to generate the 0-glycosyl trichloroacetimidate (see Scheme 1). This reaction is generally high-yielding and, because of its reversibility, high anomeric control can often be achieved; in addition, competing reactions with non-anomeric hydroxy groups are not observed. In the second step, under mild acid/Lewis acid catalysis glycosylation of various nucleophiles (for instance, of alcohols, etc.) can be readily achieved; in general high yields and high anomeric control are obtained. The anomeric control is derived from the glycosyl donors’ anomeric configuration (inversion or retention), from anchimeric assistance, from the influence of the solvent, or from thermodynamic or kinetic effects [ 1- 131. Since 1980 (up to the end of 1998), application of the trichloroacetimidate method has been reported in more than 1100 papers and over 1200 different 0glycosyl trichloroacetimidates have been described in the literature. Obviously, the wealth of this chemistry cannot be compiled in a relatively short book chapter. A

6

*OH

2 Trichloroacetimidates

.

Base Catalysis

CI&

Reaction with CI&-CN not observed

- C - NH, .

:: t

Mild Aci? Catalysis

4 +

+OyNH

CCI,

HOR

Scheme 1.

quite comprehensive review [ 71 on the trichloroacetimidate method appeared in 1994, and this chapter is, therefore, essentially devoted to the literature since 1994. Of 686 references (1994-1998) on the use of the trichloroacetimidate method, because of space limitation, only ca 220 references are considered in this chapter. The typical examples selected illustrate the application of this method in different areas.

2.2 Methods The reported methods [ 141 for the formation and reaction of trichloroacetimidates are well established. Use of sodium hydride or cesium carbonate as a base for the reaction of 1-0-unprotected sugars with trichloroacetonitrile yields generally the thermodynamically favored 0-a-glycosyl trichloroacetimidates, whereas use of potassium carbonate yields 0-p-glycosyl trichloroacetimidates because of kinetic reaction control. Use of 1,8-diazabicyclo[5.4.01undec-7-ene (DBU) often yields a/p mixtures, mostly favoring the a-trichloroacetimidates. Recently, application of new base systems has been reported using 7-methyl-l,5,7-triazabicyclo[4.4.0]dec-5-ene (MTBD) [ 151 or phase transfer catalysis (50% potassium hydroxide, dichloromethane, tetra-n-butylammonium hydrogen sulphate) which yield exclusively atrichloroacetimidates in very good yields (96%) [ 161. Besides the well established catalysts for glycosylation reactions with trichloroacetimidates, some new catalysts have been investigated. The use of silver trifluoromethanesulfonate in the reaction with phthalimido-protected glucosamine trichloroacetimidate gave higher yields and smaller amounts of a-glucoside than the use of boron trifluoride etherate; this procedure also avoids acyl migration in the acceptor molecule; but stoichiometric amounts of the promoter are required [ 17,

2.3 Q-Glycosides

7

IS]. Use of lithium perchlorate in dichloromethane requires a large excess of promoter [ 19, 20 1, but with lithium trifluoromethanesulfonate only 0.05 equivalents are required [21]. Both catalysts give high yields, but the a/P selectivity is low and has to be improved before general application is possible. The synthesis of other 0-glycosyl imidates, i.e. O-glucopyranosyl dichlorocyanoacetimidates, 2,3,5,6-tetrafluoropyridin-4-y1 glucopyranosides, etc., is under investigation [22, 231 thus variations in reactivity and selectivity are expected.

2.3 O-Glycosides 2.3.1 Synthesis of Oligosaccharides

P-Glucosides, P-Galactosides, a-Rhamnosides, etc. 1,2-trans-glycosides (P-glucosides, 0-galactosides, P-xylosides, a-mannosides, arhamnosides, etc.) are usually prepared with glycosyl donors containing participating neighboring groups (acetyl, benzoyl, pivaloyl, etc. ) at 2-OH. Some examples for Galp(1-3)Gal [24], GalP(1-3)GalN [25], and ~-Rhaa(l-2)Glc[26] are shown in Scheme 2. Instead of the methyl glycoside of azidogalactose, the corresponding glycosyl fluoride has also been used as acceptor [27]. In general glycosylation yields are high, irres,pective of the anomeric configuration of the trichloroacetimidate, and only one anomer is obtained. Occasionally, when glycosyl acceptors with more than one free OH group are used, high regioselectivity is obtained because of the different reactivity of the OH groups. It was found that trichloroacetimidate donors are especially appropriate for deactivated (fluorine containing) donors in contrast to other glycosyl donors (thioglycosides, etc.) [28]. A highly convergent synthetic method has been developed for the synthesis of cellooctaose and other cello-oligosaccharides via di- and tetrasaccharides which can be converted either into glycosyl donors by deallylation and transformation into the trichloroacetimidates, or into acceptors by deacetylation (Scheme 3) [29].The use of p-methoxybenzyl groups (PMB) instead of acetyl groups was less successful. The combination of acetyl, pivaloyl, and benzyl groups leads to the proper balance between stability and reactivity of trichloroacetimidates as donors and the acceptor substrates. High yields were achieved by using a specially developed high-vacuum system for glycosylations under anhydrous conditions. Trichloroacetimidates have also been used for the glycosylation of more complex carbohydrate acceptors (Scheme 4). Galp( 1-3)GlcN and other disaccharides were synthesized and converted into fluorescein-labeled acceptors which are used for sialyltransferase assays [301. Glucosyl diosgenine has been used as carbohydrate acceptor for further glycosylations with various trichloroacetimidates (Scheme 4) [31, 321. All glycosides were obtained in high yields and high stereoselectivity because of participating neighboring groups at 2-OH in the synthesis of dioscin, polyphyllin D, balanitin 7, and gracillin.

8

2 Trichloroacetimidates

Baw Bzo

OBI

+

BZO

Hr&

OCH3

75%

OBZ

BzO

OBz

HO

OBZ

GalP(1-3)Gal OBZ

Ph

OAc

OBz

BZO~ O ! & & O C H ,

Aug6 et al. 1241

O YCCI, NH

AcO

.

CH TMSOTf CI ,O°C

Ph

TMSOTf CICH,CH,CI,,

-1 8%

Ga@(1-3)GalN

AcO OAc

Bock et al. [251

CCI,

N,

CCI, Ad)

OANH

BFS. OEtz A

c O AcO

H

+

0 OAll

CH,CI,, -20°Cb

AcO OAll

OAc

L-Rhaa(1-2)Glc

95% Liu et al. "261

A c O AcO R q

+

6

HO BnO

0

PhtN

I

O YCCI, NH

R = OAC, H, F

R

I cot

TMSOTf CH2C12,00C CHzCIz,OoC

*

93%

93%

A

C

~ W AcO

~

Karnerling et al. [281

,OBn

O

-

~ PhtN

I

BZ& BnO BnO

0C8H17

Scheme 2.

Aminosugar Trichloroacetimidates The use of azidosugars and N-phthalimido-protected aminosugars is well established in carbohydrate chemistry. But many new protective groups have been proposed to enable deprotection under milder conditions or to apply orthogonal protection strategies (Scheme 5). For the synthesis of P-glycosides of aminosugars, protective groups with amide structure are mainly used; these are capable of neighboring group participation. But formation of stable oxazolines reduces the reactivity of the glycosyl donors, and phthaloyl groups are therefore preferred. They can be cleaved only under strongly basic conditions, for instance, with hydrazine. As an alternative, the use of the tetrachlorophthaloyl (TCP) group has been proposed [33];this can be more readily removed with ethylenediamineor with sodium borohydride followed by acid treatment (pH 5). The glycosylation reaction gives excellent yields and in acetonitrile as solvent exclusively the P anomers are obtained. Further applications of TCP-protected trichloroacetimidates are shown below (glycosyl amino acids, C-glycosides).

2.3 0-Glycosides

9

OAll

1. Se02IAcOH

bpiv

OPiv

HBnOo

~

O A OPiv

I

I

BnO OPiv

/

disaccharide-

OPiv

\

disaccharide-

BnO

OPiv

OPiv

OPiv

tetrasaccharideimidate

OPiv

tetrasaccharideacceptor CH&,

O°C

Nishimura, Nakatsubo 1291

OPlV ACO BnO

OPlV

BnO OPlV

OPlV

OPlV

OPlV

OPlV

Scheme 3.

Similar properties have been observed for 4,5-dichlorophthaloyl (DCPhth) protected trichloroacetimidates [ 341. Deprotection can be performed with ethylenediamine. The trichloroethoxycarbonyl (Teoc, Troc-both abbreviations are in use) group has already been applied successfully to aminosugars [35].Because it can be cleaved under reductive conditions (with zinc) it can be used for base-labile compounds and more importantly for demanding orthogonal protection strategies. For glycosylation reactions with N-Teoc-protected trichloroacetimidate-activated aminosugars very high yields and stereoselectivities are reported [36]. Application was also shown for the synthesis of some complex Lipid A analogs [37-391. Glycosylation of acceptors with Teoc-protected trichloroacetimidates, both containing long chain fatty acid residues, was performed in high yields and stereoselectivities. Further applications (glycolipids, glycopeptides, bacterial polysaccharides, solid-

10

2 Trichloroucetimidates

HO HO&

"Po AcHN

BF,.OEt,

+

CCI,

CH~CI, 76%

-

1. deprotection

HO

OH

2. fluorescein isothiocyanate

O(CH,),NHCbz NHAc

Slim et al. [301

HO

CCI,

AAcOc O ~ 6 ; * o I c c , , OAc

AcO

OH

OAC CCI,

L-Rhaa(l-2)GlcP(l-O)diosgenine (dioscin)

Xyl~(l-3)Glc~(1-2)Glc~(l-O)diosgenine (balanitin 7)

L-Araa(1-2)GlcP(1-0)diosgenine (polyphyllin D)

Glc~(1-3)Glc~(l-O)diosgenine

I

Rhaa(1-2)

(gracillin)

Hui et al. [31,321

Scheme 4.

phase synthesis) of Teoc-protected trichloroacetimidates are shown below. Similar results were obtained with p-nitrobenzyloxycarbonyl (PNZ) protected trichloroacetimidates [40]. Deprotection can be performed under reductive conditions either with hydrogen and palladium-on-carbon or with sodium dithionite under neutral or slightly alkaline conditions. If protective group manipulations are incompatible with basic or reductive conditions, use of the 2,5-dimethylpyrrole group is proposed; this can be treated with hydroxylamine hydrochloride to liberate the amino group [41, 421. Glycosylations with dimethylpyrrole containing trichloroacetimidates perform well, leading selectively to P-glycosides (1,2-trans-glycosides). Recently the dimethylmaleoyl (DMM) group has been proposed as an amino-protecting group which can be cleaved under mild aqueous basic and then acidic conditions (pH 5 ) [43]. Because of the anchimeric assistance and the electron-withdrawing character of the DMM group the corresponding trichloroacetimidates give stereoselectively P-glycosides in high yields. The dithiasuccinoyl (Dts) group [44] for amino group protection is discussed below (glycopeptides).

11

2.3 0-Glycosides BnO A AcO c o ~ O y C C l , +

TCPN

OH

n BnO O OBn q CH,CN, 91% r.t.

B

NH

81%

BnO 0

BnO

CI OBn

TCP = tetrachlorophthaloyl,

Castro-Palomino, Schmidt [33]

CI

0

CI

1. TMSOTf HO DCPhthN

73%

OMP

BnO DCPhthN

NH

0

OBn

AcNH

3. AcpO, pyridine 84%

wz:

DCPhth = 4,5-dichlorophthaloyl,

b

2.ethylenediamine

Ogawa et al. [34]

0 BnO

--,

0

Teoc = trichloroeihoxycarbonyl,

-k!

R = (S)-3-(tetradecanoyloxy)tetradecanoyl = (S)-3-(dodecanoyloxy)tetradecanoyl R" = (S)-3-(benzyloxy)tetradecanoyl

-OCH,CCI,

(Troc)

% % ;A

+

PNZ'

BnO

1. TMSOTf

NH

1. BF,.OEt, CHZCI,, -3OC

&

HO BnO

2. NaOMe, 79% quant.

OBn

OBn

'yNH

.

3. Na,S,O, quant. 4. Ac,O, pyr., 93% C

H NO,

OBn

AcNH

0

CCl,

PNZ = p-nitrobenzyloxycarbonyl, - - O

Kusumoto et al. [37-391

Quian, Hindsgaul [40]

~

1. TMSOTf

CCI,

--

OCH,

~~~&o,cc13 DMMN

6

HO BnO

AcO

NH

DMM = dimethylmaleoyl,

OBn

+

1. TMSOTf C 78% H,CI,

2.NaOH 3.HCI 4. Ac,O 97%

0 $CH3 CH,

Scheme 5.

OBn

CHZCI,, -25C 78% 2. NEt,,MeOH, H,Ob 3. NH20H, 8OC 4. Ac,O, pyr.

r. t.

.

AcO

BnO OCH.

Boons et al. [41]

AEoa:o+OBn AcNH

OBn

Schmidt et al. [43]

12

2 Trichloroacetimidates CONH,

CONH,

ACO

31% OAc 0

Welzel et al. 1461

TMSOTf CH,Cldhexane, 60%

r.?

Bnd

Scheme 6.

Finally, the synthesis of moenomycin-type analogs is described (Scheme 6). Use of the trichloroacetyl protecting group for aminosugar trichloroacetimidates as previously proposed [45] gave the desired disaccharide (31% yield) [46]. Because of silyl group migration under the glycosylation conditions, the tri-TBS-protected acceptor, the diglycosylated acceptor, and the oxazolidine of the donor were also formed as by-products. The trichloroacetyl group can be converted into the acetyl group either with zinc-acetic acid or with tributyltin hydride in the presence of AIBN. Better results were obtained with a 2-(trimethylsilyl)ethanesulfonyl-protected trichloroacetimidate [47].The desired P-glycoside was obtained stereospecificallyin 60% yield and further transformations are under investigation. Trichloroacetimidate derivatives for the preparation of aminosugar a-glycosides (1,2-cis-glycosides)are rare. The use of azidosugars is well established (examples are given below), but unfortunately they are not readily available. Azidoglucose, for example, is prepared either from glucose via glucal by azidonitration [48]or from glucosamine via diazo transfer with trifluoromethanesulfonyl azide [49]. The N-l(4,4-dimethyl-2,6-dioxocyclohexylidene)ethyl (Dde) group has been proposed as non-participating neighboring group for the synthesis of 1,2-cis-glycosides(Scheme 7) PI.

,OAc

Kellam et al. [501

ACO

DdeNH

~~

. .. .

0

Dde = N - l - ( 4 , 4 - d i m e t h y l - 2 , 6 ~ 1 x ~ c y c l ~ h e x ~ ~ ~ e ~ e ~ ~ ~ h y Scheme l a e n e ) e7.t 0

2.3 0-Glycosides

13

65%

CCl,

Furstner, Konetzki 1541

Scheme 8.

P-Mannosides

The synthesis of p-mannosides is crucial because p-mannoside formation is disfavored because of stereoelectronic and steric reasons. Besides the already known direct mannosylation reactions some indirect approaches which have been reported are based on the preparation of I)-glucosides and then inversion of the configuration at C-2 to yield the corresponding P-mannosides (Scheme 8). According to a previously introduced method [ 5 11, a p-glucoside, synthesized using a trichloroacetimidate as glycosyl donor, was converted into its 2-0-triflate [ 521. Intramolecular nucleophilic substitution afforded a Mano( 1-4)GlcNp( 1-4)GlcN trisaccharide occurring in the core-region of glycoproteins. In other examples 2-0-triflate intermediates were subjected to intramolecular nucleophilic substitution under inversion of the configuration by reaction with tetrabutylammonium nitrite [53] (see below, Scheme 26) or with tetrabutylammonium acetate [54]. The reaction has been significantly improved by means of ultrasound, but very good yields have been also obtained previously via a sulfonyl imidazylate [55]. 2-Deoxyglycosides

Direct glycosylation with 2-deoxyglycosyl donors is not employed because the corresponding glycosyl donors are quite labile and they lack stereocontrol by anchimerically assisting groups. A recent application of a previously developed method [56] has been reported; it uses a 2-deoxy-2-thiophenylglucosyltrichloroacetimidate which can be readily prepared from glucal (Scheme 9) [57]. After glycosylation the thiophenyl group can be removed by reduction methods. Because glucal is not readily available as starting material and the stereoselectivity of the glycosylation of carbohydrate acceptors is critically dependent on substituent X

14

2 Trichloroacetimidates

ROH, TMSOTf CH& -78% &I3

& '*OR TBDMSO

- -re$- - -b

Roush et al. [571

X = OTos, Br, H

BnO BnO

TBDMSO -&OR

PhS

a -unu- I

OCH.

1. TMSOTf

91% L 2. BuqSnH, AlBN 74%

-

BnO I

OCH.

Castro-Palomino,Schmidt I581

Scheme 9.

at the 2-position, a new method has been proposed. Glycosylation reaction with 2-O-thiobenzoyl-protected O-glucosyl trichloroacetimidates gave exclusively the 0-glucosides which could be converted into the 2-deoxy-P-glucosides by reduction with tributyltin hydride [58]. Starting from mannosyl trichloroacetimidates the 2-deoxy-a-glucosides were obtained in the same manner [ 5 8 ] . Miscellaneous Compounds

The synthesis of some rare glycofuranosyl donors has also been reported (Scheme 10). The trichloroacetimidates of D-glucofuranose [59], D-mannofuranose [59], and D-galactofuranose [59, 601 have been prepared and used for the synthesis of furanosyl glycosides. With the heptosyl trichloroacetimidate some neoglycoconjugates containing Hep( 1-4)Kdo and Hep(1-5)Kdo-4-phosphate-(2-6)GlcNAcresidues were synthesized which are part of the inner core region of bacterial lipopolysaccharides [61]. A 5-thio-~-fucosyltrichloroacetimidate was used for the synthesis of 5-S-~-Fuca( 1-2)Gal and other disaccharides which have inhibitory activity against a-L-fucosidases [621. Some 5-thiomannose-containing oligosaccharide mimetics were prepared with a 5-thiomannosyl trichloroacetimidate; their binding constants to concanavalin A were determined [63]. An example of intramolecular glycosylation has been reported with a Glca(12)Glc trichloroacetimidate [64, 651. Reaction with boron trifluoride-diethyl ether gave the cyclo-a(1-2)glucobiose (cyclokojibiose) with concomitant expulsion of the O-benzyl protecting group. 2-O-debenzylation previous to cyclization is not necessary because the oxygen is sufficiently nucleophilic and the reaction is entropically favored. A similar example has been reported with a fucose moiety [66]. Complex Oligosaccharides

Complex oligosaccharides are major parts of many biomolecules (glycolipids, glycoproteins, bacterial lipopolysaccharides, mucopolysaccharides, etc. ) and they have a variety of functions. Much interest has been focussed on Lewis antigen structures which are found in glycosphingolipids and on glycoproteins. Many syntheses have

2.3 0-Glycosides

15

AcO AAco & * I

O YCCI, NH

CCI,

LOB"

Plusquellec et al. 1591

Roy et al. 1601

Kosma et al. (611 HO O ,H

&

A d

OAc NH A d

OAc

+

Hashimoto et al. 1621

OAll

OH

BF3.OE1, CH2C12,-2OOC 86%

*

--b H : + HO

OH

,.OAc

Scheme 10.

been described for the Lewis' antigen (LeX)structure which has been found to be a tumor-associated antigen, and for the sialyl LewisX(s LeX)epitope which is involved in cell-adhesion and inflammation processes mediated by its binding properties to selectins. Several synthetic strategies with various building blocks and/or different protective group patterns have been employed, but in recent investigations two alternatives have been favored. A suitably protected azidolactose building block which is available from lactose via lactal, has been chosen (Scheme 11) because it already contains the desired Galp( 1-4)Glc:N linkage [67]. Because of the non-participating neighboring group at 2-OH, fucosylation proceeded exclusively to the a-glycoside and gave the LeXtrisaccharide repeating unit in very good yield (97%). Transformation into the trichloroacetimidate and coupling with the lactose moiety gave spacer-linked Le" which could be extended with the trisaccharide trichloroacetimidate into the spacerlinked dimeric LeXantigen. Following the alternative strategy (Scheme 12), an azidoglucose building block was chosen as starting material and glycosylated first with fucosyl trichloroacetimidate and after benzylidene ring opening with galactosyl trichloroacetimidate [68]. Both glycosylations were highly stereoselective and gave, because of suitable neighboring groups and reaction conditions, exclusively the anomers desired. The high yields were based on the use of the 'inverse procedure' (i.p.), i.e. addition of the trichloroacetimidate donor to the acceptor-catalyst mixture. The LeXtrisaccharide repeating unit was then transformed into the trichloroacetimidate donor and into the glycosyl acceptor and then coupled with trimethylsilyl triflate as catalyst in

16

2 Trichloroucetimidutes

NH

OBI

OBn

97%

Aco

OBZ

Windrnuller, Schmidt [671

Scheme 11.

~

h

T t HO

-

~

N3

~

~

~

TMSOTf, i.p. Et,O, r.t. 85%

TMSOTf, i.p. Et20, r.t. 71%

trisaccharide

hexasaccharide irnidate acceptor N3

imidate 081

dodecasaccharide imidate

n = 2 : dim.LeX n = 3 : trim.LeX n = 4 : tetram. LeX

Scheme 12.

Schmidt et al. 68

17

2.3 0-Glycosides

acetonitrile as solvent to yield stereospecifically the P-connected hexasaccharide by means of the 'nitrile effect'; this effect is particularly efficient with 2-azido sugars. The nonasaccharide and dodecasaccharide were prepared in the same way by coupling the hexasaccharide trichloroacetimidate with the tri- or hexasaccharide acceptor. Transformation into the trichloroacetimidates and reaction with the lactose acceptor yielded the octa-, undeca-, and tetradecasaccharides. After coupling with azidosphingosine and further transformations the dimeric, trimeric, and tetrameric Le" antigens were obtained. The synthesis strategy is highly convergent and allows variations in the saccharide structure and the lipid moiety. The ready a.vailability of glucosamine was reason to transfer this synthesis strategy to glucosamine building blocks which can be obtained directly from glucosamine. The N-tetrachlorophthaloyl (TCP) [69], N-trichloroethoxycarbonyl (Teoc) [70], and N-phthalimido (Pht) [711 protected glucosamine acceptors were glycosylated with the 0-galactosyl and 0-fucosyl trichloroacetimidates; both reactions proceeded with high stereoselectivity. The resulting Le" trisaccharide repeating unit was transformed to the corresponding trichloroacetimidates which were used for the glycosylation of the lactose acceptor to give, owing to neighboring group participation, the p( 1-3)-linked Le" pentasaccharide. The Lewisa (Lea) antigen is a blood group antigen; because of its structural similarity to the Re" antigen it has been synthesized in a similar manner (Scheme 13) [72]. The benzylidene-protected azidoglucose acceptor was first glycosylated with galactosyl trichloroacetimidate and, after benzylidene ring opening, with fucosyl trichloroacetimidate; both glycosylations could again be performed with high diastereoselectivity. Transformation into the trichloroacetimidate and reaction with the CCI,

CCI,

OTBS

TMSOTf Et20, r.t. 88%

N3

Ph T HO a O

T

D

BnO

TMSOTf, i.p. Et20, r.t.

AcO

*

OAc

78%

S

I

0KCC'3 NH

BF, OEt2 CH,C12, hexane

H

Schmidt et al. [;TO]

o

#

~

o

~

O

OBn B n

'

61%

FUCCU(~-4) HHO

O

aOTDS NTCP

Schmidt et al. [69]

Hio&

OTBS

I

Gal~(I-3)GlcNAc~(1-3)Gal~(1-4)Glc

N,

Toepfer, Schmidt [72]

Lea Toepfer, Schmidt [72]

Scheme 13.

18

2 Trichloroacetimidates

lactose acceptor in a non-polar solvent as catalyst with diethyl ether-boron trifluoride gave the Lea pentasaccharide. The p-linkage was stereospecifically formed under inversion of the configuration; typical conditions for this reaction type were used. The N-trichloroethoxycarbonyl (Teoc) protected glucosamine acceptor was proceeded in the same way [70]. The Lea trisaccharide was also prepared from the 3,4-di-O-unprotectedazidoglucose [72] and N-tetrachlorophthaloyl (TCP)-protected glucosamine [69] acceptors by attachment of the two sugar residues directly one after the other. The blood group antigen H and the LewisY (Ley) antigen, a stage-specific embryonal antigen (SSEA), have been synthesized using the same azidolactose building block (Scheme 14) [67]. By regioselective monofucosylation and by difucosylation of the 3,2'-di-O-unprotected azidolactose acceptor, the H-antigen trisaccharide and the Ley tetrasaccharide were synthesized in high yields. Transformation into the trichloroacetimidate and coupling with the lactose moiety furnished the spacer-

CH,Cl$hexane BF,.OEt, -10%

70%

!

o(cHd&~H, HO

OBZ

Ga@(1-4)GlcNAcp(1-3)Ga@(1-4)GlcP(1-0)Sp FuW( 1-2)

H-antigen

I

BFS.OEt2 CH,Cldhexane

H O + O - Y

OBZ

-1O°C

71%

F u c ~1(-2) F u c ~1(-3) Windmuller, Schmidt I671

Scheme 14.

2.3 0-Glycosides

19

-

trisaccharide imidate

BnO

--+ trisaccharide N3

TMSOTf. i.p. Et,O

TMSOTf, i.p. EtPO

85%

89%

acceptor BnO

TMSOTf CHSCN, -4OC 80%

I

COOCH,

FUctr(1-3)

dim. sLeX

1:

octa-

hexasaccharide imidate

Fuca(1-2) Galp(l-4)GlcNAcp(l-3) alp(l-4)Glcp(l-I)Cer

[

F U C ~-3) ('~

I

dim.LeY

BnO

Hummel, Schmidt [73]

Scheme 15.

linked H-antigen and Ley antigen, respectively. Another strategy has been applied to the synthesis of the dimeric Ley antigen (Scheme 15) [73]. According to the previously described synthesis of dimeric LeX (Scheme 12, Ref. [68].) the Le' trisaccharide has been prepared from the azidoglucose acceptor, the fucosyl and the galactosyl trichloroacetimidate, and the derived trisaccharide was used as trichloroacetimidate donor and as acceptor for the synthesis of the corresponding hexasaccharide. Coupling with the lactose moiety gave the octasaccharide which was subjected to a protective group change at the terminal galactose moiety. Regioselective protection of the 3-OH, 4-OH, and 6-OH groups and fucosylation at 2-OH gave the desired nonasaccharide. After coupling with azidosphingosine and further transformations the dimeric Ley antigen was obtained. For the synthesis of the sialyl Lewis' (sLe') antigen, the terminal galactose moiety of the octasaccharide was 6-0-benzyl-protected. Regioselective glycosylation of the 2,3,4-tri-O-unprotected acceptor with neuraminic acid phosphite as donor gave the desired 3-0-sialylated oligosaccharide which was coupled with azidosphingosine and transformed into the dimeric sLe' antigen by use of established methods. Further syntheses of the sLe' determinant follow the same strategy, but applying other glycosylation methods (Scheme 16). Fucosylation of a similar O-benzylideneprotected glucosamine acceptor performed well with a thioglycoside donor [74-761, but the galactosylation was critical because of the steric hindrance of the 4-OH group. The methylthio galactoside donor gave a 16% yield [74]; the trichloroaceti-

20

2 Trichloroacetimidates

HNAC

Ga@(1-4)GlcNAcP-X NeuAca(2-3) Fuca(1-3)

= NH2

Kretzschmar, et al. [74-761

Scheme 16.

midate was, therefore, again favored. Regioselective sialylation with the thioglycoside donor gave the sLeXdeterminant bearing an amino group or a spacer at the reducing end for attachment to peptides or for the generation of other conjugates. The synthesis was suitable for large-scale preparations 1761 and enabled the use of sialic acid derivatives with modifications at C-5 to synthesize sLeXanalogs and to investigate their binding properties to E- and P-selectins [75]. A similar strategy for the synthesis of the sLeXdeterminant was used by applying the glycal method (Scheme 17) [77]. The 3,4-di-O-unprotected glucal was regioselectively fucosylated ( a : = 7.5: 1) with fucosyl fluoride and then galactosylated with 0-galactosyl trichloroacetimidate to yield the Le" trisaccharide glycal. The 2,3,4-tri-O-unprotected galactose acceptor was then regioselectively sialylated with sialyl chloride as donor. Then application of the glycal method gave either the sLeX trisaccharide or-by extension with the tributyltin derivative of the lactal acceptor -the sLeXpentasaccharide glucal derivative which can be activated to provide a donor for the synthesis of conjugates. The synthesis is efficient and convergent, and it is worth mentioning that for the crucial galactosylation of the sterically hindered 4-OH group the trichloroacetimidate method has again been employed. The synthesis of LeXor sLeXanalogs or simplified structures, so-called mimetics, is of great interest for binding studies with selectins, thus high affinity ligands concomitantly with convenient chemical syntheses for novel anti-inflammatory drugs are evaluated. The synthesis of some only slightly modified Le" analogs containing Glc instead of GlcNAc [78], GalNAc instead of Gal [79],or 5-thiofucose instead of Fuc [801 has been similarly performed as described above using trichloroacetimidate donors. It has been found that sialylations are always crucial and expensive processes, therefore many syntheses for sLeXmimetics have been developed in which

2.3 0-Glycosides

21

OTDPS OBn

OTDPS

A

c

O

g

O

&

BZO

AgC104 SnCI, DTBP 59%

OBn

BF,~OEt, CH& -78OC 73%

Bzo

Danishefsky el al. 071

Gal~(1-4)GlcNAc~(l-3)Gal~(l-4)Glucal

Galp(1-4)GlcNAc

NeuAca(2-3) Fuca(1-3)

NeuAca(2-3) Fuca(1-3)

Scheme 17.

AcO

BnO

AcO

a

O

B

OAc

Ph q 'HO a O A l l 0 0 ~

TMSOTf CH& 87%

I

AgOTf Bu,SnCI, 90%

b

F u c ~1-4) (

n

I

AcO W o OAc & O A I I

3'-phospho-Lea-analog AcO

OBn OAc

Kiessling et al. [Sll

3'-C-carboxymethyl OBZ OBn

DMTST TMSOTf benzene, 6OC CH,CN, -1OOC 86% 67%

OBn

b

'~6

BnOoBn @OBn

I

Kiso el al.[831

NHAc

HO

Scheme 18.

Thoma el al. I841

22

2 Trichloroacetimidates

HO

OH

Wong et al. IS81

Scheme 19.

neuraminic acid is replaced by other acids (Scheme 18). A 3’-phospho-Lea analog [81] and a 3’-sulfo-Lex analog [82] were prepared from 3-0-unprotected 4,6-0benzylidene-glucose derivatives. As well established in the synthesis of LeXand sLeX (see above), for all glycosylations with galactose the trichloroacetimidates were successfully used as donors. For the synthesis of the 3’-C-carboxymethyl LeXanalog [831 the synthesis and use of the required galactose-derived trichloroacetimidate donor has been reported. A more rigidly fixed carboxylic group is contained in the sLeXmimetic with a six-membered ring fused to galactose bearing an equatorial carboxylic acid group [84]. The required branched galactose moiety has been prepared and used as trichloroacetimidate-activated donor. Further sLeXmimetics with other simplifications have been proposed (Scheme 19). The GlcNAc moiety has been replaced by cyclohexanediol which was used as acceptor and glycosylated first with the galactosyl trichloroacetimidate and then with the corresponding fucosyl donor [85]. Then a malonic acid derived group was introduced at the 2-OH of galactose. In an even more simplified structure the galactose moiety has been replaced by an aliphatic chain [86]. The acceptor which was prepared from glucal is first glycosylated with the fucosyl trichloroacetimidate and then converted into the malonic acid derivative. In a related mimetic, the malonic acid group is attached more rigidly at a benzene ring [87]. Glycosylation with the

2.3 0-Glycosides

A*

OAC

COOCH,

OBn

23

OBn

(GM, intermediate) pentasaccharide imidate

Gal~(I-3)GalNAc~(1-4)Gal~(l-4)Glc~(l-1)Cer NeuAca(2-3)

Schmidt et al. [891

Scheme 20.

fucosyl trichloroacetimidate resulted in a low yield only (35%), because of steric hindrance. A c:ompletelydifferent structure has been chosen with a terpenoid system [88]. Glycosylation of isosteriol with the mannosyl trichloroacetimidate gave a mimetic which selectively inhibits P-selectin, but not E- and L-selectins. Glycolipids other than the Lewis antigens have also been synthesized with trichloroacetimidate donors. For the synthesis of the ganglioside GM1 several strategies have been suggested [89, 901. Applying a convergent strategy (Scheme 20) [89] a 3,4-di-O-unprotected lactose acceptor was first sialylated with neuraminic acid phosphite as donor to give a GM3 intermediate, and then glycosylated with a Galp( 1-3)GalN trichloroacetimidate as donor; the latter can be easily prepared from azidoglucose and galactosyl trichloroacetimidate. Coupling with azidosphingosine by application of trichloroacetimidate activation of the pentasaccharide residue and then further transformations yielded GMI . Another strategy via a C A I intermediate performing sialylation at a later stage has also been reported [90]. The GM3 intermediate is also suitable for the synthesis of GM2 [91]Glycosylation of the GM3 intermediate with the Teoc-protected galactosamine trichloroacetimidate, which can be more readily synthesized from galactosamine than azidogalactose from galactal, and application of the established transformations, yielded GM2 which is required in a cancer immunization study with GM2-containing vaccines. A glycolipid of the neolucto series named 3-0-sulfo glucuronyl paragloboside is also of interest because of its immunological properties (Scheme 21) [92]. The syn-

24

2 Trichloroucetimidutes COOMe

k0.d-

H:p&

BzOO

OBn OSE

HO

8

BnO

gn&

OBn OSE

OBn

CCl,

OSE

94% TMSOTf

CH,CI,, r.1.

pentasaccharide imidate

N3

BF,.OEt, CHZCI,, O°C 72%

00~S-0-3GlcAP(1-3)GalP(I-4)GIcNAc~(1-3)Gal~(1-4)Glc~(l-l)Cer

3-0-sulfo glucuronyl paragloboside

Hasegawa et al. [921

Scheme 21.

thesis is based on glycosylation of a lactose acceptor with phthalimidoglucosyl trichloroacetimidate. Glycosylation with a disaccharide trichloroacetimidate, which was prepared from a glucuronic acid trichloroacetimidate and a galactose acceptor, and the usual transformations including introduction of the sulfo group yielded 3-0-sulfo glucuronyl paragloboside. A more straightforward synthesis of a related structure has been reported using a Galb( 1-4)GlcN disaccharide (lactosamine) which is contained in the neolucto structure as starting material (Scheme 22) [93]. Glycosylation of the lactose acceptor with the azidolactosyl trichloroacetimidate donor followed by sialylation and application of the standard azidosphingosine glycosylation procedure gave the neolacto ganglioside LM1. All the above mentioned examples contain P-glycosidic linkages which can be obtained by means of participating neighboring groups or-for azidosugars-by means of the nitrile effect. The synthesis of gangliosides of the globo series which contain also an a-galactose linkage is more difficult. From the azidogalactosyl trichloroacetimidate and a galactose acceptor the 2-0-benzyl-protected disaccharide was prepared (Scheme 23) [94, 951; this was converted into the trichloroacetimidate and used as donor for the glycosylation of the lactose acceptor. Using typical conditions favoring s,~-type reactions (non-participating neighboring group, strong Lewis acid, diethyl ether as solvent), the a-connected tetrasaccharide was obtained with high selectivity. Ensuing galactosylation, sialylation, and application of the

25

2.3 0-Glycosides TMSOTf CH2C12,-4OOC (SOYO) tetrasaccharide

b

I OBn

58%

OBn

OEt

pentasaccharideimidate

5

lF3.0Et2 CH2C12,O°C 43%

O h

NeuAca(2-3)Gal~(l-4)GlcNAc~(l-3)Gal~(1-4)Glc~(I-1)Cer Tietze et al. [931

LMl

Scheme 22.

Ph

Ph

Ph

TMSOTf

&:s

Ho BnO

OAll

HO

N,

OBn

BnO

OBn

OBn

1

CCI,

0

Y

NH

Y CCI,

Ph

Ph

TMSOTf, i.p EbO, -1OOC

Bds+E;ohloBn TMSOTf I

-'I

Ell0

AcOO

vNH I CCI,

TMSOTf CH,CN, -4OOC-I (a$= 3.5:l)

%En

AcO

62%

AcN

COOCH,

dAc

hexasaccharide imidate

N,

TMSOTf OBZ

Neu5Aca(2-3)Gal~(l-3)GalNac~(l-3)Gala(l-4)Gal~(1-4)Glc~(1-1)Cer Schmidt et al. 194,951

Scheme 23.

SGG

26

2 Trichloroacetirnidates

azidosphingosine glycosylation procedure yielded the sialylgalactosylgloboside (SGG), which bears the stage-specific embryonic antigen 4 (SSEA-4) structure. Another example of the globo series is the globo H (human breast cancer) antigen. It has been prepared by means of a similar strategy [96]. Although in glycopeptide synthesis, most strategies are based on the synthesis of the complete oligosaccharide moiety and then its attachment to the amino acid or peptide, glycosylations of glycosylated amino acid acceptors have also been reported (Scheme 24). Thus, glycosylation of the GalNa( I-0)Thr acceptor with

Ph

'%$

+

H

o

1. TMSOTf, CH2CI,, O°C 02%

b

2. Zn, &OH, AC20, THF 3. AqO. pyr. (80%)

TeocN

b

O YCCI, N H

core3

Fmoc[O X HC0,Bu' 3

FmocNH

Bock et al. [971 Ph

+

Ph

"q

Bock et al. 1441

1. TMSOTf, CH2C12,40°C 67% 2. Zn, 7OoC THF, AC20,HOAc)

AcNH

H

Ph

Ph

Ph

CCI,

Qiu, Koganti I981

w T s 0

'no

0

II

-15% 69%

AcN

OBn

Scheme 24.

OBn

OBn

OBn

2.3 O-Glycosides

27

N-trichloroethoxycarbonyl (Teoc)- [97] or the N-dithiasuccinyl (Dts)- [44] protected glucosamine trichloroacetimidates, respectively, afforded the core 3 building blocks. The P-glycosides were stereospecifically obtained by means of neighboring group participation. Use of the N-acetyl-protected galactosamine trichloroacetimidate gave stereoselectively the a-connected core 5 structure [98]. Thus, core 1, 2, 3 , 4, 5 , sialyl 5 , 6, and F1-a building blocks were synthesized [44, 97-99]. A sialylated glycosy1 serine building block was prepared with a disaccharide trichloroacetimidate donor and used in the solid-phase synthesis of the B-chain of human u2HS glycoprotein [loo, 1011. In the synthesis of branched complex type N-glycans the oligosaccharide part is usually prepared first; it is then attached to the amino acid, usually asparagine. For investigations of the function of glycoproteins the synthesis of libraries with variations of the antennary structures is required. A highly convergent chemoenzymatic approach has recently been reported (Scheme 25) [ 102-1041. The core trisaccharide Mar$( 1-4)GlcNP( 1-4)GlcN was synthesized via the P-glucoside and then inversion of the configuration by intramolecular nucleophilic substitution as described above (see Scheme 8:1, thus avoiding the crucial direct P-mannosylation. The trisaccharide acceptor is suitable for the attachment of various antennary structures with linear or branched trichloroacetimidate donors which were again synthesized by application of the trichloroacetimidate method. Several examples were prepared and used for investigations on further glycosylations with galactosidases and sialidases. Another versatile strategy enables the synthesis of diantennary and bisected diantennary oligosaccharides (Scheme 26) [ 531. The azidoglucose acceptor was glycosylated with the glucosyl and mannosyl trichloroacetimidates. The P-glucoside of the resulting trisaccharide was converted into the P-mannoside moiety by inversion of the configuration at C-2 via triflate formation and substitution with tetrabutylammonium nitrite. Further glycosylations with the mannosyl and the azidoglucosyl trichloroacetimidates gave the bisected diantennary pentasaccharide which was used as acceptor for extension with lactosamine trichloroacetimidate. After transformation of the nonasaccharide into the trichloroacetimidate the terminal glucosamine residue was attached at the reducing end. Thus, the synthesis is highly variable in the connection to the protein and in the antennary structures. Further syntheses applying direct 0-mannosylation have also been reported [ 105- 1071, and it was found, that the trichloroacetimidate method was superior to other glycosylation methods, especially for the synthesis of u-mannosides. The acidic mucopolysaccharides, which consist of disaccharide repeating units containing D-glucuronic or L-iduronic acid and glucosamine or galactosamine, respectively, are also members of the class of glycoproteins or proteoglycans. Because L-iduronic acid is not readily available it must be prepared. A new synthesis from Dglucose was recently described by formation of the exo-glucal, diastereoselective hydroboration, and oxidation of C-6 [ 108). Galactosamine is also a very expensive starting material and it has therefore been prepared from glucosamine by inversion of the configuration of the 4-OH group via the triflate, and used for the synthesis of dermatan sulfate fragments [ 1091. In another approach [ 1101 an L-idose acceptor was glycosylated stereoselectively (P: u = 2.5 : 1) with an azidogalactosyl trichloroacetimidate. The P-selectivity was achieved, without neighboring group participa-

28

2 Trichloroacetimidates

85% b

BnO

BnO

F

0

\

NPht

BF3 OEt2 CH2C12 84%

Ad)

Ad)

BF, OEt2 CH2CI2 -45% 76% Aco AcO

OAC

Aco Aco

GkNAc&l-6)

I MW(1-6)

I

GkNAcS(1-2)

1

Man~(1-4)GICNACP(1-4)GlcNA@N3

Unverzagt [102-1M]

I

GlcNAcP(l-P)Mana(l-3)

Scheme 25.

tion, in a non-polar solvent at low temperature by an SN2-type reaction. The resulting disaccharide was oxidized to the iduronic acid derivative and converted into the GalNp( 1 -4)IdoA trichloroacetimidate and into a corresponding acceptor which were then used for the next glycosylation step to provide a dermatan tetrasaccharide fragment; thus, because of neighboring group participation an a-glycosidic bond was formed stereospecifically. The tetra-, hexa-, and octasaccharide fragments of hyaluronic acid (Scheme 27) [ 1111 have been prepared by repeated glycosylation of a GlcNP( 1 -4)GlcA acceptor with a disaccharide trichloroacetimidate which was prepared from a trichloroacetyl

29

2.3 0-Glycosides

ACO AcO OBn

I

TMSOTf CH2C12,hexane 93%

Et20,-4OOC

68%

Bu~N'NO,

I

GlcNAc~(l-4)Man~(l-4)GlcNAc~(1-4)GlcNAc

I

Ga@(1-4)GlcNA1:b(1-2)Mana(1-3) Weiler, Schmidt 1531

Scheme 26.

(TCA)-protected glucosamine trichloroacetimidate and a glucuronic acid acceptor. All reactions gave, because of neighboring group participation, exclusively pglycosides in high yields. Some chondroitin fragments have been prepared starting from an azidoglucosyl trichloroacetimidate and a glucose acceptor (Scheme 28) [ 1 12, 1131. Despite the use of a non-participating neighboring group, the p-glycoside was obtained exclusively because in a non-polar solvent at low temperature with boron trifluoride as the catalyst an S,z-type reaction was favored. With trimethylsilyl triflate lower stereoselectivity (p :CI = 2 : 1) was observed. The disaccharide was oxidized to the glucuronic acid derivative and converted into the

30

2 Trichloroacetimidates

+

PhT-+

Ho* BzO

OMP

TCANH

OBz

O Y N H dCl,

q

89%

~

~

~

&

O

M

~

HO

TCANH

COOMe

-4 87%

Ph

TMSOTf

zqsy-;*oMe TCANH

Cm+ie

TCANH

COOMe

n=

hyaluronic acid

Blatter, Jaquinet

1,2,3

I1 111

Scheme 27.

GalNAc(1-4)GlcA trichloroacetimidate and the corresponding acceptor which were used again for glycosylation to yield a tetrasaccharide fragment. The di- and tetrasaccharide trichloroacetimidates were used for the glycosylation of a Galf3(13)Galf3(1-4)Xyl acceptor which represents a typical linkage of mucopolysaccharides to proteins. The resulting penta- and heptasaccharide trichloroacetimidates were used for the glycosylation of serine to yield chondroitin fragments with one or two repeating units; these compounds are of interest as acceptors for glycosyltransferases in enzymatic studies. Great biological interest has been shown in heparan sulfate or analogs of heparin, containing glucosamine and D-glucuronic or L-iduronic acid. A convergent synthesis has been reported (Scheme 29) [114], starting from an azidoglucose acceptor which was glycosylated with two iduronic acid trichloroacetimidates. The resulting two disaccharides were converted into the trichloroacetimidates and used consecutively for glycosylation of the suitably protected iduronic acid acceptor. Both aglycosidic bonds were formed stereospecifically with inversion of the configuration. Thus, a heparan trisulfate fragment has been prepared and used for binding studies to fibroblast growth factor-2. Syntheses of C-bridged heparan sulfate analogs [ 1151 and of heparin-like glycoconjugates which were used for thrombin binding studies [ 116, 1171, have also been reported. The synthesis of bacterial polysaccharides is of great importance for the development of antibacterial vaccines. A terminal a( 1-2)-linked rhamnose trisaccharide

2.3 0-Glycosides

31

YPh

Ph

Ph I

P' h

BF3.OEtz toluene. -5OOC

r

71%

1

5.0-70%

GalNAcP(1-4)GlcAP( n=1,2

Scheme 28.

HNFmoc

1-3)GalP(1-4)XylP(1 -0)Ser chondroitin

Ogawa et al. 1112,1131

32

2 Trichloroacetimidates

~

'

"

not reported

3

I

Lea

qn%

Lev0

OAc

w

HO

KCci3 84% TBDMSOTf NH

OAc

HO

I

OAll

toluene, -20°C

TBDMSOTf CHZCI,, -2OOC

IdoAa(l-4)GlcNa(l-4)ldoAa(i-4)GlcNa(l-4)IdoAal-OMe

6 0 , '

Sinay et al. [I141

Ao3Q

Scheme 29.

of the group B Streptococcus antigen has been synthesized with a 2-0-acetyl protected rhamnosyl trichloroacetimidate by repeated glycosylation and deprotection of, the 2-OH group [118]. A hexasaccharide fragment of the a(l-4)-linked 3-0methylmannose polysaccharide of Mycobacterium smegmatis has been convergently prepared, via a disaccharide building block [ 1191. From a 4-0-acetyl protected 3-0methylmannosyl trichloroacetimidate and a suitable acceptor a 3-0-methylmannose disaccharide was prepared which was converted into the trichloroacetimidate and used for repeated glycosylation and then deprotection of the 4'-OH group. All glycosidic bonds were stereospecifically formed owing to the participation of neighboring groups. Most bacterial polysaccharides are much more complex. Ninety 90 serotypes of Streptococcus pneumoniae are already known. The synthesis of a pyruvated tetrasaccharide repeating unit of Streptococcus pneumoniae type 27 has been described (Scheme 30) [ 1201. The disaccharide trichloroacetimidate donor was prepared from acetobromoglucose and the pyruvated glucosamine acceptor applying the KoenigsKnorr method. The a-linked disaccharide acceptor was prepared from a galactosyl trichloroacetimidate and a rhamnose acceptor. The use of a strong Lewis acid and solvent (diethylether) favoring an S,I -type reaction mechanism, gave the a-glycoside with high selectivity (a: = 9 : 1). Glycosylation of the disaccharide acceptor with the disaccharide trichloroacetimidate gave the tetrasaccharide repeating unit carrying a spacer which is suitable for the preparation of glycoconjugates. The syntheses of another tetrasaccharide repeating unit of this type [121] and four tetrasaccharide repeating units of Streptococcus pneumoniae type 6B [ 1221 have also been reported. The chemoenzymatic synthesis of a decasaccharide consisting of two repeating units of Streptococcus capsular polysaccharides type I11 group B (GBSP 111) (Scheme 31) [123] has been described. Glycosylation by the Koenigs-Knorr method of a 3,4-

a

2.3 0-Glycosides

61Yo

33

TMSOTf

6OoC

scheme 30.

di-O-unprotected glucosamine acceptor with acetobromogalactose gave regiospecifically the lactosamine disaccharide which was converted into an acceptor and into a trichloroacetimidate donor. Glycosylation of the disaccharide acceptor with acetobromoglucose, again under Koenigs-Knorr conditions, gave the trisaccharide building block which was transformed into the trichloroacetimidate donor. The disaccharide trichloroacetimidate was used for the glycosylation of a galactose acceptor; the resulting linear trisaccharide was glycosylated with a lactosyl trichloroacetimidate to give the pentasaccharide building block. Glycosylation with the trisaccharide trichloroacetimidate finally gave the desired octasaccharide. All pglycosidic bonds were formed stereospecifically, but the synthesis seems not to be straightforward because the tetrasaccharide repeating unit was not used twice as a building block. Finally, the neuraminic acid residues were enzymatically attached with a(2-3)sialyltransferase to yield the desired decasaccharide fragment. The synthesis of an octasaccharide mimic of two repeating units of the Streptococcus capsular polysaccharide type I11 group B (GBSP 111) with S-1-carboxyethyl groups replacing the siailic acids has also been reported [124]. A convergent synthesis has been reported for a hexasaccharide fragment of the Streptococcus group A cell-wall polysaccharide (Scheme 32) [ 1251. The 3 - 0 unprotected rhamnose acceptor was first glycosylated with the N-phthaloylprotected glucosamine trichloroacetimidate and then with a suitably protected

34

2 Trichloroucetimidutes

ACO

HS(CN12 tol. CHsNO2,

ACO

OSE

72%

.A;&oHg

OSE

Pht

NPhth

OAc

A

disaccharide acceptor

disaccharide imidate

J

-OAc

Ad)

Oh43

TMSOTf

Glc~(1-6)GlcNAc~(1-3)Gal~(l-4)Glc~(1 -6)GlcNAcP(1-3)GalPl-OMe

I

I

GaiP(1-4)

GalP(1-4)

1

a(2-3)sialyNransferase CMP-NedAc

Glc~(l-6)GlcNAc~(l-3)Gal~(l-4)Glc~(l-6)GlcNAc~(l-3)Gal~l-OMe

I

NeuAca(2-3)GalB(l-4)

I

NeuAca(2S)Ga@(1-4)

GBSP 111 Jennings et al. [I231

Scheme 31.

rhamnosyl trichloroacetimidate. The resulting trisaccharide was converted into the trichloroacetimidate and into an acceptor; coupling yielded the hexasaccharide with high stereoselectivity. A participating neighboring group is not necessary for the formation of the a-rhamnoside bond. An efficient synthesis has also been reported for the hexadecasaccharide fragment

35

2.3 0-Glycosides OAll

B

z

O

H

TESOTf CH2CI2, -75OC 58%

r

I '

Bz

OAc

trisaccharide imidate

Rhaa(l-2)Rhaa(1-3)-Rhaa(l-2)Rha GlcNAcp(1-3)

OBz

BZO

CH2CI2, TESOTf,-75OC 84%

TESOTf, CH2C12, r.t. 43%

GlcNAcp(1-3)

trisaccharide acceptor Pinto et al. [125]

Scheme 32.

of the Shigelln dysenteria type-1 0-specific polysaccharide (Scheme 33) [ 126, 1271. For the synthesis of the tetrasaccharide repeating unit by subsequent glycosylation of the rhamnose acceptor with azidoglucosyl chloride, methylthio galactoside, and rhamnosyl trichloroacetimidate, several glycosylation methods were applied. For the ensuing gl ycosylations with the tetrasaccharide as donor, the trichloroacetimidate method was favored. Repetitive glycosylation of the hexanoic acid acceptor gave the octa-, dodeca-, and hexadecasaccharides in high yields. Coupling with the linker to human serumalbumin (HSA) enabled the preparation of materials for vaccine studies. Further oligosaccharide fragments of the Rhizobium leguminosarum [ 128, 1291, CCI, CCI,

OANH N, SPh BnO

I

AgOTf DBMP CH&12, O°C

64%

OMBn I

I

MeOTf DBMP Et20, r.1. 58%

Brio

ClAcd

BF .OH

bez b

cH3c12,200c 8940

Rhaa(1-2)Gala(l-3)GlcNAca(l-3)Rhaa 1-O(CH,),COOMe n = 2-4

BnO

OAC

HO(CH,),COOMe

Pozsgay [126,1271

Scheme 33.

36

2 Trichloroacetimidates

BnO

OH

OBn

dBn B

n

En0 O q

68%1Sn(OTI)Z Et20/CH2CI,. r.t.

+

CCI,

Sn(OTI), Et20/CH2C12,r.t.

+

En0 B

n

O

g

0 OBn

I D-Gala(1-1)-D-rnyo-inositol

(Galactinol)

Mayer, Schmidt [ 1 1

Scheme 34.

Mycobacterium gornoae [ 1301, Mycobacterium arium [ 1311, and Vibrio cholerae [ 1321 polysaccharides have also been prepared with the help of trichloroacetimidates as glycosyl donors.

2.3.2 Inositol Glycosides The synthesis of galactinol gala( 1-l)-~-myo-inositol]which is involved in the biosynthesis of the raffinose family, has been reported (Scheme 34) [133]. The readily available L-myo-inositol derivative was glycosylated with the 0-benzyl protected galactosyl trichloroacetimidate. Because of reaction conditions favoring an S,I -type reaction mechanism, the a-glycoside was formed exclusively. In the same way D-Gala(1-1)-D-myo-inositol was obtained from the corresponding D-myoinositol derivative. The glycosylphosphatidyl inositols (GPIs) are an interesting class of compounds which serve as anchors for proteins and glycoproteins in membranes. They consist of many variants in the carbohydrate moiety and in the lipid part. The first total synthesis of a GPI anchor from yeast has been reported; it contains a ceramide residue with a phytosphingosine constituent (Scheme 35) [ 134, 1351. The synthesis has been performed in a convergent manner by glycosylation of 6-0-unprotected Dmyo-inositol acceptor with the azidoglucosyl trichloroacetimidate yielding stereospecifically the a-glycoside as a result of s,~-type reaction conditions. A similar

37

2.3 0-Glycosides

BBilOn 0

0

OAll

O-(-)-Mnt

"

G

TMSOTf

NH

o&.~~,

85% Et,O. r.t.

Pl

0

BnO

TMSOTf

00

;;oq I

uoo~OyO(-)-Mnt 0

cc13

TMSOTf

O YCCI, NH

HO

"YNH CCI,

HN

0

08 0- 6Mana(1-P)Mana(1 -P)Mana(l-4)GlcNa(1-6)lnositol-l-o,

H,Nlco':/-

I

0

Mana(l-2)

8

OH

Schmidt et al. 1134,1351

Scheme 35.

approach with a benzyl-protected azidoglucosyl trichloroacetimidate in dichloromethane gave much lower stereoselectivity [ 1361. The mannose tetrasaccharide building block was prepared by the consecutive glycosylation of the 6-0-unprotected mannose acceptor with three suitably protected O-mannosyl trichloroacetimidates. The two building blocks were coupled by glycosylation of the GlcNyx(1-1)inositol acceptor with the tetramannosyl trichloroacetimidate in very good yield. All a-

38

2 Trichloroacetimidates

mannoside bonds were stereospecifically formed. The synthesis was completed by attaching the phosphate and the ceramide phosphate moieties at the suitably deprotected positions. Other inositolglycans are of pharmaceutical interest because of their insulin-like effect. The synthesis of a 0-linked inositol glycoside using a linear strategy has been reported [ 1371. The azidoglucose acceptor was first glycosylated with mannosyl trichloroacetimidate. The resulting disaccharide was again converted into the trichloroacetimidate and used for the glycosylation of the inositol acceptor. The pseudotrisaccharide was stereoselectively (p :a = 3 : 1) obtained and then consecutively glycosylated with the three suitably protected mannosyl trichloroacetimidates. Introduction of the phosphate groups gave the desired inositolglycans; some related derivatives have also been synthesized.

2.3.3 Glycosylation of Sphingosine Derivatives and Mimics The azidosphingosine glycosylation method [ 138, 1391 is well established. It has been found that azidosphingosine is a better glycosyl acceptor than ceramide or other sphingosine derivatives. Some examples of the synthesis of glycosphingolipids with complex oligosaccharide trichloroacetimidates and azidosphingosine have already been shown (see above). Even when other glycosylation methods have been used for the synthesis of the oligosaccharide moiety, the trichloroacetimidate method has been preferentially chosen for the coupling to azidosphingosine (Scheme 36) [140]. A tetrasaccharide trichloroacetimidate of the ganglio series containing three neuraminic acid residues was used for the glycosylation of 3-0-protected azidosphingosine. Reduction of azide, attachment of the fatty acid, and deprotection

N.

TMSOTf CH,CI,. O°C 65%

I NeuAca(2-8)NeuAca(2-3)

Scheme 36.

OTBDPS

I

NeuAca(2-3)

GT'a

Hasegawa et al. [14W

2.3 0-Glycosidrs

6BZ

39

Higuchi et al. [154]

Scheme 37.

yielded the ganglioside GT1,. With the help of the azidosphingosine glycosylation method many other glycosphingolipids have been synthesized, ie. GD1, [ 1401, GD2 [141], GT3 [142], GT4 [142], GQlb [143, 1441, GQlba [145], sialylated globosides [ 1461, sulfated glucuronyl paraglobosides [ 1471, sulfated sLeX glycosphingolipids [ 1481, ganglioside lactams corresponding to G M I , GM2, GM3, and GM4 [ 1491, and some xylose analogs of glycosphingolipids [ 1501. Although the glycosylation of ceramides has been found to proceed with low yields [ 151, 1521, this method has been applied a few times, especially when special ceramide derivatives are required and their azido derivatives were not available (Scheme 37). ‘The synthesis of astrocerebroside A, which has been isolated from starfish and which is of interest because of its pharmacological activity, has been reported via glycosylation of its ceramide moiety with 0-glucosyl trichloroacetimidate [153]. A hematoside-type ganglioside called GAA-6 which has been also found in starfish, was prepared in the same way via glycosylation of the phytosphingosine derivative with the trisaccharide trichloroacetimidate [ 1541. To find potent selectin binding inhibitors, mimics not only of the carbohydrate part of glycosphingolipids, especially of LeXand sLeX(see above), but also mimics of various lipid analogs have been prepared. 2-Tetradecylhexadecanol has been proposed as lipid moiety, and was glycosylated with the sialylated lactosyl trichloroacetimidate to yield the GM3 analogous neoglycolipid [ 1551. With the same lipid moiety were also prepared the GM4 [155], LeX[156], 3’-C-carboxymethyl LeX [ 1571, and some sulfated [ 158- 1601 and phosphorylated [ 1601 neoglycolipids. Another lipid moiety containing a spacer of variable length was glycosylated with the required trichloroacetimidate to prepare sLeXneoglycolipids [ 1611. A double-chain bis-sulfone has been used for the glycosylation with the corresponding trichloroacetimidates in the synthesis of a globotriosyl neoglycolipid and some deoxy derivatives [162, 1631. Other lipid moieties have been used for various purposes (Scheme 38). The hexadecanediol derivative was glycosylated with the trisaccharide trichloroacetimidate [ 1641. The resulting glycoside was transformed into the 16-mercaptohexadecanyl glycoside and used for biosensor studies. Glycosylation of a trivalent acceptor with mannosyl trichloroacetimidate furnished a triantennary cluster mannoside; its

2 Trichloroucetimidutes

40

TMSOTf CH2C12,r.t. 71%

1 \OAc

AcO

?Oq

no

0

NH +

Y CCI,

TMSOTf CH2C12, -65°C b 74%

HO$

no

Lindhorst et al. 11651

+

m 2 w p e x a n e , r.t.

no

~

49-72%

-OH

*R

Schmidt et al. [1661

Scheme 38.

binding properties to lectins were investigated [ 1651. Long-chain diols were glycosylated with glucosyl trichloroacetimidate to synthesize bis-glycosides, furnishing new types of non-ionic surfactants of the 2/1-type (two sugar head groups, one lipophilic chain) [ 1661. Surfactants of the 2/2-type have also been prepared [ 1661. 2.3.4 Glycosylation of Amino Acids The well established method [167] for the preparation of a-glycosylated serine and threonine derivatives has also been applied to the attachment of complex oligosaccharides (Scheme 39). Glycosylation of the serine acceptor with the P-configured trisaccharide trichloroacetimidate gave the a-glycosylated product stereospecifically, thus furnishing a derivative of the Fla-antigen [ 1681. Other derivatives including the sialyl T (ST) tumor antigen [ 1691, have also been prepared. The use of N-acetyl-protected and 4,6-O-benzylidene-protectedgalactosamine trichloroacetimidates has been proposed for the synthesis of a-glycosides [170]. Because the cisbicyclic structure disfavors oxazoline formation, a-glycosides are formed exclusively even from the a-trichloroacetimidates. Thus the Thomsen-Friedenreich (TF) antigen has been prepared from the Galp( 1-3)GalNAc trichloroacetimidate and serine or threonine as acceptor [ 1711.

2.3 0-Glycosides

41

HNFmoc HO

GalP(1-4)GlcNP(l-G)GalNa(1-3)Ser

Danishefsky et al. [I681 NH

Ph HNFmoc HO . hCOOCH,COPh

BF3.0Etz,THF, -20%

Ad3

Koganti et al. [1711

68% kl,

HtjFmoc

rCm,

HO

m AGO $

R = H, CH3 R' = All, Pfp

R

TMSOTf, CH2CIz,r.t.

TeocN

HNAc

80-90% HoYNH CCI,

Saha, Schmidt 11721

k0% AcO

Ad3

TMSOTf, CHzCIz, 74%

b

I O YCCI, N H

'

O

H

% A +

DtsN

Fmoc-lle-Ser-Gly-lle-Gly- OCH,

A

Fmoc-lle-S~r-Gly-Ile-Gly

I

I

TMSOTf, CH2CI2, 78%

G Resin e +

A%$

0

I

O YCCI, N H

MeMal et a1.[1731

Fmoc-Ile-Ser-Gly-lle-Gly- OCH,

Scheme 39.

The preparation of Gala(1-3)Ser and its use for further syntheses has also been reported [98, 99, 170, 1711. The N-trichloroethoxy (Teoc)-protected glucosamine trichloroacetimidate has been employed for the synthesis of P-glycosidically linked amino acids [ 1721. Glycosylation of various protected serine and threonine esters gave P-glycosides, because of the presence of participating neighboring groups. The N-tetrachlorophthaloyl (TCP)-protected derivatives have been used similarly. After conversion into the N-acetyl derivatives, the building blocks are suitable for glycopeptide synthesis. A completely different approach for the synthesis of glycopeptides has been reported by glycosylation of the serine-containing peptides on the solid-phase [ 1731. A polyoxypropylene cross-linked polyethylene glycol resin (POEPOP, polyoxyethylene-polyoxypropylene) was chosen because it contains only ether bonds and no amide bonds, and several linkers were investigated. The use of an Ndithiasuccinyl (Dts)-protected glucosamine trichloroacetimidate as donor gave the P-glycosylated peptide whereas the use of the azidoglucosyl trichloroacetimidate

42

2 Trichloroacetimidates

gave, as expected, the a-glycosylated peptide. Galactosyl, mannosyl, and fucosyl trichloroacetimidates have been used for the synthesis of libraries of peptide acceptors with more than one serine residue. 2.3.5 Polycyclic and Macrocyclic Glycosides Glycosides of polycycles or macrocycles (anthracyclines, chalicheamicin, macrolactones, etc.) are of great interest because of their antibiotic and antitumor activity. The synthesis of calichearubicin A and B, which have the same carbohydrate moiety as calicheamicin, has recently been reported (Scheme 40) [174, 1751. The phenolic acceptor was stereospecifically glycosylated with rhamnosyl trichloroacetimidate in very good yield. The resulting a-glycoside was transformed into the trichloroacetimidate donor which was used for the glycosylation of the anthracycline acceptor with silver triflate as catalyst. The a-glycosidically linked calichearubicin A was stereoselectively ( a :P = 5 : 1) obtained. The P-glycosidically linked calichearubicin B was similarly prepared from the same trichloroacetimidate donor with boron trifluoride-diethyl ether as the catalyst and, despite the lack of a participating neighboring group, again the P-stereoselectivitywas high (P : a = 5 : 1, 35%). The syntheses of a modified carbohydrate moiety, a head-to-tail dimer of the calicheamicin oligosaccharide [ 1761, the 1,1'-disaccharide (trehalose) containing the carbohydrate moiety of everninomycin [ 1771, and the vancomycin disaccharide CH,

BF,.OEtz H " ; iTBDMGH3 ; ; . CH30

+

HO

OAc

OCH,

OCH,

CH2CI2 95%

OCH,

OCH,

TBDMSO

0%

\/ ?

H,CO

CU

OCH,

H:q Danishefsky et al. [174,1751

CH30

OH

OH

calichearubicinA

Scheme 40.

HO

a

43

2.3 0-Glycosides OTESMo

0

TMSOTf dioxaneflol., O'C 75% (a$=8:1)

OH

+

CCI,

HNWoOPMB b o T B D M s I

Nicolaou et al. [181,1821

OTBDMS

TMSOTf CH2CI , -78'C 91 % (i:a=5:1)

eleutherobin

OH

TMSOTf CH2CI2, -4O'C not reported F CCI,

C I A c O ) TBDMSO PhS

iPrC0

M &

HO

I

Me0

v

'Me

OH OH

HO

HO HO &

o

o

0

u iPrC0

iPrCO O L *k

OH OH

olivomycin A

Roush et al. [1831

Scheme 41.

[ 1781 have also been reported. The total synthesis of neocarcinostatin including the preparation of the N-methylfucosamine trichloroacetimidate and the glycosylation of the endiyne-type aglycone were recently described [ 179, 1801. The synthesis of eleutherobin, which has antitumor activity and cytotoxic properties similar to those of taxol, has also been reported (Scheme 41) [181, 1821. The carvone-derived acceptor was stereoselectively glycosylated with the arabinosyl trichloroacetimidate ( a :p = 8: 1). The resulting glycoside intermediate was treated as for sarcodictyins A and B to yield eleutherobin and eleuthosides A and B. Studies of the synthesis of olivomycin A have been reported [ 1831. Glycosylation of a glycal acceptor with a trichloroacetimidate and an acetate donor gave a trisaccharide which was transformed into the trichloroacetimidate and used for the glycosylation of the acyloin acceptor. The resulting P-glycoside is a model compound for the approach to the synthesis of olivomycin A.

44

2 Trichloroacetimidates

OH

80% (1:1)

I

BFa.OEt2 CH2C123 -80c

no

Jiang, Schmidt [184,1851

calonyctinA

Scheme 42.

The macrolactone calonyctin A, a plant growth regulator, has been synthesized (Scheme 42) [ 184, 18.51. The quinovose acceptor was glycosylated subsequently with the suitably protected quinovosyl, rhamnosyl, and quinovosyl trichloroacetimidates; all steps could be performed stereospecifically. Transformation of the tetrasaccharide into the trichloroacetimidate and glycosylation of the racemic hydroxy acid derivative gave two epimeric compounds. Separation and lactonization gave calonyctin A. A convergent strategy has been applied to the synthesis of Tricolorin A, a natural herbicide with cytotoxic activity against cultured P-388 and human breast cancer cells (Scheme 43) [186, 1871. The acceptor was prepared by glycosylation of the hydroxy acid derivative with the galactosyl and the glucosyl trichloroacetimidates. It was coupled with the disaccharide trichloroacetimidate which had been prepared before from the suitable rhamnose donor and acceptor. All steps proceeded with high stereoselectivity. The resulting tetrasaccharide was transformed into the macrolactone Tricolorin A. The synthesis of the disaccharide macrolactone lonitoside 1 has also been reported [ 1881. 2.3.6 Glycosides of Phosphoric and Carboxylic Acids

Trichloroacetimidates can be used for glycosylation of phosphoric and carboxylic acids without additional Lewis acid. Reaction of the fucosyl a-trichloroacetimidate

45

2.3 0-Glycosides TMSOTf CH,C12, r, t. 79%

AgOTf CH2C12,r. t. a4%

phT& PivO

0

A"Oo CCI,

YNH CCI,

CCI,

0

TMSOTf CH2C12,r.t. 75%

BnO

OBn

0 0

0

HO

Hd Larsson, Heathcock [186,1871

tricolorin A

Scheme 43.

with dibenzyl phosphate gave the b-fucosyl phosphate with stereospecific inversion of configuraticin (Scheme 44) [189]. Deprotection and coupling with GMP morpholidate yielded GMP fucose which was purified by preparative HPLC. Reaction of the galactosyl trichloroacetimidate with a uridine phosphonate derivative furnished a phosphonate analog of UDP-galactose [22] which might serve as glycosyl donor or as inhibitor for galactosyltransferases. Some nucleoside monophosphate sugars have also been prepared directly by reaction of 0-glycosyl trichloroacetimidates with nucleoside monophosphates [ 1901. Reaction of the glucosyl trichloroacetimidate with a ceramide- 1-phosphate derivative gave exclusively the P-glycoside, owing to the participating neighboring group [ 1911. Reaction of the benzyl-protected glucosyl P-trichloroacetimidate with a malonic acid derivative as acceptor gave, with inversion of configuration, the a-linked malonyl derivative which was designed to mimic the pyrophosphate moiety of UDP-glucose and to serve as inhibitor for glycosyltransferases [ 1921. 2.3.7 Solid-Phase Synthesis

Although solid-phase chemistry is well developed, progress has not yet been fully extended to oligosaccharide chemistry, because of the high demands on the polymer support and the lack of powerful analytical tools for monitoring reactions on solid phases. The synthesis of an a(1-2)-linked pentamannose moiety has recently been reported (Scheme 45) [193]; a Merrifield resin, a thio-linker, and a 2-0-acetyl protected mannosyl trichloroacetimidate were used. The glycosylation reaction was

46

2 Trichloroucetimidutes CCI,

0

Schmidt et al. [IS91 0 A*

OAc

AcO!&&oyCCI, OAC

0

0

II

+

NH

Schrnelzer 1221

HO AcO

OAc

Schmidt et al. [I911

CCI,

It II Wong et al. [I921

x

0

0

HO

OH

Scheme 44.

monitored by MALDI-TOF and, after oxidative cleavage from the solid phase, the oligosaccharide was obtained by preparative HPLC. The method described has been applied to other examples, i.e. the synthesis of the branched-core pentasaccharide moiety of complex type N-glycans [ 194, 1951. The same 2-0-acetyl-protected mannosyl trichloroacetimidate was used in a polymer-supported solution-phase synthesis of a tetramannose moiety with polyethylene glycol monomethyl ether (MPEG) as polymer-support [ 1961. The use of controlled-pore glass (CPG) as solid-phase support is under investigation in the synthesis of oligosaccharides [ 1971 and glyconucleotide conjugates [ 1981 with trichloroacetimidates as donors. The synthesis of polylactosamine oligosaccharides was performed with a Merrifield resin and a benzyl-type linker (Scheme 46) [199].The glycosylation was stereospecifically performed with a phthalimido-protected lactosamine trichloroacetimidate bearing a levulinoyl group as temporary protecting group which can be removed with hydrazinium acetate under mild conditions. After cleavage from the linker the octasaccharide was obtained in good yield (overall 42%).

47

2.3 Q-Glycosides r

1

bR

Rademann, Schmidt [I931

R = H, CH,

Scheme 45.

Heparan sulfate-like oligomers have been prepared by polyethylene glycol monomethyl ether (MPEG)-supported solution-phase synthesis (Scheme 47) [200]. Glycosylation of the succinyl linked acceptor (linkage not via the anomeric position) was performed with the disaccharide trichloroacetimidate up to five times, again using the levulinoyl group as temporary protecting group. The yield of the coupling step could be increased to 95% by optimizing the reaction temperature to 10 "C, and only the a-glycoside was obtained. Cleavage from the linker with lithium hydro-

,

Ph,hN

OAC

BnO

,

,

~ BnO

~

~

o

NH

A

C

C

~

NH,NH,.AcOH EtOH

,

OHn

1

TrtBF,

CH,CI,.

En0

OBn

OBn

r.t.

Ogawa et al. 11991 n = 1-3

Scheme 46.

48

2 Trichloroacetimidates

OMe

Dreef-Tromp et al. I2001

n = 2-5

Scheme 47.

peroxide and further transformations yielded the hexa-, octa-, deca-, and dodecasaccharide fragments of heparan sulfate. In a recently introduced method for the synthesis of glycopeptides the peptide was synthesized from monoglycosylated amino acid derivatives followed by glycosylation reactions on the same solid phase (Scheme 48) [201]. A polyethylene glycol N,N-dimethylacrylamide copolymer (PEGA) resin was used for the synthesis of the nonapeptide acceptor. When the resin-bound acceptor was thoroughly washed and dried before subsequent glycosylation with the benzoyl-protected trichloroacetimidate, the core 3 peptide derivative was stereospecifically obtained in good yield (62%). The synthesis of a-glycosides with trichloroacetimidates bearing no participating neighboring groups is much more difficult. The reactivity of the donor has to be well adapted. Nevertheless, glycosylation of the octapeptide acceptor with the benzoyl-protected azidogalactosyl trichloroacetimidate was stereospecific, but gave the core 5 peptide derivative in only 20% yield. Use of the corresponding less reactive 0-acetyl-protected trichloroacetimidate did not work. Further examples employed disaccharide trichloroacetimidates or performed two subsequent glycosylations to synthesize branched trisaccharides. Although the direct glycosylation of OH-unprotected serine-containing peptides did not work on this resin, the use of another resin, containing no amide bonds, gave good results (see above, Scheme 39) 11731. Further applications of polymer-supported glycosylations for the synthesis of

2.4 S-Glycosides

49

Ph ,002

I

Ac-Glu-Pro-Thr-Thr-Thr-Pro-lle-Thr-Thr7 Ac-Glu-Pro-Thr-Thr-Thr-Pro-lle-Thr-Thr-

I

‘Bu

l

‘Bu

l

‘Bu

I I

‘Bu ‘Bu

Ac H Resin

CH&

AcN H i

“I3 TMSOTf “I3

b Ac-Glu-Pro-Thr-Thr-Thr-Pro-Ile-Thr-Thr-NH2

-15OC

core 3

62%

*

TMSOTf CH2CI2, -1 5OC 20%

Ac-Pro-Thr-Thr-Thr-Pro-Ile-Thr-Thr-NH,

core 5

Bock et al. iZOl1

Scheme 48.

oligosaccharide libraries by a combinatorial approach have been reported; to this end, polymer-bound acceptors with several free hydroxy groups were glycosylated randomly with trichloroacetimidates [2021.

2.4 S-Glycosides Thio-linked analogs of oligosaccharides, i. e. Lewis’, sialyl Lewis’, etc., are of interest because of their improved stability to glycosidases. The synthesis of several examples by the trichloroacetimidate glycosylation method has been described (Scheme 49) [204, 2051. Glycosylation of the 3-thio galactose acceptor with a fucosyl trichloroacctimidate gave the a-disaccharide stereospecifically; this was converted into the 3,4-dithio glucose derivative via the triflate and then substitution with inversion of the configuration. Conversion into the disaccharide trichloroacetimidate and glycosylation of heptyl mercaptan gave the thio-linked disaccharide acceptor which was coupled with the neuraminic acid disaccharide to yield the sLeX analog containing only thio linkages and a glucose instead of a glucosamine moiety. For synthesis of the LeXanalog, the disaccharide intermediate was coupled with acetobromogalactose and transformed into the trisaccharide trichloroacetimidate. Glycosylation of the thio-linked acceptor, which was prepared from the lactosyl trichloroacetimidate and heptyl mercaptan, yielded the Le’ analogdus pentasaccharide. All p-glycosidic bonds were stereospecifically formed owing to the participation of neighboring groups. Similarly the thio-linked Lewisa [205]and ganglioside GM3 [206] epitopes have been prepared.

50

2 Trichloroucetimidutes

OH OH HO

LeXanabg

Eisele, Schmidt I2031

Scheme 49.

The benzyl-protected mannosyl a-trichloroacetimidate has been used for the glycosylation of 2-mercaptopyridine (Scheme 50) [207], but the stereoselectivity was low. The P-glycoside was separated and oxidized to the 2-pyridyl sulfone which was employed as glycosyl donor for the synthesis of C-glycosides. The corresponding azidogalactose and N-acetylgalactosamine derivatives have also been prepared [208]. An S-glycosylated amino acid derivative has been synthesized by glycosylation of a D-cysteine derivative with the L-rhamnosyl trichloroacetimidate [209]. The Sglycosyl amino acid was used for the synthesis of an 5'-glycopeptide derivative of a luteinizing hormone-releasing hormone (LH-RH) antagonist. The S-glucosylated L-

2.6 C-Glycosides

51

(pa= 5:4) Beau et al. 12071

CCI,

c

HS

50%

c

Kessler et al. [2091

Scheme 50.

cysteine has been prepared applying the same method [210]. Some glycosylations of thiophenol [21 I ] and thiobenzyl alcohol [212]derivatives with trichloroacetimidates have also been reported.

2.5 N- and P-Glycosides Trichloroacetimidates are also suitable glycosyl donors for the synthesis of Nglycosides. For instance, reaction of N-phthaloyl-protected glucosamine trichloroacetimidate with trimethylsilyl azide gave, owing to neighboring group participation, only the p-glycosyl azide (Scheme 51) [213]. In the synthesis of Nglycopeptides the glucosamine azide is a suitable building block at the reducing end. After reduction of the azide to the amino group it can be linked to aspartate via amide bond formation. Another example of the synthesis of N-glycosides is the reaction of the ribopyranosyl trichloroacetimidate with 2-(3-pyrazolyl)pyridine [2141. The resulting nucleotide was employed for investigations on non-helical supramolecular nanosystems. A P-glycoside has been prepared with the hemiacetal-type trichloroacetimidate and trimethyl phosphite [2151. The resulting diastereomeric phosphonates can be considered as P-analogs of uronic acids, and their glycosides are of interest in investigations with glycosidases.

2.6 C-Glycosides Application of trichloroacetimidates to the synthesis of aromatic C-glycosides, i. e. vitexin, isovitexin, isoembigenin, etc. which are of interest because ot their physiological properties, is well established [216, 2171. A furan derivative was recently

52

2 Trichloroucetimidutes

0

&lo phq *'

NPht

CCI,

KNH

TMSN3 TMSOTf, CH2CI,, -78'C) 700,"

lto et al. k131

NPht

Eschenrnoser et al. [2141 CCI,

BnLv OPWH,),

P(OCH& TMSOTf, CH3CN, 4'C 43%

BnO

w *

OAll

+

Bgo%oAl,

OAll

OPWH,),

1:l

Vasella et al. [2151

Scheme 51.

glycosylated with glucosyl trichloroacetimidate to yield the P-glycoside stereospecifically; this served as intermediate in the synthesis of visnagine analogs (Scheme 52) [218]. The synthesis of bis(C-glycosyl) compounds [219] and of glycosides with glucosamine [220] has also been reported. Occasionally the 0-glycoside intermediates could be isolated and subjected to Fries-type rearrangement to furnish the final C-glycosides. _OBn

+

O n% 'h

O Y N H

TMSOTt CH2C12,-300C,

&CH,

0

'

OH

g

EilO

77% OBn

iCI, BnO 0 n

Schmidt et al. [2181

TMSCN

Aa&o$cl, Ad)

OAc

CH2C12, TMSOTfr.t. qu

CN

Shiozaki et al. [222,223]

References

53

A known method [221] for the preparation of C-glycosyl cyanides from trichloroacetimidates has been applied to the reaction of the 0-acetyl-protected 0glucosyl trichloroacetimidate [222, 2231. Reaction with trimethylsilyl cyanide gave the a-glycoside, a valuable intermediate in the synthesis of carboxylic acid derivatives or other C-glycosides, in quantitative yield. The reaction of trichloroacetimidates with silylenol ethers is also well established (2211. Another interesting example, reaction of benzyl-protected galactosyl trichloroacetimidate with a silyl enol ether, which can be considered as a homoalanine carbanion equivalent [224], led to the a-glycoside; as expected from previous results [221] it was formed with high stereoselectivity. The P-glycoside could be obtained by base-catalyzed isomerization. Further transformations were performed to obtain the a- and P-C--galactosyl-L-serinederivatives.

2.7 Conclusion and Outlook 0-Glycosyl trichloroacetimidates have proven to be valuable glycosyl donors for efficient syntheses of various types of glycosides. Their ease of formation and their high glycosyl donor properties released under mild acid catalysis has led to extraordinarily wide application. Not surprisingly, 0-glycosyl trichloroacetimidates have often afforded the best results in demanding glycosylation studies. The trichloroacetimidate method will, therefore, continue to advance efficient synthesis, particularly of the complex glycoconjugates which are in demand for biological and clinical studies. Thus, the trichloroacetimidate method will also contribute strongly to the future development of glycoscience.

References 1. R. R. Schmidt, Angew. Chem. 1986, 98,213-236; Angew. Chem. Int. Ed. Engl. 1986,25, 212235. 2. R. R. Schmidt, in Stereochemistry of Organic and Bioorganic Transformation-Workshop Conferences Hoechst (W. Bartmann, K. B. Sharpless, Eds), Vol. 17, pp 169-189, VCH, Weinheim, 1987). 3. R. R. Schmidt, Pure Appl. Chem. 1986,61, 1257-1270. 4. R. R. Schmidt in Comprehensive Organic Synthesis, Vol. 6, B. M. Trost, I. Fleming, E. Winterfeld, Eds., Pergamon Press, Oxford 1991, pp. 33-64. 5. R. R. Schmidt, in Carbohydrates-Synthetic Methods and Application in Medicinal Chemistry, A. Hasegawa, H. Ogura, T. Suamin, Eds., Kodanasha Scientific, Tokyo 1992, pp. 66-88. 6. K. Toshima, K. Tatsuta, Chem. Rev. 1993, 93, 1503-1531. 7. R. R. Schmidt, W. Kinzy, Adv. Carbohydr. Chem. Biochem. 1994, 50, 21-123. 8. R. R. Schmidt in Modern Methods in Curbohydrate Synthesis, S. H. Khan, R. A. O’Neill, Eds., Hanvood Academic Publishers, Chur, Schweiz, 1996, pp. 20-54. 9. R. R. Schmidt, K.-H. Jung in Preparative Carbohydrate Chemistry, S. Hannessian, Ed. Marcel Dekker, Inc., New York, 1997, pp. 283-312.

54

2 Trichloroacetimidates

10. R. R. Schmidt in Glycosciences: Status and Perspectives, H.-J. Gabius, S. Gabius, Eds., Chapman and Hall, Weinheim, 1997, pp. 31-53. 11. R. R. Schmidt, Pure Appl. Chem. 1998, 70, 397-402, 12. R. R. Schmidt, J. C. Castro-Palomino, 0. Retz, Pure Appl. Chem., in print. 13. F. Barresi, 0. Hindsgaul in Modern Synthetic Methods, B. Ernst, C. Leumann, Eds., Verlag Helvetica Chimica Acta, Basel, 1995, pp. 283-330. 14. R. R. Schmidt and J. Michel, Angew. Chem. 1980,92, 763-764; Angew. Chem., Znt. Ed. Engl. 1980, 19, 731-732. 15. R. Hoos, J. Huixin, A. Vasella, P. Weiss, Helo. Chim. Acta 1996, 79, 1757-1784. 16. V. J. Patil, Tetrahedron Lett. 1996, 37, 1481-1484. 17. S. P. Douglas, D. M. Whitfield, J. J. Krepinsky, J. Curbohydr. Chem. 1993, 12, 131-136. 18. G. G. Cross, D. M. Whitfield, Synlett 1998, 487-488. 19. H. Waldmann, G. Boehm, U. Schmid, H. Roettele, Angew. Chem. 1994,106,2024-2025 (See also Angew. Chem., Int. Ed. Engl., 1994,33, 1994-1996). 20. U. Schmid, H. Waldmann, Liebigs Ann.lRec1. 1997, 2573-2577. 21. A. Lubineau, B. J. Drouillat, J. Curbohydr. Chem. 1997, 16, 1179-1186. 22. U. Schmelzer, Dissertation, Universitat Konstanz, 1997. 23. U. Huchel, C. Schmidt, R. R. Schmidt, Eur. J. Org. Chem. 1998, 1353-1360. 24. A. Lubineau, K. Basset-Carpentier, C. Auge, Curbohydr. Res. 1997, 300, 161-167. 25. H. Paulsen, S. Peters, T. Bielfeldt, M. Meldal, K. Bock, Curbohydr. Res. 1995, 268, 17-34. 26. M.-Z. Liu, H.-N. Fan, Z.-W. Guo, Y.-Z. Hui, J. Carbohydr. Chem. 1996,15, 1139-1145. 27. T. Tsuda, S.-I. Nishimura, Chem. Commun. (Cambridge) 1996, 2779-2780. 28. J. A. L. M. van Dorst, C. J. van Heusden, A. F. Voskamp, J. P. Kamerling, J. F. G. Vliegenthart, Carbohydr. Res. 1996,291, 63-83. 29. T. Nishimura, F. Nakatsubo, Carbohydr. Res. 1996, 294, 53-64. 30. G. Limberg, G. C. Slim, C. A. Compston, P. Stangier, M. M. Palcic, R. H. Furneaux, Liebigs Ann. 1996, 1773-1784. 31. Ch. Li, B. Yu, M. Liu, Y. Hui, Curbohydr. Res. 1998, 306, 189-195. 32. S. Deng, B. Yu, Y. Hui, Tetrahedron Lett. 1998, 39, 6511-6514. 33. J. C. Castro-Palomino, R. R. Schmidt, Tetrahedron Lett. 1995,36, 5343-5346. 34. M. Lergenmuller, Y. Ito, T. Ogawa, Tetruhedron 1998, 54, 1381-1394. 35. H. Paulsen, B. Halpap, Curbohydr. Res. 1991, 216, 195-212. 36. W. Dullenkopf, J. C. Castro-Palomino, L. Manzoni, R. R. Schmidt, Carbohydr. Res. 1996, 296, 135-147. 37. W.-C. Liu, M. Oikawa, K. Fukase, Y. Suda, H. Winarno, S. Mori, M. Hashimoto, S. Kusumoto, Bull. Chem. SOC.Jpn. 1997, 70, 1441-1450. 38. M. Oikawa, A. Wada, H. Yoshizaki, K. Fukase, S. Kusumoto, Bull. Chem. SOC.Jpn. 1997, 70, 1435-1440. 39. K. Fukase, Y. Fukase, M. Oikawa, W.-C. Liu, Y. Suda, S. Kusumoto, Tetrahedron 1998, 54, 4033-4050. 40. X. Qian, 0. Hindsgaul, Chem. Commun. 1997, 1059-1060. 41. S. G. Bowers, D. M. Coe, G.4. Boons, J. Org. Chem. 1998,63,4570-4571. 42. J. C. Castro-Palomino, Dissertation, Universitat Konstanz, 1998. 43. M. R. E. Aly, J. C. Castro-Palomino, El-Sayed I. Ibrahim, El-Sayed H. El-Ashry, and R. R. Schmidt, Eur. J. Org. Chem. 1998, 2305-2316. 44. E. Meinjohanns, M. Meldal, A. Schleyer, H. Paulsen, K. Bock, J. Chem. SOC.,Perkin Trans. 1 1996, 985-993. 45. G. Blatter, J.-M. Beau, J.-C. Jacquinet, Carbohydr. Res. 1994,260, 189-202. 46. R. Kosmol, L. Hennig, P. Welzel, M. Findeisen, D. Muller, A. Markus, Van J. Heijenoort, J. Prukt. Chem. Chem.-Ztg. 1997,339, 340-358. 47. 0. Ritzeler, L. Hennig, M. Findeisen, P. Welzel, D. Mueller, Tetrahedron 1997, 53, 16651674. 48. R. U. Lemieux, R. M. Ratcliffe, Can. J. Chem. 1979,54, 1244-1251. 49. A. Vasella, C. Witzig, J.-L. Chiara, Helv. Chim. Acta 1991, 74, 2073-2077. 50. I. Toth, G. Dekany, B. Kellam, PCT Znt. Appl., WO 9838197 A1 19980903, 42 pp., (Chem. Abstr., 1998, 129, 216855).

References

55

51. H. Kunz, W. Gunther, Angew. Chem. 1988, 100, 1118-1119; Angew. Chem. Int. Ed. Engl. 1988,27, 1086-1087. 52. J. Seifert, C. Unverzagt, Tetrahedron Lett. 1996, 37, 6527-6530. 53. S. Weiler, R. R. Schmidt, Tetrahedron Lett. 1998, 39, 2299-2302. 54. A. Furstner, I. Konetzki, Tetrahedron Lett. 1998, 39, 5721-5724. 55. S. David, A. Malleron, C. Dini, Carbohydr. Res. 1989, 188, 193-200. 56. R. Preuss, R . R. Schmidt, Synthesis 1988,694-697. 57. W. R. Roush, D. P. Sebesta, R. A. James, Tetrahedron 1997,53, 8837-8852. 58. J. C. Castro-Palomino, Schmidt, R. R., Synlett 1998, 501-503. 59. M. Gelin, V . Ferrieres, D. Plusquellic, Carbohydr. Lett. 1997, 2, 381-388. 60. A. K. Choudhury, N. Roy, Carbohydr. Res. 1998, 308, 207-211. 61. H. Sekljic, N. Wimmer, A. Hofinger, H. Brade, P. Kosma, J. Chem. Soc., Perkin Trans. 1 1997, 1973-1982. 62. M. Izumi, 0. Tsuruta, S. Harayama, H. Hashimoto, J. Org. Chem. 1997, 62, 992-998. 63. H. Yuasa, S Matsuura, H. Hashimoto, Bioorg. Med. Chem. Lett. 1998, 8, 1297-1300. 64. E. P. Dubois, A. Neszmelyi, H. Lotter, V. Pozsgay, Tetrahedron Lett. 1996, 37, 3627-3630. 65. V. Pozsgay, E. P. Dubois, H. Lotter, A. Neszmelyi, Carbohydr. Res. 1997, 303, 165-173. 66. M. Ludewig, D. Lazarevic, J. Kopf, J. Thiem, J. Chem. Soc., Perkin Trans. 1 1998, 17511752. 67. R. Windmueller, R. R. Schmidt, Tetrahedron Lett. 1994, 35, 7927-7930. 68. A. Toepfer, ‘W. Kinzy, R. R. Schmidt, Liebigs Ann. Chem. 1994, 449-464. 69. L. Lay, L. Manzoni, R. R. Schmidt, Carbohydr. Res. 1998,310, 157-171. 70. L. Manzoni, L. Lay, R. R. Schmidt, J. Carbohydr. Chem. 1998,17,739-758. 71. N. L. Pohl, I-. L. Kiessling, Tetrahedron Lett. 1997, 38, 6985-6988. 72. A. Toepfer, IR. R. Schmidt, J. Carbohydr. Chem. 1993, 12, 809-822. 73. G. Hummel, R. R. Schmidt, Tetrahedron Lett. 1997,38, 1173-1 176. 74. U. Sprengard, G. Kretzschmar, E. Bartnik, C. Huels, H. Kunz, Angew. Chem., Int. Ed. Engl. 1995, 34, 990-993. 75. G. Kuznik, W. Hoersch, G. Kretzschmar, C. Unverzagt, Bioorg. Med. Chem. Lett. 1997, 7, 577-580. 76. G. Kretzschmar, W. Stahl, Tetrahedron 1998, 54, 6341-6358. 77. S. J. Danishefsky, J. Gervay, J. M. Peterson, F. E. McDonald, K. Koseki, D. A. Griffith, T. Oriyama, St. P. Marsden, J. Am. Chem. Soc. 1995, 117, 1940-1953. 78. L. Lay, R. Windmueller, S. Reinhardt, R. R. Schmidt, Carbohydr. Res. 1997, 303, 39-49. 79. J. Bartek, R. Muller, P. Kosma, Carbohydr. Res. 1998, 308, 259-273. 80. M. Izumi, 0. Tsuruta, H. Hashimoto, S. Yazawa, Tetrahedron Lett. 1996, 37, 1809-1812. 81. D. Manning, C. R. Bertozzi, S. D. Rosen, L. L. Kiessling, Tetrahedron Lett. 1996, 37, 19531956. 82. R. Roy, W. K. C. Park, 0. P. Srivastava, C. Foxall, Bioorg. Med. Chem. Lett. 1996, 6, 13991402. 83. H. Ishida, H. Hosokawa, H. Kondo, M. Kiso, A. Hasegawa, Carbohydr. Res. 1997,303, 131133. 84. G. Thoma, F. Schwarzenbach, R. 0. Duthdler, J. Org. Chem. 1996, 61, 514-524. 85. G. Kretzschmar, A. Toepfer, C. Huels, M. Krause, Tetrahedron 1997, 53, 2485-2494. 86. A. Toepfer, G. Kretzschmar, S. Schuth, M. Sonnentag, Bioorg. Med. Chem. Lett. 1997, 13171322. 87. G. Kretzschmar, Tetrahedron 1998, 54, 3765-3780. 88. T. Ikeda, T. Kajimoto, H. Kondo, C.-H. Wong, Bioorg. Med. Chem. Lett. 1997, 7, 24852490. 89. U. Greilich, R. Brescello, K.-H. J u g , R. R. Schmidt, Liebigs Ann. 1996, 663-672. 90. T. Stauch, U. Greilich, R. R. Schmidt, Liebigs Ann. 1995, 2101-2111. 91. J. C. Castro-Palomino, G. Ritter, S. R. Fortunato, S. Reinhardt, L. J. Old, R. R. Schmidt, Angew. Chern., Int. Ed. Engl. 1997,36, 1998-2001. 92. Y. Isogai, T. Kawase, H. Ishida, M. Kiso, A. Hasegawa, J. Carbohydr. Chem. 1996,15, 10011023. 93. L. F. Tietze, C. 0. Janssen, J.-A. Gewert, Eur. J. Org. Chem. 1998, 1887-1894.

56

2 Trichloroacetimidates

94. J. M. Lassaletta, K. Carlsson, P. J. Garegg, R. R. Schmidt, J. Org. Chem. 1996, 61, 68736880. 95. J. M. Lassaletta, R. R. Schmidt, Tetrahedron Lett. 1995, 36, 4209-4212. 96. J. M. Lassaletta, R. R. Schmidt, Liebigs Ann. 1996, 1417-1423. 97. N. Mathieux, H. Paulsen, M. Meldal, K. Bock, J. Chem. Soc., Perkin Trans. I 1997, 23592368. 98. D. Qiu, Koganty, R. Rao, Tetrahedron Lett. 1997, 38, 961-964. 99. D. Qiu, Koganty, R. Rao, Tetrahedron Lett. 1997, 38, 45-48. 100. Y. Nakahara, Y. Nakahara, Y. Ito, T. Ogawa, Tetrahedron Lett. 1997, 38, 7211-7214. 101. Y. Nakahara, Y. Nakahara, Y. Ito, T. Ogawa, Curbohydr. Res. 1998,309, 287-296. 102. C. Unverzagt, Angew. Chem. 1994, 106, 1170-1173 (See also: Angew. Chem., Int. Ed. Engl., 1994,33, 1102-1 104. 103. C. Unverzagt, Angew. Chem., Int. Ed. Engl. 1996, 35, 2350-2353. 104. C. Unverzagt, Angew. Chem., h t . Ed. Engl. 1997, 36, 1989-1992. 105. H. Paulsen, M. Springer, F. Reck, E. Meinjohanns, I. Brockhausen, H. Schachter, Liebigs Ann. 1995, 53-60. 106. Z. Szurmai, L. Janossy, Curbohydr. Res. 1996,296, 279-284. 107. Y. Nakahara, S. Shibayama, Y. Nakahara, T. Ogawa, Curbohydr. Res. 1996,280, 67-84. 108. L. Rochepeau-Jobron, J.-C. Jacquinet, Curbohydr. Res. 1998, Volume Date 1997, 305, 181191. 109. L. Rochepeau-Jobron, J.-C. Jacquinet, Carbohydr. Res. 1997,303, 395-406. 110. H. Uzawa, M. Gerz, Y. Ito, T. Ogawa, Curbohydr. Lett. 1996,2, 197-204. 111. G. Blatter, J.-C. Jacquinet, Curbohydr. Res. 1996,288, 109-125. 112. K. W. Neumann, J.4. Tamura, T. Ogawa, Glycoconjugute J . 1996, 13, 933-936. 113. J.-i. Tamura, K. W. Neumann, T. Ogawa, Liebigs Ann. 1996, 1239-1257. 114. J. Kovensky, Ph. Duchaussoy, M. Petitou, P. Sinay, Tetrahedron: Asymmetry 1996, 7, 31193128. 115. A. Helmboldt, M. Petitou, J.-M. Mallet, J.-P. Herault, J.-C. Lormeau, P.-A. Driguez, J.-M. Herbert, P. Sinay, Bioorg., Med. Chem. Lett. 1997, 7 , 1507-1510. 116. R. C. Buijsman, W. H. A. Kuijpers, J. E. M. Basten, E. Kuyl-Yeheskiely, G. van der Marel, C. A. A. van Boeckel, J. H. van Boom, Chem. Eur. J. 1996,2, 1572-1577. 117. J. E. M. Basten, C. M. Dreef-Tromp, B. De Wijs, C. A. A. Van Boeckel, Bioorg. Med. Chem. Lett. 1998, 8, 1201-1206. 118. J. C. Castro Palomino, M. Hernandez Rensoli, V. Verez Bencomo, J. Curbohydr. Chem. 1996, 15, 137-146. 119. W. Liao, D. Lu, Curbohydr. Res. 1996,296, 171-182. 120. G. Schuele, T. Ziegler, Tetrahedron 1996, 52, 2925-2936. 121. G. Schuele, T. Ziegler, Liebigs Ann. 1996, 1599-1607. 122. M.-J. L. Thijssen, M. H. G. Bijkerk, J. P. Kamerling, J. F. G. Vliegenthart, Curbohydr. Res. 1998,306, 111-125. 123. W. Zou, H. J. Jennings, Bioorg. Med. Chem. Lett. 1997, 7 , 647-650. 124. W. Zou, J.-R. Brisson, Q.-L. Yang, M. van der Zwan, H. J. Jennings, Curbohydr. Res. 1996, 295, 209-228. 125. F.-I. Auzanneau, F. Forooghian, B. M. Pinto, Curbohydr. Res. 1996,291, 21-41. 126. V. Pozsgay, Angew. Chem., Int. Ed. 1998,37, 138-142. 127. V. Pozsgay, J. Org. Chem. 1998,63, 5983-5999. 128. T. Ziegler, E. Eckhardt, D. Keller, J. Curbohydr. Chem. 1997, 16, 719-738. 129. A. Banaszek, Curbohydr. Res. 1998, 306, 379-385. 130. M. K. Gurjar, S. Adhikari, Tetrahedron 1997, 53, 8629-8634. 131. I. Bajza, K. E. Kover, A. Liptak, Curbohydr. Res. 1998, 308, 247-258. 132. Y. Ogawa, P. Kovac, Glycoconjugute J. 1997, 14,433-438. 133. T. G. Mayer, R. R. Schmidt, Liebigs Ann./Recl. 1997, 859-863. 134. B. Kratzer, T. G. Mayer, R. R. Schmidt, Eur. J. Org. Chem. 1998, 291-298. 135. T. G. Mayer, B. Kratzer, R. R. Schmidt, Angew. Chem. 1994, 106, 2289-2293. 136. J. Ye, W. T. Doerrler, M. A. Lehrman, J. R. Falck, Bioorg. Med Chem. Lett. 1996, 6, 17151718.

References

57

137. W. Frick, G. Mueller, Ger. Offen., 1998, DE 19649350 A1 19980604,20 pp. (Chem. Abstracts 129 : 54536). 138. R. R. Schmidt and P. Zimmermann, Angew. Chem. 1986,98,722-723; Angew. Chem., Int. Ed. Engl. 1986,25, 725-726. 139. P. Zimmerniann, R. Bommer, T. Bar, R. R. Schmidt, J. Curbohydr. Chem. 1988, 7, 435452. 140. H. Ishida, H . Ando, H. Ito, H. Ishida, M. Kiso, A. Hasegawa, J. Curbohydr. Chem. 1997, 16, 413-428. 141. A. Hasegawa, H. Ishida, Y. Isogai, H. Ishida, M. Kiso, J. Curbohydr. Chem. 1993, 12, 12171222. 142. H. Ando, H. Ishida, M. Kiso, A. Hasegawa, Curbohydr. Res. 1997, 300, 207-217. 143. H. Ishida, H . Ishida, M. Kiso, A. Hasegawa, Curbohydr. Res. 1994, 260, ClLC6. 144. H. Ishida, H . Ishida, M. Kiso, A. Hasegawa, Tetrahedron; Asymmetry 1994, 5, 2493-2512. 145. K. Hotta, H. Ishida, M. Kiso, A. Hasegawa, J. Curbohydr. Chem. 1995, 14, 491-506. 146. H. Ishida, R. Miyawaki, M. Kiso, A. Hasegawa, J. Curbohydr. Chem. 1996, 15, 163-182. 147. Y. Isogai, H. Ishida, M. Kiso, A. Hasegawa, J. Curbohydr. Chem. 1996, 15, 1119-1137. 148. K. Yoshino. H. Ohmoto, N. Kondo, H. Tsujishita, Y. Hiramatsu, Y. Inoue, H. Kondo, H. Ishida, MI. Kiso, A. Hasegawa, J. Med. Chem. 1997, 40, 455-462. 149. M. Wilstermann, L. 0. Kononov, U. Nilsson, A. K. Ray, G. Magnusson, J. Am. Chem. Soc. 1995,117,4742-4754. 150. M. Wilstermann, G. Magnusson, J. Org. Chem. 1997,62, 7961-7971. 151. S. Fujita, M. Numata, M. Sugimoto, K. Tomita, T. Ogawa, Curbohydr. Res. 1994,263, 181-196. 152. M. Iida, A. Endo, S. Fujita, M. Numata, K. Suzuki, S. Nunomura, T. Ogawa, Glycoconjugute J. (1996), 13,203-21 1. 153. N. Chida, N. Sakata, K. Murai, T. Tobe, T. Nagase, S. Ogawa, Bull. Chem. Soc. Jpn. 1998, 71, 259-272. 154. T. Sugata, Y. Kan, Y . Nagaregawa, T. Miyamoto, R. Higuchi, J. Curbohydr. Chem. 1997,16, 917-925. 155. A. Hasegawa, N. Suzuki, H. Ishida, M. Kiso, J. Curbohydr. Chem. 1996,15,623-637. 156. Y . Wada, T. Saito, N. Matsuda, H. Ohmoto, K. Yoshino, M. Ohashi, H. Kondo, H. Ishida, M. Kiso, A. Hasegawa, J. Med. Chem. 1996,39,2055-2059. 157. H. Ishida, H Hosokawa, H. Kondo, M. Kiso, A. Hasegawa, Curbohydr. Res. 1997,303, 131133. 158. E. Tanahashi, K. Murase, M. Shibuya, Y. Igarashi, H. Ishida, A. Hasegawa, M. Kiso, J. Curbohydr. Chem. 1997, 16, 831-858. 159. T. Ikami, H. Hamajima, T. Usui, T. Mitani, H. lshida, M. Kiso, A. Hasegawa, J. Curbohydr. Chem. 1997,16, 859-875. 160. T. Ikami, N. Tsuruta, H. Inagaki, T. Kakigami, Y. Matsumoto, N. Tomiya, T. Jomori, T. Usui, Y. Suzuki, H. Tanaka, D. Miyamoto, H. Ishida, A. Hasegawa, M. Kiso, Chem. Phurm. Bull. 1998, 46, 797-806. 161. K. Adachi, Y. Yamada, H. Wada, A. Kameyama, H. Ishida, M. Kiso, J. Curbohydr. Chem. 1998, 17, 595-607. 162. 2. Zhang, G. Magnusson, J. Org. Chem. 1995,60, 7304-7315. 163. 2. Zhang, G. Magnusson, J. Org. Chem. 1996,61, 2383-2393. 164. P. I. Kitov, C:. Railton, D. R. Bundle, Curbohydr. Res. 1998, 307, 361-369. 165. S. Kotter, U. Krallmann-Wenzel, S. Ehlers, T. K. Lindhorst, J. Chem. Soc., Perkin Truns. I 1998,2193-2200. 166. R. R. Schmidt, K. Jankowski, Liebigs Ann. 1996, 867-879. 167. W. Kinzy, R R. Schmidt, Curbohydr. Res. 1987,164,265-276. 168. X.-T. Chen, ID. Sames, S. J. Danishefsky, J. Am. Chem. Soc. 1998, 120, 7760-7769. 169. D. Sames, X.-T. Chen, S. J. Danishefsky, Nature (London) 1997,389, 587-591. 170. J. E. Yule, T. C. Wong, S. S. Gandhi, D. Qiu, M. A. Riopel, Koganty, R. Rao, Tetrahedron Lett. 1995, 36, 6839-6842. 171. D. Qiu, S. S. Gandhi, Koganty, R. Rao, Tetrahedron Lett. 1996, 37, 595-598. 172. U. K. Saha, R. R. Schmidt, J. Chem. Soc., Perkin Truns. 11997, 1855-1860.

58

2 Trichloroacetimidates

173. A. Schleyer, M. Meldal, R. Manat, H. Paulsen, K. Bock, Angew. Chem., Int. Ed. Engl. 1997, 36, 1976-1978. 174. K. M. Depew, S. M. Zeman, S. H. Boyer, D. J. Denhart, N. Ikemoto, S. J. Danishefsky, D. M. Crothers, Angew. Chem., Int. Ed. Engl. 1997, Volume Date 1996,35, 2797-2800. 175. S. H. Olson, S. Danishefsky, Tetrahedron Lett. 1994, 35, 7901-7904. 176. K. C. Nicolaou, K. Ajito, H. Komatsu, B. M. Smith, P. Bertinato, L. Gomez-Paloma, Chem. Commun. (Cambridge) 1996, 1495-1496. 177. K. C. Nicolaou, F. L. van Delft, S. R. Conley, H. J. Mitchell, Z. Jin, R. M. Rodriguez, J. Am. Chem. SOC.1997,119,9057-9058. 178. K. C. Nicolaou, H. J. Mitchell, F. L. Van Delft, F. Rubsam, R. M. Rodriguez, Angew. Chem., Int. Ed. 1998, 37, 1871-1874. 179. T. Kaneko, K. Takahashi, M. Hirama, Heterocycles 1998, 47, 91-96. 180. A. G. Myers, J. Liang, M. Hammond, P. M. Harrington, Y. Wu, E. Y. Kuo, J. Am. Chem. SOC.1998,120, 5319-5320. 181. K. C. Nicolaou, F. L. van Delft, T. Ohshima, D. Vourloumis, J. Xu, S. Hosokawa, J. Pfefferkorn, S. Kim, T. Li, Tianhu, Angew. Chem., Int. Ed. Engl. (1997), 36, 2520-2524. 182. K. C. Nicolaou, T. Oshima, S. Hasokawa, F. L. van Delft, D. Vourloumis, J. Y. Xu, J. Pfefferkorn, S. Kim, J. Am. Chem. SOC.1998, 120, 8674-8680. 183. W. R. Roush, K. Briner, B. S. Kesler, M. Murphy, D. J. Gustin, J. Org. Chem. 1996, 61, 6098-6099. 184. Z.-H. Jiang, R. R. Schmidt, Liebigs Ann. Chem. 1994, 645-651. 185. Z.-H. Jiang, A. Geyer, R. R. Schmidt, Angew. Chem., Int. Ed. Engl. 1995,34, 2520-2524. 186. D. P. Larson, C. H. Heathcock, J. Org. Chem. 1997,62, 8406-8418. 187. D. P. Larson, C. H. Heathcock, J. Org. Chem. 1996, 61, 5208-5209. 188. R. T. Brown, S. P. Mayalarp, J. J. Watts, Chem. Soc., Perkin Trans. 11997, 1633-1637. 189. R. R. Schmidt, B. Wegmann, and K.-H. Jung, Liebigs Ann. Chem. 1991, 121-124. 190. R. R. Schmidt, H. Braun, K.-H. Jung, Tetrahedron Lett. 1992, 33, 1585-1588. 191. T. J. Martin, G. Dufner, B. Kratzer, R. R. Schmidt, GlycoconjugateJ . 1996, 13, 547-553. 192. R. Wang, D. H. Steensma, Y. Takaoka, J. W. Yun, T. Kajimoto, C.-H. Wong, Bioorg. Med. Chem. 1997,5,661-672. 193. J. Rademann, R. R. Schmidt, J. Org. Chem. 1997, 62, 3650-3653. 194. J. Rademann, A. Geyer, R. R. Schmidt, Angew. Chem., Znt. Ed. 1998,37, 1241-1245. 195. J. Rademann, R. R. Schmidt, Tetrahedron Lett. 1996,37, 3989-3990. 196. L. Jiang, R. C. Hartley, T.-H. Chan, Chem. Commun. (Cambridge) 1996, 2193-2194. 197. A. Heckel, E. Mross, K.-H. Jung, J. Rademann, R. R. Schmidt, Synlett 1998, 171-173. 198. M. Adinolfi, G. Barone, L. De Napoli, A. Iadonisi, G. Piccialli, Tetrahedron Lett. 1998, 39, 1953-1956. 199. H. Shimizu, Y. Ito, 0. Kanie, T. Ogawa, Bioorg. Med. Chem. Lett. 1996, 6, 2841-2846. 200. C. M. Dreef-Tromp, H. A. M. Willems, P. Westerduin, P. van Veelen, C. A. A. van Boeckel, Bioorg. Med. Chem. Lett. 1997, 7, 1175-1180. 201. H. Paulsen, A. Schleyer, N. Mathieux, M. Meldal, K. Bock, J. Chem. SOC.,Perkin Trans. 1 1997,281-293. 202. Y. Ding, J. Labbe, 0. Kanie, 0. Hindsgaul, Bioorg. Med. Chem. 1996,4,683-692. 203. Th. Eisele, R. R. Schmidt, Liebigs Ann.jRecl. 1997, 1303-1313. 204. Th. Eisele, A. Toepfer, G. Kretzschmar, R. R. Schmidt, Tetrahedron Lett. 1996, 37, 13891392. 205. Th. Eisele, R. Windmuller, R. R. Schmidt, Carbohydr. Res. 1998, 306, 81-91. 206. Th. Eisele, R. R. Schmidt, Liebigs Ann.jRecl. 1997, 865-872. 207. T. Skrydstrup, 0. Jarreton, D. Mazeas, D. Urban, J.-M. Beau, Chem. Eur. J. 1998, 4, 655671. 208. D. Urban, T. Skrydstrup, J.-M. Beau, J. Org. Chem. 1998, 63, 2507-2516. 209. K. Michael, V. Wittmann, W. Konig, J. Sandow, H. Kessler, Int. J. Pept. Protein Res. 1996, 48, 59-70. 210. L. Kaesbeck, H. Kessler, Liebigs Ann./Recl. 1997, 165-167. 211. E. Bozo, S. Boros, J. Kuszman, Carbohydr. Res. 1998, 311, 191-202. 212. H. C. Hansen, G. Magnusson, Carbohydr. Res. 1998,307, 243-251.

References

59

213. A. Dan, M. Lergenmuller, M. Amano, Y. Nakahara, T. Ogawa, Y. Ito, Chem. Eur. J. 1998, 4, 2182-2190. 214. A. Eschenmoser, C. Miculka, N. Windhab, H.-U. Hoppe, PCTZnt. Appl., 1998, WO 9825943 A1 19980618, 26 pp., (Chem. Abstracts, 129: 67987). 215. R. Hoos, J. Huixin, A. Vasella, P. Weiss, Helv. Chim. Acta 1996, 79, 1757-1784. 216. J.-A. Mahling, R. R. Schmidt, Synthesis, 1993, 325-328. 217. J.-A. Mahling, K.-H. Jung, R. R. Schmidt, Liebigs Ann. 1995, 461-466. 218. E. El Telbani, S. El Desoky. M. A. Hammad, A. R. H. A. Rahman, R. R. Schmidt, Eur. J. Org. Chem. 1998,2317-2322. 219. E. El Telbani, S. El Desoky, M. A. Hammad, A. H. Abdel Rahman, R. R. Schmidt, Carbohydr. Rex 1998, 306, 463-461. 220. J. C. Castro-Palomino, R. R. Schmidt, Liebigs Ann. 1996, 1623-1626. 221. M. Hoffmdnn, R. R. Schmidt, Liebigs Ann. Chem. 1985, 2403-2419. 222. M. Shiozaki, T. Mochizuki, T. Wakabayashi, S.4. Kurakata, T. Tatsuta, M. Nishijima, Tetrahedron Lett. 1996, 37, 7271-7274. 223. M. Shiozaki, S.4. Kurakata, T. Tatsuta, H. Maeda, M. Nishijima, Tetrahedron 1997, 53, 16041-1 6060. 224. A. Dondoni, A. Marra, Chem. Commun. (Cambridge) 1998, 1741-1742.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

3 Iterative Assembly of Glycals and Glycal Derivatives: The Synthesis of Glycosylated Natural Products and Complex Oligosaccharides Lawrence J. Williams, Robert M. Garbaccio, and Samuel J. Danishefsky

3.1 Introduction Polypeptides, oligonucleotides and oligosaccharides are three classes of oligomers that play key roles in transacting the processes of life. While the functions of proteins and DN,4 have been appreciated, at least qualitatively, for over fifty years, basic information as to the vital missions of carbohydrates, beyond that of energy storage, are still being elucidated. Thus, the roles of oligosaccharides in cellular adhesion, protein folding, cell signaling, fertilization, pathogen binding and inflammation have been recognized only recently. The detection, purification and structure determination of complex oligosaccharides have been critical in the emergence of glycobiology. The continuing maturation of this science has added to the incentives for preparing complex oligosaccharide conjugates in homogeneous form. Among these incentives is the possibility of evaluating vaccines comprised of synthetic, cell free, carbohydrate-based tumor antigens for enhancing immune response to cancers which bear such cell surface epitopes. In addition to the necessity for such homogeneous constructs, the preparation of complex oligosaccharides poses no small challenge to the science of chemical synthesis. While the successes achieved in oligonucleotide and oligopeptide synthesis arose only after long term and exacting research, the issues associated with oligosaccharide synthesis are of a much higher level of complexity. In assembling either amide bonds for polypeptide construction, or internucleotide phosphate bonds in DNA synthesis, no new stereochemical issues are encountered. By contrast, each glycosylation reaction in oligosaccharide construction involves a question of stereochemical outcome at the anomeric site (a or P-linkage, Figure 1). Moreover, every glycosylation event requires identification of a specific hydroxyl group to serve as an acceptor center (Scheme 1, A). In a typical hexose, the pinpointing of an acceptor site obliges one to identify one of the five potential hydroxyl groups to serve as the reaction locus for glycosylation. In some instances, this specificity or selectivity of action may reflect inherent reactivity margins between free

62

3 Iterative Assembly of Glycals and Glycal Derivatives

Oligopeptides

Oligonucleotides

Oligosaccharides

RO

Glycoside Bond

1-

RO RO RO

OR

Figure 1.

hydroxyl groups. Alternatively and much more commonly, oligosaccharide construction involves extensive reliance on the use of protecting group strategies. For purposes of reiteration, a specific protecting group at the reducing end of an oligosaccharide, having been brought in at the glycosyl acceptor stage, would then be converted into a donating group. In addition to the requirement for manipulating the acceptor and donor character of the terminal hexose with each elongation cycle, the need for inclusion of N-acetyl functions, as well as occasional C-methyl, 0-methyl, 0-acyl, or sulfated alcohol groups, not to speak of sialic acid units, serves to further complicate an already daunting set of challenges. It was with these considerations in mind that we began to explore the possible advantages of a new strategy, ie. glycal assembly, to facilitate oligosaccharide construction [l]. While the use of glycals as direct glycosyl donors or precursors of glycosyl donors had been practiced long before our entry to the field in 1988, it seemed that several potentially interesting and promising synthetic opportunities had not been recognized. Upon casual inspection, it appeared that the complexities associated with differential hydroxyl protection would be significantly reduced in a glycal. Thus, in a hexose glycal, only three hydroxyls need be distinguished (Scheme 1, B). Further-

3.1 Introduction

63

A

1

1

a

-

O

H _L Po

HO

--+bop :bop OP

PO

OH

+

PO

Po

OP

OP

HO

Ohgosaccharide

4

X PO

OP Po

OP

B Po

E+

PO Po

Po

Reiterate Ohgosaccharide

4

Po

Scheme 1.

more, each hyclroxyl could well differ from the others in its expected reactivity, since one is primary.,one is allylic, and the other is a more hindered secondary alcohol. It seemed, therefore, that the need for selective protection could be lessened and those protections which are needed could be simplified in the context of glycals. The most significant point of departure of ‘glycal assembly’ originated from the use of glycals as both glycosyl donors and glycosyl acceptors. This two modality

64

3 Iterative Assembly of Glycals and Glycal Derivatives

usage sets up the possibility of synthesizing polysaccharides by the assembly of glycals in an iterative fashion. In this approach, the activated, glycal-derived donor would be captured by a glycal acceptor in a glycosylation reaction (Scheme 1, B). The glycal linkage of the newly generated disaccharide, though initially introduced as a glycosyl acceptor, could now serve as a donor for the next elongation cycle. Thus, glycals would have the potential advantage that coupling maneuvers could be implemented, and reiterated, with minimal protecting group manipulation. It is well to emphasize that the success of the glycal paradigm could not have been assumed in advance of experiment. Thus, employment of a glycal-housed hydroxyl as the glycosyl acceptor means that the sensitive enol ether linkage, bearing a potential allylic leaving group at C3, must withstand the action of the promoter required for the glycosylation event. Yet, glycal linkages bearing such leaving groups are often subject to Ferrier rearrangement under conditions similar to those that would be used in glycosylation. Hence, it was necessary to find promoter conditions that were effective for glycosylation while consistent with survival of the acceptor glycal. Through the continued investigation of these ideas, we have developed a broad based approach to synthesize complex oligosaccharides, including glycosylated natural products, via glycals. In the context of specific target structures, the program has served as an incentive for exploring new issues of chemical reactivity. For the purpose of this review, the program description will be confined to examining, and demonstrating by example, the feasibility of these theorems in the context of selected total syntheses of polysaccharides and their conjugates. While there is a unified theme governing this chapter, i.e. that of fostering progression of the methodology and strategy of synthesis of complex glycans, the discussion is divided into several parts. For purposes of presentation, each of the first four sections introduces a key methodological advance in glycal technology, typically under the umbrella of the natural product that inspired them. The oligosaccharide targets discussed in the last three sections demonstrate the integrated application of these methodologies, and the iterative glycal assembly strategy, to access tumor-associated oligosaccharides used in our expanding anti-cancer vaccine program.

3.2 Ciclamycin 0 The ciclamycin complex of antibiotics possesses high inhibitory activity against experimental tumors [2] and is claimed to be of value against human neoplasia [3]. These antibiotics are largely or entirely composed of q- and 8-pyrromycinones joined at C7 by glycosidic linkages to variously permuted carbohydrate domains [4, 51. One component of this complex, which is particularly difficult to obtain, is ciclamycin 0. At the time of our investigation, the structural assignment for 1 was based solely on spectroscopic arguments. Our strategy focused on construction of an appropriately protected trisaccharide and its use as a donor to achieve glycosylation of the anthracyclinone nucleus.

3.2 Ciclamycin 0

OH 0

OH

65

P

e &eM

1 Ciclamycin 0

0

Figure 2.

Given our interest in iterative assembly, a trisaccharide glycal was envisioned to be the target. The requirement for a-linked 2-deoxy hexoses prompted the utilization of the iodoglycosylation reaction developed by Lemieux [6] and Thiem [7] (Scheme 2). Adapting this methodology to iterative glycal assembly required strict control of the roles of glycals, since substochiometric amounts of iodine could, in principle, react with either the donor glycal or the acceptor glycal resulting in coincident symmetrical coupling and possibly polymerization. It had long been known that acyl protecting groups lower the reactivity of glycosy1 donors [gal, but the idea of exploiting this effect in the coupling of two potential glycosyl donors had not been conceptualized until the seminal experiments of Fraser-Reid [Sb]. Consequently, it was hoped that the nature of such a coupling could be controlled by the resident substituents on two different glycals [9]. Glycal6 has its oxygen atoms alkylated, while those of 7 are acylated. Consistent with expectations, glycal 6 proved to be more reactive towards the iodonium electrophile, thereby generating the active glycosyl donor selectively in the presence of acceptor 7 (Scheme 3). Glycal 7 has a free hydroxyl group and, therefore, can act as the only acceptor. Iodinative coupling of 6 and 7, via trans-diaxial attack, gave rise to iodoglycoside 8 (58%). The C3 hydroxyl, allylic to the double bond of the glycal, is the most obvious position where the electronic difference between and acyloxy and an alkoxy group might be manifested. However, the clean and successful coupling of 6 and 9 to give 10 (76%) indicates that even without an acyloxy group at C3, subtle effects can be responsible for an ordered glycosylation.

2

Scheme 2.

3

4

66

3 Iterative Assembly of Glycals and Glycal Derivatives

B BnOn

O

g

+

no

O

S

/

+

O

e

b -

I+ 58%

BZos HO

6

B BnOn

O

7

6

BB n

HBZO

9

e

+

8

10

-

BZO

11

Scheme 3.

A true test for this strategy of controlling glycal-glycal coupling arrived with the total synthesis of ciclamycin 0 [lo], which embodied the additional feature of containing 2-deoxyhexoses (Scheme 4) [l 11. Coupling of benzoylated acceptor 14 to benzylated donor 13, followed by Ph3SnH reduction of the resultant iodide gave the 2-deoxy-disaccharide 15 in 57% yield. Following protecting group adjustment, silylated donor 16 was coupled to the same benzoylated acceptor 14 followed again by Ph$nH reduction, in 54% overall yield. Manipulation of the resulting trisaccharide into a suitable glycal donor was followed by a successful coupling to Epyrromycinone (19), which gave a mixture of a-linked iodides 20 in 60% overall yield. Protecting group removal and iodide reduction afforded ciclamycin 0. Up to this time, the direct coupling of an antrhacyclinone to a fully synthetic oligosaccharide en route to a naturally occurring antibiotic had not been accomplished

WI. 3.3 Allosamidin Allosamidin (21, Figure 3) was identified as a selective and powerful chitinase inhibitor in 1987 [13]. As such, it generated interest as a potential insecticide and fungicide. Allosamidin is composed of a 3,3’-epi-chitobiose P-linked to a novel aglycone sector termed ‘allosamizoline’ [ 141. This unique structure is believed to impart to it the function of a transition-state mimic for the hydrolysis of chitin [15]. The possibility that the electrophilic nature of the oxazoline linkage could allow it

3.3 Allosamidin

>

-&

67

->

OH Bno

14

13

MePo --

1. I+(symcoll)*cIo,&

2. PhaSnH, 54%

BnO 16

15

14

BnO OH 0

Mw 0

I+(symcoll)2c104-

>

19

18

20

60% M0

e

d

o

Scheme 4.

to function as a potential locus for covalent attachment to the chitin-degrading enzyme was also considered [16]. The combination of the interesting structure, valuable biological activity and scarcity in nature of allosamidin prompted our attempts at its total synthesis. In the contlext of carbohydrate synthesis, a key feature of this effort included a novel and efficient synthesis of axial glycal derivatives utilizing a Ferrier-type rearrangement [ 171 followed by a [2,3]-sigmatropic rearrangement [18, 191. The allosamidin effort also spurred development of the method of sulfonamidoglycosylation [20] for the construction of P-linked 2-aminohexoses. Following the invention and

21 Allosamidin

Figure 3.

3 Iterative Assembly of Glycals and Glycal Derivatives

68

50% overall SPh 22

OPC

23

25

24

Scheme 5.

application of methods addressed to these ends, an expeditious synthesis of allosamidin was enabled. A general objective of the efforts in this area was an efficient route to glycals bearing C3 axial alcohol derivatives in the context of a range of substituents at C4. Glycals of this type had been difficult to synthesize [21] and rapid access to such derivatives would broaden the expanding collection of glycosyl donors and acceptors that could be incorporated into the assembly program. We found that Ferriertype rearrangement of tri-0-acetyl-D-gluca122 with thiophenol yielded 23 (Scheme 5) [22]. Oxidation of the a-(pheny1thio)pseudoglycal with DMDO and exposure of the resultant sulfoxide (24) to diethylamine triggered a [2,3]-sigmatropicrearrangement which resulted in the desired allal 25 in 96% yield. This sequence was expanded to include both allal and gulal derivatives as well as a range of substituents at C4. The issue of access to 2-aminohexose moieties has been impressively addressed by the azidonitration protocol of Lemiuex [23] and relies on the character of the C2 substituent to direct the outcome of any subsequent glycosylation. In this regard, azidonitration, while far from ideal in terms of stereoselection, remains one of the most useful methods to access the 2-amino-a-linkage. We, however, sought a new, direct method for introducing equivalent functionality through which stereoelectronic factors, rather than local steric factors, would determine the outcome of subsequent glycosylation. To this end, a key reaction, sulfonamidoglycosylation, was developed [22] (Scheme 6).

I(syn-Colli)zCi04

RRO o

> -HzNSO2P or h BrZNSOzPh

e

NH41

1

Scheme 6.

LTMP, AgOTf > -

RO

Ph02SNH 27 X = Br, I

26

28

R:q

29

30

3.4 KS-502 and Rebeccamycin

Bny% BnO

69

OSugar Ph02SNH 31

PhOzSNH

32 52%

34

64%

23%

Scheme 7.

This method, which involves trans diaxial addition of an N- halobenezesulfonamide to a glycal, leads to the 1-aminohexose 27. Alternatively, the action of IDCP (iodonium syn-collidine perchlorate) followed by benzenesulfonamide provides the same type of 2-halo-1-aminohexose. Treatment of 27 with base, likely proceeds through the 1,2-sulfonyloaziridine 28, which then reacts quickly and selectively with acceptors to form a P-linkage at the anomeric carbon. The scope of this general method was shown to include glycal- and nonglycal-containing carbohydrates as potential acceptors. Donors could include di- or higher saccharides terminating in a glycal linkage (Scheme 7). An additional benefit of this method was the finding that the intermediate halosulfonamides were often stable crystalline solids. The efficiency of coupling was later enhanced by the combination of stannylalkoxides in the presence of AgBF4 for the actual glycosylation reaction [24]. Furthermore, an iodosulfonamide can be readily converted to a corresponding ethyl thioglycoside, which in turn functions as a useful and alternative glycosyl donor. The adaptability of the sulfonamidoglycosylation to the iterative mode (glycals as donors and acceptors) made it particularly attractive for the synthesis of allosamidin (Scheme 8) [25]. The route to allosamidin utilized the sulfonamidoglycosylation for both glycosylation reactions, as well as the [2, 31 sigmatropic rearrangement methodology for both hexoses. Bromosulfonamidation of 36 (63%) followed by direct rollover with KHMDS in the presence of glycal 38 gave the P-linked disaccharide 39 in 81% yield. In turn, 39 then served as a donor in a second azaglycosylation with the aglycon derivative 41 to afford, after several steps, allosamidin (21). The efficiency of this synthesis served to reflect the power of iterative glycal assembly.

3.4 KS-502 and Rebeccamycin Glycosylated natural products, and other carbohydrate conjugates, present a challenging problem to synthetic chemists who must modify traditional carbohydrate

3 Iterative Assembly of Glycals and Glycal Derivatives

70

BnO

B n O KHMDS, 30%

-

-w

21 Allosarnidin

Scheme 8.

domain methodology to tolerate greater diversity present in the non-carbohydrate sector. In principle, incompatibility of synthetic methods to reach one domain can often limit options along a proposed route. KS-502 (43) and rebeccamycin (44) posed precisely this challenge and became the targets of our synthetic efforts in 1992-1 993. KS-502 (43) was isolated in 1989 [26] and identified as a specific inhibitor of calmodulin-dependent cyclic-nucleotide phosphodiesterase (CaM-PDE). The important roles of CaM and CaM-PDE in cultured cells and living systems underscored the utility of KS-502, perhaps as a valuable biological tool. Its structure comprises an interesting aryl furanoside for which new methodology was devised. Rebeccamycin (44)was shown to have antitumor activity against P388 leukemia, L1210 leukemia and B16 melanoma implanted in mice [27]. It is believed to impart this effect through the single strand cleavage of DNA. Water-soluble analogues Ho*

n-C7H 15

OH 42 KS 501 R = H 43 KS 502 R = C02H

Figure 4.

3.4 KS-502 and Rebeccamycin

71

have shown increased potency and are currently advancing through clinical trials [28]. The rebeccamycin structure consists of a symmetrical indolocarbazole chromophore P-N-linked to a 4-0-methylglucose residue. Synthetic efforts had concentrated on the aglycon portion of rebeccamycin, and an effective general methodology for the glycosylation step was sought to facilitate the total synthesis of this family of antitumor agents. While a unifying solution was pursued for these two glycoconjugate problems, total synthesis was not the sole inspiration for such efforts. Though sulfonamidoglycosylation provided routes from glycals to 2-aminoglycosides [ 14, 161, there was a more general need for a reliable and far-reaching route that would serve to convert glycals to common glycosides of glucose, galactose and mannose. The trichloroacetimidate donor methodology advanced by Schmidt [29], probably the single most powerful glycosylation method available, can often provide selective access to either a- or P-glycosides. We alternatively focussed on the application of glycal epoxides to tackle these and other problems. Such epoxides can participate in iterative strategies and can provide, in some circumstances, strikingly direct routes to polysaccharides. The existing record of a-epoxides as stereoselective glycosylating reagents was, at best, unimpressive when we launched our study. Access to such 1,2-anhydrosugar derivatives was awkward and their applications to glycosylations were beset by serious problems in stereoselectivity [30]. We first sought a direct and reliable route to 1,2-anhydrosugars. Furthermore, we hoped to gain advantages at the level of stereoselectivity by careful selection of resident functions. In particular, it was felt that resident groups bearing neighboring group participating functions, would tend to undercut clean formation of P-glycoside from a-epoxides. A major advance toward this goal came with the recognition that a variety of glycals react smoothly with 3,3-dimethyldioxirane (DMDO) [311. Glucal derivative 45 reacted with DMDO to give near quantitative yields of 46 and with >20: 1 aselectivity (Scheme 9). Galactal derivative 47 yielded the same result. In the allal series, glycal 49 underwent selective epoxidation instead from the P-face, while the gulal derivative 51 yielded a 1 : I mixture of epoxides. With these results in hand, the glycosyl donating abilities of these epoxides was explored (Scheme 10). Considerable experimentation indicated that ZnClz promoted selective p-glycosylation with alcohol acceptors, (e.g. 54, 55) whereas basic conditions promoted effective P-glycosylation only with moderately acidic acceptors (e.g. 56), such a s phenols. In some instances stannyl derived acceptors, conveniently generated in situ, afforded superior results. Subsequent to these investigations, it was found that oxirane 46 is actually among the poorest of donors [32], and that glycosylation yields could be improved by constraining the C3-C4 or C4-C6 oxygens into a cyclic motif. An interesting and useful variation on this procedure arises from the AgBF4 promoted reaction of an epoxide with a stannyl ether, which gives a-linked product. For example, treatment of 46 with 57 in the presence of AgBF4, followed by acetylation, cleanly affords 58 [33]. This method is presently limited to primary alcohol acceptors. These constraints underscore the need for a comprehensive solution to the challenging problem of stereospecifically generating a-glucosides.

3 Iterative Assembly of Glycals and Glycal Derivatives

72

0%

> -

>95%

45

0 % > -

TBSO

>95%

47

a:D

> 20:l

0% > -

>95%

TBSO

TBSO

49

Ph 0% > I

TBSO 51

>95%

50 > 2O:l P:a

?G

TBSO

52 1:l 8:a

Scheme 9.

R O *RO

OH

0-sugar

53

88%

42%

55

56

""d-""' /

Scheme 10.

67% (KzC03,18-C-6)

3.4 KS-502 and Rebeccamycin

73

Roao

1 .DMSO AqO

RO

pyridine F

RO 59

RO

61

60

Scheme 11.

The glycal epoxide method of glycosylation results in the placement of a free hydroxyl group at the 2-position of the hexose following coupling. This feature of the method enabled two alternative and useful advances (Scheme 11). First, the unique C2-hydroxyl group could be removed by the free-radical reduction of the derived pentafluorophenyl thionocarbonate (Barton deoxygenation). This sequence provides a synthesis for 2-deoxy-~-glycosides(60) [32, 341. Alternatively, the C-2 hydroxyl group could be inverted via an oxidation (AclO/DMSO)/reduction (NaBH4) protocol furnishing P-mannosides (61) [33, 351. Furthermore, the efficient and reliable route to a-glycal epoxides, facilitates numerous other methodologies leading to established donor functions (Scheme 12). Of particular importance is the fluoro-sugar type (cf. 66). Following benzylation of the uniquely free C2 hydroxyl group, derivatives such as 67 are produced. This method provides access to glycosyl donors ready for the powerful Mukaiyama promoter system [36]. We have often applied this protocol to allow for the stereoselective introduction of a-glycosides, as in 68.

62

63

BnNb

TBAF

<:

ZnCl2.71%

/

6

Bno BnO

NHB

OH 64

I

B\%c V

65

Scheme 12.

HX

X

Bno& BnO

AR'OH S & OR

66R=H 67 R = Bn

74

3 Iterative Assembly of Glycals and Glycal Derivatives " 5 ; H

nC5H11

184-6

H % H

q$(c!F?: En0

81% 69

"OH

15

,q

Rn6,H

OH

71

42 KS 501 R = H 43 KS 502 R = COfl

3 --L

48% 72

73

Rebeccamycin

Scheme 13.

The finding that the acceptor for a glycal epoxide can itself be a glycal established the basis for an iterative strategy for the synthesis of repeating p-glycosides. This concept has been a central component of our approach to oligosaccharide synthesis and has been adapted to solid support strategies as well [37]. The glycosidic bonds present in KS-502 and rebeccamycin provided especially attractive contexts for evaluation of the glycal epoxide method (Scheme 13). Treatment of the furanose glycal with DMDO gave the furanose epoxide 69 stereoselectively [38].Reaction of this epoxide with the phenol 70 afforded an 81% yield of the desired aryl-glycoside 71. Subsequent steps completed the synthesis of KS501 (42) and KS-502 (43). These furanoid epoxides were also used in the synthesis of nucleosides [39]. The extension of this procedure to the glycosylation of indoles was pursued in the context of rebeccamycin [40]. In this example, deprotonation of the differentiated indolocarbazole 72 followed by exposure of the resultant anion to glycal-epoxide 73 gave a 48% yield of the glycosylated product 74. Subsequent 2,2'-bis indole bond formation and deprotection yielded the glycoconjugate rebeccamycin (44).

3.5 Extension to Thioethyl Donors Galactal derived epoxy donors, such as 48, are particularly effective in enabling pgalactosylation. By contrast, glucal derived epoxides, lacking a comparable inter-

3.5 Extension to Thioethyl Donors

75

Lewis Acid

OH RO 75

77

RO 76

Scheme 14.

locking pattern between C3 and C4, are considerably more labile to typical Lewis acid promoters. This heightened instability brings with it significant limitations to the scope of their effectiveness as p-glycosyl donors. When the acceptor reactivity of the hydroxyl group in ROH is diminished, the yields of the glycosylation are compromised. In addition, the glycal epoxide method had not provided an efficient route to mannosides. For these reasons, we sought to extend the glycal epoxide method and turned to the use of thioethyl donors (75) for glycosylation and mannosylation (Scheme 14). Thioethyl and thioaryl glycosyl donors have previously been shown, in impressive studies by Garegg [41], to provide access to PI-4 glucosyl linkages in good yield. Donor activation is accomplished by either heavy metal salts or, directly and more efficiently, by other thiophilic reagents. The stereoselectivity of the glycosylation reaction is determined by the nature of the protecting group on C2. Indeed, it was found that glycals can be smoothly converted to thioethyl donors via glycal epoxides [42] (Scheme 15). The thioethyl glycosides thus produced, could be converted into the mannosides (75 + 85, Scheme 16). This overall inversion of configuration at C2 proceeded via an oxidation-reduction sequence to give, exclusively, the mannose derivative. The

EtSH cat. TFAA

B nO

71%

>95% 70

BnO& B nO

>95% w

BnOO 46

Scheme 15.

EtSH

-*cat. TFAA 78%

s OH . 81

EtSH cat. TFAA

03

BnO 6 BnO

U

-0TIPS

82

OH

80

79

E

45

BnO

-

U

84

76

3 Iterative Assembly of Glycals and GlycaI Derivatives 1. protect

R RO

asEt

O

OH

2. MeOTf DTBP 4A MS

75 1. ACZO DMSO pyridine 76 % 2. NaBH4

77

R

RO

O

MeOTf b -

F&SEt

85 R' = Ac, Bz, PIV

R O = L

S

* rR RO

DTBP 4A MS

a7

RORO -

Scheme 16.

resultant C2 hydroxyl groups can be protected with a variety of groups of known participatory capacity. The pivalate (Piv) derivatives were generally found to be the neighboring groups of choice for installing the desired linkage in that they provide very good margins of p-stereoselection without incurring problems of orthoester formation in competition with glycosylation (Scheme 17) [42].

3.5 Lewisy The antigenic blood group determinant LewisY (Ley) [43] is over-expressed in glycolipid and mucin glycoprotein form on many human tumor cells including those found in colon, lung, breast, and ovarian cancers. With this in mind, we set out to fashion vaccines based on the lipid and protein forms of Ley (Figure 5) [44]. To achieve these goals, we settled on the following target systems: (i) the natural product Ley-ceramide 94, (ii) a conjugate form for vaccination purposes 95, and a (iii) mucin mimic en route to a polypeptide-based vaccine (vide infra). The Ley tetrasaccharide determinant consists of a bis-fucosylated N-acetyl lactosamine. We reasoned that the iodosulfonamidation methodology could be used to install the amine function and would favor the subsequent incorporation of additional glycal units. It also seemed likely that both the Ley-ceramide target and the Ley-conjugate could be prepared from a common, but late, glycal epoxide intermediate. Concise construction of the Ley tetrasaccharide 98 commenced with lactal (96, Scheme 18) [44, 451. Protection of the primary alcohols, followed by engagement of the syn-vicinal diols as a cyclic carbonate, provided the lactal acceptor 97. Stereoselective introduction of the fucose moieties was achieved by applying a modifica-

77

3.5 LewisY MeOTf OTIPS B BnO n O

a

Brio*

DTBP

s E OPlV

t

OTIPS

OSugar

BnO

b *

OPlV

88

89

91

80%

BnO

MeOTf TIPSO n o % ,., BnO

4AMS DTBP ~

~

Bno

TIPSO BnO

93

92

0-Sugar

78%

91%

76%

Scheme 17.

tion of the original Mukaiyama conditions, using either 99 or 100. By starting with the glycal, the protecting group manipulations were minimal (2 steps). It remained however to convert the glycal function to the required lactosamine and to effect glycosylation to the pentasaccharide. In the event, 98 could be cleanly converted, by application of the iodosulfona-

OH

Figure 5.

94; R = Ceramide 95 R = Ally1

78

3 Iterative Assembly of Glycals and Glycal Derivatives

I

V

IDCP, 94% Oh

OBn

&

OB"

p

PhS02HN

102

75%

Scheme 18.

midation protocol, to 101 bearing the anomeric sulfonamide in excellent yield (94%). The latter underwent rearrangement to 102, using the 3-stannyl ether of 6Tips galactal, in 75% yield. This product contained the necessary lactosamine core linked in the desired p-configuration to the terminal galactal unit and constituted a protected Ley-pentasaccharide. After conversion of 102 to peracetate 103, the stage was set for divergent pathways to the ceramide (94) as well as to the ally1 glycoside (95). The later was the form we selected for conjugation to protein carriers, bovine serum albumin (BSA) and keyhole limpet hemocyanin (KLH). These were obtained from 103 upon treatment with DMDO followed by addition of the corresponding acceptors to the glycal epoxide. By application of iodosulfonamidation and the glycal epoxide methodologies, the targets were readily obtained. These syntheses paved the way for preclinical, and eventually clinical, investigations based on the Ley-pentasaccharide [46]. In these first generation vaccines, clinical evaluation of the immune response revealed that the antibodies displayed an apparent selective binding to Leyglycolipids, while relatively low antibody reactivity was noted with Ley-mucin pro-

3.5 Lewis”

79

teins. Implicitly, these immunological results posed the challenge of designing and synthesizing a closer mimic of a Ley-bearing m u c h glycoprotein. Before any effort to construct a mucin mimic could be launched, we were faced with the issue of glycoprotein-targeted vaccine design. We set out, therefore, to construct the simplest, and yet most realistic, molecules that would serve to direct the immune response against mucins. However, unlike Ley-ceramide, which is a discrete molecule, glyco-patterns in mucins may vary along the backbone and from one protein to the next. Hence, the uncertainties regarding glycoprotein structure render it quite difficult to design, with high confidence, molecules that will stimulate an immune response which targets mucin bearing cancer cells. The complexities of mucin structure notwithstanding, we proposed to synthesize several Ley glycopeptide-based vaccine candidates [45b, c]. To achieve this, we elected to incorporate the key common features of a mucin [43d, 471, including a primary structure characterized by sequentially arranged, and glycosylated, serine or threonine residues. The first carbohydrate of the glycodomain is a-0-linked GalNAc, and a number of intervening Gal and GlcNAc spacer units may insulate the core from the terminal region, which may bear, for instance, a blood group determinant like Ley. The glycopeptide domains of three of our mucin vaccines are shown in Figure 6 (104-106). Clustered pentasaccharide 104 satisfies the minimum requirements of the Ley tetrasaccharide attached to GalNAc-a-0-Ser. The GalNAc-P-0-Ser derivative 105, containing unnatural p-linkages, was also prepared. We wondered if such a ‘non-self’ structure might prove more immunogenic than 104 and still elicit antiLey antibodies. The hexasaccharide cluster, 106, has the natural GalNAc-a-0-Ser core as well as a spacer Gal between the Ley tetrasaccharide and the first carbohydrate and as such might be considered an even more realistic mimic of mucin structure.

B

R

Figure 6.

R

80

3 Iterative Assembly of Glycals and Glycal Derivatives

Having settled on a design, we returned to our original synthesis of Ley and identified the pentasaccharide 103 as a subgoal. If this pentasaccharide could be converted to the appropriate glycosyl amino acid bearing constructs, derivatives 104-106 should be accessible through standard solution phase peptide coupling. While iodosulfonamidation of glycals is often high yielding, and the adduct ready for subsequent glycosylation, removal of the 2-benzenesulfonamido group requires Birch (metal/ammonia) reduction conditions. This protocol was developed to accommodate the glycal moiety in the context of a complex oligosaccharide and achieves removal of benzyl groups without recourse to hydrogenation. However, dissolving metal reduction is not compatible with the base- and acid-sensitive glycoamino acid function. Thus, the iodosulfonamidation method could not be appropriated for the synthesis of Ley pentasaccharide bearing glycopeptides. To gain access to 104 and 105, pentasaccharide 103 would have to be transformed to a suitable donor substrate. The latter would be combined with the amino acid core for eventual peptide coupling (Scheme 19). In practice pentasaccharide 103 was subjected to Lemeiux azidonitration [48].This transformation was followed by reduction of the nitrate ester, and conversion of the resultant anomeric alcohol to the Schmidt trichloroacetimidate [49]. Donor 107, thus prepared, was ready for coupling with a serine acceptor. Compound 107 and FmocSerOBn (108), under promotion by TMSOTf, afforded the desired O-linked glycoamino acid, albeit in a ratio of 2.7: 1, a(109):b(ll0). After considerable experimentation, it was found that by altering the solvent composition, the relative ratio could be reversed to 1 : 1.7, a(109) :P(ll0). Thus both building blocks necessary for the construction of 104 and 105 were obtained from a common donor. Further transformations afforded the acids 111 and 112, respectively. These products were ready now for peptide coupling. In several instances, as illustrated above, glycosylation of elaborated 2-azido galactosyl donors seemed to be dependent upon the nature of the donor and the identity of the serine or threonine acceptor. In this context an acceptor such as 113 was preferred (Scheme 20). Assuming pentasaccharide 103 could be converted into a suitable donor, the glycoamino acid required for assembling hexasaccharide cluster 106 could be obtained by direct glycosylation with acceptor 113 [50]. To prepare 119, pentasaccharide glycal 103 was subjected to the action of DMDO, generating the glycal-epoxide donor. However, under normal promotion conditions acceptor 113 failed to couple with the donor epoxide. In order to enhance the reactivity of the donor, the epoxide was converted to the thioethyl donor 114, which, after protecting the newly generated 2-hydroxyl group as the benzoate (116), was coupled with 113 under MeOTf promotion. The glycosylation did proceed, but in low yield and was accompanied by N-methylation of the acetamide functionality, affording both the desired 118, as well as, the undesired methyl imidate. Thus, methyl triflate promotion of thioethyl donors, found to be fully compatible with glycal acceptors, becomes complicated in the presence of acetamides and this difficulty appears to be general. Though the imidate, formed as byproduct in the methyl triflate reaction, could be hydrolyzed back to the acetamide, other more direct methods were pursued. In this regard, it was found that in the presence of NIS/TfOH [51], the desired coupling was achieved in minutes and in excellent yield. Following further optimization, the preferred route to 118 is as follows. Pen-

3.5 LewisY

81

1. CAN, Nab, (66%) b

103

OCNHCCI,

2. PhSH, iP?!NEt (92%)

3. CbCCN. I(,CO,

(89%) 107

> 75%, 1:1 .7 109:110

75%#,2.7~1

WFmcc 108

0 109

AcHN

NiFmoc

OAc

,&&

OAc d 0 r A c

0 OAc

112

ACOoAc

Scheme 19.

tasaccharide glycal 103 was subjected to the action of DMDO and subsequently converted to the pentenyl donor 115 [52]. The 2-hydroxy glycoside was then protected as the benzoate. In the event, NIS/TfOH-promoted coupling of donor 117 with acceptor 113 cleanly afforded the desired serine-O-linked hexasaccharide 118. After manipulating the protecting groups, the glycoamino acid 119 was obtained. Sequential peptide coupling of 111, 112, and 119 provided the glycopeptide domains 104--106 and eventually several vaccine candidates. It was gratifying in this regard to find that preclinical immunological evaluation demonstrated that we had indeed constructed a vaccine which provokes an antibody response targeting

82

3 Iterative Assembly of Glycals and Glycal Derivatives

oac Ho

oms

-%J--.

1. DMDO

103

2. HSEI or 4-penlen-1-01

AcHN

'NHF~~:

113

116 w117

f

3

114: R ,

= SEI. & = H

115: R ,

= 0-4-Penteny1, R,= H

BzCl

(45% from 103)

116 RI = SEt. R2 = Bz

1

117 R1 = 0-4-Penteny1, R, = Bz

Scheme 20.

Ley-mucin bearing cells. Clinical evaluation of an anti-cancer vaccine based on 106 is currently well underway.

3.7 Globo H The hexasaccharide designated globo H (120, Figure 7) has been identified on human prostate, breast and small cell lung carcinomas, as well as in a restricted number of normal epithelial tissues [53]. Globo H was originally isolated in ceramide form from the human breast cancer cell-line MCF-7 1541. The enhanced expression of antigens assumed to be globo H on both primary and metastatic prostate cancer specimens drew us to the possibility of initiating a globo H-based vaccine strategy in prostate and other cancers 1551. At the planning stage of the synthesis 1561 it seemed reasonable that, similar to the Ley synthesis, late stage divergence from a hexasaccharide glycal epoxide could pro-

83

3.7 Globo H

H type 111 trisaccharide

Globotriose

120: R = Ceramide 121: R = Ally1

Figure 7.

vide both the ceramide form and the ally1 glycoside (121) for immuno-conjugation to KLH. As i1.s name indicates, globo H contains globotriose and an H type I11 module. The chemical union of two trisaccharides, a globo trisaccharide acceptor with an H type donor, would constitute a highly convergent approach this complex target. The galactosamine unit constituted added complexity, since iodosulfonamidation had not been attempted on a galactal unit. Additionally, a final 3 3 coupling using a galactal donor derived from an iodosulfonamide could serve to provide a dramatic demonstration of the method. The galactosyl units, including the galactosamine, might be derived from a common galactal building block. Similarly, the glucose unit of the lactosyl ceramide would come from glucal, while the terminal fucose could be installed at any convenient stage in the synthesis. Compound 122 serves as an excellent acceptor, due in part to its allylic nature, and does not suffer from competing glycosylation at the axial 4-position (see 123, Scheme 21) [56].In keeping with our earlier precedents, conversion of 122 to cyclic carbonate 124 followed by treatment of 124 with DMDO generates the expected glycal epoxide 125. This intermediate was coupled with 122 and 127 to give disaccharide building blocks 126 and 128 respectively. Disaccharide 125 was carried forward without protecting group manipulations and was fucosylated under Mukaiyama conditions (125 + 130, Scheme 22). This step required selective functionalization of the unreactive C2'-hydroxyl over the,

+

(123) (not observed)

Scheme 21.

126

128

84 0

3 Iterative Assembly of Glycals and Glycal Derivatives 0

OTPS

OH 0T1p

-

AgC104, SnCl2

\ k k & O e OH 125

130: R = H

48%

131: R = Protecting Group

BnO BnO

129

Scheme 22.

presumably less reactive, axial C6hydroxyl. In the event, under optimized conditions, the desired product was obtained in 48% yield. The undesired regioisomer was a minor byproduct. Still, this sequence provided the most direct route to the target trisaccharide glycal (a total of four steps from the common precursor, 6-Tips galactal). Protection of the lone hydroxyl was achieved in a straightforward manner (130 -+ 131). As will be shown, an even more serious unanticipated turn of events arose in attempting to use this glycal as a donor. The acceptor for the projected 3 + 3 coupling was prepared from fluoro-donor 134 and lactal 132 (Scheme 23). Compound 127 was readily converted to lactal acceptor 132, while 134 was prepared from selectively protected galactal 133. It seemed prudent to convert 133 to the more reactive 0-derivative 134 by direct opening of the epoxide with TBAF, followed by protection of the resultant hydroxyl. Mukaiyama coupling of 132 with 134 proved troublesome. While the yield was high (72%), the major impediment was selectivity ( a :p, 3: 1). The desired

I33

I27

OH OBn

o&

BnO-

Bno Bm 134

135

132

AgC104, SnCI2 132

+ I34 M 135

(88%, l O : l , a:p)

136: R = PMB 137: R = H

Scheme 23.

3.7 Globo H

85

product was obtained in only about 50% yield. Furthermore, as the reaction proceeded the reactive p-derivative 134 was slowly converted into the unreactive aderivative 135. Eventually, application, with slight modification, of the Suzuki [ 571 promotion procedure, using the superiluorophile CplZr(OTf)2, provided the desired product in 10: 1 ratio in an overall yield of >80% [ 5 8 ] . In fact, this promoter allowed for smooth conversion of either 134 or 135 to the desired product. This finding simplified the sequence considerably, in that treatment of 133 with DMDO could be followed by HF:pyridine effecting clean and rapid conversion to a-derivative 135. After removing the protecting group from 136, the acceptor trisaccharide (137) was ready for the key glycosylation reaction. As depicted in Scheme 24, compound 131 underwent smooth conversion to iodosulfonamide 138 in good yield. However, in practice, 138 failed to couple with trisaccharide 137. Undaunted we pursued the complementary route. Thus iodosulfonamide 138 was converted to the thioethyl donor 139. The 3 3 coupling of 139 and 137 was effected using methyl triflate in excellent yield, particularly considering the sophistication of the j’oined partners. A small quantity of an isomeric hexasaccharide was obtained. At the stage of hexasaccharide (140) it was difficult to rigorously assign the structure of the product, and it was primarily precedence upon which we base the structural assignment. Only after eventual conversion of oligosaccharide 140 to the deprotectetl ceramide, 141, did we realize that in the 3 + 3 coupling gave the undesired a-glycoside. This unpredicted turn of events is actually rather general [591. It was found that even monsaccharides consisting of 2-sulfonamide thioethyl galactosides resulted in glycosylation with poor selectivity. In the case at hand, modifying the protecting group on the C4-hydroxyl (see ‘R’ in 139) failed to reverse selectivity. Eventually, we would conduct the key glycosylation without any protecting group on the C4-position. However, C4 deprotection of 139 to give 143 (see Schemes 24 and 25) proved to be no simple matter, regardless of the identity of the protecting group. After considerable experimentation, it appeared as if the iodosulfonamidation and the ‘roll over’ reaction would have to be performed on the unprotected glycal 130.

+

131

IDCP L (82%. PhSqNH,) W

O

B

BnO OBn

13’,

Hexasaccharlde

139 b

60%

140

... Scheme 24.

UH

n 139

86

3 Iterative Assembly of Glycals and Glycal Derivatives 0 OTIpS 130

IDCP, PhSO N 2

50%

2

Oxo&-

p

LHMDS, HSEt L

PhS0,NH

79x

oy*o& p

om SEt !‘hSO,NH

143

I43

137, MeOTf L

70 - 85%

L

L

I20 and121

Scheme 25.

While this approach did eventually succeed, the path was not traversed without further obstacles (Scheme 25). Treatment of 130 with IDCP gave the desired iodosulfonamide. However, unlike similar derivatives, the immediate product was not stable to purification or storage and isomerized to an unproductive stereoisomeric donor. Usually, instead of attempting to purify the intermediate, 142 was taken on to the ‘roll over’ step providing the desired donor 143. Remarkably, methyl triflate promoted coupling of 143 with 137 gave the desired product in good yield. This compound was accompanied by a small amount of an undesired isomer. That the major product was indeed hexasaccharide 144 was demonstrated ultimately by its conversion to the natural product. In the end, 144 was converted to a peracetate glycal, which was converted to globo H-ceramide 120 and the ally1 glycoside 121. Compound 121 was taken on to vaccination trials in the form of a KLH conjugate and continues to be studied in various clinical settings with encouraging early results. While hard won, the lessons garnered in the globo H venture paid large dividends, in that a number of related carbohydrates were accessible using ‘unprotected 4-hydroxy’ donors similar to 143. Thus, asialoGM1 [59], GM1 [60], and the demanding fucosyl GM1 [61] have each been prepared for study. In retrospect, globo H was synthesized by implementing most of the glycal methods available, including glycal donors, acceptors, epoxides, and fluorides, as well as by extension of the iodosulfonamidation protocols.

3.8 KH-1 KH-1 (145, Figure 8) is one of the most formidable carbohydrate-based tumor antigens characterized to date [ 621. This nonasaccharide has been identified on the cell surfaces of all human adenocarcinoma cells thus far studied. Furthermore, its pres-

3.8 KH-1

P

AcHN

LeY

OH

AcHN

LeX

OH

87

OH

Lactose

145: R = Ceramide 146: R = Ally1

Figure 8.

ence has never been detected in normal colonic extracts. In keeping with our program goals we targeted both natural ceramide (145) and ally1 glycoside (146) forms for total synthesis [63]. Structurally. KH-1 can be viewed as a lactosyl ceramide bearing the Lewis' (Le') trisaccharide upon which is mounted the Ley tetrasaccharide [62]. In the synthetic planning stage, however, we set, as the cardinal strategic objective, maximum relief from blocking group manipulations. In an effort to implement a highly convergent route, we hoped to introduce all three fucosyl units in a single operation. The subgoal would then consist of assembling, in an iterative fashion, glycal disaccharides to form the hexasaccharide core. Importantly, the three oxygen centers to be fucosylated would be formulated such that a single deprotection maneuver would serve to unveil them concurrently. Pursuit of the target began with glycal building blocks 147-149, and from these three glycals, and fucosyl donor 150, the entire molecule would be constructed (Scheme 26). TES ether protection, for example resident in 151, was chosen as the mask for sites of eventual fucosylation. Thus, treatment of 147 with DMDO cleanly afforded the a-epoxide, to which the acceptors 148 and 149 could be added, in the usual way, in a parallel sequence giving glycals 151 and 152. Reiteration of glycal assembly was attempted after independently applying either the acetate (153) or an additional TES ether (154) to the newly generated hydroxyl of 151. Compound 152 was destined to house the lactose bearing the ceramide unit and was therefore converted to trio1 155. While 155 has several potential glycosylation sites, in principle, the C3' hydroxyl is the most reactive position. A suitable donor could be fashioned by subjecting 153, the core of the Le' sector, to IDCP iodosulfonamidation (see 156). Addition of ethanethiolate effected the roll over reaction and the donor was ready for coupling with 155. In the event, treatment of 157 with 155 under promotion of MeOTf did, in fact, give the desired tetrasaccharide 158, albeit often in low yield (35-55%). In addition, the reaction mixture appeared to contain significant quantities of at least one regioisomer.

88

3 Iterative Assembly of Glycals and Glycal Derivatives

\KO&

?go& m, TESO

OH En0

\,

151

KzC03

80%

I,,":

c

c

LHMDS. HSEt OYKo%

-*

TESO TESO PhS0,HN I60

PhS02HN 156

I59 R=SnBu,

oy&q Eso

Aco

SEl PhSO,HN

91%

1

I 159, AQEF,

V

I55 R = H

81%

/

157

155, MeOTf

35 - 55%

Scheme 26.

We wondered if the donor, one of our most activated, might be so reactive as to be insensitive to the subtle differences in the other hydroxyl units of the acceptor. To test this hypothesis, we attempted the union of 156 with the tin ether of the triol, 159. Direct 'roll over' proved to be quite slow, but very sensitive to the differences in hydroxyl reactivity in the acceptor. This chemistry constituted a very pleasing and attractive solution to the assembly of 158. The next plateau to be reached was seen to be that of the full hexasaccharide core (Scheme 26). In an effort to maintain the tempo of the synthesis, we proceeded on to another assembly iteration with minimal protecting group-based transformations. The use of a thioethyl donor was explored; however, in the end the direct roll over route proved more reproducible. To this end, iodosulfonamidation of 154 afforded 160. While presumably even more demanding than the earlier transformation, we opted to attempt the selective coupling of the tin ether of pentaol 161 with 160 (Scheme 27). This acceptor was readily available from the tetrasaccharide by simple removal of the cyclic carbonate (158 +.161). The reaction between 160 and 162 was performed, as before, in the form of the tin ether acceptor. Fortunately, the required

3.8 KH-1

89

161 R = H

162 R = SnBu,

163 R=H.R,=TES 164 R = A c . R ( = H

\\,

Scheme 27.

coupling could be executed in good yield (>60% of 163) and without detectable formation of side product arising from glycosylation at alternative acceptor sites. The culminating stage of the synthesis was confronted after protecting the four remaining hydroxyls and removing the silyl ether protection (163 + 164). The critical step, threefold fucosylation, was indeed accomplished through the reaction of 150 and 164. An exemplary yield (60%) of the nonasaccharide 165 was obtained. In the final steps, direct epoxide opening with ally1 alcohol proved optimal for preparing 146. Our alternative procedure was used to install the ceramide. Thus, thiolation of the epoxide generated from the reaction of 165 with DMDO. This was followed by protection of the resultant free hydroxyl to give donor 166 ready for coupling with a suitable acceptor. Following this sequence, the glycal was successfully taken on to the natural product (145). Conjugation of 146 to KLH provided a vaccine and pre-clinical immunological testing is under way. Application of iterative glycal methodologies requires glycal compatible promotion systems, and these promoters are now available. The ability to attenuate donor reactivity with the acceptor is frequently desirable, especially in the search for optimized routes, and was used to effect in the case at hand. By applying the full complement of glycal technology the target was obtained in a highly convergent and efficient way. In summary, the synthesis of KH-1 serves as a cogent demonstration

90

3 Iterative Assembly of Glycals and Glycal Derivatives

of glycal assembly and the minimized protecting group manipulations that this strategy can provide.

3.9 Concluding Remarks As shown above the logic of glycal assembly has been field tested in a variety of different contexts. While our focus here has been on the total syntheses themselves, clearly, much has been learned about the chemistry of glycals and their amenability to iterative coupling with suitable acceptors and donors. At this stage, with the required compounds in hand our attention is directed toward vaccine performance, with respect to eliciting active immunity against various cancers.

Acknowledgments

This work was supported by the National Institutes of Health (Grant Numbers: CA28824, A116943, HL25848). Postdoctoral Fellowship Support is gratefully acknowledged by L.J.W. (Grant Number: NIH: F32CA79120) and R.M.G. (NIH F32CA85894). We gratefully acknowledge Dr. George Sukenick of the SloanKettering Institute’s NMR Core Facility (Grant Number: CA-08748) for NMR and Mass Spectral Analyses. References 1. S. J. Danishefsky, M. T. Bilodeau, Angew. Chem. Int. Ed. 1996, 35, 1380-1419. 2. V. Q. G. De Lima, C. A. Albert, 0. G. De Lima, An. Acad. Bras. Cienc. 1964,36, 317. 3. J. J. Afora, C. F. Santana, 0. G. De Lima, Ann. CVIIth Int. Congr. of Hematology and Hemotherapy, Paris, 1978, p. 165. 4. (a) 0. G. Da Lima, F. D. Monache, I. L. D’Albuquerque, G. B. Marini-Bettolo, Tetrahedron Lett. 1969,471; (b) L. W. Bieber, A. A. Da Silva Filho, J. F. De Mello; W. Von Der Saal, 0. G. De Lima, Rev. Znst. Antibiot. (Recife) 1982-1983, 21, 27. 5. T. Oki, Anthracycline Antibiotics [Pap,. Symp. On Anthracylines, Aug 24-25,1981, New York] Khakeme, E. Ed.; Academic Press: New York, 1982; p. 75. 6. (a) R. U. Lemiew, S. Levine, Can. J. Chem. 1964,42, 1473; (b) R. U. Lemieux, A. R. Morgan, ibid 1965, 43, 2190. 7. (a) J. Thiem, H. Karl, J. Schwentner, Synthesis 1978, 696; (b) J. Thiem, H. Karl, Tetrahedron Lett. 1978, 4999; (c) J. Thiem, P. J. Ossowowski, Carbohydr. Chem. 1984, 3, 287; (d) J. Thiem, A. Prahst, I. Lundt, Justus Liebigs Ann. Chim. 1986, 1044; (e) J. Thiem, W. Klaffke, J. Org. Chem. 1989, 54, 2006; (f) J. Thiem, Trends in Synthetic Carbohydrate Chemistry, D. Horton, L. D. Hawkins, G. L. McGarvey (Eds.), ACS Symposium Series #396, American Chemical Society, Washington, D. C., 1989, Ch. 8. 8. (a) H. Paulsen, Angew. Chem. Int. Ed. Engl. 1982,21, 155; (b) D. R. Mootoo, P. Kondradsson, U. Udodong, B. Fraser-Reid, J. Am. Chem. SOC.1988,110, 5583. 9. (a) R. W. Friesen, S. J. Danishefsky, J. Am. Chem. SOC.1989,111, 6656; (b) R. W. Friesen, S. J. Danishefsky, Tetrahedron 1990, 46, 103. 10. K. Suzuki, R. Friesen, G. Sulikowski and S.J. Danishefsky, J. Am. Chem. SOC. 1990,112,8895.

References

91

11. J. Gervay and S.J. Danishefsky, J. Org. Chem. 1991, 56, 5448. 12. For an even more concise synthesis of ciclamycin 0, see: S. Raghavan, D. Kahne, J. Am. Chem. Sac. 1993, 11'5,1580. 13. (a) D. Koga: A. Isogai, S. Sakuda, S. Matsumoto, A. Suzuki, S. Kimura, A. Ide, Agric. Biol. Chem. 1987, 51, 471; (b) S. Sakuda, A. Isogai, S. Matsumoto, A. Suzuki, J. Antibiot. 1987, 40, 296. 14. B. M. Trost, D. L. Van Vranken, J. Am. Chem. SOC.1990, 112, 1261. 15. M. L. Sinnot, Chem. Rev. 1990, 90, 1171. 16. (a) T. Iversen, D. R. Bundle, J. Chem. SOC.Chem. Commun. 1981, 1240; (b) L. J. Mathias, Synthesis 1979, 561; (c) L. A. Clizbe, L. E. Overman, Org. Synth. 1978, 58, 4. 17. R. J. Ferrier, N. Prasadi, J. Chem. Soc. C 1969, 570. 18. (a) P. Bickart, F. W. Carson, J. Jacobus, E. G. Miller, K. Mislow, J. Am. Chem. Soc. 1968, 90, 4869; (b) R. Tang, K. Mislow, J. Am. Chem. Soc. 1970, 92, 2100. 19. (a) D. A. Evans, G. C. Andrews, C. L. Sims, J. Am. Chem. SOC.1971, 93, 4956; (b) D. A. Evans, C. A. Bryan, C. L. Sims, J. Am. Chem. SOC.1972, 94, 2891; (c) D. A. Evans, G. C. Andrews, Arc. Chem. Res. 1974, 7, 147. 20. D. A. Griffith, S. J. Danishefsky, J. Am. Chem. Soc. 1990, 112, 5811. 21. S. J. Danishefsky, Chemtracts Org. Chern., 1989,2, 273. 22. M. D. Wittman, R. L. Halcomb, S. J. Danishefsky, J. Org. Chem. 1990, 55, 1979. 23. R. U. Lemieux, T. L. Nagabhushan, Can. J. Chem. 1968,46,401. 24. S. J. Danishefsky, K. Koseki, D. A. Griffith, J. Gervay, J. M. Peterson, F. E. McDonald, and T. Oriyama, J. Am. Chem. SOC.1992, 114, 8329. 25. D. A. Griffith and S. J. Danishefsky, J. Am. Chem. Soc. 1991, 113, 5863. 26. Isolation: (a) S. Nakanishi, K. Ando, I. Kawamoto, W. Kase, J. Antibiot. 1989, 42, 1049; Structure determinaton: (b) T. Yasuzawa, Y. Saitoh, H. Sano, J. Antibiot. 1990, 43, 336. 27. J. A. Bush, B. H. Long, J. J. Catino, W. T. Bradner, K. Tomita, J. Antibiot. 1987, 40, 668. 28. (a) T. Kaneko, H. Wong., J. Utzig, T. W. Doyle, J. Antibiot. 1990, 43, 125; (b) A dibromo analogue of rebeccamycin has been prepared K. Sing Lam, D. R. Schroeder, J. M. Veitch, J. A. Matson, S. Forenza, J. Antibiot. 1991, 44, 934. 29. R. R. Schmtdt, Angew. Chem., Znt. Ed. Enyl. 1986, 25, 212. 30. (a) W. J. Hickinbottom, J. Chem. SOC.1928, 3140; (b) E. Hardegger, J. de Pascual, Helv. Chim. Acta 1948, 31, 281; (c) L. L. Klein, W. W. McWhorter, Jr. S. S. KO, K.-P. Pfaff, Y. Kishi, J. Am. Chern. SOC. 1982, 104, 7362; (d) V. Bellosta, S. Czernecki, J. Chem. Soc. Chem. Commun. 1989, 199. 31. S. J. Danishefsky, R. L. Halcomb, J. Am. Chem. Soc. 1989, 111, 6661. 32. (a) C. M. Timmers, G. A. van der Marel, J. H. van Boom, Red. Trau. Chim. Pay-Bus 1993, 112, 609; (b) Ibid. Chem. Eur. J. 1995, I , 159. 33. K. K.-C. Liu, S. J. Danishefsky, J. Org. Chem. 1994,59, 1892. 34. (a) A direct method for the formation of 2-deoxy-P-glycosides from glycals employing Ph3P.HBr has been developed: V. Bolitt, C. Mioskowski, S.-G. Lee, J. R. Falck, J. Org. Chem. 1990, 55, 5812; (b) D. H. R. Barton, J. C. Jaszberepyi, Tetrahedron Lett. 1989, 30, 2619. 35. H. B. Boren, G. Ekborg, K. Eklind, P. J. Garegg, A. Pilotti, C.-G. Swahn, Acta. Chem. Scand. 1973,27, 2639. 36. T. Mukaiyama, Y. Murai, S. Shoda, Chem. Lett. 1981, 431. 37. (a) J.T. Randolph and S.J. Danishefsky, Angew. Chem. Int. Ed. 1994, 33, 1470; (b) K. F. McClure, J. T. Randolph, R. Ruggeri, S. J. Danishefsky, Science 1993, 260, 1307; (c) S. J. Danishefsky, J. Y. Roberge, Pure Appl. Chem. 1995,67,-1647;(d) J. Y. Roberge, X. Beebe, S. J. Danishefsky, Science 1995,269, 202. 38. R. G. Dushin, S. J. Danishefsky, J. Am. Chem. Soc. 1991, 114, 655. 39. K. Chow, S. J. Danishefsky, J. Org. Chem. 1990, 55, 4211. 40. M. Gallant, J. T. Link, S. J. Danishefsky, J. Org. Chem. 1993, 58, 343. 41. P. Fugedi, P. J. Garegg, H. Lonn, T. Norberg, Glyccoonj. J. 1987, 4, 97. 42. P. Seeberger, M. Eckhardt, C. Gutteridge and S.J. Danishefsky, J. Am. Chem. SOC.1997, 119, 10064. 43. (a) K. 0. Lloyd, E. A. Kabat, E. J. Layug, F. Gruezo, Biochem. 1966, 5, 1489; (b) T. Kaizu, S. B. Levery, E. Nudelman, R. E. Stenkamp, S. Hakomori, J. Biol. Chem. 1986, 261, 11254; (c) S. B. Levery, E. Nudelman, N. H. Anderson, S. Hakomori, Carbohydr. Res. 1986, 151, 311;

92

3 Iterative Assembly of Glycals and Glycal Derivatives

(d) B. W. T. Yin, C. L. Finstad, K. Kitamura, M. J. Federici, M. Welshinger, V. Kudryashov, W. J. Hoskins, S. Welt, K. 0. Lloyd, Int. J. Cancer 1996, 65, 406. 44. V. Behar, S. J. Danishfefsky, Angew. Chem. Int. Ed. Engl. 1994, 33, 1468. 45. (a) S. J. Danishefsky, V. Behar, J. T. Randolph, J. Am. Chem. SOC.1995, 117, 5701; (b) P. W. Glunz, S. Hintermann, J. B. Schwarz, S. D. Kuduk, X.-T. Chen, L. J. Williams, D. Sames, S. J. Danishefsky, V. Kudryashov, K. 0. Lloyd, J. Am. Chem. SOC.1999, 121, 10636; (c) Glunz, P. et al. Unpublished Results. 46. (a) V. Kudryashov, H. M. Kim, G. Ragupathi, S. J. Danishefsky, P. 0. Livingston, K. 0. Lloyd, Cancer Imunol. Immunother. 1998, 45, 281; Clinical data: (b) P. Sabbatini, V. Kudryashov, S. Danishefsky, P.O. Livingston, G. Ragupathi, W. Bornmann, M. Spassova, A. Zatorski, D. Spriggs, C. Aghajanian, S. Soignet, M. Peyton, C. O’Flaherty, J. Curtin, K.O. Lloyd, International Journal of Cancer, In Press. 47. (a) I. Brockhausen, in Glycoproteins, Montreuil, J., Schachter, H., Vliegenthart, J.F.G. Eds.; Elsevier: Netherlands 1995; pp. 201-259; (b) K. 0. Lloyd, J. Burchell, V. Kudryashov, B. W. T. Yin, J. T. Taylor-Papadmitrou, J. Biol. Chem. 1996,271, 33325; (c) S. Muller, S. Goletz, N. Packer, A. Gooley, A. M. Lawson, F. G. Hanisch, J. Biol. Chem. 1997, 272, 24780; (e) P. M. Rudd, R. A. Dwek, Crit. Rev. Biochem. Mol. Bio. 1997, 32, I ; (f) I. Carlstedt, J. R. Davies, Biochem. SOC.Trans. 1997,25, 214; (g) P. van den Steen, P. M. Rudd, R. A. Dwek, G. Opdenakker, Crit. Rev. Biochem. Mol. Bio. 1998, 33, 151. 48. R. Lemieux, R. M. Ratcliffe, Can. J. Chem. 1979, 57, 1244. 49. R. R. Schmidt, Angew. Chem. Int. Ed. 1986, 25, 212; R. R. Schmidt, W. Kinzy, Adv. Carbohydr. Chem. Biochem. 1994, 50, 84. 50. (a) X.-T. Chen, D. Sames, S. J. Danishefsky, J. Am. Chem. SOC.1999, 121, 10636; (b) J. B. Schwarz, S. D. Kuduk, X.-T. Chen, D. Sames, P. W. Glunz, S. J. Danishefsky, J. Am. Chem. SOC.1999,121, 2662. 51. (a) G. H. Veeneman, S. H. van Leeuwen, J. H. van Boom, Tetrahedron Lett. 1990, 31, 1331; (b) P. Konradsson, U. E. Udodong, B. Fraser-Reid, Tetrahedron Lett. 1990, 31, 4313. 52. D. Gordon, S. J. Danishefsky, Curb. Rex 1990,206, 361. 53. (a) P. 0. Livingston, Cancer Biol. 1995, 6, 357; (b) S. Zhang, C. Cordon-Cardo, H. S. Zhang, V. E. Reuter, S. Adluri, W. B. Hamilton, K. 0. Lloyd, P. 0. Livingston, Int. J. Cancer 1997, 73, 42; (c) S. Zhang, H. S. Zhang, V. E. Reuter, S. F. Slovin, H. I. Scher, P. Livingston, Clin. Cancer Res. 1998, 4 , 295. 54. (a) R. Kannagi, S. B. Levery, F. Ishijamik, S. Hakomori, L. H. Schevinsky, B. B. Knowles, D. Solter, J. Biol. Chem. 1983, 258, 8934; (b) E. G. Bremer, S. B. Levery, S. Sonnino, R. Ghidoni, S. Canevari, R. Kannagi, S. Hakomori, J. Biol. Chem. 1984,259, 14773. 55. (a) G. Ragupathi, S. F. Slovin, S. Adluri, D. Sames, I. J. Kim, H. M. Kim, M. Spassova, W. G. Bornmann, K. 0. Lloyd, H. I. Scher, P. 0. Livingston, S. J. Danishefsky, Angew. Chem. Int. Ed. 1999, 38, 563; (b) S. F. Slovin, G. Ragupathi, S. Adluri, G. Ungers, K. Terry, S. Kim, M. Spassova, W. G. Bornmann, M. Fazzari, L. Dantis, K. Olkiewicz, K. 0. Lloyd, P. 0. Livingston, s. J. Danishefsky, H. I. Scher, Proc. Natl. Acad. Sci. USA 1999,96, 5710; (c) Z.-G. Wang, L. J. Williams, X.-F. Zhang, A. Zatorski, V. Kudryashov, G. Ragupathi, M. Spassova, W. Bornmann, S. F. Slovin, H. I. Scher, P. 0. Livingston, K. 0. Lloyd, S. J. Danishefsky, Proc. Nat. Acad. Sci., 2000, 97, 2719. 56. T. K. Park, I. J. Kim, S. Hu, M. T. Bilodeau, J. T. Randolph, 0. Kwon, S. J. Danishefsky, J. Amer. Chem. SOC.1996, 118, 11488. 57. (a) K. Suzuki, H. Maeta, T. Suzuki Tetrahedron Lett. 1989, 30, 6879; (b) T. Matsumoto, H. Maeta, K. Suzuki, G. Tsuchihashi, G. Tetrahedron Lett. 1988, 29, 3567; (c) K. Suzuki, H. Maeta, T. Matsumoto, G. Tsuchihashi Tetrahedron Lett. 1988,29, 3571. 58. J. R. Allen, J. G. Allen, X.-F. Zhang, L. J. Williams, A. Zatorski, S. J. Danishefsky Eur. J. Chem. 2000, 1366. 59. 0. Kwon, S. J. Danishefsky, J. Am. Chem. SOC.1998,120, 1588. 60. S. K. Bhattacharya, S. J. Danishefsky, J. Org. Chem. 2000, 65, 144. 61. J. Allen, S. J. Danishefsky, J. Am. Chem. SOC.1999, 121, 10875. 62. E. Nudelman, S. B. Levery, T. Kaizu, S . 4 . Hakomori, J. Biol. Chem. 1986,261, 11247. 63. (a) P. P. Deshpande, S. J. Danishefsky, Nature 1997, 387, 164; (b) P. P. Deshpande, H. M. Kim, A. Zatorski, T.-K. Park, G. Rapgupathi, P. 0. Livingston, D. Live, S. J. Danishefsky, J. Am. Chem. SOC.1998,120, 1600.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

4 Thioglycosides Stefan Oscarson

4.1 Introduction Thioglycosides are not very common in nature, only a few simple alkyl and aryl thioglycosides have been found as constituents of antibiotics from Streptomyces species [ 1-31. Although not really thioglycosides, glucosinolates with an anomeric sulfur are also well-known natural compounds [4, 51. In spite of their low natural abundance, thioglycosides and their chemistry have for a long time been a field of interest for many researchers, the first thioglycoside was synthesized in 1909 [6], and their basic chemistry was early reviewed [4, 71. But it is only recently that their excellent glycosyl donor qualities have been recognized. In the beginning of the seventies the first successful synthesis of a disaccharide was performed [S], but it was not until the middle of the eighties that enough efficient promoters were discovered to make glycosylation with thioglycoside donors into a general and accepted method. Since then and in a very short time-span, thioglycosides have become one of the most used type of donors, especially in glycosylations with oligosaccharide donors (block synthesis). There are already some reviews on the subject [9-121, and in the most recent complete list of published glycosylations during one year (1994) [ 131, thioglycosides were used as donors in about 20% of all reported glycosylations and in 35% of all couplings using oligosaccharide donors, competing well with the long established halide donors and the trichloroacetimidate method. The strength of thioglycosides as glycosyl donors is their easy synthesis (Section 4.2) combined with their stability, withstanding most reaction conditions used in e.g. protecting group manipulations and glycosylations, and at the same time their effective acth ation using chemoselective thiophilic promoters (Sections 4.3.2 and 4.4.1) and also easy transformation into other types of glycosyl donors (Section 4.3.1). Hence, they are perfectly fitted to be used in block synthesis of oligosaccharides, where stable oligosaccharide donors are a prerequisite. Their stability also makes them very apt in synthetic schemes where a number of chemical modifications are performed just prior to the glycosylations, as, e.g., in various types of

94

4 Thioglycosides

internal glycosylations (Section 4.4.2). In this article the synthesis of thioglycosides and their activation and application as glycosyl donors are discussed.

4.2 Synthesis of Thioglycosides Obviously one important feature of attractive glycosyl donors are their easy availability. The synthesis of the most used types of thioglycoside donors, mainly simple alkyl and aryl glycosides, is straight-forward and can be performed on a large scale using cheap chemicals. Often the product is crystalline and can be purified without chromatography. Numerous ways to prepare thioglycosides have been described [4, 10, 111, but for the easy production of thioglycosides mainly two synthetic methods are commonly employed, namely the reaction of a peracetylated glycose with a thiol in the presence of a Lewis acid and the reaction of an acylated glycosyl halide with a thiolate.

4.2.1 From Anomeric Acetates As depicted in Scheme 1 an anomeric acetate is usually easily substituted with a thiol under Lewis acid promotion to give thioglycosides. A standard procedure is to react a peracetylated aldose dissolved in CH2C12 with a slight excess of a thiol using a hard Lewis acid, e.g. BF3-etherate [ 141 or FeC13 [ 151, as promoter to give high yields of the 1,2-truns-product (neighbouring group participation). @-Acetateprecursors usually react faster [16], which is an advantage since this diminish the formation of 1,2-cis-impurities through Lewis acid catalysed anomerisation of the pthioglycoside [ 171. a/p-Mixtures are often obtained with less reactive mercaptans [18-201 or glycosyl acetates [21, 221. Variations of this method include the utilization of trialkylstannyl [23] or silyl derivatives [24] of the mercaptan. Advantages are less odor and sometimes higher yields owing to less formation of (1,Zcis- and dithioacetal) side products, disadvantages are the accessibility and cost of the reagents. Also, thioacetic acid can be used as nucleophile in this approach, to give anomeric thioacetates, which, in a two step-procedure, can be selectively S-deacylated and alkylated to give thioglycosides, a sequence often used in the synthesis of thiooligosaccharides [251.

A AcOc

O

Scheme 1.

GOAC OAc

BF3-Et20 EtSH CHC13 83%

AcO

SEt OAc

4.2 Synthesis of Thioglycosides ,OAc

95

,OAc

5 Et,O/H,O ACO*~~,, AcO AcO Br

84%

OAc

Scheme 2.

4.2.2 From Glycosyl Halides

This is a classical way to thioglycosides [6], in which an acylated glycosyl halide is reacted with a thiolate in a polar solvent to yield the 1,2-truns-thioglycoside (Scheme 2). Both alkyl-and aryl mercaptans have been used and good yields are usually obtained, although, especially with the more basic alkyl thiolates, a reacetylation step is often necessary prior to work-up and isolation of the product. If controlled conditions and an aprotic polar solvent are used, a S~2-typedisplacement reaction can be performed to give the inverted anomeric configuration in the product, e.g. a-glycosides from p-glycosyl halide precursors, in spite of the presence of participating 2-@protecting groups (Scheme 3) [26, 271. A modification is the use of phase-transfer conditions, allowing the use of free thiols and non-polar solvents in the reaction [28-301. Also with glycosyl halides as donors, alkylstannyl derivatives of the thiol have been employed as acceptors [23]. The lower reactivity of these reagents necessitates the use of elevated temperature or a promoter in these reactions, which often result in alp-product mixtures. Syntheses of thioglycosides are sometimes possible using Koenigs-Knorr glycosylation conditions (Scheme 4) [3l], although mercury(I1) salts have been rather frequently used to promote glycosylation with thioglycoside donors (Section 4.3.2). Indirect formation of thioglycosides can also be performed from glycosyl halides. Reaction with thiourea yields a pseudothiouronium salt, which (as the thioacetate above) can be hydrolyzed under mild, selective conditions to give the anomeric thiol, which then can be alkylated in an efficient, non-smelling procedure to produce

Scheme 3.

Scheme 4.

96

4 Thioglycosides

czA,& " 3 S

Aito+

acetone A c o ~ r 80%

AcO OAc

+

H20

Aco&sH AcO

YNH2 90% NH2

OAc

Scheme 5.

I IV

OH

Scheme 6.

thioglycosides (Scheme 5) [321, however hampered by the believed carcinogenity of thiourea (and alkyl halides).

4.2.3 Protecting Group Manipulations in Thioglycosides All common protecting groups such as esters (acetates, benzoates, pivaloates, leviulonates), ethers (benzyl, allyl, trityl, silyl), acetals (isopropylidene, benzylidene, cyclohexane-l,2-diacetal (CDA)) and orthoesters can effectively be introduced, manipulated and removed. Problems encountered are the removal of benzyl groups using catalytic hydrogenolysis (when Birch reduction conditions is an alternative), since the sulfur generally contaminates the catalyst, and sometimes participating of the anomeric sulfur to the 2-position and a consecutive 1 -+ 2-migration of the thio group. In an attempt to introduce a 2,3-orthoester into a rhamnose thioglycoside, using a trimethylorthoester in CH3CN the methyl 2-thio-2-deoxy glycoside was found to be the main product (Scheme 6) [33]. This migration could be avoided by changing the solvent to DMF or by removing the methanol formed in the reaction [34]. Similar 1,Zmigrations of thioglycosides have been utilized to give 2-thio-2-deoxy glycosides or glycosyl fluorides, which can be processed to 2-deoxy-glycosides (Scheme 7) [35, 361. Most functional group conversions abundantly used on carbohydrate derivatives are also compatible with thioglycosides, e.g. SN~-displacementsof sulfonates [37], azo transfer [ 381, and deoxygenation by tributyltin hydride reduction of thiocarbonyl carbamates [39]. During oxidations care has to be taken so not to obtain the sulfoxide derivative (if that is not desired). DMSO and also PDC-mediated oxidations can usually be performed without problems [40]. The thioglycosides are also stable towards various metal reagents such as Grignard and alkyl lithium reagents, which allows carbon elongation of carbohydrates and is a handy way to higher carbon sugar donors [41-431.

4.3 Glycosylutions with Thioglycoside Donors

97

1) i.BupSnO

MeOH

ii. BnBr, CsF

DMF, 66%

SEt

OH

SEt

2)CIC(S)OPh DMAP, CH$N

80% BnO BnO

OMe OBn

OMe OBn

Scheme 7.

4.3 Glycosylations with Thioglycoside Donors The stability of thioglycosides was early recognized, as was also their very mild conversion into glycosyl halides. Hence, thioglycosides were initially used as convenient precursors for various glycosyl halides, which then could be used as donors in glycosylation reactions. Later, it was found that thioglycosides could be activated directly by the use of chemoselective thiophilic promoters, and this discovery opened up a new field of applicable donors and glycosylation reactions. 4.3.1 A Two-step Procedure: Transformation of Thioglycosides into Other Types of Glycosyl Donors Glycosyl bromides and chlorides are effectively produced using bromine [44, 451 (also I-Br [46]) or chlorine [47] (also I-Cl [46]), nowadays often with CHzClz as solvent. Under these mild conditions even unstable glycosyl halides can be produced, as e.g. furanosyl [48, 491 and alkylated oligosaccharide bromides [50], and various acid-labile protecting groups are allowed as well. Glycosyl fluorides have been produced using numerous reagents [51], e.g. dimethyl(methy1thio)sulfonium tetrafluoroborate (DMTSB) [52], NBS-DAST [ 531, HF/pyridine [53]and Selectfluor [54]. Glycals can be obtained from phenyl thioglycosides by treatment with lithium naphtalenide under conditions compatible with acid4abile groups [ 551. Trichloroacetimidates may be obtained by the hydrolysis of the thioglycoside followed by the subsequent transformation into the imidate using trichloroacetonitrile and a base [56). Hydrolysis have been performed using various promoters (e.g. AgNO:3 [57], DMTSB [58],NOBF4 [59], and DMTST (dimethyl(methylthi0)sulfonium triflate) [43])in the presence of water. Sulfoxides are formed by oxidation using MCPBA or dimethyldioxirane [60].

98

4 Thioglycosides

/

J

-cl S

L

-0Ac o

7

W

N

O HN&I,

Scheme 8.

Anomeric esters have been synthesized through the action of silver [61] or mercuric [62]carboxylates, NOBF4/Ac20 [59] or NIS/RCOOH [63] on a thioglycoside. Since it is possible to activate a thioglycoside in the presence of a pentenyl glycoside [12], it should be possible to transform a thioglycoside into a pentenyl glycoside, but to my knowledge this sequence has not been explored. This convertibility is still an attractive feature of thioglycosides and the two-step glycosylation procedure is continuously being used in glycosylations and is a good complement to the direct one-step activation of thioglycosides. It is an important asset, since in oligosaccharide synthesis it is difficult to foresee which donor that will function best in which coupling, and, furthermore, since it allows thioglycosides as the sole source of donors, which simplifies the synthesis and diminishes the number of precursors that has to be made. Of the possible donors it is mainly the halides that have been employed. As mentioned, this was the original way to use thioglycosides in glycosylation reactions and already in the early publications on the conversion to halides, alkyl 0-glycosides was prepared by activation of the formed halide by a halide promoter, Ag2C03 [49] or Hg(CN)2 [64], in the presence of an alcohol (Scheme 9) both in situ and after purification of the halide. Both procedures have been used also with glycosyl acceptors. The in situ procedure [65-671 is advantageous if the isolation of the glycosyl halide is a problem. However, with the mild conditions used, this is normally not a problem and the halide is most often crudely purified before the glycosylation reaction, which, at least in our experience, generally gives a cleaner coupling reaction with less side-products. The in situ method, of course, also disqualifies thioglycoside acceptors in orthogonal couplings to build thioglycoside blocks (Section 4.4.1).

1) clp, CHpCIp

Ac N

SEt

benzene 69%

Scheme 9.

~

2) EtOH,Hg(CN)P

OEt

4.3 Glycosylutions with Thioglycoside Donors

99

BnO

SMe

BnO

OBn

OBn

AgOTf 63%

BnO

OBn

Scheme 10.

These transformations are possible to perform on complex oligosaccharides and a number of large, biologically important structures have been synthesized with this approach both using bromides 1501 (Scheme 10) and fluorides [68]. Treatment of perbenzylated thioglycosides with iodine in the presence of an acceptor gave good yields of 0-glycosides, but there is no discussion of glycosyl iodides as intermediates in the reactions [69]. No additional promoter was necessary, and acetylated thioglycosides were not affected.

4.3.2 Direct .4ctivation of Thioglycoside Donors Heavy Metal Salt Promotors The first efforts to use thioglycosides directly as donors were performed mainly with mercury(I1) salts as promoters. It was early recognized that mercury had a high affinity for the sulfur functionality [4], and effective hydrolysis of 1-thiofuranosides using mercuric chloride as catalyst instead of a mineral acid was reported [70]. An early attempt to methanolyze ethyl tetra-0-acetyl-1 -thio-P-D-mannopyranoside using HgC12 was nct successful [71], but when Ferrier et al. tried alcoholysis on unprotected phenyl 1-thio-D-glucopyranosides using Hg(0Ac)z or HgClz/HgO as promoters good yields of the corresponding alkyl glycosides with inverted anomeric configuration was obtained [8, 721. Using the perbenzylated phenyl 1-thio-p-Dglucopyranoside as donor and HgS04 as promoter the first disaccharide was produced (Scheme 11) 181. Other similar promoters were then tried, i.a. HgC12 1731, Hg(0Bz)z [74] and PhHgOTf [75],but especially with less reactive donors, e.g. peracylated, the yields were low and it was difficult to find a general high-yielding glycosylation method.

100

4 Thioglycosides

HgS04 THF, reflux, 3h 54%

Scheme 11.

However, if a heterocyclic aryl thioglycoside was used as donor a more efficient activation with metal salts was accomplished. HgClz-promoted methanolysis of phenyl thioglycosides had been found to be faster than for the corresponding ethyl thioglycoside (compare Section 4.4.1), and with the heteroatom in the aryl system an even faster metal-ion promoted reaction took place, indicating that the heteroatom facilitated the activation of the thioglycoside (remote activation concept [76]). Mainly 2-pyridyl thioglycosides have been employed (Scheme 12), but also e.g. 2-pyrimidinyl [771 and benzothiazole thioglycosides have been used, and also different metal salt promoters, e.g. Hg(N03)Z [78], CuOTf2 [79], Pb(C104)2 [80, 811, or AgOTf [80, 82, 831. More recently, Mereyela et al. in a number of papers introduced methyl iodide as a promoter for 2-pyridyl thioglycosides [ 84-87]. Halonium, sulfonium and carbonium type promoters Owing to the many advantages of thioglycoside donors and the promising but so far limited success with heavy metal salt promoters a quest for more generally effective promoters was launched. Starting with NBS in 1983 [88] and especially with methyl triflate (MeOTf) in 1984/5 [89-911 and DMTST in 1986 [92] a seemingly never ending spectrum of new promoters has been reported (Table l), most examples variation of a common theme, soft electrophile reagents (halonium or sulfonium) with non-nucleophilic counter ions (triflate or perchlorate), but also electrochemical and radical activation have been described (Section 4.3.2). In spite of the array of different promoters and thioglycoside donors developed and reported, only a few have so far been proven generally useful in more extensive oligosaccharide synthesis. Among donors the methyl, ethyl and phenyl thioglycosides are by far the most utilized, and MeOTf, IDCP(T), DMTST and NIS alone or

OBn Ac0-q

Me0

scheme 12.

N

+ HO& Me0

Pb(C'04)2 THF,8h 59%, C@J 3:l

MeO

4.3 Glycosylations with Thioglycoside Donors

101

Table 1. Thiophilic promotors. Promoter

References

MeOTf DMTST/DMTSB NBS N BS/TfOH NBS/BQNOTf NBS/LiC104 NIS, NIS/TfOH or TMSOTf NIS/AgOTf or I3t3SiOTf IDCP IDCT Methylsulfenium triflate/bromide (MeSOTf/Br) 12

IBr Iz/AgOTf PhSOTf PhSeOTf NOBF4 N-PhSePhthOTf S02C12/TfOH PhIO-Tf20

in combination with triflic acid or silver triflate cover more or less completely the spectrum of more frequently used promoters. Numerous highly complex structures have been synthesized using these combinations verifying the impressive utility of the method. A few examples, exhibiting the synthesis of structures from N-linked glycans [91], the Salmonella LPS-core [ 1lo], and gangliosides [ 1111, are shown in Schemes 13-15. The methodology are compatible with almost all types of donors, and thioglycoside donors of e.g. furanoses, heptoses, Kdo, neuraminic acid, deoxy sugars, and uronic acids have been prepared and used successfully in glycosylation reactions. These promoters also complement each other nicely regarding reactivity, from the least reactive IDCP and MeOTf through the intermediate DMTST up to the most reactive NIS/TfOH. This promoter reactivity difference in combination with the different reactivity of donors and acceptors allows many practical orthogonal couplings between thioglycosides (Section 4.4.1). Notwithstanding the efficiency of these promoters and thioglycosides as donors, problems can, of course, be encountered in various applications. The problems are connected both to the type of thioglycoside used as well as to the promoter employed. Regarding the promoter, MeOTf can, if the acceptor is unreactive, give methylation instead of glycosylation of the acceptor hydroxyl group [ 1121, although this alkylation is normally surprisingly slow in the absence of base. The use of DMTST together with N-acetamido sugars might lead to thiomethylation of the amide group [113]. NIS, due to the

102

4 Thioglycosides

MeOTf

1

W OH L n OBn

Scheme 13.

nucleophilicity of the counter ion (as compared to triflate in most other promoters), can, once more in the case of unreactive acceptors, give rise to N-succinimide glycosides of the donor as the major product (Scheme 16) [39, 114-1 161. The obvious considerations that soft nucleophilic centres, like e.g. double bonds, should not be allowed in the donor/acceptor, since these would compete with the sulfur in its reaction with the electrophilic promoter, is not as adamant as perhaps expected. Accordingly, glycosidation with thioglycoside donors and allyl and pentenyl groups both in the donor and the acceptor have been performed using controlled conditions (Scheme 17) [ 117-1 191. Problems have been encountered when the acceptor hydroxyl group is adjacent to an allyl protecting group, which have resulted in internal addition to an activated double bond being the major reaction (Scheme 18) [97, 1131. Common problems in oligosaccharide synthesis not directly related to the method of direct activation of thioglycoside donors, such as for example transacylations or formation of orthoesters, when using 2-0-participating groups, or glycal or 1,6-anhydro derivative formation of the donor can, of course, also be encountered [90, 1lo]. As is the case with most glycosylation reactions, i.a. because of their complexity, no truly thorough mechanistic investigation has been performed on the direct activation of thioglycosides. An obvious general mechanism (Scheme I9), very similar to the one suggested for other types of glycosylation reactions, in which the soft

4.3 Glycosylutions with Thioglycoside Donors

OSP

BnO

BnO

BnO

OAc SM.3

BzO BzO

AcO

EtNBr, 90%

HO

OAc

OH

0

Scheme 14.

103

104

4 Thioglycosides

BzO

BzO?'$ko&sph BzO

:

r

s

E

e

OBn OBn

OB+oBn

'f

1) NlSRfOH 86%

NPhth

OBn

2) NaOMe

OBn

OBn

OBn

SMe

OAc

') DMTST 2) Ac20, pyridine, DMAP 3) CAN 55%

OBz

AcO

OBn

NlSKfOH 60%

OBn

OBn

OBn OBn

Scheme 15.

SE = CH2CH2SiMe3

OBn

4.3 Glycosylutions with Thioglycoside Donors

105

BnO

BnO O(CH2)&02Me NHAc

HO

Scheme 16.

DMTST

EQO OSE

92% dp 2.3/1 or NIS/AgOTf (75%)

N3

OBn

Scheme 17.

DMTST

OBn

SEt

SEt

Scheme 18.

X-

A

+HX

Scheme 19.

electrophile E+ (e.g. I+ (NIS, NBS, IDCT(P), Mef (MeOTf) or MeS+ (DMTST(B), MeSOTf(Br)) attaches to the sulfur functionality, whereafter the anomeric carbonsulfur bond is broken to generate an oxycarbenium ion (with some counterion X-, e.g. TfO- or C104), which then is attacked by the acceptor nucleophile, sometimes preceded by the formation of a cyclic acyloxonium ion if 2-0-participating groups are used, to give the 0-glycoside, is more or less commonly accepted, but the intrinsic details, like the exact nature of the intermediates or rate-determining step, are not known, and are also probably very different for different combinations of acceptor/donor/promoter and reaction conditions. No kinetic studies have been performed. Some studies of the anomerisation of alkyl p-D-thioglycosides with IDCP have been carried out showing this to be an intermolecular process and bulky

106

4 Thioglycosides

SEt

DMTST Et20 76% a

OMe

OMe

OBz

63% p

Scheme 20.

alkyl groups to be inert (however possible to use as donors) [17, 1201, and Crich and Sun have identified a stable anomeric triflate as intermediate when activating thioglycosides with PhSOTf at -78 “C [loll, but mostly mechanistic suggestion are based on product structure and distribution in combination with the understanding about mechanisms in similar reactions. Since the major outlines of the mechanism is the same, with common intermediates, as for most other types of glycosylations, much of the knowledge about the stereochemical outcome of glycosylation reactions with other type of donors is valid also for thioglycoside donors. Consequently, the use of 2-0-participating groups will lead to 1,2-trans-glycosides, if there is not a mismatch between donor and acceptor [ 1211. Furthermore, with non-participating groups, the use of diethyl ether as solvent will improve the alp-ratio [ 1221, whereas the use of acetonitrile will give predominantly P-glycosides, especially with primary acceptors [ 1231. Boons et al. advocate the use of toluene/dioxane as an even better a-directing solvent mixture [124].The stereochemical outcome is also dependent on the promoter used and on the protecting group pattern (even apart from the 2-0-protecting group). To generalize, IDCP and DMTST are good a-directing promoters whereas MeOTf and NIS are not (Scheme 20 [ 1251) [ 124, 1261. Bulky groups at 0-6 of glucopyranoside donors [ 1271 and electron-donating ester groups at 0-4 of galactopyranosides [ 1281 have been shown to increase the a//p-ratio in the disaccharide products due to steric and participating effects, respectively. As so often in glycosylation reactions these findings are only guidelines, and an optimisation is normally necessary for each specific reaction. Single-Electron Activation

Thioglycosides can also be activated by a single-electron transfer mechanism (Scheme 21). In these reactions a radical cation is generated by the transfer of a single electron from the sulfur to a suitable electron acceptor. Subsequent heterolytic cleavage of the anomeric linkage yields a sulfur radical, which dimerize to a disulfide, and an oxocarbenium ion, which, as in “normal” glycosylations, can undergo a nucleophilic attack by an alcohol to generate an 0-glycoside. The single electron transfer can be accomplished either photochemically [ 1291, electrochemically [130, 1311, or chemically [132]. Aryl thioglycosides in CH3CN/MeOH (9: 1) were irradiated in the presence of

107

4.3 Glycosylutions with Thioglycoside Donors single-

+I12 RSSR Scheme 21.

1,4-dicyanonaphtalene to give methyl glycosides [ 1291. However, the choice of protecting groups was a problem. Benzyl protected thioglycosides gave mainly other products, whereas acetylated derivatives were not activated. Hence, only permethylated compounds were compatible with the reaction conditions to give the desired glycosides. Electrochemical glycosylation was possible using both phenyl (also p-methoxyphenyl [131]) and ethyl thioglycosides [130, 1331. Acetylated donors were found to be less efficient whereas both benzylated and benzoylated donors gave good yields of disaccharides (Scheme 22). Acylated donors gave, as expected, exclusively the 1,2-truns-products, and also the benzylated donors gave predominantly the pproducts, probably due to the use of CH3CN as solvent in the reactions. Chemically, single electron transfer glycosylations have been effected by the use of the commercially available electron transfer reagent tris(4-bromo-pheny1)ammoniyl hexachloroantimonate (BAHA) as promoter. Once more, both ethyland phenyl thioglycosides, benzylated and benzoylated, could be utilized as donors to afford disaccharides in high yields (Scheme 23) [ 1321. The efficient synthesis of a trisaccharide have been reported [134],but, in spite of the promising results no other oligosaccharide syntheses have been published using this methodology. Glycosylations performed with an electron transfer quencher (1,2,4,5-tetramethoxybenzene) present showed that the reactions are completely quenched in CH2C12 but only partly in CH3CN (28% yield), giving evidence for the single-electron transfer mechanism, but also suggesting alternative mechanistic possibilities in CH3CN [135].

BBnOn

O

a SPh + OBn

Brio%

HO

anodic oxidation BBnO n -e__.___) CH3CN

BnO OMe 73%, a/p 1.3

OBz Bzo*SPh BzO

OBz

+

anodic oxidation -eCH&N

&fo&$,

~

BnoOMe

Scheme 22.

45%

O BnO

G

o

6

BnO BnoOMe

B5:oGo&$, BzO

Brio

BnoOMe

108

4 Thioglycosides

Q BnO& BnO

SPh +

OBn

::o

4

*+

SbCIi

a

Br B r * BBnO n BnO CH3CN OMe 86%, a l p 1 :2

OBz BzO Bzo$&SEt

OBz

O BnO

G 0 BnO

G BnoOMe

OBz

+

:to+

CH&N BAHA BnoOMe

81%

Bi!o&oG BzO

Brio BnoOMe

Scheme 23.

Other Types of Donors With an Anomeric Sulfur

Although not proper thioglycosides, various other derivatives with an anomeric sulfur have been synthesized and used as glycosyl donors. Kotchetkov et al. introduced, as a complement to their cyanoorthoester 1,2-truns glycosylation method, (3-glucopyranosyl thiocyanates as 1,2-cis-glycosyl donors with tritylated acceptors and trityl perchlorate as promoter [136]. With a nonparticipating group at 0-2 the glycosylation is completely stereoselective to give the inverted anomeric configuration in the product (Scheme 24). Also polycondensation using this approach has been reported [137]. Anomeric carbodithioates, both 0-and N-linked have been investigated. The 0ethyl derivative, i.e. the ethyl S-glycosyl xhantate, of 2-azido-galactopyranoseswas found to be an efficient donor activated by either Cu(0Tf)z or DMTST [138, 1391, but especially the same type of derivative of N-acetyl-neuraminic acid has proven to be an excellent alternative as a sialic acid donor, which has found general applications activated by thiophilic promoters as DMTST or MeSOTf (Scheme 25) [ 1401421. The analogous anomeric N-piperidine derivatives could be activated not only by thiophilic promoters (MeOTf and DMTST) but also by AgOTf and various Lewis acids, e.g. SnC14 and FeC13, to allow orthogonal couplings with thioglycosides and the building of thioglycoside donor blocks (Scheme 26) [ 1431.

AcoG,,, + Tf)o&oAc

AcO

OBn

= +

OAc

CHzC12 77%

OAc AcO

Scheme 24.

4.4 Applications of Thioglycosides OAc OAc COOMe - - ...-

-

Acol*%

OEt

'dS\

AcHN AcO

MST CH3CN CH&

~

109

AcHN AcO

Scheme 25.

AAcOc

O

a

s

OAc

K

Ng CH& 79%

'OBn

Scheme 26.

Also anomeric phosphorothioates have been explored. Perbenzylated S-glycosyl phosphorodiamidimidothioates promoted by 2,6-lutidinium p-toluenesulfonate gave good yields of disaccharides and 1,2-cis-selectivity(Scheme 27) [ 1441. If BQNI was used as an additive in the glycosylation even higher 1,2-cis-selectivitywas obtained. S-glycosyl phosphorodithioates as glycosyl donors are discussed in a number of articles, both 2-deoxy derivatives activated by AgF [ 1451 or iodonium promoters [ 1461 to give alp-mixtures of disaccharides, and 2-0-pivaloyl derivatives activated by MeOTf to give 1,2-trans products (Scheme 27) [147]. Although stable under some reaction conditions, none of the donors above exhibit the superior stability of thioglycosides and are therefore not an alternative to

BnO toIuene 85%

'OBn

BnO

BnO& BnO bpiv

Scheme 27.

-

MeOTf

CH2CIz 55%

-

PivO

110

4 Thioglycosides

thioglycosides, e.g. in block synthesis, but rather an addition to the spectrum of less stable glycosyl donors, which might find use in applicable cases.

4.4 Applications of Thioglycosides The stability of thioglycosides in combination with their easy activation makes them highly suitable as donors in most glycosylation strategies and technologies, e.g. in block synthesis as well as in techniques where the donor is manipulated before glycosylation and in solid phase synthesis. 4.4.1 Block Syntheses, Orthogonal Glycosylations The concept of orthogonal glycosidations will be discussed in more detail in other chapters in this book, but since the possibility of oligosaccharide block syntheses is one of the major advantages of thioglycosides, a few aspects of orthogonal glycosidations involving thioglycosides will complementarily be discussed also here. Thioglycosides as Acceptors

Most other donors can effectively be used in couplings to thioglycoside acceptors to give thiooligosaccharide blocks, e.g. halide donors using halide-assisted conditions or heavy-metal salt promoters, trichloroacetimidate donors with hard Lewis acid promoters, selenoglycosides with AgOTf activation and sulfoxide donors with TMSOTf or triflic anhydride as promoter. The efficiency in the couplings are dependent on the reactivity of the acceptor and donor and complications can arise especially with activated thioglycoside acceptors, when transglycosylation of the thiol is an often observed side reaction (Scheme 28) [89, 148, 1491. Taking the knowledge of the protecting group influence on the reactivity into account, it is normally possible to design an efficient way to thiooligosaccharide blocks, however, the protection group manipulations might be tedious and surprises might be encountered (Scheme 29) [ 1171.

Hg(CN)z or HgBr2 or . . O M B r OAc

Scheme 28.

+

A c O W S E t toluene

OAc 6O-8OYo

4.4 Applications of Thioglycosides

I

tES,&&ThTBDMSO P

11 1

bBZ

c AgOTf --

BnO

SEt

Br

OBz

Scheme 29.

Thioglycosides as Both Donors and Acceptors

The protecting group pattern is known to have a large influence on the reactivity of the donor (and of the acceptor). This has been recognized for a long time, but only been possible to utilize effectively in synthetic schemes with the event of stable donors, which has allowed the formation of donors with almost any type of protecting groups and pattern. This reactivity difference can be utilized to permit orthogonal glycosylations between thioglycoside donors (reactive, alkyl protecting groups) and acceptors (inreactive, acyl protecting groups) exploiting less reactive promoters, which only react with activated thioglycosides (Scheme 30) [97]. The idea of using the reactivity difference more systematically was developed by Fraser-Reid et al. in orthogonal couplings between armed and disarmed pentenyl glycosides [ 1501. This concept has been further developed by quantitative determination of the reactivity of donors by the use of competitive glycosylations [ 1511. Recently, fifty thioglycosides with different protecting group patterns have been synthesized and their reactivity has been determined. Through this approach it has been possible lo assign a reactivity constant to each thioglycoside, which allows the planning of synthetic schemes involving orthogonal glycosylations between thioglycosides and one-pot synthesis of oligosaccharides and also a possible computation of the selection process [ 1521. However, with thioglycosides there is an additional possibility to change the reactivity, namely through changing the alkyl (aryl) group of the anomeric thiol. This anomeric activation/deactivation prospect might be compared to the reactivity dif-

B"O& BnO

SEt

OBn

Scheme 30.

&-

+BB'po

IDCP

OBz

SEt (CICH2)2/Et20 84%

BnO SEt OBz

1 12

4 Thioglycosides OBn

SEt + H O

BnO OBn

OBn

72%

Scheme 31.

ference between halide sugars (iodo > bromo > chloro > fluoro) and the inactivation of pentenyl glycosides by bromination of the double bond, but the spectra allowed with thioglycosides are broader. Both steric effects and electronic effects can be utilized to influence the reactivity. Hence, thioglycosides of sterically crowded mercaptans, as e.g. 1,l-dicyclohexylidenemethyl mercaptan, show lower affinity for thiophilic promoters and can be used as acceptors in condensations with ethyl thioglycosides as acceptors [ 201. The higher stability of phenyl thioglycosides as compared to ethyl thioglycosides with IDCP as promoter has been shown [ 1191, and this reactivity difference have been further proved by glycosylations with p-tolyl thioglycoside donors and thioethyl acceptors using NIS/TfOH as promoter (Scheme 31) [153]. By introducing various substituents into the aromatic ring of phenyl thioglycosides a variety of reactivity can be obtained. On the extreme of the unreactivity side is the p-nitrophenyl thioglycoside, where activation does not have to be feared even with efficient promoters. On the contrary, sometimes deactivation of the acceptor hydroxyl group might be a problem and for continued use as donor the p-nitrophenyl group has to be activated as the acetamido analogue. This sequence has been investigated [ 154, 1551, but problems often arise and the scope of the methods appears to be limited, however effective with sialyl derivatives [ 1561. p-Halogenphenyl thioglycosides might be interesting compounds of intermediate reactivity. With methyl triflate or DMTST as promoters, acylated derivatives are not activated, whereas the use of NIS/triflic acid smoothly allows the use of these derivatives as glycosyl donors [ 1571. 4.4.2 Intramolecular Glycosidations To obtain desired regio- and stereoselectivity in glycosidations, new glycosylation methodology are continuously developed. Recently two intramolecular variants, internal acceptor delivery and intramolecular glycosidations of spacer prearranged glycosides, were described. In both of these approaches a number of reactions are performed on the acceptor/donor pair before the glycosidation reaction, and therefore thioglycosides as donors have been an obvious choice. In the internal delivery approach the acceptor hydroxyl group is linked to the 2-OH of a thioglycoside donor via a labile acetal linkage. Activation of the thioglycoside also breaks the acetal bond and the acceptor is transferred stereospecifically to the anomeric site of the activated donor to give high yields of

References

1 13

1,Zcis-glycosides, also the otherwise difficult obtainable P-mannopyranosides and P-fructofuranosides [ 158-1621. In the intramolecular glycosidation using prearranged glycosides the donor and acceptor are linked through a spacer via other hydroxyl groups than the ones that will be part of the new glycosidic linkage. By using various attachment sites and various spacers, regio- and stereoselective glycosylations can be performed [ 1631651. 4.4.3 Solid Phase Synthesis Since thioglycosides are possible to synthesize with almost any type of protecting group pattern and since they are shelf-stable, they should be very suitable as donors in a commercial approach to solid-phase synthesis of oligosaccharides. That they are excellent donors also in a solid-phase approach is shown, i.e., by the publications from Nicolaou et al. in which up to dodecasaccharides are synthesized on polystyrene beads using thiophenyl and thiomethyl glycoside donors and DMTST as promoter [1166, 1671. References 1. R. R. Herr, J , Am. Chem. Soc., 1967,89,2444-2447. 2. H. Hoeksema, B. Bannister, R. D. Birkenmeyer, F. Kagan, B. J. Magerlein, F. A. MacKellar, W. Schroecler, G. Slomp, R. R. Herr, J. Am. Chem. Soc., 1964,86, 4223-4224. 3. H. Hoeksema, J. Am. Chem. Soc., 1964,86, 4224-4225. 4. D. Horton, D. H. Hutson, Adv. Carbohydr. Chem., 1963, 18, 123-199. 5. V. Gil, A. J. MacLeod, Phytochern., 1980,19, 2071. 6. E. Fischer, K. Delbriick, Ber., 1909, 42, 1476-1482. 7. A. L. Raymond, Adv. Carbohydr. Chem., 1945, 1, 129-145. 8. R. J. Ferrier, R. W. Hay, N. Vethaviyasar, Carhohydr. Res., 1973, 27, 55-61. 9. P. Fugedi, P. J. Garegg, H. Lonn, T. Norberg, Glycoconj. J., 1987, 4, 97-108. 10. T. Norberg, in S. H. Khan, R. A. O’Neill (Eds.): Modern methods in Carbohydrate Synthesis, Harwood Academic Publishers 1995, p. 82-106. 11. P. J. Gareg,g, Adti. Carbohydr. Chem. Biochem., 1997, 52, 179-205. 12. G. H. Veeneman, in G.-J. Boons (Ed.): Carbohydrate Chemistry, Blackie Academic & Professional 1998, p. 98-174. 13. F. Barresi, 0. Hindsgaul, J. Carbohydr. Chem., 1995, 14, 1043-1087. 14. R. J. Ferrier, R. H. Furneaux, Carbohydr. Res., 1976, 52, 63-68. 15. F. Dasgupta, P. J. Garegg, Acta Chem. Scand., 1989,43,471-475. 16. R. U. Lemieux, Can. J. Chem., 1951,29, 1079-1091. 17. B. Lindberg, B. Erbing, Acta Chem. Scand., 1976, B30, 611-612. 18. W. V. Dahlhoff, Liebigs Ann. Chem., 1990, 1025-1027. 19. S. A. Galema, J. B. F. N. Engberts, H. A. v. Doren, Carbohydr. Res., 1997, 303, 423-434. 20. G. J. Boons, R. Guertsen, D. Holmes, Tetrahedron Lett., 1995, 31, 6325-6328. 21. P. J. Garegg, L. Olsson, S. Oscarson, J. Org. Chem., 1995, 60, 2200-2204. 22. T. Nakano, Y. Ito, T. Ogawa, Tetrahedron Lett., 1990, 31, 1597-1600. 23. T. Ogawa, M. Matsui, Carbohydr. Res., 1977, 54, C17-C21. 24. V. Pozsgay, H. Jennings, Tetrahedron Lett., 1987,28, 1375-1376. 25. H. Driguez, Topics Curr. Chem., 1997, 187, 85-1 16. 26. M. Appani, M. Blanc-Muesser, J. Defaye, H. Driguez, Can. J. Chem., 1981, 59, 314-320.

114 27. 28. 29. 30.

4 Thioglycosides

M. Blanc-Muesser, J. Defaye, H. Driguez, Carbohydr. Res., 1978, 67, 305-328. J. Bogusiak, W. Szeja, Pol. J. Chem., 1985, 59, 293-298. F. D. Tropper, F. 0. andersson, C. Grand-Maftre, R. Roy, Synthesis, 1991, 734-736. S. J. Meunier, F. 0. andersson, M. Letellier, R. Roy, Tetrahedron: Assymetry, 1994, 5, 23032312. 31. M. Gerz, H. Matter, H. Kessler, Angew. Chem. Znt. Ed. Engl., 1993, 32, 269-271. 32. D. Horton, Methods Carbohydr. Chem., 1963,2,433-437. 33. F. I. Auzanneau, D. R. Bundle., Carbohydr. Res., 1991, 212, 13-24. 34. V. Pozsgay, Carbohydr. Res., 1992,235, 295-302. 35. K. C. Nicolaou, T. Ladduwahetty, J. L. Randall, A. Chuchulowski, . I Am. Chem. Soc., 1986, 108. 2466-2467. 36. H. M. Zuurmond, P. A. M. v. d. Klein, G. A. v. d. Marel, J. H. v. Boom, Tetrahedron Lett., 1992,33, 2063-2066. 37. M. Nilsson, T. Norberg, J. Chem. Soc. Perkin Trans. I , 1998, 1699-1702. 38. T. Buskas, P. J. Garegg, P. Konradsson, J.-L. Maloisel, Tetrahedron: Asymmetry, 1994, 5, 2187-2194. 39. S. Oscarson, U. Tedebark, D. Turek, Carbohydr. Res., 1997, 159, 159-164. 40. P. J. Garegg, L. Olsson, S. Oscarson, J. Org. Chem., 1995, 60, 2200-2204. 41. G. J. P. H. Boons, G. A. v. d. Marel, J. H. v. Boom, Tetrahedron Lett., 1989, 30, 229-232. 42. G. J. P. H. Boons, Doctoral thesis, Leiden University 1991. 43. C. Bernlind, S. Oscarson, J. Org. Chem., 1998,63, 7780-7788. 44. W. A. Bonner, J. Am. Chem. Soc., 1948, 70,770-772. 45. F. Weygand, H. Ziemann, H. J. Bestmann, Chem. Ber., 1958,91,2534-2537. 46. R. Kartha, R. Field, Tetrahedron Lett., 1997, 38, 8233-8236. 47. M. L. Wolfrom, W. Groebke, J. Org. Chem., 1963,28, 2986-2988. 48. M. L. Wolfrom, P. McWain, R. Pagnucco, A. Thompson, J. Org. Chem., 1964,29, 454-457. 49. F. Weygand, H. Ziemann, Liebigs Ann. Chem., 1962, 657, 179-198. 50. K. Ekelof, S. Oscarson, J. Org. Chem., 1996, 61, 7711-7718. 51. P. J. Card, J. Carbohydr. Chem., 1985, 4 , 451-487. 52. L. Blomberg, T. Norberg, J. Carbohydr. Chem., 1992, 11, 751-760. 53. K. C. Nicolau, R. E. Dolle, D. P. Papahatjis, J. L. Randall, J. Am. Chem. Soc., 1984, 106, 4189-4192. 54. M. D. Burkart, Z. Zhang, S.-C. Hung, C.-H. Wong, J. Am. Chem. Soc., 1997, 119, 1174311746. 55. A. Fernandez-Mayoralas, A. Marra, M. Trumtel, A. Veyriirres, P. Sinay, Carbohydr. Res., 1989,188,81-95. 56. R. R. Schmidt, W. Kinzy, Ado. Carbohydr. Chem. Biochem., 1994,50, 21-117. 57. P. J. Garegg, H. Hultberg, C. Lindberg, Carbohydr. Res., 1980, 83, 157-162. 58. S. Nilsson, H. Lonn, T. Norberg, Glycoconjugate J., 1991, 8, 9-16. 59. V. Pozsgay, H. Jennings, J. Carbohydr. Chem., 1990, 9, 333-343. 60. D. Kahne, S. Walker, Y. Chang, D. v. Engen, J. Am. Chem. Soc., 1989,111,6881-6882. 61. C . Pederson, H. G. Fletcher, J. Am. Chem. Soc., 1960, 82, 3215-3217. 62. R. J. Ferrier, S. R. Haines, Carbohydr. Res., 1984, 127, 157-161. 63. G. H. Veeneman, S. H. v. Leeuwen, J. H. v. Boom, Tetrahedron Lett., 1990,31, 1331-1334. 64. M. L. Wolfrom, H. G. Garg, D. Horton, J. Org. Chem., 1963,28, 2989-2992. 65. J. 0. Kihlberg, D. A. Leigh, D. R. Bundle, J. Org. Chem., 1990, 55, 2860-2863. 66. S. Sato, M. Mori, Y.lto, T.Ogawa, Carbohydr. Res., 1986, 155, C6-C10. 67. F. 0.Andersson, P. Fugedi, P. J. Garegg, M. Nashed, Tetrahedron Lett., 1986,27, 3919-3922. 68. K. C. Nicolau, H. Ueno, in S. Hanessian (Ed.): Preparative Carbohydrate Chemistry, Marcel Dekker, Inc. 1997, p. 314-338. 69. R. Kartha, M. Aloui, R. Field, Tetrahedron Lett., 1996, 37, 5175-5178. 70. M. L. Wolfrom, K. Anno, J. Am. Chem. SOC.,1953, 75, 1038-1039. 71. J. Fried, D. E. Waltz, J. Am. Chem. SOC.,1949, 71, 140-143. 72. L. Hough, M. I. Taha, J. Chem. Soc., 1956,2042-2048. 73. T. Y. R. Tsai, H. Jin, K. Wiesner, Can. J. Chem., 1984,62, 1403-1405. 74. J. W. v. Cleve, Carbohydr. Res., 1979, 70, 161-164.

References

1 15

75. P. J. Garegg, C. Henrichson, T. Norberg, Curbohydr. Res., 1983, 116, 162-165. 76. S. Hanessian, in S. Hanessian (Ed.): Preparative Carbohydrate Chemistry, Marcel Dekker, Inc. 1997, p. 381-387. 77. Q. Chen, F. Kong, Carbohydr. Rex, 1995,272, 149-157. 78. S. Hanessian, C. Bacquet, N. Lehong, Carbohydr. Rex, 1980, 80, C17-C22. 79. T. Mukaiyama, T. Nakatsuki, S. Shoda, Chem. Lett., 1979, 487-490. 80. R. B. Woodward, e. al., J. Am.Chem. Soc., 1981, 103, 3215-3217. 81. P. G. M. U'uts, S. S. Bigelow, J. Org. Chem., 1983, 48, 3489-3493. 82. K. Tsuboyama, K. Takeda, K. Torii, M. Ebihara, J. Shimizu, A. Suzuki, N. Sato, H. Furuhata, Chem. Pharm. Bull., 1990, 38, 636-638. 83. Z. Zhiyuan, G. Magnusson, Carbohydr. res., 1994,262, 79-101. 84. G. V. Reddy, V. R. Kulkarni, H. B. Mereyala, Tetrahedron Lett., 1989, 30, 4283-4286. 85. H. B. Mereyala, G. V. Reddy, Tetrahedron, 1991, 47, 6435-6448. 86. H. B. Mereyala, V. R. Kulkarni, D. Ravi, G. V. M. Sharma, B. V. Rao, G. V. Reddy, Tetrahedron, 1992, 48, 545-562. 87. H. B. Mereyela, V. R. Gurijala, Curbohydr. Rex, 1993,242, 277-280. 88. K. C. Nicolaou, S. P. Seitz, D. P. Papahatjis, J. Am. Chem. Soc., 1983, 105, 2430-2434. 89. H. Lonn, Chem. Commun. Stockholm Univ., 1984, No2, 1-30. 90. H. Lonn, Carbohydr. Res., 1985, 139, 105-113. 91. H. Lonn, C'arbohydr. Rex, 1985, 139, 115-121. 92. P. Fugedi, P. J. Garegg, Curbohydr. Res., 1986, 149, C9-Cl2. 93. M. Sasaki, K. Tachibana, H. Nakanishi, Tetrahedron Lett., 1991, 32, 6873. 94. K. Fukase, A. Hasuoka, S. Kusumoto, Tetrahedron Lett., 1993, 34, 2187-2190. 95. K. Fukase, A. Hasuoka, Kinoshita, S. Kusumoto, Tetrahedron, 1995, 51, 4923-4932. 96. P. Konradsson, U. E. Udodong, B. Fraser-Reid, Tetrahedron Lett., 1990, 31, 4313-4316. 97. G. H. Veeneman, J. H. v. Boom, Tetruhedron Lett., 1990,31, 275-278. 98. G. H. Veeneman, S. H. v. Leeuwen, Z. H, J. H. v. Boom, J. Curbohydr. Chem., 1990,9, 783796. 99. F. Dasgupta, P. J. Garegg, Curbohydr. Rex, 1988, 177, C13-Cl7. 100. F. Dasgupta, P. J. Garegg, Curbohydr. Rex, 1990,202, 225-238. 101. D. Crich, S. Sun, J. Am. Chem. SOC.,1998,120,435-436. 102. D. Crich, S. Sun, Tetrahedron, 1998, 54, 8321-8348. 103. Y. Ito, T. Ogawa, Tetrahedron Lett., 1988, 29, 1061-1064. 104. Y. Ito, T. Ogawa, M. Numata, M. Sugimoto, Carbohydr. Rex, 1994,202, 165-175. 105. V. Pozsgay, H. J. Jennings, J. Org. Chem., 1987, 52, 4635-4637. 106. H. Shimizu, Y. Ito, T. Ogawa, SynLett, 1994, 535-536. 107. H. Shimizu, SynLett., 1998, 84-86. 108. H. Lonn, Glycoconjugute J., 1987, 4, 117-118. 109. K. Fukase, A. Hasuoka, I. Kinoshita, S. Kusumoto, Tetrahedron Lett., 1992, 33, 7165-7168. 110. S. Oscarson, H. Ritztn, Carbohydr. Rex, 1994,254, 81-90. 111. A. Hasegawa, M. Kiso, in S. Hanessian (Ed.): Preparative Carbohydrate Chemistry, Marcel Dekker, Inc. 1997, p. 357-379. 112. P. J. Garegg, S. Oscarson, M. Szonyi, Carbohydr. Res., 1990,205, 125-132. 113. A.-K. Tiden, Doctoral thesis, Stockholm University 1991. 114. L.-X. Wang, N. Sakiri, H. Kezukura, J. Chem. SOC.Perkin I , 1990, 1677-1682. 115. C. Krog-Jensen, S. Oscarson, J. Org. Chem., 1996, 61, 1234-1238. 116. M. Anderson, P. Konradsson, S. Oscarson, L. Oberg, J. Carbohydr. Chem., in press. 117. M. Alpe, S. Oscarson, unpublished results. 118. S. K. Das, R. Ghosh, N. Roy, J. Carbohydr. Chem., 1993, 12, 693-701. 119. H. M. Zuurmond, S. C. v. d. Laan, G. A. v. Marel, J. H. v. Boom, Carbohydr. Res., 1991, 215, clLc3. 120. G.-J. Boons, T. Stauch, Synlett, 1996, 906-908. 121. N. M. Spijker, C. A. A. van Boeckel, Angew. Chem. Znt. Ed. Engl., 1994, 30, 180-182. 122. H. Lonn, J. Curbohydr. Chem., 1987, 6, 301--306. 123. I. Braccini, C. Derouet, J. Esnault, C. H. d. Penhoat, J. M. Mallet, V. Michon, P. Sinay, Curbohydr. Rex, 1993,246, 23-41.

1 16

4 Thioglycosides

124. A. Demchenko, T. Stauch, G.-J. Boons, Synlett, 1997, 818-820. 125. P. Svahnberg, Licenciate Thesis, Stockholm University 1999. 126. P. J. Garegg, S. Oscarson, H. Ritztn, M. Szonyi, Carbohydr. Rex, 1992, 228, 121-128. 127. S. Houdier, P. J. A. Vottero, Carbohydr. Res., 1992,232, 349-352. 128. A. V. Demchenko, E. Rousson, G.-J. Boons, Tetrahedron Lett., 1999, 40, 6523-6526. 129. G. W. Griffin, N. C. Bandara, M. A. Clarke, W . 3 . Tsang, P. J. Garegg, S. Oscarson, B. A. Silwanis, Heterocycles, 1990,30, 939-947. 130. C. Amatore, A. Jutand, J.-M. Mallet, G. Meyer, P. Sinay, J. Chem. SOC.Chem. Commun., 1990, 718-719. 131. G. Balavoine, A. Greg, J.-C. Fischer, A. Lubineau, Tetrahedron Lett., 1990,31, 5761-5764. 132. A. Marra, J.-M. Amatore, C. Amatore, P. Sinay, SynLett., 1990, 572-574. 133. J.-M. Mallet, G. Meyer, F. Yvelin, A. Jutand, C. Amatore, P. Sinay, Carbohydr. Rex, 1993, 244,237-246, 134. Y.-M. Zhang, J.-M. Mallet, P. Sinay, Curbohydr. Res., 1992, 236, 73-88. 135. S. Mehta, B. M. Pinto, Carbohydr. Rex, 1998, 310, 43-51. 136. N. K. Kochetkov, E. Klimov, N. N. Malysheva, Tetrahedron Lett., 1989,30, 5459-5462. 137. N. K. Kochetkov, N. N. Malysheva, E. Klimov, A. V. Demchenko, Tetrahedron Lett., 1992, 33, 381-384. 138. A. Marra, L. K. S. Shun, F. Gauffeny, P. Sinay, Synlett, 1990, 445-448. 139. A. Marra, F. Gauffeny, P. Sinay, Tetrahedron, 1991, 47, 5149-5160. 140. A. Marra, P. Sinay, Carbohydr. Res., 1990, 195, 303-308. 141. H. Lonn, K. Stenvall, Tetrahedron Lett., 1992, 33, 115-116. 142. U. Ellervik, G. Magnusson, J. Orq Chem., 1998, 63, 9314-9322, 9223. 143. P. Fugedi, P. J. Garegg, S. Oscarson, G. Rosen, B. A. Silwanis, Carbohydr. Res., 1991, 211, 157-1 62. 144. S. Hashimoto, T. Honda, S. Ikegami, Tetrahedron Lett., 1990,31, 4769-4772. 145. H. Bielawska, M. Michalska, J. Curbohydr. Chem., 1991, 10, 107-1 12. 146. L. Laupichler, H. Sajus, J. Thiem, Synthesis, 1992, 1133-1136. 147. 0.J. Plante, P. H. Seeberger, J. Org. Chem., 1998,63, 9150-9151. 148. D. A. Leigh, J. P. Smart, A. M. Truscello, Carbohydr. Res., 1995, 276, 417-424. 149. F. Belot, J.-C. Jaquinet, Carbohydr. Res. 290, 1996, 290, 79-86. 150. D. R. Mootoo, P. Konradsson, U.Udodong, B. Fraser-Reid, J. Am. Chem. Soc., 1988, 110, 5583-5584. 151. N. L. Douglas, S. V. Ley, U. Lucking, S. L. Warriner, J. Chem. SOC.Perkin Trans. I , 1998, 51-65. 152. Z. Zhang, I. R. Ollman, X.-S. Ye, R. Wischnat, T. Baasov, C.-H. Wong, J. Am. Chem. SOC., 1999.121. 734-753. 153. A. K. Choudhury, I. Mukherjee, B. Mukhopadhyay, N. Roy, J. Carbohydr. Chem., 1999,18, 361-367. 154. S. Cao, F. Hernandez-Mateo, R. Roy, J. Carbohydr. Chem., 1998, 17, 609-631. 155. L. A. J. M. Sliedregt, K. Zegelaarjaarsveld, G. A. van den Marel, J. H. van Boom, Synlett, 1993, 335-337. 156. R. Roy, F. 0. Andersson, M. Letellier, Tetrahedron Lett., 1992, 33, 6053-6056. 157. M. Lahmann, S. Oscarson, unpublished results. 158. F. Baressi, 0. Hindsgaul, J. Am. Chem. SOC.,1991, 113, 9376-9377. 159. M. Bols, J. Chem. SOC.Chem. Commun., 1992,913-914. 160. F. Baressi, 0. Hindsgaul, Can. J. Chem., 1994, 72, 1447-1465. 161. Y. Ito, T. Ogawa, Ang. Chem. Int. Ed. Engl., 1994,33, 1765-1767. 162. C. Krog-Jensen, S. Oscarson, J. Org. Chem., 1998, 63, 1780-1784. 163. T. Ziegler, R. Lau, Tetrahedron Lett., 1995, 36, 1417-1420. 164. S. Valverde, A. M. Gomez, J. C. Lopez, B. Herradon, J. Chem. SOC.Chem. Commun., 1995, 2005-2006. 165. M. Miiller, U. Huchel, A. Geyer, R. R. Schmidt, J. Org. Chem., 1999, 64, 6190-6201. 166. K. C. Nicolaou, N. Winssinger, J. Pastor, F. DeRoose, J. Am. Chem. Soc., 1997,119,449-450. 167. K. C. Nicolaou, N. Watanabe, J. Li, J. Pastor, N. Winssinger, Angew. Chem. Int. Ed. Engl., 1998,37, 1559-1561.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

5 Glycosylation Methods: Use of Phosphites Zhiyuan Zhang and Chi-Huey Wong

5.1 Introduction

In the chemical synthesis of oligosaccharides, the overall procedure can be divided into five steps: 1) design and preparation of the building blocks (glycosyl donors and acceptors) with selected protecting groups; 2) manipulation of the protecting groups to free the desired hydroxyl group; 3) separation of products after each step; 4) stereo-controlled glycosylation; and 5 ) deprotection and product purification. Of those steps, 1-3 are the most time-consuming. Considerable efforts have thus been made to facilitate the synthetic process. For example, many methods have been developed to reduce the number of steps in the synthesis of building blocks. Recently developed orthogonal protection-deprotection strategies [ 1, 21 provide the efficiency and flexibility for the preparation of highly diversified carbohydrate libraries [ 1, 31. To minimize protecting group manipulation, the armed-disarmed strategy [4], the orthogonal glycosylation strategy [ 5 ] , and the one-pot multiglycosylation strategy [6-81 have been developed. In terms of simplifying the purification procedure, solid-phase synthesis [91 and one-pot solution-phase synthesis [7d, 81 have shown great promise. Glycosylation is the central reaction of carbohydrate chemistry. Because low yield, low regioselectivity, and low stereoselectivity are frequently encountered, it is this step which requires special attention. Though chemists might be able to synthesize any oligosaccharide [lo],there is no universally employed glycosylation [ lOc, 111. Therefore, development of new and efficient methods of glycosylation is still quite challeng,ing.

118

5 Glycosylation Methods: Use of Phosphites

The use of glycosyl phosphite as a glycosyl donor was initiated in 1992. The Wong group [12] and the Schmidt group [13] independently reported that trimethylsilyl triflate (TMSOTf) could activate glycosyl phosphites in glycosylation reactions. An advantage of this method is that it provides a very high yield and aselectivity in the synthesis of sialic acid derivatives [12, 131. This method has been further developed by these two groups and others. Different promoters have been investigated and some unusual stereoselectivities have been reported. The purpose of this chapter is to review recent developments in phosphite glycosylation and its applications.

5.2 Preparation of Glycosyl Phosphites Glycosyl phosphites are prepared from their corresponding anomeric hydroxyl compounds, either by treatment with dialkyl N,N-diethylphosphoramidites (DDP) and 1,2,4-triazole [ 141 or with dialkyl phosphorochloridites in the presence of Hunig's base (Figure 1) [ 13, 151. Most phosphitylation reagents are commercially available, and both methods give glycosyl phosphites in high yield. Examples of glycosyl phosphites that can be prepared by these methods are dibenzyl phosphites, diethyl phosphites, and dimethyl phosphites. Bicyclic glycosyl phosphites such as 4 and 5 can also be prepared. In the synthesis of an N-acetyllactosamine (LacNAc) derivative it has been shown that 4 gives better yields than the corresponding trichloroacetimidate [ 14fl. When the sugar moiety is highly electron-donating, such as the 2,6-di-deoxy derivative, its diethyl phosphite derivative could not be prepared.

Et2N-P-(OBn),

PCH

A . d

I

COOMe

1,2,4-triazole AcO

1

CI-P-(0Et)2 EtNiPr2

4

3 97%

AcO

Figure 1.

COOMe

5.3 Glycosylotion using Glycosyl Phosphites

1 19

The corresponding pinacol phosphite 5 was, however, stable enough to be isolated in 67% yield [16]. Several phosphitylation reagents are now available for the preparation of glycosyl phosphites with desirable reactivity. Factors known to affect the anomeric ratio in the preparation of glycosyl phosphites include the anomeric composition of the starting hydroxyl compound. For example, a mixture of galactosyl phosphite with an a : ratio of 90 : 10 was obtained with an a : ratio of 90: 10. A different from 2,3,4,6-tetra-O-benzylgalactopyranose anomeric mixture ( a :p = 61 :39) of the phosphite product was, however, generated from an anomeric mixture of the galactopyranose ( a :p = 56:44) [17].2,3,4,6-Tetra0-acetylgalactose ( a :p = 3 : 1) and glucose ( a :p = 2: 1) gave the phosphite p isomer as a predominant product ( a :p = 1 :4 and 1 :2 respectively), whereas the 2-acetamido-2-deoxy sugar gave mainly the a-phosphite ( a :p > 7: 1) [18]. Surprisingly, only phosphite p isomers (2 and 3) were observed when sialic acid (NeuSAc) derivatives were phosphitylated using either phosphitylation method [ 12, 131. Solvents also affect the product composition. Phosphitylation of 2,3,4,6-tetra-Oacetylgalactose in CH2C12 gave the dibenzyl phosphite with an a : p ratio of 1 :2; this increased to 1 : 6 when THF was used as solvent [ 181. Effects of Lewis acid [ 191 and mild base [20] on the anomeric ratio of the phosphite have also been reported. In general the anomeric ratio of the phosphite does not have a large effect on the glycosylation step in terms of reactivity and stereoselectivity [ 171. It might, however, be a factor in the synthesis of phosphate derivatives from phosphites (see Section 5.4).

5.3 Glycosylation using Glycosyl Phosphites 5.3.1 Mechanism The first catalyst used to activate the glycosyl phosphite was TMSOTf [ 12, 131. The mechanism of the reaction has been investigated [12]. As shown in Figure 2, path a applies when TMSOTf is mixed with donor first or added into the mixture of donor and acceptor. In this method, the formation of a substantial amount of phosphonate orthoester is a considerable problem. As illustrated in path b, however, a premixed solution of TMSOTf and acceptor spontaneously generates triflic acid (TfOH). TfOH then activates the glycosyl phosphite added subsequently to give the desired product, and reduces the formation of the orthoester. The most facile method is through path c where TfOH directly catalyzes the glycosylation, and also reduces the nucleophilicity of the hydrogen phosphonate. TfOH is actually involved in each of the pathways, and is potentially the real catalyst even when TMSOTf is used [21]. In general, TfOH-catalyzed glycosylation gives the highest yield. Notably, addition of glycosyl phosphite to premixed TMSOTf and acceptor gave a better yield than addition of TMSOTf to the mixture of donor and acceptor [22]. In practice, however, it is more convenient to use TMSOTf than TfOH as catalyst. Interestingly, BF3.Et20-promoted [ 171 and TMSOTf-promoted [ 12, 13, 211 gly-

120

5 Glycosylation Methods: Use of Phosphites

a. TMSOTf reacts with donorfirst

-

TMSO-P(OBn)2

TMSOTf

HOTf

b. TMSOTfreacts with acceptorfirst

TMSOTf

TMSOR HO-P(OBn),

-

R

HP(OBn),

c. When TfOH was used

Figure 2.

cosylations using perbenzylated phosphites at low temperature (-78 "C), as reported by Hashimoto et al., gave similar stereoselectivity and yields. Both activation gave high 1,2-trans-P-selectivityeven without a participating group at the C-2 position of the glycosyl donor [ 171. BF3.EtzO-promoted glycosylations, furthermore, gave even better P-selectivity. The reason for this, and the nature of the intermediate involved in the BF3 .Et20-promoted glycosylation, remain unclear [ 171. More recently, the Hashimoto group developed a new activation method using 2,6-di-tert-butylpyridinium iodide (DTBPI) and tetrabutylammonium iodide (BmNI), and achieved high a-selectivity. The intermediate was found to be the aglycosyl iodide, which is first anomerized to the P isomer then glycosylated through a S~2-likemechanism (Figure 3). In general, a-selectivity is predominant in glycosylations using the phosphite method. Watanabe et al. [22] have investigated other Lewis acids as promoters of glycosylation using glycosyl phosphites; they found that ZnCl2, ZnCl2-AgC104, 12, and other Lewis acids can efficiently activate perbenzylated phosphites at room temperature. The stereoselectivities are generally lower than when reaction is performed out at low temperature. The effect of promoters on the glycosylation of perbenzylated glucosyl phosphite 6 with a primary alcohol was studied (Figure 4).ZnClz-

5.3 Glycosylation using Glycosyl Phosphites

121

60-95% cc/B > 90110

8 donors

Figure 3.

promoted glucosylation gave predominantly the P product, 7. In contrast, 12promoted reactions mainly gave the a product, whereas almost no selectivity was observed with the ZnClz-AgC104-promoted glycosylation [ 17b]. It is unclear why the addition of AgC104 would increase the efficiency and reduce the selectivity of the reaction. The origin of selectivity remains to be investigated [17b]. Metal perchlorate salts (e.g. LiC104), which act as weak Lewis acids, activate the glycosyl phosphite and yield the glycosyl perchlorate ion pair which leads to the product [23] This method provides a useful alternative when acid-labile protecting groups, or acid-sensitive glycosyl linkages are involved. The low yield and low stereoselectivity are potential drawbacks of this method. 5.3.2 Low Temperature-Dependent Stereoselectivity The stereoselectivity of a glycosylation is controlled by many factors, including the substituent at the C-2 position, the leaving group on the anomeric center, temperature, solvents, and the structural features both of glycosyl donor and acceptor. In the construction of a glycosyl P-linkage, a participating group at the C-2 position of the glycosyl donor is normally required. In the absence of this participating group the thermodynamically favorable a-glycosyl bond is generally formed because of the anomeric effect [24]. Thus, the stereo-controlled synthesis of the P-linkage is considered to be difficult. A solution to this problem is to use acetonitrile as solvent to provide the nitrile effect [25] to aid the formation of the P-linkage. In addition, some newly developed glycosylatioii methods using glycosyl donors with a non-participating group at the C-2 position at low temperature gave very high p-selectivities [17, 21, 22b, 26, 271. These methods include the use of sulfoxides activated with Tf20 [27, 281, glycosyl bromides activated with AgOTf [27], thioglycosides activated with RSCl-AgOTf

Lewis acid

Ph(CH&OH

B

%B

Lewis acid

6

+

BB& O(CH.&Ph

OP(O5h 7

Figure 4.

Yield

1,

92% 81/19

ZnCI,

88% 20180

ZnCI2-AgC1O4 92% 52/48

122

5 Glycosylation Methods: Use of Phosphites

I

Temp.

-78OC

-5OOC

8

3:97

8:92

11:89

1:99

2:98

3:97

-3oOc

8

BF3-Et20

Ratio

ROH

9

(b.) Watanabe eta/.

6 I

LNlS/TfOtJ CH2C12

-\

B"4 BnO 7

O(CH&Ph

3°C

-21 "C -45°C -68"C

Figure 5.

[27, 291 and glycosyl phosphites activated with TMSOTf [17, 211 or BF3.Et20 [17, 261. Low temperature seems to promote P-selectivity; the thermodynamically stable a-glycosyl triflate is formed and further reacted in a S~2-likemechanism with the added glycosyl acceptor to give the P-glycoside [27b, c]. Figure 5 shows the temperature-dependent selectivity reported by Hashimoto et al. [17] and Watanabe et al. [22b] Further study showed that P-selectivity was almost independent of solvent [ 171, although slightly dependent on the anomeric composition of the donor [17, 211. This is exemplified by the synthesis of Pgalactosyl-, P-glucosyl-, and P-glucuronyl linkages [ 14e, 16, 17, 21, 26, 301. Fucosyl phosphites behave differently under these conditions, however, and a-selectivity is predominant [21, 30, 311. The stereoselectivity issues of glycosyl phosphites will be discussed later.

5.3.3 Glycosylation of Sialyl Phosphites Sialylated oligosaccharides are often expressed on the surfaces of cells, including cancer cells, and are very attractive targets for investigation. Sialylation [ 321, especially yield and stereoselectivity, is, however, still a major problem in oligosaccharide synthesis. Synthetic difficulties stem from the tertiary-hindered anomeric center and the lack of an electron-demanding group adjacent to the anomeric center of sialylic acid (NeuAc) [ 121. Sialylation using sialyl phosphites is currently one of the most successful synthetic methods, and many sialylated compounds have been prepared by this strategy. Compounds 10-11 and 13 (see Figure 6) were prepared from dibenzyl phosphite 2 [12], and compounds 10, 12 and 13-14 were synthesized from diethyl phosphite 3 [13]. In the synthesis of sialyl Lewisa (sLea), a regioselec-

5.3 Glycosylution using Glycosyl Phosphites A

123

OOMe

AcO 10 80% (a:p6:1) from 2 70% (cx:p4:1) from 3

OMe

Ho

PcHN 11 48% only cx from 2

OpC

Ph

13 44% (a:p6:1)from 2

12 38% cx from 3

Figure 6.

tive coupling of the sialyl moiety to the 3-OH on the Gal unit of partially protected Lea gave the desired compound 14 in low yield (18%). Also isolated was 35% of the lactonized product 15 [33]. This lactone side-product 15 was isolated as the only product (37%) when the same acceptor and a different glycosyl donor and promoter were used [34].

5.3.4 Glycosylation of C-2-Acylated Glycosyl Phosphites In general, the 1,2-tuans P-glycosyl linkage is formed by use of a C-2 participating protecting group. Catalytic amounts of TMSOTf (0.5 equiv.) can promote the glycosylation of peracetylated glycosyl phosphites. Donors utilized include fucosyl phosphites, galactosyl phosphites, glucosyl phosphites, glucuronyl phosphites, and glucosaminyl phosphites. Reaction of these donors with different glycosyl acceptors yield the desired P-product with yields of 40-70% [21, 351. T h s method has also been used for the synthesis of some P-C-glycosides and P-S-glycosides [21, 301. A few side-reactions are, however, encountered when using this method; examples include the formation of glycosyl phosphonate orthoesters [21, 30, 361, which are less reactive intermediates, the ester transformation problem [21, 361, and in some instances the formation of silylated acceptor byproducts [21]. Side-reactions often depend on the type of catalyst used in the reaction. Perbenzoylated glucosyl phos-

124

5 Glycosylution Methods: Use of Phosphites

Oh

BzO 16 R' =Bn, R2,R3 = PhCH, %YO, a only 17 R' =R3 = Bn, R2= H, 73%, a:p1:1 18 R' =R2=H, R3= Bn 88%, p only

BzO 19

loo%, aonly

20 93%, ponly

Figure 7.

phite in the presence of BF3.Et20 (1 equiv.), for example, gave 24% of the phosphonate orthoesters whereas use of 1 equiv. TMSOTf at -50°C gave the desired 1,2 trans-p product in excellent yield, with no traces of the aforementioned byproducts [36]. Other promoters such as ZnC12, ZnC12-AgOTf, and metal perchlorate salts have not been used for the activation of acylated glycosyl phosphites. Stereo-controlled synthesis of fructofuranosides is regarded as a difficult problem, even with a participating group next to the anomeric center [36]. The study of TMSOTf-catalyzed glycosylation using fructofuranosyl phosphite showed that a- or P-selectivity was highly dependent on the properties of the acceptor (Figure 7) [36]. 5.3.5 Glycosylation with C-2-0-Benzylated Glycosyl Phosphites

The glycosylation reaction using glycosyl phosphites without a participating protecting group at C-2 is more complicated. The stereoselectivity seems to depend on many factors, including reaction temperature, the core structure of the glycosyl phosphite, and the properties of acceptors and the promoters. Glycosylation using Glucosyl Phosphites with a Benzyl Group at C-2 Glycosylution with 0-selectivity Glucosylation of a primary hydroxyl acceptor with perbenzylated glucosyl phosphite, promoted by TMSOTf at -78"C, gave nearly exclusively the P-product 8 (66%, a : P = 1 :99 in CH2C12; and 78%, a : P = 2: 99 in EtCN). The P-selectivity decreased, giving the anomeric mixture 9 (83%, a : P = 32: 68 in CH2C12; and 65%, a : f3 = 17:83 in EtCN) when a secondary alcohol acceptor was used [17, 211. Of the many promoters studied, BF3.Et20 gave products in the highest yield and Pselectivity, even with secondary hydroxyl acceptors (a:P = 5 :95) [ 171. ZnC12-promoted glucosylations [22] gave lower 0-selectivity than the lowtemperature procedure (TMSOTf or BF3.EtzO), most probably because of the higher reaction temperature. ZnC12-AgC104 [22] was, however, successfully employed in the total synthesis of paeoniflorin reported by Corey and Wu [37]. In this instance glycosylation of the very hindered acceptor 22 (a mixture of enantiomers) with perbenzylated glucosyl phosphite 21 gave the enantiomerically pure 23 (1 8%,

5.3 Glycosylation using Glycosyl Phosphites

'& B

TIP93

22

OP(OMe)2

BnO

BB *

ZnC12-AgClO4

21

x

Y

@

BnO

M

125

e TIPS3

23 18% enantiomeric pure p

Figure 8.

see Figure 8). This process was more successful than any other glycosylation method, including the TMSOTf-catalyzed phosphite procedure. Glycosylation with a-selectivity From the results discussed above it seems that glycosyl phosphites are not adequate glycosyl donors for the formation of a-glycosyl linkages. Despite this, glycosylations using the perbenzyl glucosyl phosphite donor, in the presence of I2 at room temperature, gave the a-glycosylated compounds 7 and 24 as major products. Glucosylation of a hindered acceptor with the glucosyl phosphite, under ZnCl2AgC104 conditions, gave 9 (88%, a : p = 7 : 3, see Figure 9) with lower selectivity [22b]. The best a-selectivity was achieved under DTBPI-Bu4NI conditions to give product 9 in a 95 : 5 a : p ratio. Glycosylation using Galactosyl and Fucosyl Phosphites with a Benzyl Group at C-2

0-Galactosylation using perbenzylated galactosyl dibenzyl phosphite, in the presence of TMSOTf or BF3.Et20 at -78"C, gave nearly pure products [17, 211. In addition, ZnC12-promoted galactosylation of secondary and primary alcohols using galactosyl diethyl phosphite gave the 0 product with selectivity (a: p 1 :4)in the same range as that obtained in glucosylations [22]. To achieve high a-selectivity, 60-acetyl galactosyl dibenzyl phosphite 25 was used to regioselectively glycosylate the 3-0-position of a partially protected acceptor in CH2C12 at - 15"C to give predominantly product 26 in 60% yield (Figure 10) [30]. Fucosylation using fucosyl dibenzyl phosphites has been reported in the synthesis of Lea and Ley catalyzed by TfOH [31]. The phosphite and trichloroacetimidate

s

-

BnO B

O(CH&Ph 7 92%, a l p = 81/19

Figure 9.

24 91%, a@ = 81/19

9 88%, cr/p = 69/31

5 Glycosylution Methods: Use of Phosphites

126

TMSOTf

(0.2equiv) -15"C

B

oP(ow2

25

BB g E # %

b

OBl

60%

HO 26 cL/p = 85/15

Figure 10.

methods have been compared for the synthesis of Lea (Figure 1 1 ) [35].It was concluded that 0-glycosyl phosphites nicely complement 0-glycosyl trichloroacetimidates, especially in the higher reactivity range where trichloroacetimidates tend to be less stable. In the course of development of sialyl LewisXmimetics, the Wong group conducted extensive studies on the fucosylation of more than ten different N-acylated p-hydroxyl amino acids. They found that fucosylation of benzyl-protected amino acids with perbenzylated fucosyl phosphite gave a greater yield and better aselectivity than the corresponding ethyl-protected amino acids [30, 311. Figure 12 depicts representative syntheses of these compounds. In addition, the DTBPIBWNI-promoted fucosylation reported by Hashimoto's group also gives excellent a-selectivity [ 17b]. Glycosylation using other Glycosyl Phosphites with a Benzyl Group at C-2

The perbenzylated glucuronyl phosphite has shown preference for 1,2-truns-Pselectivity, similar to that of glucosyl phosphite, when coupled with primary and secondary alcohols in the presence of BF3.Et20 at -65°C [22]. Synthesis of the fi-mannosyl linkage is a major problem in glycosylation reactions. Many efforts have been made to improve the p-selectivity of mannosylation using, for example, insoluble Ag silicate in glycosylation [38],inversion at the C-2 of p-glucosyl derivative after glycosylation [391, and intramolecular glycosylation methods [40].Recently, p-mannosylation was also acheved at low temperature through the triflate intermediate which is generated by different activation methods [27](see Section 5.3.2). A similar result might be expected from activation of mannosy1 phosphite with TMSOTf or TfOH, because it might proceed through the

A Ac

E

x TMSOTf

____)

Et20 A

AcO 27 X = OP(OCH$CI& 28 X = OC(NH)CC13

Figure 11.

sRq

OBl

TMSOTf

___)

OTDS EtpO

opc 29 64%

88%

N3

AC

Opc

30 88% 78%

5.3 Glycosylution using Glycosyl Phosphites

127

3 2 H 2 1 L

-

BocH

31

CHJ

TfOH, CH2Cl2, -15 "C

OP(QW2

33 84% R= Bn, c(/p=99/1 34 72% R= Et,dD=71/29

Figure 12.

glycosyl triflate intermediate. Indeed, the Hashimoto group recently showed that 2,3-di-O-benzyl-4,6-O-benzylidine-mannosyl diethylphosphite preferentially gives P-selectivity [ 17~1,although ZnCl2- or ZnClz-AgC104-catalyzed mannosylations from the glycosyl phosphites have been reported to afford a- or P-selectivity. When the perbenzylated mannosyl phosphite is glycosylated with cyclohexanol under ZnCl2 conditions, slight preference for the p-anomeric linkage (83%; a : P 2 : 3 ) is obtained. When the glycosylation is performed with methyl 2,4,6-tri-O-benzyla-glucopyranoside and ZnClz-AgCIO4 activation, the product is pure a-linked disaccharide in 76% yield [22]. The phosphite method has also been used in the synthesis of di-mannosylated phosphatidylinositol with two a-glycosyl linkages, as shown in Figure 13. In the first glycosylation, ZnClz-AgC104 was used to activate perbenzylated glycosyl phosphite to give a 2: 1 alp mixture of 36a. The same conditions gave the pure a product 36b in 70% yield when a 2-OAc mannosyl phosphite was used as donor. Interestingly, at room temperature TMSOTf-catalyzed mannosylation using the perbenzylated phosphite gave a-glycoside 37 in 74% yield [41]. It is not clear how the temperature affects the stereoselectivity of glycosylations employing benzylated mannosyl phosphite with TMSOTf or TfOH as promoter. N

5.3.6 Glycosylation with 2-Deoxy Glycosyl Phosphites The synthesis of 2-deoxy oligosaccharides with controlled a- or P-selectivity remains a difficult problem [lOc, 421 and many methods have been developed to solve it.

@-

ZnC12AgC104

P

"'OH Et20

TIP<S

OH RO

35

Figure 13.

oh

36a: R=Bn 65% a:p2:1 36b: R=Ac 70% a only

5 Glycosylation Methods: Use of Phosphites

128

D-A Temp. Yield a:O

38-43 -78 "C 91% 19:81 38-43 -50

"C

87% 29:71

39-43 -94

"c

93% 13187

40-43 -94

"C 97% 1436

41-43 -94 "C 95% 23:77 42-43 -94 "C 96% 67:33 41-44 -94 "C 86% 69:31

38

39

40

Bn9 ,ow B

w (EtOhPO F

41

OH

BBs%q

43

44

B* OMe

Figure 14.

Examples include electrophile-mediated addition of acceptor to glycals [42a], and glycosylation of the C2-substituted glycosyl donor [42b] followed by reductive removal of the C-2 substituent [ lOc]. Direct stereoselective glycosylation using a 2deoxy glycosyl donor is, however, desirable. 'Remote-controlled' [43] activation of the 2-pyridine-thioglycoside by AgOTf [44]provides a very efficient method for the a-selectively controlled synthesis of 2-deoxy glycosides. Use of insoluble silver silicate for the activation of 2-deoxy glycosyl halides provides one of the best methods of preparation of 2-deoxy-0-glycosides [45]. The Schmidt group has studied the glycosylation reaction using 2-deoxy glycosyl phosphites activated by BF3 .Et20 in CH2Cl2-hexane or in Et20 [36]. Preferential a-selectivities were obtained in the formation of both the 2-deoxy galactosyl linkage and the 2-deoxy glucosyl linkage. In the glycosylation of 2-deoxy glycosyl phosphites under TMSOTf activation [12, 131, the Hashimoto group [26] recently found that the reaction temperature is crucially important in the control of P-selectivity (see Figure 14, entries 1-3). The selectivities were not significantly affected by the reactivities of the glycosyl donors (entries 1, 4, and 5; a :P = 10: 90, 13 :87, and 14: 86, respectively). They are, however, affected by other structural properties of the glycosyl donors. For example, compounds 41 and 42 with axial substituents at C-4 gave lower p-selectivities (entry 6, a : P = 23: 77 and entry 7, a : P = 67: 33). The p-selectivity was observed to increase with increasing reactivity of the acceptor alcohol (compare entries 6 and 8). In the synthesis of the hexasaccharide fragment of landomycin A, the desirable p-glycosyl linkage of the 2,6-dideoxy unit was successfully formed by use of the TMSOTf-promoted phosphite method (Figure 15) [ 161.

5.4 Other Applications of Glycosyl Phosphites Besides the 0-,S-, C-glycosylations and 0-glycopeptide synthesis discussed above, other applications of glycosyl phosphites are discussed below.

5.4 Other Applications of Glycosyl Phosphite

129

TMSOTf -94 OC AcO

HO

c,

AcO

47

&*+

OMP

TMSOTf, Toluene,

0 M

48 R' = AC (42%, a:p 5 2 ~ 8 )

A d

Figure 15.

5.4.1 Synthesis of CMP-NeuAc

The synthesis of CMP-NeuSAc remains of interest, because it is a glycosyl donor in glycosyltransferase-catalyzed reactions. Three chemical synthetic methods based on the phosphite strategy have been reported. The Wong group employed a glycosyl phosphoamidate (Figure 16A) [12], whereas the Halcomb group used a nucleoside phosphoamidate in the synthesis of CMP-NeuSAc (Figure 16B) [46a]. The latter method seems preferable, because of its high-yielding deprotection step. The Schmidt group utilized the diethyl glycosyl phosphite and the nucleoside phosphate 53 in its acid form to give P-CMP-NeuAc 54 in 50% yield (Figure 17) [46b].The reaction seems to be mediated by acid catalysis.

5.4.2 Synthesis of GDP-Fucose

GDP-fucose is a glycosyl donor substrate for fucosyltransferases. Although it has been synthesized by many chemical methods [47],the use of fucosyl phosphite in the synthesis seems the most practical [48]. A recent improvement of this method includes the addition of tetrazole to the morpholidate coupling reaction [48b]. This method might also be applicable to the synthesis of other nucleoside diphosphate sugars (Fig-ure 18). Together with the synthesis of CMP-NeuSAc discussed above, glycosyl phosphite chemistry indeed provides a useful alternative for the synthesis of natural and unnatural donor substrates for the study of enzyme-catalyzed glycosylations.

130

5 Glycosylation Methods: Use of Phosphites NHBz 1. HOR 2. BuOOH 12 Yo

50

51

52

Figure 16.

HQ

PH

HAC

CH ,CN/TH FIDMF

0 3

AcO 01% 53

(1:l:l)

-15 "C to rt.,

AcO AcO 54 50%

Figure 17.

1.30% H202 2. H2IPd-C F 0 $ O N a h AcO F z z B n 0 ) 2 3. NaOWH20 HO OH 55 63% 54 cr:p =1:10 tetrazolej Pyr

57 91%

Figure 18.

Ho

OH

OPC

5.4 Other Applications of Glycosyl Phosphite Ac

131

TMSOP(OMe)* TMSOTf CHzCI2

PCH

A d

ACQ 58 63 Yo Ponly

3 Figure 19.

5.4.3 Formation of Glycosyl Phosphonate Glycosyl phosphites can be converted by Arbuzov-type fragmentation to the corresponding glycosyl phosphonates as analogs of glycosyl phosphates [49]. Glycosyl phosphonates are more stable than glycosyl phosphates [50]and can act either as inhibitors [5I] or as regulators in metabolic processes where natural phosphates function. Figure 19 shows the synthesis of a sialyl phosphonate starting from the glycosyl phosphite [52]. 5.4.4 Transformation to other Types of Glycosyl Donor Phosphate Oxidation [I 8, 481 of glycosyl phosphites provides glycosyl phosphates which have been used for the synthesis of oligosaccharides [53] and sugar nucleosides [12, 81 (Figure 20).

rA-OP(ORh

oxidation

8

* a O F ' ( O R h

HOR TMSOTf

OR

Figure 20.

Phosphorimidate More recently, a new glycosylation method has been reported using glycosyl phosphorimidates as glycosyl donors. Thus, compounds were prepared from the corresponding glycosyl phosphites (Figure 21) 1541.

Figure 21.

132

5 Glycosylation Methods: Use of Phosphites

In summary, use of glycosyl phosphites in a new glycosylation method has received much attention for several reasons, including the high reactivity of the glycosyl phosphite at low temperature, the high efficiency of glycosylation with stereo-hindered donors or acceptors, and the high stereoselectivity of the reaction. Applications of glycosyl phosphites in the synthesis of glycosyl phosphates and glycosyl phosphonates are, furthermore, especially useful. Applications and modifications of glycosyl phosphite chemistry will surely be of interest to many chemists in the future.

References 1. Wong, C.-H., Ye, X.-S., and Zhang, Z . J. Am. Chem. SOC.1998,120, 7131. 2. T. Wubberg, C. Kallus, T. Opatz, S. Henke, W. Schmidt, H. Kunz, Angew. Chem. Int. Ed. Engl. 1998,37, 2503-2505. 3. Random glycosylation: (a) 0. Kanie, F. Barresi, Y. Ding, J. Labbe, A. Otter, L. S. Forsberg, B. Ernst, 0. Hindsgaul, Angew. Chem. Znt. Ed. Engl. 1995, 34, 2720. (b) Y. Ding, J. Labbe, 0. Kanie, 0. Hindsgaul, Bioorg. Med. Chem. 1996, 4, 683. Latent-active glycosylation: G.-J. Boons, B. Heskamp, F. Hout, Angew. Chem. Znt. Ed. Engl. 1996,35,2845. Solid-phasemethod: R. Liang, L. Yan, J. Loebach, M. Ge, Y. Uozumi, K. Sekanina, N. Horan, J. Gildersleeve, C. Thompson, A. Smith, K. Biswas, W. C. Still, D. Kahne, Science 1996,274, 1520. 4. (a). D. R. Mootoo, P. Konradsson, U. Udodong, B, Fraser-Reid, J. Am. Chem. SOC.1988,110, 5583; (b). B, Fraser-Reid, Z. Wu, U. D. Udodong, H. Ottosson, J. Urg. Chem. 1990, 55, 60686070. 5. Y. Ito, T. Ogawa, J. Am. Chem. SOC.1994,116, 12073. 6. (a). S. Raghavan, D. Kahne, J. Am. Chem. SOC.,1993,115, 1580. (b) H. Yamada, T. Harada, H. Miyazaki, T. Takahashi, Tetrahedron Lett., 1994, 35, 3979. (c) H. Yamada, T. Harada, T. Takahashi, J. Am. Chem. SOC.1994, 116, 7919. (d) H. K. Chenault, A. Castro, Tetrahedron Lett., 1994, 35, 9145. (e) R. Geurtsen, D. S. Holmes, G.-J. Boons, J. Urg. Chem. 1997, 62, 8145. 7. (a) S. V. Ley, H. W. M. Priepke, Angew. Chem. Int. Ed. Engl. 1994,33,2292. (b) P. Grice, S. V. Ley, J. Pietruszka, H. W. M. Priepke, E. P. E. Walther, Synlett 1995, 781. (c) P. Grice, S. V. Ley, J. Pietruszka, M. I. Osborn, W. M. Henning, H. W. M. Priepke, S . L. Warriner, Chem. Eur. J. 1997, 431. (d). N. L. Douglas, S. V. Ley, U. Lucking, S. L. Warriner, J. Chem. Soc. Perkin Trans. 1, 1998, 51. (e). L. Green, B. Hinzen, S. J. Ince, P. Langer, S. V. Ley, S . L. Warriner, Synlett. 1998, 4, 440. 8. Z. Zhang, I. R. Ollmann, X.-S. Ye, R. Wischnat, T. Baasov, C.-H. Wong, J. Am. Chem. SOC. 1999, 121, 734. 9. Solid-phase synthesis: For review see, (a). P. H. Seeberger, S. J Danishefsky, Acc. Chem. Res. 1998, 31, 685. (b). J. S. Fruechtel, G. Jung, Angew. Chem., Int. Ed. Engl. 1996, 35, 17; For general articles see, (c) M. J. Sofia, Comb. Chem. Mol. Diversity Drug Discovery, Eds: E. M. Gordon, J. F. Kenvin Jr., Wiley-Liss, New York, 1998, 243-269; (d). R. R. Schmidt, Pure Appl. Chem. 1998, 70, 397-402; (e). Z.-W. Guo, Y. Nakahara, Y. Nakahara, T. Ogawa, Angew. Chem., Znt. Ed. Engl. 1997, 36, 1464; (f). R. Rodebaugh, S. Joshi, B. Fraser-Reid, H. M. Geysen, J. Org. Chem. 1997, 62, 5660; (8). S. J. Danishefsky, K. F. McClure, J. T. Randolph, R. B. Ruggeri, Science 1993,260, 1307; For enzymatic methods see, (h). M. Schuster, P. Wang, J. C. Paulson, C.-H. Wong, J. Am. Chem. SOC.1994,116, 1135. 10. For the recent review articles on oligosaccharide synthesis see: (a) H. Paulsen, Angew. Chem. Int. Ed. Engl., 1990, 29, 823; (b) J. Banoub, Chem Rev. 1992, 92, 1167; (c) K. Toshima, K. Tatsuta, Chem. Rev. 1993, 93, 1503. (d) R. R. Schmidt, W. Kinzy, Adv. Carbohydr. Chem. Biochem. 1994, 50, 21. (e) S. J. Danishefsky, M. T. Bilodeau, Angew. Chem. Int. Ed. Engl., 1996,35, 1380. 11. H. Paulsen Angew. Chem. Int. Ed. Engl., 1982,21, 155.

References

133

12. H. Kondo, Y. Ichikawa, C.-H. Wong, J. Am. Chem. SOC.1992,114, 8748. 13. T. J. Martin, R. Schmidt, Tetrahedron. Lett. 1992, 33, 6123. 14. (a). L. I. Smirnova, M. A. Malenkovskaya, D. A. Predvoditelev, E. E. Nifantev, Zh. Org. Khim. 1980, 16, 1170; (b). J. W. Perich, R. B. Johns, Tetrahedron Lett. 1987,28, 101; (c). K. L. Yu, B. Fraser-Reid, Tetrahedron Lett. 1988, 98, 979; (d). R. L. Pederson, J. Esker, C.-H. Wong, Tetrahedron, 1991, 47, 2643; (e). S. Aoki, H. Kondo, C.-H. Wong, Methods in Enzymology, 247, 193; (f). Qiao, B. W. Murray, M. Shimazaki, J. Schultz, C.-H. Wong, J. Am. Chem. SOC. 1996, 118, 1653. 15. H. Paulsen, H. Tietz, Carbohydr. Res. 1984, 125, 47. 16. Y. Guo and G. A. Sulikowski, J. Am. Chem. SOC.1998,120, 1392. 17. (a). S. Hashimoto, K. Umeo, A. Sano, N. Watanabe, M. Nakajima, and S. Ikegami, Tetrahedron Lett., 1995,36, 2251; (b). H. Tanaka, H. Sakamoto, A. Sano, S. Nakajima, S. Hashimoto, Chem. Commun., 1999, 1259; (c). unpublished result, personal communication with Prof. Hashimoto. 18. M. M. Sim, H. Kondo and C.-H. Wong, J. Am. Chem. SOC.1993,115,2260. 19. R. R. Schmidt, B. Wegmann, K.-H. Jung, Liebigs Ann. Chem. 1991, 191, 121. 20. S. Sabesan, S. Neira, Carbohydr. Res. 1992, 223, 169. 21. H. Kondo, S. Aoki, Y. Ichikawa, R. L. Halcomb, H. Ritzen, C.-H. Wong, J. Org. Chem. 1994, 59, 864. 22. (a). Y. Watanabe, C. Nakamoto, S. Ozaki, Synlett, 1993, 115-116; (b). Y. Watanabe, C. Nakamoto, T. Yamamoto, S. Ozaki, Tetrahedron, 1994, 50, 6523. 23. H. Schene, H. Waldmann, Eru. J. Ory. Chem. 1998, 1227. 24. Lemieux Abstr: Pap. Am. Chem. SOC.1958. 133, 31N; (b) J. P. Praly, R. U. Lemieux, Can. J. Chem. 1987,65,213. 25. (a). R. R. Schmidt, M. Behrendt, A. Toepfer, Synlett, 1990, 694; (b). A. J. Ratcliffe, B. FraserReid, J. Chem. Soc., Perkin Trans. 1 1990, 747; (c). P. Sinay, Pure Appl. Chem. 1991, 63, 519. 26. S. Hashimoto, A. Sano, H. Sakamoto, M. Nakajima, Y. Yanagiya, S. Ikegami, Synlett., 1995, 1271. 27. (a). D. Crich, S . Sun, J. Ory. Chenz. 1996, 61, 4506-4507: (b). D. Crich, S . Sun, J. Am. Chem. SOC.1997, 119, 11217; (c). D. Crich, S. Sun, J. Am. Chem. SOC.1998, 120, 435. 28. (a). D. Kahne, S. Walker, Y. Cheng, D. V . Engen, J. Am. Chem. SOC.1989, 111, 6881; (b). L. Yan, D Kahne, J. Am. Chem. SOC.1996, 118, 9239. 29. (a). V. Martichonok, G. Whitesides, J. Am. Chem. SOC.1996, 118, 8187; (b). V. Martichonok, G. Whitesides, J. Org. Chem. 1996, 61, 1702. 30. C.-C. Lin, M. Shimazaki, M.-P. Heck, S. Aoki, R. Wong, T. Kimura, H. Ritzen, S. Takayama, S.-H. Wu, G. Weitz-Schmidt, C.-H. Wong, J. Am. Chem. SOC.1996, 118, 6826. 31. S.-H. Wu, M. Shimazaki, C.-C. Lin, L. Qiao, W. J. Moree, G. Weitz-Schmidt, C.-H. Wong, Angew. Chem. Int. Ed. Engl., 1996, 35, 88. 32. For sialylations see ref. 12, and 13 and references therein. 33. N. E. Nifantev, Y. E. Tsvetkov, A. S. Shashkov, L. 0. Kononov, V. M. Menshov, A. B. Tuzikov, a n d N . V. Bovin, J. Carbohydr. Chem. 1996, 15, 939. 34. U. Sprengard, G. Kretzschmar, E. Bartnik, C. Hiils, and H. Kunz, Angew. Chem. lnt. Ed. Engl. 1995, 34, 990. 35. T. Muller, G. Hummel, R. R. Schmidt, Liebigs Ann. Chem. 1994, 325. 36. T. Miiller, R. Schneider, R. R. Schmidt, Tetrahedron Lett. 1994,35,4763. 1993, 115, 8871. 37. E. J. Corey, Y.-J. Wu, J. Am. Chem. SOC. 38. H. Paulsen, 0. Lockhoff, Chem. Ber. 1981,114, 3102. 39. H. Kunz, W . Giinther, Angew. Chem. Int. Ed. Engl. 1988, 27, 1086. 40. (a). F. Barresi, 0. Hindsgaul, J. Am. Chem. SOC.1991, 113,9376; (b). G. Stork, G. Kim, J. Am. Chem. SOC.1992, 114, 1087; (c). Y. Ito, T. Ogawa, J. Am. Chem. Soc. 1994, 116, 12073. 41. (a). Y. Watanabe, T. Yamamoto and S. Ozaki, J. Org. Chem. 1996,61, 14-15; . (b). Watanabe, T. Yamamoto and S. Ozaki, Tetrahedron, 1997, 53, 903. 42. (a). J. Thiem, W. KlaKke, Top. Curr. Chem. 1990, 154, 870; (b). William Roush and Chad E. Bennett J. Am. Chem. SOC. 1999, 121, 3541. 43. S. Hanessian, C. Bacquet, N. Lehong Carbohydr. Res. 1982, 80, c17. 44. Z. Zhang, G. Magnusson, Carbohydr. Res. 1994,262, 79.

134

5 Glycosylation Methods: Use of Phosphites

45. (a). P. J. Garregg, S. Kopper, P. Ossowski, J. Thiem, J. Carbohydr. Chem. 1986, 5, 59; (b). R. W. Binkley, D. J. Koholic, J. Org. Chem. 1989, 54, 3577; (c). Z. Zhang, G. Magnusson, J. Org. Chem. 1996,61, 2383. 46. (a). M. D. Chappell, R. L. Halcomb, Tetrahedron 1997, 53, 11109; (b). T. J. Martin, R. R. Schmidt, Tetrahedron Lett. 1993, 34, 1765. 47. U. B. Gokhale, 0. Hindsgaul, M. M. Palcic, Can. J. Chem. 1990, 68, 1063; (c). R. R. Schimidt, B. Wegmann, K.-H. Jung Liebigs. Ann. Chem. 1991,191, 121. 48. (a). Y. Ichikawa, M. M. Sim, C.-H. Wong, J. Org. Chem. 1992, 57, 2943-2946; (b). V. Wittmann, C.-H. Wong J. Org. Chem. 1997, 62, 2144. 49. R. Meuwly, A. Vasella, Helv. Chim. Acta. 1986, 69, 25. 50. (a). J. Gao, V. Martichonok, G. M. Whitesides, J. Org. Chem. 1996, 61, 9538; (b). T.-H. Chan, Y.-C. Xin J. Org. Chem. 1997, 62, 3500; (c). B. Miiller, T. J. Martin, C. Schaub, R. R. Schmidt, Tetrahedron Lett. 1998, 39, 509; (d). M. J. Ruira, M. J. Pkrez-Pkrez, Jan. Balzaerini, M. Camarasa, Synlett. 1998, 177. 51. C. L. White, M. N. Janakiraman, W. G. Laver, C. Philippon, A. Vasella, G. M. Air, M. Luo, J. Mol. Biol. 1995,245, 623. 52. M. Imamura, H. Hashimoto, Tetrahedron Lett. 1996, 37, 1451. 53. S. Hashimoto, T. Honda, S. Ikegami, J. Chem. Soc. Chem. Commun. 1989, 685. 54. (a). S. Pan, H. Li, F. Hong, B. Yu, K. Zhao, Tetrahedron. Lett. 1997, 38, 6139; (b). H. Li, M. Chen, K. Zhao, ibid., 1997,38, 6143.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

6 Glycosylation Methods: Use of n-Pentenyl Glycosides Bert Fraser-Reid, G. Anilkumar, Mark R. Gilbert, Subodh Joshi, and R a y Kraehmer

6.1 Introduction The use of protecting groups in syntheses of complex oligosaccharides is currently inevitable. Electronic or steric factors may enable regioselectivity between competing hydroxyl groups, but for the foreseeable future such interventions will be exceptions to the rule. If protecting groups are to be necessary, it would be advantageous for them to serve additional roles. In our laboratory this concept was first appreciated with the development of n-pentenyl glycosides (NPG), in which the pentenyl group protects the sensitive anomeric center, and survives (most of) the chemical transformations tolerated by normal alkyl glycosides. Treatment with an electrophile converts the pentenyl moiety into a leaving group, however, thereby enabling replacement reactions to occur at the anomeric center. This dual protection/activation role is a fundamental property of NPGs, and this duality was heavily exploited in the synthesis of the rat brain Thy-1 glycosylphosphatidylinositol membrane anchor by the retrosynthetic plan outlined in Scheme 1 [I]. NPGs have, however, also been valuable in mechanistic investigations designed to probe events at the anomeric center. For example, NPGs were instrumental in the demonstration that protecting groups can profoundly alter reactivities of glycosyl donors-and glycosyl acceptors. These observations indicated the importance of fine-tuning reactivity of donor/acceptor partners, and efforts to contribute to this insight continue in our laboratory.

6.2 Fundamental Reactions A summary of (most of) the fundamental reactions of n-pentenyl glycosides is presented in Scheme 2 [ 2 ] .Being simple alkyl glycosides, NPGs can be readily prepared

6 Glycosylation Methods: Use of n-Pentenyl Glycosides

136

V

0

HO

HO-

-

Thy1 Anchor from Rat Brain

OBn BnO

Scheme 1. Retrosynthetic plan for rat brain Thy-1 GPI membrane anchor.

by standard Fischer glycosidations, e.g., 4 4 5 . Where one or other anomer is specifically required, however, the standard procedures used for simple alkyl glycosides can be applied as summarized in Scheme 3. The role as glycosyl donor proceeds through the standard oxocarbenium intermediate 2, which can be trapped by nucleophiles. In aqueous medium the product is an aldose (1, Nu = OH). When such oxidative hydrolyses were probed it was found that with C2-OAcyl substrates the reaction occurs much more slowly than with C2OAlkyl. This observation has been introduced into a chemoselective coupling [3] protocol that has come to be known as the 'armed-disarmed strategy.' [4]Thus as illustrated in Scheme 4a, the partners 10 and 11 gave disaccharide 12a, with no evidence of the product of self-coupling of 11, nor of further reaction of disaccharide 12a with the acceptor 11, both of which would have occurred at disarmed anomeric

RN"& p-1 Y

1 Nu = OR, OSugar, Br, NHCOPeptide

/

6.2 Fundamental Reactions

p!

Y

Y

2

3 furanylium ion

glycosyl donor

137

J

6

free sugar

NPG

I disaccharide or higher

Zn or Smlp or Nal

glycosyl acceptor

cyclic halonium

I

+

XsNu'

9 bromohydrin Nu=OH dibromide Nu=Br

Scheme 2. The chemistry of n-pentenyl glycosides (NPG).

centers. Transformation to the armed counterpart 12b, however, enabled ready formation of trisaccharide 13 [3]. Glycoside hydrolysis is of immense mechanistic interest because of the insight it provides into enzymatic cleavage or assembly of oligosaccharides [5, 61. The option of oxidative hydrolysis meant that the anomeric center of NPG could be cleaved without the use of acids and, furthermore, that acid-labile protecting groups could be employed to enhance mechanistic insight. In the course of such investigations, it became clear that frequently used cyclic protecting groups could also deactivate glycosyl donors [7]. Thus, the advantage of simultaneous protection of two hydroxyl groups might be offset by the torsional constraints imposed upon the bicyclic array. Theoretical calculations supported this finding [81.

138

(Hqn-k>0H

6 Glycosylution Methods: Use of n-Pentenyl Glycosides

camphor Pent-OH

(HO)n{>O(/

sulfonic acid

I

OPent

Scheme 3. Preparative routes to NPG.

The reality of torsional armed-disarmed effects was demonstrated (Scheme 4b) with the glucosides 14 and 15, which gave the product of cross coupling, 16, but with no evidence of 17, the product from self-coupling of 15 [9].

6.3 Determination of Relative Reactivities The evidence in Scheme 4 makes it clear that success in oligosaccharide coupling can be dependent on protecting groups present on the reactants. A procedure for direct comparison of glycosyl donor reactivity would be advantageous, and such a methodology was therefore developed [lo]. As shown in Figure 1, eq. (l), equivalent amounts of the two NPGs, S1 and S2, compete for one equivalent of NBS in aqueous solvent. As eq. (2) shows the method requires that the NBS must be completely consumed, and so the reaction must continue until a negative test is obtained with potassium iodide-starch paper, shown independently to be accurate to within 2%. Because both anomers are in the same reaction medium, the NBS terms in the rate expressions cancel out, leading eventually to eq. (3). After work-up, the ratio of unreacted NPGs in the crude reaction product is determined, eq. (4). For our purposes, HPLC was convenient for these determinations. All the terms in eq. ( 5 ) are now known or knowable. And so the relative reactivity, kl/k2, of the two NPGs can be obtained by solving eq. (3). Some comparisons of reactivities, as determined by the procedure in Figure 1, are assembled in Scheme 5. That the choice of anomeric configuration of the donor might be important is apparent from the results for 18 and 19. In both the anomer

6.3 Determination of Relative Reactivities

k M

J+

0

,""I?,..

b T a

9

139

6 Glycosylation Methods: Use of n-Pentenyl Glycosides

140

S1

+ S2 + NBS

H20 ,Products CH,CN

(1 eq. of each) [end point determined by negative KI-starch test paper]

Figure 1. Calculating the relative rates of two glycosyl donors.

reacts more quickly, the higher ratio in 19 being evidence of anchimeric assistance resulting from neighboring-group participation in the cleavage reaction [ 1I]. The classical electronic armed-disarmed situation (Scheme 4a) is seen with the p glycosides 20a and b. Notably, 20b is less reactive than 20a despite the anchimeric assistance which must be occurring in the ester (cf 19). The results for the p glucoside pairs 20 and 21 confirm that the benzylidene ring can be disarming-much more so for acyl protecting groups (21, R = Ac compared with R = Bn). Couplings involving benzylidinated or acetonated donors might, therefore, occur less readily than with their torsion-free counterparts.

BnO

OPent

AcO

,OAc

OBn 18

OPent

R:)ope,,

= 1.59

OR

20

19

18p/ M a = 1.70

20d20b

\

a R=Bn

19p/ 19a= 5.19

b R=Ac R RO o R ~ O F ’ e n pl

~

~

~

~2 0 d 2 l a O = 1.6

\

OR

OK

20

21

a R=Bn

a R=Bn

b R=Ac

b R=Ac

Scheme 5. Relative reaction rates for some glycosyl donors.

F20b/ 2lb m= 10.4

6.4 n-Pentenyl Orthoesters as Glycosyl Donors

141

Although these results are based on NPG, they should apply equally to any pair of donors, irrespective of the anomeric leaving group. This assumption is based on general relevance of armed-disarmed factors across the spectrum of glycosyl donors 1121.

6.4 n-Pentenyl Orthoesters as Glycosyl Donors Alkyl orthoesters, 23, like cyclic acetals, also simultaneously protect two hydroxyl groups. Their use in carbohydrate synthesis [13] is greatly enhanced by a regioselective, acid-catalyzed rearrangement, whereby the resulting ester is axial rather than equatorial [ 141. Glycosyl (or 1,2-) orthoesters are readily prepared by Lemeiux’s procedure, in which acylated glycosyl bromides are treated with a hindered base, 22 23 [ 151 (Scheme 6). Mild acid treatment causes a rearrangement, 23 -, 26, in which the alkoxy unit is transferred to the anomeric center [16] (Scheme 6). The group so transferred can vary from simple alkoxy to an oligosaccharide [ 171. n-Pentenyl orthoesters (23, S = Pent) have an additional mode of reaction. Thus, treatment with an electrophile leads to furanylium ion 27, and thence to the same dioxolenium intermediate 25 (Scheme 6). The discharged entity, 28, however, is not nucleophilic and so a nucleophile present in the reaction medium faces no competition for reaction with 25 leading to 29. Notably, 26c and 29a are both higher saccharides, but the strategies for their preparation offer a choice between protonand electrophile-driven reactions. Again, n-pentenyl orthoester 23 (S = Pent) and NPG 26b are both glycosyl donors, but they have very different properties, notably towards acids. Further donor versatility comes from the fact that titration with bromine affords bromide 29b, the classic glycosyl donor, ready for reaction with silver triflate [ 181. To go from one glycosyl bromide to another, 22 +29b, might seem redundant, but the advantage of this possibility is readily appreciated from Scheme 7, which summarizes two approaches to trimannan 36, a building block for a synthesis of the Thy-1 GPI rnembrane (see Scheme 1) [19]. The tetrabenzoyl mannosyl bromide 30 was converted into orthoester 31a [15]. One portion was silylated [20] and then benzylated to give 31b, which upon treatment with bromine led to the glycosyl bromide 32. The remainder of 31a was completely benzylated, and one portion of the resulting material, 31c, was processed to give the acceptor 33a whereas the other was brominated to give glycosyl bromide 35. Coupling of 32 and 33b, led to 34a and debenzoylation to acceptor 34b which reacted with 35 to give trimannan 36 P11. Glycosyl bromides 32 and 35, being unstable, are best generated and used in situ. An approach to the same trimannan in which all reacting participants are stable is, however, outlined in Scheme 7b. The pathway takes advantage of some of the unique aspects of n-pentenyl glycosides. Thus, a major development in NPG chemistry emanated from mechanistic investigations which showed that intramolecular reaction of the cyclic halonium compound leading to the furanylium ion, --f

142

6 Glycosylation Methods: Use of n-Pentenyl Glycosides

v1

c

x 8 *

8

%

Scheme 7. Two routes to a trimannan from an n-pentenyl orthoester. (i) Pent-OH, 2,6-lutidine, CHzC12; (ii) tBuPh2SiCl; imidazole, THF; (iii) BnBr, NaH, B U ~ N ITHF; , (iv) 31b or c; Br2, CH2C12, 0 ° C ; (v) 31c CSA, CH2C12, 5OoC, 6 h; (vi) NaOMe; (vii) AgOTf, CHzCl,; (viii) TBAF, THF then Ac20, DMAP, CHzC12; (ix) Br2, Et4NBr, CH2C12; (x) NIS, Et&OTf, CH2Clz;(xi) NH3, MeOH; (ClAc)20, Et3N, CH2CI2; (xii) Zn, BQNI, EtOH then (ClAc)20.

U

144

6 Glycosylation Methods: Use of n-Pentenyl Glycosides

(Scheme 2, 6-+3), could be offset by the bimolecular process leading to 9, in the presence of excess nucleophile. For example, added bromide ion led to a vicinal dibromide; because, however, reductive elimination readily restored the double bond, dibromides 7 and 8 represent ‘sidetracked’ forms of an NPG. Nevertheless, the sidetracked material enables an hydroxyl group to be liberated, thereby creating a glycosyl acceptor 8 capable of reacting with a donor to give 7. The utility of the ‘sidetracked’ capability is illustrated in Scheme 7b. A portion of NPG 33a was debenzoylated to 33b and then ‘sidetracked’ to give acceptor 38. Coupling with orthoester 37 then gave dimannan 39a, which underwent routine protecting-group alterations to produce disaccharide acceptor 39b. Coupling with the remainder of 33a now produced 40. The major process for obtaining the previously described 36 now involved reductive debromination to regenerate the pentenyl moiety [22].

6.5 n-Pentenyl Orthoesters as Latent C2 Esters It is apparent from Scheme 7 that proton- or electrophile-driven reaction leads to specifically C2-0-acylated products. C2-Esters are important for controlling stereoselectivity in anomeric replacement reactions [ 1 11, and so the use of orthoesters as specific latent C2-esters has been explored in the context of a protected form of the ‘linkage region’ of proteoglycans, which contains four 1,2-trans diequatorial linkages (Scheme 8). For the C2-acylated galactoside acceptor, a simple route begins with galactosyl bromide 41 (Scheme 8a). Orthoester formation required the use excess pentenyl alcohol and tetrabutylammonium iodide as promoter to obtain 42a and thence 42b. Acid-catalyzed reaction of the latter in the presence of excess pentenyl alcohol, then benzylidination gave acceptor 43a. The xylosyl acceptor 46b (Scheme 8b) was obtained similarly from bromide 44 via the n-pentenyl orthoester 45. Diol 46a was subjected to selective silylation at C4, paving the way to 46b. It was expected that the uronate moiety would be problematic because such 2-0acyl donors usually lead preponderantly to orthoester products. Of several donors that were tested, trichloroacetimidates were found to give the least amount of orthoesters. Donor 49 was therefore prepared from commercially available benzyl P-glucoside, 47a, via methyl ester 48 as summarized in Scheme 8c. Coupling with galactoside 43a gave the disaccharide donor 50 as the major product, the orthoester (51) being minimized at 25%. The disaccharide acceptor 52b was obtained by coupling the ‘sidetracked’ xyloside 46b to the n-pentenyl galactoside 43b, followed by dechloroacetylation. The latter reaction was problematic and required use of the van Boeckel procedure [23]. The two disaccharides were then coupled to give 53a. Iodide-induced debromination [24], a milder alternative to the zinc reduction used for 42 -+ 38 in Scheme 7, afforded 54b. Finally, coupling with the protected serine 54 [25] occurred smoothly to give the desired ‘linkage region’ construct 55 [26].

6.5 n-Pentenyl Orthoesters as Latent C2 Esters

145

146

6 Glycosylution Methods: Use of n-Pentenyl Glycosides

6.6 Protecting Groups The presence of one free amine and four amide functions in the Thy-1 anchor (Scheme 1) sensitized us to the need for new, easily removable amine protecting groups. The n-pentenyl based protecting groups for alcohols and vicinal diols shown in Scheme 9a-c are outgrowths of NPG chemistry [27], all of which are cleaved or installed with halogen electrophiles. Notably, the acetalating agents in Scheme 9d can be put in place without disturbing acid-sensitive groups already present [28]. Glycosyl n-pentenoyl esters (Scheme 9e) have been found to be good glycosyl donors, a circumstance that provides evidence for ready loss of the pentenoyl group [29]. In the light of these precedents, amine protection via n-pentenoylation was examined. As shown in Scheme 9f installation for primary and secondary amines is routine, and cleavage occurs readily with iodine. C2-N-acylated sugars are, however, prone to give oxazolines as shown in Scheme 9g. Lemieux showed that this problem could be obviated by employing phthaloyl protecting groups [30], de-Nphthaloylation being effected by prolonged treatment with hydrazine or ethylenediamine. It has been shown [31, 321 that electron-withdrawing groups on the aromatic ring facilitate cleavage remarkably, as indicated by the difference for X = C1 and X = H in Scheme 9h. Cleavage of pentenoyl or tetrachlorophthaloyl (TCP) groups is therefore seen to occur very readily, and the very different procedures employed enable these protecting groups to be used orthogonally. This aspect was explored in the context of lipid A, 56 [33] (Scheme lo), a molecule in which all sites on the disaccharide need to be differentiated. The glucosaminide functions would be protected by either npentenoyl or TCP; but use of the former on the acceptor rather than the donor was dictated by the oxazoline problem noted in Scheme 9g. In the event Koenigs-Knorr coupling of 57 or 58 produced 59 in good yield. Comparison with lipid A, 56, shows that differentiation between the three acetylated sites of 54 would be necessary for a more useful precursor. Deacetylation with base would not be tolerated by the TCP group; but acid-catalyzed deacetylation was successful and benzylidination led to the more attractive counterpart 60 with the five critical sites being differentially accessible. With respect to the protected amines, treatment with ethylenediamine liberated one and iodine the other, giving 61 and 62 respectively [34]. Notably, cleavage of the benzylidine group with sodium cyanoborohydride [35] is tolerated by TCP [36].

6.7 Solid-Phase Iterative Couple-Deprotect-Couple Strategy Scheme 2 shows that NPGs have the potential to function as glycosyl donors and also as glycosyl acceptors via sidetracked intermediates (e.g. 8). The latter property

6.7 Solid-Phase Iterative Couple-Deprotect-Couple Strategy

h I

QfB I

1 (0 0

+ I '

8

q)==J 0

I

\

141

148

6 Glycosylation Methods: Use of n-Pentenyl Glycosides

Lipid A /OAC

AcO AcO BzO

Ad3

BzO

Oknt

-

57

DIFFERENTIALLY ACCESSIBLE SITES

Scheme 10. Orthogonal protecting groups for a lipid a precursor. (i) AgOTf; (ii) Me2CO/H20/HCl, 60"C, 20 h; PhCH(OMe)2/PF'TS/dioxane, reflux; (iii) ethylenediarnine, 60°C 2 h; (iv) 12 (2 equiv.), HzO/THF, 8 min.

was instrumental in a process used to assemble the high-mannose nonamannan component via an iterative deprotect-couple-deprotect strategy [37]. A solid-phase version of this procedure would have the advantage of minimizing transformations on the bound saccharide(s). An attempt to develop such an iterative strategy, shown in Scheme 11, used the known [38], bound pentenyl alcohol 63 for a KoenigsKnorr reaction with 64, by following the procedure developed by Frechet and Schuerch [39]. To test the viability of the cleavage reaction, compound 65 was treated with benzyl alcohol and NIS/TESOTf in wet dichloromethane to afford the

R=Bn

b)

OPent

+

BnO

ii

70

BnO

BnO

65

BnO

66

BnO

71

BnO

67

Scheme 11. A solid-phase version of NPG. (i) THF; (ii) pyridine, CH2C12; (iii) BnOH, NIS, TESOTf, wet CH2C12, 2 h; (iv) NaOMe/MeOH/THF; Br2, Et4nBr, CH2C12; (v) NIS, TESOTf, CHzC12, 2 h; (vi) 0.1 M SmI2, THF, 30 mix; (vii) thiourea, methoxyethanol, 8 0 T , 6 h.

68 R=Ac

a)

Bn?% BnO

63

OH

u

P

150

6 Glycosylation Methods: Use of n-Pentenyl Glycosides

benzyl glycoside 66 in 60% yield and, presumably, the bound iodomethyltetrahydrofuran 67. For a more elaborate test case, the sidetracked acceptor 69 was prepared and coupling with 68a proceeded smoothly to give a dibromide. Zinc could not be used to restore the double bond, because of the solid-solid interaction that would be involved. A one-electron reductive elimination with samarium I1 iodide [24] proceeded smoothly, however, to give 70. Unfortunately, efforts to cleave the disaccharide from the solid support to produce 71 were unsuccessful. FTIR data suggested that addition to the double bond overwhelmed the cyclization reaction. The above results meant that the oxidative cleavage worked sometimes (65-+ 66), but not always (70 -+ 71). The use of sidetracked acceptors such as 69 might, therefore, be unreliable. Nevertheless, use of NPG as donors was still an option and so the oxidative cleavage aspect was replaced by a photocleavable alternative (Scheme 12). Thus, the Rich-Guwara linker 72A [40]was investigated. Success was had with a linear trimannan, and so a branched trimannan was tested, because this would be more demanding in that chemoselectivity deprotection would be required. This study is reported in Scheme 12. The Rich-Guwara photolinker was successfully coupled with diester 73, and chemoselective deacylation of the bound monosaccharide, 74a +74b, was found to proceed smoothly. Although coupling with donor 68b gave 15a in only moderate yields, ranging from 58-76'3'0, subsequent deprotection of the C6 acetate and iterative coupling with donor 68b proceeded without incident to give the resin-bound trisaccharide 76. Irradiation of the A-series trisaccharide required three days for recovery of optimum (10%) amounts of trisaccharide 77. On the assumption that this low yield was because of problems with the photocleavage reaction, a new linker, 72B, was designed. Advantages over 72A were (a) that linkage to the sugar was through a secondary, rather than primary, hydroxyl, and (b) that the electron-withdrawing amide carbonyl was remote from, rather than attached to, the benzene ring, where its deactivating effect would be irrelevant to the photocleavage event [41]. Indeed with 76B, the cleavage was very much faster and the overall yield of 77 was very much better [41]. An effort has also been made to synthesize polymer bound, fully deprotected saccharides by the couple-deprotect-couple iterative strategy [42]. A crucial problem with this approach was, however, that normally used benzyl groups, as in Scheme 12, would be problematic, because hydrogenolytic cleavage would involve a solid-solid interaction. Dissolving metal options were examined, but were also problematic in that the resin backbone apparently acted as an electron sink. Ester protecting groups were the only other option, and so concerns about their deactivating affect on the NPG donors were dispelled by success with preliminary experiments. The plan in Scheme 13 was therefore tested. The known NPG 78 [43] was converted into the differentially protected counterpart 79, and coupling with the polystyrene resin, A, led to 80a which reacted with 81 to give disaccharide 82a. Another decouple-couple iteration led to trisaccharide 83. Global deprotection by treatment with ethylenediamine in methoxyethanol at 70 "C for 4 h gave the fully deprotected bound trisaccharide 84.

6.7 Solid-Phase Iterative Couple-Deprotect-Couple Strategy

wI . -

P OD

\ .-

'-

I

+

Q

t

15 1

152

6 Glycosylution Methods: Use of n-Pentenyl Glycosides

..-

.-.-

;.i d

4 >

P

d 0 \

?

References

153

The sequence could be repeated with the tentage1 macrobeads, B, and the grafted crown linkers, C. The last was particularly valuable in that acetylation followed by photocleavage afforded 85, the identity of which was verified by independent synthesis [42].

References 1. For a concise review see: A. J. Bridges in Chem Tracts-Organic Chemistry, 1996, 9, 215-223. 2. B. Fraser-Reid, U. E. Udodong, Z. Wu, H. Ottosson, R. Merritt, C. S. Rao, C. Roberts and R. Madsen. Synlett, 1992, 927-942. 3. D. R. Mootoo, P. Konradsson, U. Udodong and B. Fraser-Reid. J. Am. Chem. SOC.,1988, 110, 5583-5584. 4. G. J. Boons, Tetrahedron, 1996, 52, 1095-1121. 5. M. L. Sinnott. Advances in Physical Organic Chemistry. 1988, 24, 113-204. 6. P. Deslongchamps. Stereoelectronic Efsects in Organic Chemistry. Pergamon, Oxford. 1983. 7. A. J. Ratcliffe, D. R. Mootoo, C. W. Andrews and B. Fraser-Reid. J. Am. Chem. Soc., 1989, I l l , 7661-7662. 8. C. W. Andrews, J. P. Bowen and B. Fraser-Reid. J. Chem. SOC. Chem. Comm., 1989, 19131916. 9. B. Fraser-Reid, Z. Wu, C. W. Andrews, E. Skowronski and J. P. Bowen. J. Am. Chem. Soc., 1991, 113, 1434-1435. 10. B. G. Wilson and B. Fraser-Reid. J. Org. Chem, 1995, 60, 317-320. 11. H . L. Frush and H. S. Isbell, J. Res. Nutl. Bur. Stds, 1941, 27, 413. 12. S. Oscarson and U. Tedebark. J. Carbohydr. Chem., 1996, 15, 507-512. 13. G. J. Boons, Contemp. Org. Syn., 1996, 3, 173-200. 14. J. F. King and A. D. Allbutt. Can. J. Chem., 1970, 48, 175441767, 15. R. U. Lemieux and N. J. Chu. Abstracts of Papers. Am. Chem. SOC., 1958,133, 31N. 16. A. F. Bochkov and G. E. Zaikov. Chemistry of the 0-Glycosidic Bond, Chapter 2, Pergamon Press, Oxford, 1979. 17. For some recent examples see: W. Wang and F. Kong. J. Org. Chem. 1998, 63, 5744-5745. L. Snaidman, C. Johnson, C. Crasto and S. M. Hecht. J. Org. Chem., 1995,560,3942-3943. 18. S. Hanessian and J. Banoub. Carbohydr. Res. 1977, 53 C13-Cl6. 19. A. S. Campbell and B. Fraser-Reid. J. Am. Chem. SOC.,1995,117, 10387-10388. 20. S. Hannesian and P. Lavallee, Can. J. Chem. 1975, 53,2975-2978 21. C. Roberts, C. L. May, and B. Fraser-Reid. Carbohydr. Lett., 1994, 1, 89-93. 22. C. Roberts, R. Madsen and B. Fraser-Reid. J. Am. Chem. Soc., 1995,117, 1546-1553. 23. C. A. A. van Boeckel and T. Beetz. Tetrahedron Lett. 1983,24, 3775. 24. J. R. Merritt, J. S. Debenham and B. Fraser-Reid. J. Carbohydr. Chem., 1996, 15, 65-72. 25. M. Meldal and K. J. Jensen. J. Chem. SOC.,Chem. Commun., 1990,483-484. U. K. Saha and R. R. Schmidt. J. Chem. Soc., Perkin I, 1997, 1855-1858. 26. J. G. Allen and Bert Fraser-Reid. J. Am. Chem. Soc., 1999, 121, 468-469. 27. Z. Wu, D. R. Mootoo and B. Fraser-Reid. Tetrahedron Lett., 1988, 29, 6549. 28. R. Madsen and B. Fraser-Reid. J. Chem. SOC.,Chem. Commun., 1994, 749-750. R. Madsen and B. Fraser-Reid. J. Org. Chem., 1995, 60, 772-779. 29. J. C. Lopez and B. Fraser-Reid. J. Chem. SOC.Chem. Comm., 1991, 159-161. 30. R. Lemieux, T. Takeda, B. Chung. Synthetic Methods for Carbohydrates (Ed. H. S. El Khadem, ACS Symposium Series, Washington, D. C., 1976, vol. 39, pp. 90-115. 31. J. S. Debenham, R. Madsen, C. Roberts, B. Fraser-Reid. J. Am. Chem. Soc., 1995,117, 33023303. 32. J. C. Castro-Palomino, R. R. Schmidt, Tetrahedron Lett. 1995, 36, 5343-5346. H. Shimizu, Y . Ito, Y. Matsuzaki, H. Iijima, and T. Ogawa. Biosci. Biotech. Biochem, 1996, 60, 73-76. 33. C. R. H. Raetz. J. Bacteriology, 1993, 175, 5745-5753. 34. S. L. Griffiths, R. Madsen, and B. Fraser-Reid. J. Org. Chem., 1997, 62, 365443658,

154

6 Glycosylation Methods: Use of n-Pentenyl Glycosides

35. P. T. Garegg. Pure Appl. Chem., 1984, 56, 845-858. 36. J. S. Debenham, S. D. Debenham, and B. Fraser-Reid. Bioorg. Med. Chem., 1996, 4, 19091918. 37. J. R. Merritt, E. Naisang and B. Fraser-Reid. J. Org. Chem., 1994,59, 4443-4449. 38. M. Bernard and W. T . Ford. J. Org. Chem., 1983, 48, 326-332. 39. J. M. Frechet and C. Schuerch. J. Am. Chem. Soc., 1972, 94,604-609. 40. D. H. Rich and S. K. Guwara. Tetrahedron Left., 1975, 16, 301-304. 41. R. Rodebaugh, B. Fraser-Reid and H. M. Geysen. Tetrahedron Lett., 1997,38,7653-7656. 42. R. Rodebaugh, S. Joshi, B. Fraser-Reid and H. M. Geysen. J. Org. Chem., 1997, 62, 56605661. 43. J. S. Debenham, R. Rodebaugh and Bert Fraser-Reid. Liebigs AnnJRecueil, 1997, 796-802.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

7 Glycosylidene Diazirines Andrea Vasella, Bruno Bernet, Martin Weber, and Wolfgang Wenger

7.1 Introduction Glycosylidene diazirines are useful as precursors of glycosylidene carbenes [I]. These cyclic, amphiphilic singlet alkoxycarbenes enable the formation of two bonds at the anomeric center. Although many aspects of the chemistry of pyranosylidene diazirines have been reviewed [2-51, furanosylidene diazirines [ 61 are hardly known because of their high reactivity and concomitant poor stability. Furanosylidene carbenes or carbenoids have been generated from the readily available lactone sulfonyl hydrazones [7].Whereas diazirines generate carbenes either thermally or photochemically under mild conditions, and simultaneously only produce dinitrogen, sulfonyl hydrazones, however, also generate potentially reactive side-products [7, 81. Glycosylidene carbenes, like other intermediates, are reactive and versatile. Their precursors-glycosylidene diazirines (= 1-azi-l,5-anhydroglycitols)-are thus expected to be useful. They have been devised primarily to serve as universal, highly reactive glycosylating reagents responding to the kinetic acidity of glycosyl acceptors, irrespective of their nucleophilic properties and of steric hindrance, and requiring no promoter. Their advantages will be discussed, as will their two disadvantages, uiz. their restricted stability, and their preparation which requires five (generally high-yielding) steps from hemiacetals.

7.2 Synthesis of Glycosylidene Diazirines The original route, depicted in Scheme 1, for the synthesis of the benzylated glucosylidene diazirine 7 [6, 91, is still used with minor variations, as evidenced by the preparation of the GlcNAc-derived analog 14 [3]. Details have been published, and some comments will prove useful.

156

7 Glycosylidene Diuzirines

BnO% BnO

i_ 1

B BnOn

B BnO n O

6 OBn N O

B BnO .

H

0

6 NHOH

BnO OH

O

h

2

%,

y

v

BnO BnO

BnO

Brio

7

iv

BnO& BnO BnO

NH

6

vi

BnO% BnO AcHN

NH

B BnO n 0

6

iii

N

O

H

AcHN

OH

8

/OR

E45 RR==MHs

B BnOn

~

N

O

6NHOH

AcHN

10

9

vii

c xF*n:niE

BnO% BnO y

ix

AcHN

AcHN

14

13

NH

&: ;B AcHN

/OR N

Scheme 1. Reagents: (i) Na, 96% EtOH, then NH3OH C1 and 1 , 7 h at 60°C, quant. (variable ratio of 2/3, ( E / Z )cu 65:35); (ii) NaOAc, NaI04, EtOH/H20, 19 h at 70°C, 79% from 1; (iii) pyridine, MsCl, CH2C12, 12 h at 0 ---t room temp., 87%; (iv) sat. NH3 in MeOH, CH2C12, 5 days at room temp., 76%; (v) MqN, 12, CH2C12, 2-5 h at -5O"C, 83%; (vi) Na, MeOH, then NH30HCl and 8, 33 h at 60"C, quant. (variable ratio of 9/10); (vii) NaOAc, NaI04, EtOH/H20, 18 h at 50°C, 70% from 8; (viii) Et3N, TsCl, CHzC12, 3 h at O T , 100%; (ix) sat. NH3 in MeOH, MeOH, 7 days at -2O"C, ca 70%; (x) Et3N, 12, CH2C12/MeOH, 3 h at O T , 68% from 12.

Formation of the oximes-generally obtained as ( E / Z )isomers and, most clearly [ 10, 111 (see also Ref. [ 121) and 2-acetamidofor 2,3,4,6-tetra-0-benzyl-~-glucose 3,4,6-tri-O-benzyl-2-deoxy-~-glucose oximes [3], as a mixture with the cyclic p-Dpyranosyl hydroxylamines (3 and 10, respectively)-proceeds readily at a pH of ca 6. Yields are high. The best procedure for the oxidation to the hydroximolactones depends on the structure of the oximes. Several oxidants have been tested. MnOz is generally reliable when prepared according to Attenburrow [ 131; commercial Mn02 has led to low yields [ 141.Oxidation is slightly exothermic. Fresh batches of MnOz might have to be added during the oxidation of large batches. Starting material and product have very similar RF values, but stain differently (H2S04, mostain) on the TLC

7.2 Synthesis of Glycosylidene Diazirines

157

plate. If the structure of the starting material enables formation of 1,4- and 1,5hydroximolactones, both might be formed [ 141. The relative amounts occasionally depend upon the oxidant (e.g. Hg(OAc)2 us 02/CuC12 in pyridine). NCS in the presence of DBU at low temperatures (ca -40°C) is a faster oxidant than MnO2. Low temperature is required for the oxidation of 0-acetyl oximes to avoid migration of the acetyl group from OC(4) to OC(5) [15]. Occasionally, NaI04/NaOAc has proven useful. 0-Sulfonylation of the hydroximolactones is unproblematic. Sulfonyl chlorides or anhydrides in the presence of base react rapidly at low temperature. The products often crystallize readily. A range of sulfonates has been tested [3]. Mesylates are often well suited; triflates are sometimes too unstable, although Beckmann rearrangement does not readily occur [ 141. Formation of the diaziridine requires a protic solvent; MeOH has been used as a standard. The concentration of NH3 required for the transformation depends on the nature of the sulfonyl group and on the acceptor properties of the 0-protecting groups. For the transformation of benzyl-protected sulfonates, MeOH must be saturated with NH3 at 0 "C in a thick-walled flask (the volume increases by ca 30% and the solution contains ca 24.5 g NH3 in 100 mL MeOH; it is stable for months at -20 "C); the reaction is performed at 22-26 "C, and the pressure is maintained by closing the flask with a septum; alternatively an autoclave may be used. The 2,3,4,60-pivaloylated glucose-derived hydroximomesylate reacts much faster [3]. The 2,3,4,6-0-acetylated glucose-derived hydroximotriflate reacted with half-saturated NH3/MeOH/CH*C12 at -20 "C without deacetylation [ 151. The diaziridines are ninhydrin-positive, stain dark blue with mostain, and oxidize KI to 12 in the presence of acid. They often crystallize from EtzO, and their purification by crystallization leads to improved overall yields. They are acid-labile; chromatography on silica gel must be rapid. It has led to slight losses even in the presence of base (Et3N); acetone is often a suitable (co)solvent. Crystalline diaziridines can be kept for months at -2O"C, and their solutions are stable in the absence of acid and oxygen. At 25°C and in the air, the diazirines turn red, presumably by forming increasing amounts of tetrazines. The diaziridines are usually obtained as mixtures of the two trans diastereoisomers, readily equilibrating in the presence of traces of acids. The ratio of the isomers depends on the glycosylidene residues. Oxidation of the diaziridines to diazirines requires some skill; anhydrous conditions are mandatory. Oxidation with I2 in the presence of a tertiary amine is the standard procedure. It was originally performed in MeOH and in the presence of 1-azi-glucose 7 crysEt3N. Under these conditions, the 2,3,4,6-tetra-O-benzylated tallizes from solution and can be isolated by filtration under anhydrous conditions [9]. For some diazirines, however, it is advantageous to perform the oxidation in CH2C12, using I2 and Me3N. The reaction corresponds to a titration, it is stopped when the solution remains slightly yellow. Because excess 12 is not readily removed, it is best to add a small amount of the starting diaziridine to discharge the color of the solution. The product is purified by filtration through a short Si02 column and thoroughly dried under high vacuum; a low boiling solvent (CH2C12) is recommended. On TLC, the diazirines stain purple with a solution of 4-(4-nitro-

158

7 Glycosylidene Diazirines

benzy1)pyridine; this is a specific and useful way of detection. Acyl-protected diazirines are more stable than their benzylated analogs and can be purified by flash chromatography at 0°C; solvents should be low boiling. The fractions are cooled, and condensation of water must be avoided [3, 151. The diazirines remain in the crystalline form or in solution for several weeks at -78 "C; 0-acyl protected diazirines can be stored at -20 "C.

7.3 Stability of the Glycosylidene Diazirines The stability of l-azi-l,5-anhydrohexitols depends on the substituents (protecting groups), the configuration, and the conformational flexibility. It has been quantified by thermolysis of the glycosylidene diazirines in methanol at 298 K [3, 15, 161; the results are given in Table 1. Their relative stability has been rationalized by postulating sequential (non-synchronous) heterolytic breakage of the C (1),N bond, leading to a cationic intermediate and requiring a conformational change of the pyranose ring whereby the pseudoequatorial C-N bond becomes pseudoaxial. Hence, acceptor substituents should increase the stability, and so should conformational locking, as by an intramolecular NH. . .OC(1) hydrogen-bond in the N-acetylallose derivative 25.

7.4 Glycosidation by Glycosylidene Diazirines 7.4.1 General Aspects Deprotonation of the glycosyl acceptor by a glycosylidene carbene ought to lead to a highly reactive, initially unsolvated ion pair, where both donor and acceptor are activated and should react readily with each other. Stereoelectronic effects dictate that deprotonation of the acceptor occurs in the a-plane of the carbene (more or less equivalent to the plane of the pyranoside ring), whereas addition of the oxyanion to the oxycarbenium cation has to occur in the mplane, i.e. more or less perpendicular to the a-plane, and preferentially from the axial direction (illustrated in Scheme 2 by the carbene derived from the benzylated glum-diazirine 7). Glycosidation by diazirines has been studied by using equivalent amounts of glycosyl acceptor and donor. It is expected that yields of the products increase with excess diazirine; occasionally this was shown to be so. Solvents have not systematically been optimized, but toluene, THF, and dioxane seem well suited; early experiments were mostly performed in dichloromethane. The course of the glycosidation of hydroxy compounds by glycosylidene diazirines is mainly dictated by four factors:

159

7.4 Glycosidution by Glycosylidene Diazirines Table 1. Kinetic parameters for the thermolysis of 1-aziglycosesin MeOH.

Diazirine

Compound

logA

15 R = Piv 16 R = AC 17 R = 4-F-Bn

23.0 25.0 21.6 23.7

13.4 14.1 11.6 13.8

33 202 181 43

22.4 24.0 21.0 23.1

1.7 4.8 -7.5 2.7

18

22.2

12.4

110

21.7

-3.2

19 R = B n 20 R = A c

23.2 20.5

14.2 11.4

I 45

22.6 19.9

5.5 -8.2

20.0

11.4

23

19.5

-8.2

-3.1

7 R = Bn

N

MeophT%r

BnO

AH= AS= (298 K) (298 K) [kcal [cal mol-' mol-'1 K-'1

E, [kcal mol-'1

z

(298 K) [min]

N

r *FCR

N

*::kS+yr

ND"

ND"

23

ND"

ND"

24

ND"

ND"

l4

22.6

12.6

112

22.0

25

28.1

15.1

4159

27.2

26 R = B n

20.8 22.3

11.4 12.1

88 202

20.2 21.7

BnO N OPiv PivO

Bn$Ss-,,

N

AcHN N

8.1

N

//"

RO OR

Ro

N

27 R = A c

-8.4 -5.0

160

7 Glycosylidene Diuzirines

Table 1 (continued) Diazirine

Compound

B n O - ' BnO f $ O a , N OBn BnO BnO

logA

21.8

12.2

T

(298 K) [min]

79

BnO BnO

OH

AS= (298 K) [cal mol-I K-'1

21.2

-4.8

- 100°C.

&:]

BnO% BnO

AH= (298 K) [kcal mol-'1

28

"Not determined. bVery unstable; slow decomposition even at

BnO

E, [kcal mol-l]

-

BnO

p-D

1

11 (OBn

(OBn

7

I

1

/OBn

I

OBn

BnO BnO

30

BnO

Scheme 2. Glycosidation of alcohols by gl ycosylidene carbenes derived from diazirines.

7.4 Glycosidution by Glycosylidene Diazirines

161

1) The kinetic acidity of the hydroxy group (inversely proportional to its basicity and more or less proportional to its nucleophilic properties). A sufficiently high kinetic acidity ensures rapid protonation of the carbene. Even an oxyanion derived from a poorly nucleophilic alcohol reacts readily with the simultaneously generated oxycarbenium cation. This property constitutes a major advantage of glycosylidene diazirines in the glycosidation of acids [ 171, phenols [9], highly fluorinated alcohols [ 181, hydroperoxides 1191, and hydrated Ti02 [ 151. Slow protonation of the carbene by a hydroxy compound of low kinetic acidity will lead to increasing amounts of azines (30 in Scheme 2), resulting from the interaction of the carbene with the starting diazirine. The azines are formed as a mixture of diastereoisomersthat are converted, by mild acid catalysis, mostly to the ( Z / Z )isomer [3]. 2) Hydrogen-bonding (a) Hydrogen-bonding influences the kinetic acidity of hydroxy groups. H-bond acceptors (that are not simultaneously hydrogen-bond donors) have enhanced kinetic acidity, whereas H-bond donors have reduced kinetic acidity; intramolecular hydrogen-bonding might thus determine the regioselectivity of deprotonation by the glycosylidene carbene-one of the conditions to be fulfilled for regioselective glycosidation. This effect has led to the regioselective glycosidation of dihydroxybenzene derivatives, where one of the phenolic OH groups functions as an intramolecular H-bond donor to a carbonyl group and is thereby glycosylated more slowly (Figure la) [9]. (b) Intra- or intermolecular hydrogen-bonding can also influence or determine the regio- and stereoselectivity of the ion pair combination resulting from deprotonation of the glycosyl acceptor. Whereas the kinetically most acidic OH group is deprotonated in the o-plane, a second OH group, hydrogen-bonded to the deprotonated OH group (and now linked by a H-bond to the newly generated oxyanionic center, Scheme 2) might be located in or close to the n-plane and attack the intermediate oxycarbenium ion. This effect has been demonstrated for a range of diols [20-251 and triols [26-291. It has also been evidenced for N-acetylglucosamine-deriveddiazirines, for which hydrogen-bonding to the NH group lead to the predominant formation of a-D-glycosides of sufficiently strong hydrogen-bond-accepting alcohols [301. 3) Solvation of the intermediate oxycarbenium ion. Solvation of the initially unsolvated or at least not optimally solvated oxycarbenium ion might involve the solvent, the glycosyl acceptor, or the C(2) substituent of the carbenium ion [18]. In an inert solvent, such as dichloromethane, solvation will occur by the (nucleophilic) glycosyl acceptor (inter- or intramolecularly) if it is sufficiently nucleophilic, or by the C(2)O-benzyl group for the less nucleophilic glycosyl acceptors such as phenols (as evidenced by studying the 2-deoxy analog 24 [31]). THF seems to solvate the carbenium ion well, and preferentially from the axial direction, leading preponderantly to equatorial glycosides. 4) In contradistinction to other methods of glycosidation, steric hindrance does not seem to be an important factor, as long as the hydroxy group is sufficiently acidic [ 18, 291. The following transformations illustrate these points.

162

A

7 Glycosylidene Diazirines

a)

b)

C)

F

3

C

Y

OGlcBn.,

/ OGlcBn, toluene: 79%, 8915

,,11..'

'...,,,

OGlcBn( toluene: 79%. 8416

CHzCI2: 34% (hv: 55%), 5050

dioxane, 25 "C: 76%. 95: 5 DME,25 "C 80%. 98: 2 EtCN, 25 "C: 5446,9010(16% of 31) EtCN, hv, -60 "C: 25%. 8713 (55% of 31)

31

e) CF3(CF2)5CH2CH~-OGlcBn, dioxane, 25 "C: 68%. 76:24

CHF~(CFZ)&H~~GICBQ toluene: 73% 6733

dioxane: 90%. 1:l mixture of p-D-isomers only

dioxane, 25 "C: 79%,78:22

CICH2CH2CI: 62%. 65:35 (4% of regioisorner, 4555)

Figure 1. Products of the reactions of the diazirine 7 with mono- and dihydroxy compounds: solvents, total yields, and p-D/a-D ratios (GlcBm = 2,3,4,6-tetra-0-benzyl-~-glucopyranosyl).

7.4.2 Glycosidation of Strongly Acidic Hydroxy Compounds Glycosidation of Phenols Glycosidation by the 1-azi-tetra-0-benzylglucopyranose7 proceeds in yields of approximately 75-80%, and leads mostly to the P-D-anomers [9]; presumably, the use of THF will increase the anomeric selectivity (Figure la, b). The analogous I -azimannose 19 yielded mostly a-D-mannosides, but has not been thoroughly studied [9].

7.4 Glycosidation by Glycosylidene Diazirines

163

The influence of intramolecular hydrogen-bonding is evident from the regioselective monoglycosidation of methyl orsellinate (Figure 1a) [9]. Upon addition of a second equivalent of diazirine, the chelated OH group is also glycosylated (67%,),although with less anomeric selectivity, as also observed for 4-nitrophenol. The insensitivity to steric hindrance is seen from the result of glycosylating 2,6-di(tert-butyl)-4-methylphenol(Figure 1b) [9]. Thiophenols react similarly to phenols; 4-methoxythiophenol yielding 70% of a D of thioglucosides [32]. 43 : 57 C X - D / ~ - mixture Glycosidation of Fluorinated Alcohols The influence of pK and the insensitivity to steric hindrance is seen from the glycosylation of fluorinated alcohols (Figure lc) [ 181. Glycosidation of trifluoroethanol, hexafluoro-2-propanol, and hexafluoro-tert-butanol by one equivalent of the diazirine 7 in dioxane yielded the anomeric glycosides in yields of 70-74% and in ~-D/.-D ratios ranging from 77:32 to 95:5. The solvent has an influence on the anomeric ratio, which changes for hexafluoro-tert-butanol from ~ - D / C X -= D 84 : 16 (75%) in dichloromethane to 98:2 (80%) in DME. The formation, upon glycosidation in propionitrile at -60°C, of the imino ether 31 (55%) besides the glycosides evidences the formation of an oxycarbenium cation, its solvation, and the preferred axial attack of the solvent. 3,3,4,4,5,5,6,6,7,7,8,8,-tridecafluorooctanoland of Glycosidation of CHF2(CF2)&H*OH by 7 in dioxane proceeded similarly and yielded 68 and 79% of the glucosides (P-D~CX-D = 76: 24 and 78 :22).

7.4.3 Glycosylation of Weakly Acidic Hydroxy Compounds Glycosidation of Monovalent Alcohols In contradistinction to the glycosylation of hexafluoro-tert-butanol, tert-butanol is glycosylated in only 34% yield under similar conditions (Figure 1d). Glycosylation at low temperature (photolytic generation of the carbene) doubled the yield, but had no influence on diastereoselectivity (IX-D/~-Dca 1 : 1). This dependence on the mode of generating the carbene is not general, and might reflect the temperaturedependence of the association of tert-butanol. Monoalcohols with o-acceptor substituents (such as partially protected monosaccharides) are expected to be more highly acidic and, hence, be glycosylated in higher yields. Indeed, 1,2: 5,6-di-0isopropylidene-a-D-glucofuranose yielded 75% of a 2 : 3 mixture of the corresponding anomeric disaccharides (Figure 1e) [ 181. The influence of acidity is also seen in the successful glycosidation of (hydrated) Ti02 [15]. The OH groups of hydrated Ti02 are characterized by pK values of 3.8 and 6.6. Exposure of Ti02 to benzylated or acetylated azi-glucoses 7 and 16 led to formation of a monomolecular layer of strongly (presumably covalently) bonded glucosyl residues (density 1.5 0.9 glucosyl moieties nm-2). The acetylated glucosyl moieties were deacetylated without removing the glucosyl moiety.

164

7 Glycosylidene Diazirines

Glycosidation of Diols and Triols Efsects of hydrogen-bonding The enhancing influence of an (intramolecular) hydrogen-bond on the kinetic acidity of the H-bond-accepting hydroxy group is seen from the (concentration independent!) result of the glycosidation of the myo-inositol-derived orthoester (Figure I f ) [26]. The two diastereoisomers derived from monoglycosylation of one each of the two enantiotopic hydrogen-bonded OH groups were obtained regio- and diastereoselectively in high yields. The low reactivity of the isolated equatorial OH group is because of a (bifurcated) hydrogen-bond. This effect also operates in the regioselective glycosidation at OC(2) of methyl 4,6-O-benzylidene-a-~-altropyranoside, yielding 62% of a 2: 1 mixture of anomers in dichloroethane (Figure lg) [20]. As expected, the intramolecular hydrogen-bond reduces the kinetic acidity of HOC(3), but enhances its nucleophilicity and leads to glycosylation at OC(3) bromide under Lemieux halogenwith 2,3,4,6-tetra-O-benzyl-a-~-glucopyranosyl exchange conditions, resulting in 68% of the a-D-configured, 1,3-1inked disaccharide (and 10% of the anomeric regioisomers). Combined efsects of intramolecular hydrogen-bonding and of stereoelectronic control The intramolecular hydrogen-bond between HOC(3) and the anomeric methoxy group in methyl 4,6-0-benzylidene-a-~-allopyranoside is expected to lead to protonation of the carbene mostly by the kinetically more highly acidic HOC(2) (Figure 2a) [22]. This is presumably taking place. In contradistinction to the altro-diol, however, the two OH groups are cis-oriented. Stereoelectroniccontrol requires that protonation occurs in the o-plane of the carbene and attack of the nucleophile on the ensuing oxycarbenium cation proceeds in the n-plane. HOC(3) is much closer to the x-plane than -0C(2). Glycosylation in toluene at 70 "C or in dioxane at 24 "C indeed leads mostly to the a-D-anomer of the 1,3-linked disaccharide [22]. The predominant a-D-configuration reflects the stereoelectronic control. Replacing these solvents by the more nucleophilic THF at -85°C leads to participation of the solvent. THF competes with

toluene: 79%.

C(2)0/C(3)0 2080 (68:32and 28:72)

CH2C12,-85", hv: 81%, C(2)0/C(3)O/C4)05:81:14(60:40,23:77, and 3664)

THF, -85", hv: 79%, C(2)0/C(3)0 7228 (9010and 3367)

Figure 2. Glucosylation of (a) methyl 4,6-O-benzylidene-a-~-allopyranoside and (b) benzyl ribopyranoside by the diazirine 7:solvents, total yields. regioselectivity, and p-D/a-Dratios.

p-D-

7.4 Glycosidution by Glycosylidene Diuzirines

165

Scheme 3. Preferred formation of the 3-O-P-~-glucosylateddimer in the reaction of 7 with benzyl (3-D-ribopyranoside.

HOC(3) in attacking the intermediate oxycarbenium ion, again from the axial di~ of the 1,2-1inked disaccharide! rection, and the main product is now the p - anomer Similarly for benzyl P-D-ribopyranoside, preferential protonation of the carbene ought to occur by either HOC(2) or HOC(4), but HOC(3) is in a better position to attack the carbenium ion from an axial direction. The main product, isolated in 65% yield, is the a-Danomer of the 1,3-linked disaccharide (Figure 2b) [28]. The rationalization of that reaction is illustrated in Scheme 3. In THF at -78 "C, however, the main products are the p-D anomers of the 1,2- and 1,4-linked disaccharides. More examples of regioselective glycosidations resulting from the combination of these factors have been published and a particularly striking effect of hydrogenbonding is provided by the glycosidation of the N-phthaloylated allosamine derivatives. The dominant product of the a-D-configured diol is the 1,3-linked disaccharide (Scheme 4a) [23]. The anomeric configuration suggests that the regioselectivity is the result of a protonation by HOC(4) and an interception of the carbenium ion by HOC(3); indeed, ' H NMR analysis showed the presence of a very strong hydrogen-bond between HOC(3) and the phthaloyl group, and a weaker hydrogen-bond between HOC(4) and HOC(3). The former hydrogen-bond strongly reduces the kinetic acidity of HOC(3), the latter orients the protonating HOC(4) close to OC(3). In agreement with this, the a-D-configured monoalcohol is unreactive (Scheme 4b), whereas its anomer is glycosylated (Scheme 4c), because the N-phthaloyl group now adopts a different conformation; also the anomeric diol is no longer regioselectively glycosylated (Scheme 4d). The results of the glycosylation of the a-D-configured trio1 (Scheme 4e) evidences that the weakest hydrogen-bond is that from HOC(4) to HOC(3)! Intramolecular solvation of the intermediate oxycarbenium ion has also been postulated to explain the high-yielding glycosylation of the tertiary hydroxy group

7 Glycosylidene Diazirines

166 a)

80%, 90:lO

no reaction

C)

*Thp

OAll Bn4Glc0

NFt

66%,2 1 3 9

4

CHIOBn H

o ~ HO

O

A

l

l

B n 4 G l f : N OAll + OH BqGlcO

29%.a 6 5

33%,39:61

B n 4 G l c ? !

HO FNOAll

HO FtNOAll 9%,(only p-D)

OAll NFt

+

E N Bn4Glc0 FtN OAll 51%. 72:28

Scheme 4. Glucosylation of anomeric N-phthaloylatedallosamine derivatives by one equivalent of the diazirine 7 in dioxane at room temperature: total yields and a-D/p-D ratios.

of 10-0-acetylginkgolide A by a lactose-derived diazirine (Scheme 5a) [29]; intramolecular participation of the carbonyl group of the adjacent lactone ring is presumably playing an important role also in the glycosylation of ginkgolide B, where reaction with 3.5 equivalents of 7 in THF yielded 42% of the all-P-D-configured trisaccharide (Scheme 5b) [29]. Thus, regioselective glycosidation of diols and triols by diazirines is feasible and often strikingly successful. Its course can be predicted, even if the factors are complex, if the hydrogen-bonds in the glycosyl acceptor are known. Analytically, the results of glycosylation provide important clues to the location of hydrogen-bonds (Figure 3a), e.g. evaluation of the influence of weak intramolecular OH. . .F-C hydrogen-bonding in apolar solvents (Figure 3b) [24, 251. There is yet another factor, operative in carbenes derived from 1-azi-2-acetamido-2-deoxy sugars that can occasionally be exploited. Efect of intermolecular hydrogen-bondingfrom the diazirine to the acceptor Surprisingly, thennolysis of the N-acetylglucosamine- and N-acetylallosamine-derived diazirines 14 and 25 in methanol led predominantly to the 1,2-cis-configured

7.4 Glycosidation by Glycosylidene Diuzirines

Me

167

28 (1.3 equiv.)

HO 3

THF, 25°C

0 92% of B-D-anomer

b'

0

N

Me

-

Me

7 (3.5 equiv.) THF, 25°C

42%

R = GlcBn, p+l p+3 p-10

15% p+l p-3 a+10 13% a+l p+3 p-10 24% diglucosides no rnonoglucosides

Scheme 5. (a) Lactosylation of 10-0-acetylginkgolide A by 28. (b) Glucosylation of ginkgolide B by 7.

a-r>-glycosides; similarly, glycosidation of 2-propanol (2 equivalents) by 14 at -84 "C yielded 70% of the a-D-glycoside (no 0-D-anomer observed) and 13% of the dihydrooxazole [30]. At 26 "C, the dihydrooxazole (resulting from protonation of the carbene by the acetamido group) became the major product. Under analogous conditions (-84 "C), hexafluoro-2-propanol yielded between 58 and 9 1% glycosides, depending on the excess of alcohol (1 to 10 equivalents), the anomeric ratio varying between 72 :28 and 80 :20. Glycosidation at 26 "C of 2 equivalents of hexafluoro-2propanol led to higher yields of glycosides (72 compared with 59%) and higher anomeric selectivity (.-D/~-D = 88 : 12). Glycosidation of the still more highly acidic

Figure 3. Intramolecular hydrogen-bonding in (a) ally1 2-deoxy-2-phthalimido-a-~-allopyranoside and in (b) L-4-deoxy-4-fluoro-1,3,5-pentyIidyne-myo-inositol.

168

7 Glycosylidene Diazirines

Figure 4. Intermolecular hydrogen-bonding of the carbene derived from the GlcNAc diazirine 14.

4-nitrophenol (2 equiv., 26 "C) reduced the anomeric selectivity to 58 :42 (87% yield). These and analogous observations made by glycosylating the A11NAc-derived diazirine 25 were rationalized by postulating an intermolecular hydrogen-bond from the NHAc group to the glycosyl acceptor (Figure 4). This hydrogen-bond locates the alcohol below the plane of the pyranose ring, well positioned to attack the oxycarbenium and form a-D-glycosides. The hydrogen-bond depends on the conformation of the acetamido group and on the basic properties of the alcohol (higher selectivity with the less acidic, i. e. more strongly basic 2-propanol). The conformation of the acetamido group, in its turn, depends on the configuration (glum or d o ; for a detailed study of the conformational behavior of the NHAc group see Ref. [33]) and the temperature (higher c ~ - D / ~ - Dratio at a higher temperature). The temperature-dependent association of the alcohols explains the much higher yield of the glycosidation of 2-propanol at lower temperature, influenced by the higher basicity of 2-propanol than hexafluoro-2-propanol (better H-bond acceptor) and by the greater acidity of the hydrogen-bond-accepting terminal alcohol unit in a chain of associated alcohols properly located in the o-plane of the carbene (Figure 4). Protonation of the carbene, which depends on the kinetic acidity of the alcohol, is reflected by the higher yields of the hexafluoro-2-propanol- and 4-nitrophenol-derived glycosides. These results confirm the strong influence of hydrogen-bonds on the result of glycosidation and characterize these diazirines as subtle, reactivity-based tools for the investigation of intra- and intermolecular hydrogen-bonds and their influence on regio- and diastereoselective glycosidation.

7.5 Synthesis of Spirocyclopropanes Glycosylidene diazirines react with electrophilic alkenes, e.g. N-phenylmaleimide, acrylonitrile, dimethyl fumarate, dimethyl maleate, and C&-fullerene, to yield spirocyclopropanes with formation of two C-C bonds at the anomeric center [34-371.

7.5 Synthesis of Spirocyclopropanes

R = Bn: So%, 9O:lO R = Piv: 60%. 92: 8

RO

RO

.

,

"'v

169

RO

y0N , P h

Ph

0

22%

BBnO n BnO

70%

Bn ~O- n 0

BnO

52

BnO

BnO Me02C

BBnO n BnO NC

6

,,C02Me

fCo2r"72%

q ..*>CN

CN

Brio%,

C)

O

O OBn CN q BnO

11

31

Brio%. %

Brio BnO

Me BnO BnO

,,C02Me BnO BnO

BnO C02Me

,

C02Me

BnO Me02C

60

40

-

-

29

20

33

1s

le02C

C02Me

[

: 60%

C02Me

d'

z:*,

RRO O+]. RO

'%O

R=Bn:

33%

R=Piv:

72%

Me

24%

'$o

b 9%

14%

Figure 5. Products of the reaction of the benzylated and pivaolyated diazirines 7 and 15 with (a) N-phenylmaleimide, (b) acrylonitrile, (c) dimethyl fumarate and dimethyl maleate, and (d) dihyand 2,3-dihydro-4H-pyran. drofuran, 5-methyl-2,3-dihydrofuran,

Yields depend on the protecting groups; substitution of benzyl by pivaloyl groups reduces yields, in agreement with a reduction of the predominantly nucleophilic character of these ambiphilic carbenes. As illustrated by the cyclopropanation of N-phenylmaleimide, the diastereoisomer with the substituents oriented syn to the glycosyl ring oxygen are formed preferentially (ratio = 90: 10; Figure 5a); cyclopropanation with acrylonitrile confirms this (ratio = 58 :42) and shows that the nitrile function is predominantly located below the average plane of the pyranosyl ring (ratio = 83: 17; Figure 5b). This is confirmed by cyclopropanation of ethyl acrylate [38]; the major isomer, formed in ca 47% yield, has the same configuration as the major isomer resulting from addition to acrylonitrile. The stereoselectivity of cyclopropanation with dimethyl fumarate and dimethyl maleate (Figure 5c) suggests that formation of the two C-C bonds is not concerted; it is rationalized by

170

BnO

7 Glycosylidene Diuzirines

-

BnO BnO

EN pseudoaxial approach

pseudoequatorial approach

Scheme 6. Pseudoaxial and pseudoequatorial. ring closure of the zwitterion obtained upon reaction of acrylonitrile with the carbene derived from 7.

postulating a two-step process with a preferred pseudoaxial trajectory in the formation of the second C-C bond, as illustrated in Scheme 6. In agreement with the ambiphilic character of alkoxycarbenes, glycosylidene carbenes form cyclopropanes also with electron-rich alkoxyalkenes, provided their reactivity is enhanced by ring tension, such as that in dihydrofuran, that leads to 33% (benzylated diazirine 7) or 72% (pivaloylated diazirine 15) of a single isomer (Figure 5d) [39]. Steric hindrance by an additional methyl group reduces the yield. Dihydropyran reacts less readily, and open chain enol ethers lead to low yields only. The 0-benzyl protected diazirines gave lower yields than the 0-pivaolylated analogs. The transition states of the cycloadditions leading to the isomeric cyclopropanes have been calculated (semiempiric program AM1). At the beginning of the reaction, the interaction of the HOMO of the carbene with the LUMO of the enol ether dominates, but in the transition state the interaction between the LUMO of the carbene and the HOMO of the enol ether is more important. The cyclopropanation yields of the pivaloylated diazirine 15 are higher because of the more efficient competing reaction of the 0-benzylated (as compared to the 0-pivaloylated) carbene with the starting diazirine. Azines are isolated as the main by-products. Calculations show that a pseudoaxial attack of the carbene is favored.

7.6 Addition to Aldehydes and Ketones Addition of the benzylated aziglucose 7 to acetone or cyclohexanone, used as solvents, gave 63 and 78%, respectively, of a mixture of anomeric spiroepoxidesin P-D/ a-Dratios of 70 :30 and 65 : 35, respectively (Figure 6a), whereas addition of the A11NAc-derived diazirine 25 to cyclohexanone at 60°C gave 83% of the a-Dconfigured spiroepoxide and a dihydrooxazine formed either directly from the intermediate oxycarbenium cation or by opening the anomeric oxirane (Figure 6b) [40].Similarly, addition of 25 to a threefold excess of benzaldehyde gave a mixture of &,trans-related a-D-spiroepoxides (27%), epimeric dihydrooxazoles (1 2%; formally derived from the epimeric epoxides), and a range of by-products. Such spiroepoxides have been prepared in a convenient way by Nicotra and Panza [41-431.

7.7 Explorutory Use of Diuzirines

171

b)

a)

BnO

78%

83%

35

60

40

Me

Figure 6 . Products of the reactions of (a) the diazirine 7 and (b) the diazirine 25 with cyclohexanone: total yields and product ratios.

7.7 Exploratory Use of Diazirines: Formation of Glycosyl Phosphines, Stannanes, N-Sulfonylamines, Esters, Boranes, and Alanes, and of 1,l-Difluorides The high reactivity of glycosylidene carbenes enables rapid access to the first representatives of intermediates or products for which more convenient routes might subsequently be developed. Thus, the benzylated aziglucose 7 reacted with Ph2PH under thermal conditions to give anomeric glycosyl phosphines that were isolated as the corresponding phosphine oxides in yields of up to 76% (Figure 7a) [44]. The u - D / ~ - Dratios of anomers varied between 5 1 :49 and 67 :33. Generation of the carbene at -78 "C (hv) led in 55% yield to a 74:26 mixture of anomers. Glycosyl phosphine oxides were then prepared from glycosyl acetates by TMSOTf-catalyzed treatment with MeOPPh2 and subsequently reduced to glycosyl phosphines [45].

b)

a)

BnO BnO

PPhZ OBn

16%, 55:45

Bd-0-

SnBuj

SdU3

OBn

70%. 84:16

53%. 4357

C)

."in-&+,

BnO BnO &NH.,.s OBn

50%. 2575

BnO

45%

oyyco"n ri'

NH-Bw

Figure 7. Products of the reactions of (a) the diazirine 7 with diphenylphosphine, (b) the diazirines 7 and 19 with tributylstannane, and (c) the diazirine 7 with 4-toluenesulfonamide and N-Bocasparagine benzyl ester: yields and a-D/p-D ratios.

172

7-

7 Glycosylidene Diazirines

BR3

Scheme 7. Expected primary products of the reaction of diazirine 7 with trialkylboranes.

The benzylated aziglucose 7 and azimannose 19 react with tributyl or triphenyl stannane and yield glycosyl stannanes (Figure 7b) [46]. The 1,2-cis-configured anomers are formed preferentially and the 1-C-stannylated 2-0-benzyloxyglucal is the main by-product. Advantageous preparations of glycosyl stannanes-useful precursors of glycosyl lithium compounds-from glycosyl halides have been described by Kessler et al. [47-501. Exploratory experiments have shown that glycosylidene carbenes also insert into the N-H bond of sulfonamides, whereas carboxamides yield esters, as illustrated by the product obtained from an N-Boc-asparagine benzyl ester (Figure 7c) [32]. Presumably, such esters are formed by hydrolysis of the initially formed imino ether 0-glycosides, suggesting that glycosylidene diazirines may be used for (selective?) cleavage of peptidic bonds. Nucleophilic addition of glycosylidene carbenes to triligated boron compounds followed by migration of a boron substituent to the glycosidic center should lead to anomeric glycosyl boranes (Scheme 7).

Bnb,!,, BBu3:

""

Bnd

45%

20%

62%

-

C>

B-Bu:

1) BEt3.2) : ' H

B BnO n

O

h

E

t BnO

BnO HO' 'Et 1 ) BEt, 2) PPh3.02: 25%

BnO BnO

0

46%

WC'

4

C)

'Et

55%

46

:

BnO

54

Figure 8. Products of the reaction of diazirine 7 with (a) tributylborane and 2-butyl-l,3,2-dioxaborinane, (b) triethylborane followed by acid treatment, (c) triethylborane followed by oxidation, and (d) the 9-BBN-derived phenethylborinate.

7.7 Exploratory Use of Diazirines -0Bn

D*OnB R = Me: R = 'Bu:

BnO

,OBn

BnOR

30% 20%

173

R

30% 25%

Figure 9. Products of the reaction of diazirine 7 with trialkylalanes followed by deuterolysis.

Reaction of 7 with tributyl-, triethyl-, and triphenylborane, followed by oxidation, yielded the C (1)-alkylated hemiacetals and the C(1)-alkylated glycals (Figure 8a) [51, 521. The cyclic 2-butyl-l,3,2-dioxaborinane led only to the hemiacetal. The presence of the intermediate (anomeric?) diethyl glycosyl boranes was evidenced by treating them with trifluoroacetic acid, leading to migration of a second ethyl group to C(l) and formation of a cyclic stable borinate (Figure 8b). Oxidation of the intermediate glycosyl borane by oxygen in the presence of excess triphenylphosphine led to a stable ethylglycosylborinic acid (Figure 8c). We took advantage of the higher resistance to oxidation of borinates, compared with boranes, by treating the aziglucose 7 with the 9-BBN-derived phenethyl borinate (Figure 8d). The configurations of the anomeric borinates was assigned on the basis of NOE data. Trialkylalanes, similarly to trialkylboranes, react with the aziglucose 7 leading to bona jide glycosyl alanes, as was proved by deuteration to yield the axial C-glycoside and the C(1)-alkylated glucal (Figure 9) [52]. Finally, because the 1,l-difluoro-1-deoxy derivative of cellobiose was required for kinetic studies of cellobiohydrolases, we examined the reaction of the corresponding diazirine with XeF2. A method for the synthesis of monosaccharide-derived anomeric difluorides has been described by Praly [53,541, but seems unsuitable for the synthesis of disaccharide-derived analogs. Treatment of the cellobiose-derived diazirine 26 with XeF2 in dichloromethane yielded a maximum of 22% of the desired difluoride and large amounts of azines (Scheme 8) [ 161. In the presence of small amounts of MeOH, these azines were replaced by, mostly, the equatorial monofluoride. These results show that the reactivity of glycosylidene carbenes towards XeF2 is not pronounced, but sufficient to furnish the amounts of the desired difluoride required for the enzymatic tests [ 5 5 ] .

26

R=F R=H

22%

Scheme 8. Reaction of the cellobiose-derived diazirine with xenon difluoride.

174

7 Glycosylidene Diuzirines

Acknowledgments

We thank the Swiss National Science Foundation and F. Hoffmann-La Roche AG, Basel, for continuous generous support.

References 1. A. Vasella, K. Briner, N. Soundararajan, M. S . Platz, . I . Org. Chem., 1991, 56, 4741-4744. 2. A. Vasella, in Topics in Bioorganic and Biological Chemistry (Ed.: S. M. Hecht), Oxford University Press, Oxford, 1999, pp. 56-88. 3. A. Vasella, C. Witzig, C. Waldraff, P. Uhlmann, K. Briner, B. Bernet, L. Panza, R. Husi, Helv. Chim. Acta 1993, 76, 2847-2875. 4. A. Vasella, Pure Appl. Chem. 1993, 65, 731-752. 5. A. Vasella, Pure Appl. Chem. 1991, 63, 507-518. 6. K. Briner, A. Vasella, Helv. Chim. Acta 1989, 72, 1371-1382. 7. S. E. Mangholz, A. Vasella, Helv. Chim. Acta 1995, 78, 1020-1035. 8. S. E. Mangholz, A. Vasella, Helv. Chim. Acta 1991, 74, 2100-2111. 9. K. Briner, A. Vasella, Helv. Chim. Acta 1990, 73, 1764-1778. 10. B. Aebischer, Thesis, Universite de Fribourg No. 854, 1983. 11. D. Beer, Thesis, Universitat Zurich, 1989. 12. P. Finch, Z. Merchant, J. Chem. SOC.,Perkin Trans. I 1975, 1682-1690. 13. J. Attenburrow, A. F. B. Cameron, J. H. Chapman, R. M. Evans, B. A. Hams, A. B. A. Jansen, T. Walker, J. Chem. SOC.1952, 1094-1111. 14. D. Beer, A. Vasella, Helv. Chim. Acta 1985, 68, 225442274, 15. M. Weber, A. Vasella, M. Textor, N. D. Spencer, Helv. Chim. Acta 1998, 81, 1359-1372. 16. M. Weber, Thesis, ETH-Zurich No. 13212, 1999. 17. K. Briner, A. Vasella, unpublished results. 18. K. Briner, A. Vasella, Helv. Chim. Acta 1992, 75, 621-635. 19. K. Briner, Thesis, Universitat Zurich, 1995. 20. E. Bozo, A. Vasella, Helv. Chim. Acta 1992, 75, 2613-2633. 21. P. R. Muddasani, E. Bozo, B. Bernet, A. Vasella, Helv. Chim. Acta 1994, 77, 257-290. 22. P. R. Muddasani, B. Bernet, A. Vasella, Helv. Chim. Acta 1994, 77, 334-350. 23. K. Briner, B. Bernet, J. L. Maloisel, A. Vasella, Helv. Chim. Acta 1994, 77, 1969-1984. 24. A. Zapata, B. Bernet, A. Vasella, Helv. Chim. Acta 1996, 79, 1169-1191. 25. M. A. Biamonte, A. Vasella, Helv. Chim. Acta 1998, 81, 695-717. 26. P. Uhlmann, A. Vasella, Helv. Chim. Acta 1992, 75, 1979-1994. 27. E. Bozo, A. Vasella, Helv. Chim. Acta 1994, 77, 745-753. 28. P. Uhlmann, A. Vasella, Helv. Chim. Acta 1994, 77, 1175-1 192. 29. M. Weber, A. Vasella, Helv. Chim. Acta 1997, 80, 2352-2367. 30. A. Vasella, C. Witzig, Helv. Chim. Acta 1995, 78, 1971-1982. 31. Y. Takahashi, A. Vasella, Helv. Chim. Acta 1992, 75, 1563-1571. 32. T. Rajamannar, A. Vasella, unpublished results. 33. P. Fowler, B. Bernet, A. Vasella, Helv. Chim. Acta 1996, 79, 269-287. 34. A, Vasella, C. A. A. Waldraff, Helv. Chim. Acta 1991, 74, 585-593. 35. A. Vasella, P. Uhlmann, C. A. A. Waldraff, F. Diederich, C. Thilgen, Angew. Chem. 1992,104, 1383-1385. 36. P. Uhlmann, E. Harth, A. B. Naughton, A. Vasella, Helv. Chim. Acta 1994, 77, 2335-2340. 37. U. Jonas, F. Cardullo, P. Belik, F. Diederich, A. Gugel, E. Harth, A. Herrmann, L. Isaacs, K. Mullen, H. Ringsdorf, C. Thilgen, P. Uhlmann, A. Vasella, C. A. A. Waldraff, M . Walter, Chem. Eur. J. 1995, I , 243-251. 38. C. Bluchel, A. Vasella, unpublished results. 39. C. Waldraff, B. Bernet, A. Vasella, Helv. Chim. Acta 1997, 80, 1882-1900. 40. A. Vasella, P. Dhar, C. Witzig, Helv. Chim. Acta 1993, 76, 1767-1778.

References 41. 42. 43. 44. 45. 46. 41. 48. 49. 50. 51. 52. 53. 54. 55.

115

L. Lay, F. Nicotra, L. Panza, G. Russo, G. Sello, J. Curbohydr. Chem. 1998, 17, 1269-1281. L. G. Lay, F. Nicotra, L. Panza, G. Russo, Synlett 1995, 167-168. F. Nicotra, L. Panza, G. Russo, Tetrahedron Lett. 1991, 32,4035-4038. A. Vasella, G. Baudin, L. Panza, Heteroat. Chem. 1991, 2, 151-161. A. Lopusinski, B. Bernet, A. Linden, A. Vasella, Helv. Chim. Acta 1993, 76, 94-112. P. Uhlmann, D. Nanz, E. Bozo, A. Vasella, Helu. Chim. Acta 1994, 77, 1430-1440. F. Burkhart, M. Hoffmann, H. Kessler, Tetrahedron Lett. 1998, 39, 7699-7102. M. Hoffmann, H. Kessler, Tetrahedron Lett. 1997, 38, 1903-1906. M. Hoffmann, H. Kessler, Tetrahedron Lett. 1994, 35, 6067-6070. 0. Frey, M. Hoffmann, V. Wittmann, H. Kessler, P. Uhlmann, A. Vasella, Helu. Chim. Actu 1994, 77, 2060-2069. A. Vasella, W. Wenger, T. Rajamannar, Chem. CommuM. 1999, 2215-2216. W. Wenger, A. Vaseila, Helv. Chim.Acta, submitted. J. P. Praly, L. Brard, G. Descotes, L. Toupet, Tetrahedron 1989, 45, 4141-4152. J. P. Praly, G. Descotes, Tetrahedron Lett. 1987, 28, 1405-1408. D. Becker, K. S. H. Johnson, A. Koivula, M. Schiilein, M. L. Sinnott, Biochem. J. 2000, 345, 315-319.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

8 Glycosylation Methods: Alkylations of Reducing Sugars Jun-ichi Tamuru

8.1 Introduction Methods for glycosylation generally contain a procedure for activation of the anomeric position. Scheme 1 shows two possible procedures for formation of a glycosidic linkage. Conventionally, the specific affinity of a leaving group at the anomeric position and a suitable promoter accelerates a coupling reaction, as described in the previous chapter. As a result a glycosidic bond is formed by nucleophilic attack of aglyconic oxygen (b). We can also form the glycosidic linkage by nucleophilic attack of anomeric oxygen (a). This glycosylation procedure retains the anomeric oxygen (a) in the glycosidic bond. The glycosylation methods referred to in the title of this chapter are based on the latter concept. R. R. Schmidt and P. Kovac introduced the reverse concept to glycosylation methods and developed ‘anomeric O-alkylation’ and ‘glycosylation via locked anomeric configuration’, respectively. These methods, of which general procedures are shown in Scheme 2, use different approaches for stereocontrol of glycosidic bond formation. The ‘anomeric O-alkylation’ method utilizes an anomeric sodium or potassium alkoxide as a nucleophile. Stereoselectivity in the coupling reaction depends on the different reactivities of the anomeric c1 and anions. Factors controlling the reactivity, such as conformation of the nucleophile, are described in detail in Section 8.2. Review articles by Schmidt have described the ‘anomeric O-alkylation’ method [ 1-31. ‘Glycosylation via locked anomeric configuration’ employs a stannylene acetal at the 1,2-position of the hemiacetal. The stannylene acetal fixes the anomeric oxygen in the 1,2-cis configuration and leads to 1,2-ci.stype glycoside. Details are described in Section 8.3.

8.2 Anomeric O-Alkylation In addition to temperature, solvent, promoter, and additive, we must consider the configuration of the nucleophile (including roles of substituent groups) and the re-

178

8 Glycosylution Methods: Alkylations of Reducing Sugars

Electrophile (Donor)

Nncleophile (Acceptor)

L

_-___-__ . I

Scheme 1. Conventional and new glycosylation methods. L = leaving group; M = metal atom.

11

m* Electrophile

Nucleophile

I

"Anomeric0-alkylation"

Nucleophile

1

Glycoside

I

J

I "GlycosylationvM locked anomeric configuration" I Scheme 2. General procedures for 'anomeric 0-alkylation' and 'glycosylation via locked anomeric configuration'.

activity of the electrophile (primary compared with secondary triflates). This section describes the factors that lead to the stereoselective coupling reaction.

8.2.1 Anomeric 0-Alkylation of Ribofuranose with Primary Triflates: Effect of the Protecting Group at 0 - 5 of Ribofuranose Schmidt and Reichrath performed the anomeric 0-alkylation of ribofuranose with primary triflates with complete stereocontrol as shown in Scheme 3 [4]. Ribofur-

8.2 Anomeric 0-Alkylation

179

TBDMSCI 0 0

KOtBu ITHF

I

0'

.\c (2)

I (4) 89 %

(5) 86 %

I

.\c

\I*

Scheme 3. Stereoselective syntheses of Ribfu and p(1-5)Ribf and plausible complex formation. S = solvent molecule; TBDMS = t-butyldimethylsilyl.

anose 2,3-O-isopropylidene acetal (1) and the corresponding 5-0-TBDMS derivative (2) were selected as nucleophiles. Reaction of these nucleophiles with primary triflate (3) stereoselectively gave a- and P-linked disaccharides (4 and 5), respectively, in high yields. Promoter and solvent seem not to have a significant effect on the reaction, although the substituent group at 0 - 5 of ribofuranose might be important for stereocontrol. Plausible intermediates explaining this specific reaction are shown in Scheme 3, bearing in mind the steric influence of the protecting group. 8.2.2 Anomeric 0-Alkylation of Mannofuranose with Primary Triflates: The Crown Ether Effect

Schmidt et al. reported an effect of a crown ether on the glycosylation of mannofuranose. Scheme 4 shows an example of stereoselective coupling of mannofuranose 2,3 : 5,6-di-O-isopropylideneacetal (6)with primary triflate (7) in the presence of NaH [5, 61. The coupling reaction in the absence of the crown ether afforded only P-glycosides (9). Mannofuranose is different from ribofuranose in having two endo hydroxyl groups at the 2,3-positions. Thus mannofuranose could be a better ligand for complex formation. In particular, the p anion preferentially forms a complex

180

8 Glycosylution Methods: Alkylutions of Reducing Sugars

r

1

1

(10)

Scheme 4. Anomeric 0-alkylation of mannofuranose and effect of crown ether.

(11) as speculated in Scheme 4. Addition of 18-crown-6 afforded the a-glycoside (8) exclusively. It is thought that the crown ether inhibited endo-type complex formation and equilibrium moved to favor the a-anion (lo), yielding the a-glycoside. Addition of NaI resulted in an equimolar anomeric mixture of the glycosides. Perhaps partial complex formation of 18-crown-6with the sodium ion was suppressed. 8.2.3 Anomeric 0-Alkylation of Gluco- and Galactopyranoses with Primary Triflates: High fl-Selectivity as a Result of the Reactive Anomeric p-Anion Schmidt et al. examined the reactivities of electrophile and nucleophile in the anomeric 0-alkylation of 2,3,4,6-tetra-0-benzylglucose (12) [7]. Alkylation at room temperature gave decanoyl a- and P-glucosides in 7 :3 ratio, which reflected the anomer ratio of the hemiacetal in THF (2: 1). On the other hand, alkylation with less reactive methyl triflate at room temperature afforded a 2: 8 mixture of methyl a- and P-glycosides. When the reaction is performed with the less reactive electrophile, stereoselectivity reflects the reactivities of the nucleophiles instead of the original anomer ratio of hemiacetals. The effect of complex formation was also examined. Coupling of 12 and methyl triflate in THF or toluene at -50 "C gave the a/P mixture in the ratios 8 :2 and 1 :8, respectively. Addition of crown ether to the THF or toluene at the same temperature did not significantly alter the a/P ratio of

8.2 Anomeric O-Alkylution Primary triflate (3,15-19), t-BuOK or NaH OH

OBn

/ THF or dioxane

181

Bno&oR2

*

BnO OBn

(12) R' = Bn (13) R' = H (14) R ' = MTr

EL Bg% BnO

0'

I

BnO

(17) X= OBn (15) X= >CMe2 (16) X = > C H C ~ H I ~ (18) X=OMe

1) 17, NaH I THF B BnO n

O OBnGOH

(14)

2) H+

67 %

B BnOn

O

B

4

n

O

q

(20) BnO BnoOBn

Scheme 5. Anomeric O-alkylation of glucopyranoses and primary triflates. MTr = p-methoxytrityl

the products. In contrast with the two furanose examples described above, complex formation could not explain the stereoselectivity of the coupling of pyranoses with gluco and galacto configurations. Anomeric O-alkylation of benzylated glucopyranoses 12-14 with several primary triflates (3,15-19) gave p-glucosidesexclusively as shown in Scheme 5 [7,8].These pselectivities are a result of the greater reactivity of the anomeric p-anion. Although steric hindrance by the substituent group at 0-6 (MTr, Bn and H), solvent (THF and dioxane) and promoter (NaH and t-BuOK) did not affect the P-selectivity, there is one exception-glycosylation of 14, which has MTr at 0-6, with triflate (17) unexpectedly afforded the a-linked isomaltoside (20) only. Steric factors and the different reactivity on the triflates are primarily responsible for this behavior. The product from coupling of tribenzylglucopyranose (13) and triflate (19) was converted into lipoteichoic acid carrier fragment (21). Scheme 6 demonstrates the result and excellence of this anomeric O-alkylation method for the synthesis of a complex natural product (22) [2]. 8.2.4 Anomeric O-Alkylation of Acyl-Protected Nucleophiles with Primary Triflates

Despite of the presence of NaH, neither acyl migration nor orthoester formation occurred during the anomeric O-alkylation of alkaline labile acetyl-protected sugars (hemiacetal). Scheme 7 shows the results. Klotz and Schmidt reported preferential

182

8 Glycosylution Methods: Alkylutions of Reducing Sugars

Scheme 6. Synthesis of lipoteichoic acid fragment via anomeric 0-alkylation.

p-selectivities in the anomeric 0-alkylation of tetraacetylgluco- (23) and -galactopyranose (24), and of heptaacetyllactopyranose (25) with decyl triflate at room temperature [9]. This procedure is applicable to the synthesis of p-lactosyl ceramide. Klotz and Schmidt converted heptaacetyllactose (25)with the triflate and nonaflate(nonafluorobutane-sulfonate)of azidosphingosine (31 and 32) into plactosyl ceramide (33)in 49 and 42% yield, respectively (Scheme 8). In contrast, tetraacetylmannopyranose (26)was converted to decyl a-mannopyranoside (20) stereoselectivelyat -40 "C as shown in Scheme 7. This a-selectivity arises as a result of steric and electronic effects of the nucleophile and the reduced reactivity of the p-

(OAc

Scheme 7. Stereoselective anomeric 0-alkylation of acyl-protected sugars. DME = 1,2-dimethoxyethane.

8.2 Anomeric 0-Alkylution

183

y3

NaH I 1,2-diethoxyethane

OTBDMS

(31) R = Tf

Scheme 8. Synthesis of Iactosylceramide from acyl-protected lactose via anomeric 0-alkylation.

anion. When the temperature was reduced a-glycosides were also obtained from 23-25. This tendency is similar to that observed with alkyl-protected glucoses. Lubineau et al. reported an anomeric 0-alkylation (a-stereoselective Michael (GlcNAc) (34) on P-tosyloxaddition) of 3,4,6-tri-O-acetyl-N-acetylglucosamine yacrolein, as shown in Scheme 9 [lo]. The coupling product (35) was successfully converted into the corresponding glycoside with an a-linked spacer (36). The acetylprotected disaccharide nucleophile Galp( 1-3)GalNAc is also applicable to the above procedure. Very recently Lubineau et al. also accomplished stereoselective synthesis of aand P-glycosides [ 111. Allylation of acetylated GlcNAc (34) with ally bromide in the presence of NaH in CH2C12 gave the P-anomer (37b) exclusively. In contrast, addition of Bu4NI to the reaction mixture gave an a-glycoside (37a) only. This complete stereoselectivity is because the anomeric tetrabutylammonium alkoxide was

A AcO c O

L

,

,

NHAc

A AcO c

NaH, 18-crown-6 I THF

83%

O

G

AcHNo+%o (35)

1 ) Ph3P=CHCOzEt I THF 2) H2 I Pd 3) NaOEt

i

Scheme 9. Anomeric 0-alkylation of GlcNAc (a-selective Michael addition)

184

8 Glycosylation Methods: Alkylations of Reducing Sugars

A AcO c

a

,

NHAc

r

bco&ONa AcO

NHAc

AC& AcO

AcmONa

1 I

AllBr

/OAc A cAcO O-&$, AcHNOAll

Scheme 10. Stereoselective ally1 glycosylation of GlcNAc via anomeric 0-alkylation.

more reactive than its sodium counterpart and thus the alkylation occurred more rapidly than epimerization to the p-anomer (Scheme 10). 8.2.5 Anomeric 0-Alkylation of Mannopyranose with Primary Triflates: Possibility of Intramolecular Complexation of the Nucleophile The reactive p-anion of gluco- and galacto-type nucleophiles explains the pselectivities of anomeric 0-alkylations. Schmidt et al. examined this for the mannose-type nucleophile which is a C-2 epimer of glucose. Temperature-dependent stereoselectivity was observed in the coupling of tetrabenzyl- and tetraacetylmannopyranose with triflated solketal (15), behavior similar to that of the gluco and galacto series [6, 91. Alkylation of 2,3,4-tri-O-benzyl-6-O-p-methoxy-trityl mannopyranose (39) with 15 in THF, however, gave the P-mannopyranoside exclusively at temperatures in the range -30 to 0 "C [ 6 ] .Obviously, steric hindrance by the substituent group at 0 - 6 leads to P-selectivity. This result is different from that for glucopyranose. Tamura and Schmidt explained this selectivity by use of several mannopyranosyl nucleophiles with different types of protecting group [ 121. Stereoselectivities were found to be a result of the formation of more reactive intermediates derived from substituent groups on the nucleophiles. Variation of the substituent group at 0 - 6 evidently changes the yield and stereoselectivity. Coupling of triflate 15 and mannopyranosyl nucleophiles 39-41 with bulky substituent groups (R = MTr, Tr, and TBDPS) at 0 - 6 resulted in higher pselectivity [R = MTr; a : 15, f3 :74, R = Tr; a :3, p: 62, R = TBDPS; a : 14, p: 73 (%)I in THF than for the 6 - 0 unprotected nucleophile (38) [ a : 6 ,p:21 (%)I. The

8.2 Anomeric O-Alkylation

BnO

R 1-

f."

OH

I

185

(38)R=H (39) R = MTr (40)R = Tr (trityl) (41) R = TBDPS (t-butyldiphenylsilyl) rOTf

NaH

Ic) Scheme 11. Plausible complex formation (A-D) and coupling reaction of mannopyranose. Ln = solvent molecule as ligand.

equilibrium between the 1C (A) and C1 (B) conformations in Scheme 11 explained these results. In general, for thermodynamic reasons, mannopyranose with a bulky group at 0 - 6 tends to shift towards the C1 form (complex B) in the equilibrium. Thus, P-mannosides are afforded uia the reactive intermediate B. The 6-0unprotected mannopyranose, on the other hand, prefers 1C-type stable complex A formation, which leads to a lower yield in THF. MEM (methoxyethoxymethyl) at 0 - 2 certainly forms a highly stable crown ether-like complex C in THF, as depicted in Scheme 11. In addition, THF is an obligatory ligand in the formation of the above MEM ether complex. No coupling product was obtained in THF whereas alkylation in polar CH3CN afforded a- and p-mannopyranosides in yields of 23 and 48%, respectively. The combination of solvent and protecting group also affects yield and stereoselectivity. Anomeric O-alkylation of mannopyranose with isopropylidene acetal at in 0-2,3 afforded the corresponding mannopyranoside in high yield [a:9, P :61 ("A)] THF. The same reaction in CHZC12 resulted in a lower yield (27%) but exclusive pformation. Isopropylidene acetal at 0-2,3 contributed to the formation of a stable skew-boat-type complex D and CHZC12 acted as more suitable ligand than THF. 8.2.6 Anomeric O-Alkylation of KDO with Primary Triflates a-Glycosylation with KDO (2-keto-3-deoxyoctonate= 3-deoxy-~-manno-octulosonic acid) glycosyl halides, with the exception of fluorides, by the Koenigs-Knorr

186

8 Glycosylation Methods: Alkylations of Reducing Sugars

@ R' +R2 = (CH&, @=Solvent

Figure 1. Possible reaction intermediate leading to KDO glycoside.

method often gives unsatisfactory results because of elimination of the hydrogen halide and accompanying P-glycoside formation, whereas anomeric O-alkylation accomplishes a-selective KDO glycosylation [ 131. To increase the nucleophilicity of KDO it is better to convert it into the corresponding 4,5 :7,sdicyclohexylidenecarboxamide beforehand, because this compound can take a boat conformation and keep the 0 - 2 reactive quasi-equatorial configuration. Deprotonation of the carboxamide possibly stabilized the dianion species by complexation with metal ion as depicted in Figure 1. KDO glycosylation with some primary triflates gave the desired a-glycosides exclusively in 48-83% yields [ 141. Rembold and Schmidt accomplished the synthesis of lipid A with the [KDOa(2-6)GlcNP(26)GlcNI sequence which contains fatty acid and phosphate moieties [ 151. Anomeric O-alkylation can be a powerful method for KDO glycosylation. 8.2.7 Anomeric O-Alkylation of Partially Protected Aldoses with Primary Triflate Long-chain alkyl glycosides have become increasing important as surfactants because of their good physical performance and low toxicity. A shorter synthetic route for these compounds and use of simple materials are required by industry. Schmidt's group proved it was possible to reduce the substitution on the nucleophile in the anomeric O-alkylation. The use of 6-O-unprotected tribenzylglucopyranose (13) and P-selective coupling has been described in Section 8.2.3. Scheme 12 shows p-selective coupling of 2-O-unprotected tribenzylglucopyranose (42) with decyl triflate in CHzC12 to afford 43 in 60% yield [16]. These reactive positions (2- and 6OH) proved not to need protection. The amount of promoter was shown to control the generation of the regioisomer. The use of one equivalent of NaH in the alkyla(44)with triflate 15 afforded P-glycoside tion of 3,4,6-tri-O-benzylmannopyranose (45) exclusively, whereas the undesired 1,2-disubstituted isomer (46) was also obtained when two equivalents of NaH were used (Scheme 12) [ 121. TBDPS ether was employed at the primary position of glucose and lactose to increase solubility in common organic solvents [ 161. Scheme 13 shows the results of alkylation of these nucleophiles (47 and 50). Anomeric O-alkylation of the glucose (47) in toluene afforded a- and P-decylglycoside (48a, b) in 58% yield. The coupling of lactose (50) needed additional crown ether to yield the glycoside (51a, b). THF was not suitable because of formation of an undesired product (49a, b).

8.2 Anomeric 0-Alkylation ,OBn H O* ' " BnO

187

CioH21OTf NaHCH2CIZ (1.1 eq) OH

BnO& BnO o-

on

,/rb NaH ( 1 eq)

BnOB

X

o

BnO

H

(44)

I CHpClp

NaH (2 eq)

Scheme 12. Anomeric 0-alkylation of partially protected nucleophiles.

c1&210Tf, NaH

H HOo

g

o

-

OH

(48a,b)

HO

58 % (a$= 1:4)

OTBDPS

C,oH,,OTf, NaH

(47)

THF

- -&H

OH

(49a,b)

OTBDPS

(50)

OH

toluene 15-crown-5

55 %(a$ = 1:2)

- 0

OTBDPS

(51a,b)

OH

56 % (a$= 1:l)

Scheme 13. Anomeric 0-alkylation of less protected nucleophiles.

8.2.8 Anomeric 0-Alkylation of Unprotected Aldoses with Primary Triflate, Bromides, and Cyclic Sulfates We can also use 0-unprotected aldoses for anomeric 0-alkylation by employing an aprotic dipolar solvent, namely N,N'-dimethylhexahydropyrimidin-2-one (DMPU) [17]. DMPU is essential to accomplish the anomeric 0-decylation of 0-

188

8 Glycosylation Methods: Alkylations of Reducing Sugars

unprotected aldoses with didecyl sulfate in the presence of NaH. Glucose, mannose, N-benzoylglucosamine, and xylose are converted to the corresponding decylglycopyranosides in moderate yields as anomeric mixtures ( a :p = 2 : 1-1 : 10). Decyl agalactofuranoside and a/p-arabinofuranoside ( 1 : 3) were obtained exclusively from galactose and arabinose, respectively, under these reaction conditions. Benzyl and allyl glycosides of these aldoses can be obtained by this procedure with benzyl and allyl bromides. This is a much more effective method than conventional Fisher glycosylation for simple glycoside synthesis. Cyclic sulfates can be used as potent electrophiles [ 181. Cyclic sulfates with longer alkyl chains which are suitable for the lipophilic part of the surfactants were condensed with free glucose in the same manner as above in 65-72% yields (Scheme 14). Anomeric 0-alkylation of the unprotected sugar is a potent procedure for producing amphiphilic molecules in short steps and contributes to saving materials and energy.

Scheme 14. Anomeric 0-alkylation of free glucose with cyclic sulfates.

8.2.9 Anomeric 0-Alkylation with Secondary Triflates and Nonaflate Glycosidic linkage to secondary positions [e.g. Man( 1-4)GlcI can be accomplished via anomeric 0-alkylation with secondary triflate. This procedure is accompanied ~ by inversion of the stereochemistry of the electrophile as a result of the S N reaction. The use of promoters (15-crown-5and NaH) in toluene accomplished coupling of tetrabenzylglucose (12) with aliphatic triflates (isopropyl and 1-methylnonyl triflate) in 70% yield ( a :p = 1 :9) and quantitative yield respectively, whereas condensation of 12 and the triflate of 1,6-anhydr0-2,3-di-O-benzylgalactose (52) failed under these conditions. Schmidt’s group has recently employed aprotic dipolar solvent systems; DMF-HMPT and HMPT-THF to solve the problem (Scheme 15) [ 191. These new solvent systems gave the desired glycosides in high yields with a/S

(a),

(52) R = Tf (53) R = Nf

Scheme 15. Anomeric 0-alkylation procedure with secondary electrophiles. P = protecting group; HMPT = hexamethylphosphoric triamide.

8.3 Glycosylation via the Locked Anomeric Conjiguration

189

selectivities of 2 : 1-5 : 1. Exclusive a-selectivity was observed during formation of the sequence Mana( 1-4)Glc from tetrabenzylmannose and triflate 52. The corresponding nonaflate (53) is more effectivethan the triflate for this glycosylation [12].

8.3 Glycosylation via Locked Anomeric Configuration V. K. Srivastava and C. Schuerch have first introduced 1,2-0-dibutylstannylene acetal-mediated methyl and allyl P-glycosylation for mannopyranose in 1979 [20]. This has recently been successfully developed by G. Hodosi and P. Kovai: as a titled concept; Scheme 2 shows the general procedure used. Synthetically difficult 1,2-cis P-glycosides, especially p-mannopyranosides, can be obtained by this method. The possibilities and problems of the locked anomeric configuration method are discussed in this section.

8.3.1 Synthesis of Methyl, Allyl, and Benzyl Glycosides via Stannylene Acetals Stannylene-mediated glycosylation was first applied to alkyl mannosylations. Condensation of the stannylene acetal of 3,4,6-tri-O-benzylmannopyranose (44) with methyl iodide and allyl bromide in DMF afforded the desired methyl and allyl P-mannopyranosides, respectively, in almost quantitative yields [20]. Benzyl Smannopyranoside was obtained from the same stannylene acetal by use of BwNI in benzene [21].The electrophilic leaving group is responsible for the reactivity and stereoselectivity. Methylation with methyl tosylate and dimethyl sulfate gave methyl mannopyranosides as anomeric mixtures at temperatures above 75 "C [20]. Glycosylation of 3,4,6-tri-O-benzylglucopyranose (42) via stannylene acetal with methyl iodide resulted in the production of the 2-O-methyl ether (70%) and amethyl glycoside (30%) [20], presumably as a result of the formation of stannylene acetal on axial 0 - 1 and equatorial 0 - 2 . The equatorial 0 - 2 is more reactive during stannylene complexation, which leads the 2-O-methyl ether. Thus stannylenemediated alkylation of 6-O-tritylmannose and methyl a-mannopyranoside afforded 3-O-alkylated products because of the greater reactivity of the equatorial oxygen at the 3-position [20]. The reaction mechanism is discussed in detail in Section 8.3.4.

8.3.2 Epimerization at C-2 by Locked Anomeric Configuration Method Epimerization at C-2 sometimes occurs during dibutylstannylene complex formation. In particular, 1,2-cis P-per-O-acylation for formation of the stannylene acetals of free mannose, rhamnose, and lyxose is accompanied by epimerization [22]. Scheme 16 shows the epimerization process schematically. This unwanted phenomenon can be minimized by conducting the acetalation under milder conditions-brief reaction in dilute solution at low temperatures [23].

190

8 Glycosylution Methods: Alkylations of Reducing Sugars

OR I

Bu

Scheme 16. Epimerization of mannose by stannylene acetylation. P = protecting group.

We can, on the other hand, take advantage of this epimerization to obtain rare sugars. Hodosi and KovaC converted 6-0-tritylgalactose to talose in 70% yield by stannylene acetalation and C-2 epimerization [241.

8.3.3 Locked Anomeric Configuration Method for Rhamnosyl Stannylene Acetal

Locked anomeric configuration method results in complete p-selectivity in oligosaccharide synthesis [25, 261. The electrophile must be converted to an active triflate and inactive bromide, iodide and mesylate cannot be used. Addition of CsF or Bu4NF into the reaction effectively increases the solubility of the stannylene acetals. Accompanying formation of the formate derived from the primary triflate at room temperature can be suppressed by reducing the reaction temperature to -5°C. Table 1 shows results from coupling the 1,243-stannylene acetal of rhamnose (54) with primary and secondary triflates (57 and 58) in the presence of CsF.

8.3.4 Locked Anomeric Configuration Method for Mannosyl Stannylene Acetal: Isomerization of Acetal [25, 261

Anomeric oxygen on the analogous acetal of mannose is less reactive than that of rhamnose. Coupling of the stannylene acetal of mannose 55 with the primary triflate 57 in DMF at 0-25 "C afforded the desired p-linked disaccharide together with the (3-6) linked pseudo-disaccharide in 40 and 23% yield, respectively. Changing the solvent to CH3CN and addition of BQNF increased the yield and regioselectivity. Because of the low solubility and reactivity of the stannylene acetal in CH3CN, other aprotic polar solvents (e.g. DMF, abbreviated in Table 1. DMA, and DMSO) were employed for the coupling of secondary triflate 58, as shown in

8.3 Glycosylution via the Locked Anomeric Conjiqurution Table 1. Results from glycosylation via the locked anomeric configuration. DMA = N , N-dimethylacetamide. I I

191

I

Manp( 14)Glc

RhafJ( 1-4)Glc CsF I DMF 25 "C, 78 %

B z0

Tf+

Bzo OMe (58)

DMF DMA DMSO

40% 52% 59% 25 "C

Mang( 1-4)Glc CsF I DMF 25 "C, 67 %

Table 1. The secondary triflate did not react with 3-0 of the nucleophile; the anomeric oxygen was more reactive. Isomerization of the stannylene acetal from 1,2-0 to 2,3-0 and the resultant equatorial 3-0 activation rationalize the formation of the pseudo-disaccharide as shown in Scheme 17. The epimerization of the stannylene acetal seems much faster than alkylation. The use of the analogous 3-O-benzyl ether effectively prevented the unfavorable coupling. Thus, generation of the etheric by-product was suppressed in the reaction with the same primary triflate (57).

8.3.5 Locked Anomeric Configuration Method for Stannylene Acetal with the Glucose Configuration 125,261 Scheme 18 shows results obtained from use of the locked anomeric configuration method for coupling of gluco-type stannylene acetals with triflates. Reaction of the stannylene acetal of maltose (59) and the secondary triflate (58) resulted in generation of an ether (2-4) linked pseudo-trisaccharide similar to the result with glucose. It is worthy of note that the coupling of the stannylene acetal of 4,6-O-benzylideneglucose (61) with the primary triflate (57) predominantly afforded the p( 1-6) disaccharide (62). The stereochemistry of the main product is inconceivable, however; it is plausible that the anomer oxygen favored a configuration close to equatorial orientation as the most stable 1,2-0-stannylene complex conformation.

192

8 Glycosylation Methods: Alkylations of Reducing Sugars

Scheme 17. Isomerization between 1,242- and 2,3-O-stannylene acetals of mannose.

HO

H$%o&

1) BuzSnO / MeOH (59) HO

Ho

OH

HO

-$ .*

2) 58, B u g F / DMF OHO + L -

(60) 75 %

BzO

OMe

Scheme 18. Coupling reaction via locked anomeric configuration for maltose and glucose moieties.

8.4 Conclusion The advantages of the ‘anomeric 0-alkylation’ and ‘glycosylation via locked anomeric configuration’ methods are:

1) primarily we can use unprotected sugars as nucleophiles; 2) we can also perform glycosylation reactions under non-acidic conditions; 3) we can also use polar solvents such as DMF, DMA, and DMSO, and even CH3CN; and 4) inversion of configuration occurs at the 0-glycosidically linked carbon of the electrophile. These points reduced the limits of glycosylation reactions and often result in increased coupling yields, compared with conventional glycosylation. With regard to point 4,we can avoid the use of poorly reactive hydroxyl groups, which reduce coupling yields in Koenigs-Knorr-type glycosylation. ‘Anomeric 0-alkylation’ and ‘locked anomeric configuration method’ can be alternatives to traditional KoenigsKnorr-type glycosylation.

References

193

References I . R. R. Schmidt, Angew. Chem. Int. Ed. Engl., 1986,25, 212-235 (197 references). 2. R. R. Schmidt, Pure Appl. Chem., 1989, 61, 1257-1270 (69 references). 3. R. R. Schmidt, Modern Methods in Carbohydrate Synthesis, Hanvood Academic publishers, 1996, pp. 20-54 (57 references). 4. R. R. Schmidt and M. Reichrath, Angew. Chem. Int. Ed. Engl., 1979, 18, 466-467. 5. R. R. Schmidt, M. Reichrath, and U. Moering, Tetrahedron Lett., 1980,21, 3561-3564. 6. R. R. Schmidt, U. Moering, and M. Reichrath, Chem. Ber., 1982, 115, 39-49. 7. R. R. Schmidt, M. Reichrath, and U. Moering, J. Carbohydr. Chem., 1984,3, 67-84. 8. R. R. Schmidt, U. Moering, and M. Reichrath, Tetrahedron Lett., 1980, 21, 3565-3568. 9. W. Klotz and R. R. Schmidt, J. Curbohydr. Chem., 1994, 13, 1093-1101. 10. A. Lubineau, H. Bienayme, and J. Le Gallic, J. Chem. Soc., Chem. Commun., 1989, 19181919. 1 1. A. Lubineau, S. Escher, J. Alais, and D. Bonnaffk, Tetrahedron Lett., 1997, 38, 4087-4090. 12. J. Tamura and R. R. Schmidt, J. Curbohydr. Chem., 1995,14, 895-911. 13. R. R. Schmidt and A. EDwein, Angew. Chem. Int. Ed. Engl., 1988,27, 1178-1180. 14. A. EBwein, H. Rembold, and R. R. Schmidt, Curbohydr. Rex, 1990,200, 287-305. 15. H. Rembold and R. R. Schmidt, Carbohydr. Res., 1993,246, 137-159. 16. R. R. Schmidt and W. Klotz, Synlett, 1991,3, 168-170. 17. W. Klotz and R. R. Schmidt, Liebigs Ann. Chem., 1993, 683-690. 18. W. Klotz and R. R. Schmidt, Synthesis, 1996, 687-689. 19. Y. E. Tsvetkov, W. Klotz, and R. R. Schmidt, Liebigs Ann. Chem., 1992, 371-375. 20. V. K. Srivastava and C. Schuerch, Tetrahedron Lett., 1979,35, 3269-3272. 21. A. Dessinges, A. Olesker, and G. Lukacs, Carbohydr. Res., 1984, 126, c6-c8. 22. G. Hodosi and P. KovaE, XVIII Int. Carbohydr. Symposium Abstr., Milano, 1996, pp. 394. 23. G. Hodosi and P. KovaE, Carbohydr. Rex, 1997,303, 239-243. 24. G. Hodosi and P. KovaE, J. Carbohydr. Chem., 1998,17, 557-565. 25. G. HodosiandP. KovaE, J. Am. Chem. SOC.,1997,119,2335-2336. 26. G. Hodosi and P. KovaE, Curbohydr. Res., 1998, 308, 63-75.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

9 Other Methods of Glycosylation Luigi Panza and Luigi Lay

Introduction and Summary The biological relevance of oligosaccharides in molecular recognition phenomena stimulated the development of many synthetic approaches for the formation of glycosidic linkage, which is the main problem in oligosaccharides chemistry. Besides the most popular methods, which will be discussed in other chapters of the book, many other protocols have been explored.

X = "exotic"leaving group

New methods should, in general, have advantages towards the existing protocols (e.g. to be more efficient, simpler, to use cheaper reagents, to allow better selectivity, to be environmental friendly etc). This chapter will also deal with common methods not covered by other chapters, but more attention will be given to those approaches which fulfil, at least in part, the above advantages. Highlights

1. Use of a protecting group at the anomeric position easily transformed into a good leaving group.

196

9 Other Methods of Glycosylution

Example: an anomeric 2-butenyl glycoside can be isomerized into a propenyl ether which has been demonstrated to be an efficient leaving group. Introduction of the concept of latent-active glycosyl donor

OBn 78%, a/p 1:20 TMSOTf CH$N

BnO

G . J. Boons, S. Isles, Tetrahedron Lett., 1994,35, 3593.

BnO

Scheme B.

Advantages: a protecting group can be easily converted into an efficient leaving group. Disadvantages: use of mercury salts for the preparation of 2-butenyl glycosides.

2. Use of an heterogeneous acid catalyst. Example: an heteropoly acid act both as promoter and dehydrating agent in the glycosylation of an anomeric hydroxy sugar donor

Go&$ 82%, pia 1090

BnO BnO

N3

K. Toshima, H. Nagai, S. Matsunnua, Synlett, 1999, 1420

Scheme C.

Advantages: the procedure is simple, the promoter is reusable and environmental friendly. Disadvantages: the method seems to work well only with reactive donors; only a few examples available so far.

3. New achievement with old methods. Example: use of ortho esters as intermediates in regio- and stereoselective glycosylations.

9.1 En01 Ethers

197

TMSOTf CH2C12

W. Wang, F. Kong, J. 0%.Chem., 1998,63,5744.

Scheme D.

Advantages: good regioselectivity with partially protected acceptors, multiple glycosidic bond formation. Disadvantages: multistep procedure; requires AgOTf for the ortho esters formation.

N. B. Because of the vast scope of this chapter, it is almost impossible to include all the pertinent literature. The authors have made an attempt to organise in a logical manner most of the multifarious approaches for the formation of the glycosidic linkage not included in other chapters. They apologize for any omission they can have made unintentionally.

9.1 Enol Ethers This section will examine the preparation of glycosides and oligosaccharides starting from sugar derivatives which contain a double bond substituted with an ethereal or an acetalic oxygen atom. The term enol ether is therefore used in this context in a quite broad sense. The following general scheme summarizes the approaches which have been explored for synthesis of glycosides from enol ethers. The different approaches can be classified into two main groups: a) strategies in which the starting compound is an hydroxy enol ether already containing the aglycone moiety and from which the glycoside is obtained by cyclization with an electrophile (from endo-en01 ethers, retrosynthetic pathway d and exo-enol ethers retrosynthetic pathway e, Scheme 1); b) approaches in which a properly activated donor, which can be derived from or still contain an en01 ether function, reacts with a glycosyl acceptor (a sugar alcohol) in a more classical way (from exo-glycals, retrosynthetic pathway a, endo-glycals, retrosynthetic pathway b and vinyl glycosides, retrosynthetic pathway c, Scheme 1).

198

9 Other Methods of Glycosylation

&'

from exoenol ethers

I

H'R' endo-enol ethers

Y

PO

S

O

R

Scheme 1. Retrosynthetic pathways for the synthesis of glycosides from enol ethers.

Furthermore, in some of these approaches the diverse structure of the starting compound may be used to obtain different kinds of glycosides, namely aldosides or ketosides (endo-en01 ethers and endo-glycals), whereas other starting structures allow only the formation of ketosides (exo-enol ethers and exo-glycals). The following examples illustrate some applications of the different approaches shown above.

9.1.1 Endo-Enol Ethers The first example of the use of sugar hydroxy enol ethers was originally proposed by Suzuki and Mukaiyama [ 11 in 1982 for the synthesis of 2-deoxyglycopyranosides (Scheme 2). The enol ether precursors 3 were obtained by a Horner-Wittig reaction 1 with an alkoxymethyldiphenylphosphine of 2,3,5-tri-O-benzyl-~-arabinofuranose oxide 2, as an E/Z-mixture separated by careful chromatography. Treatment of each diastereoisomer with various electrophiles (Hg(I1) salts, PhSeX, NIS) induced a 6-endo cyclization. The substituent at position 2 was then reductively removed to give the desired 2-deoxydisaccharides 4 and 5. It is noteworthy that, in such an approach, the factors which control the stereochemistry of the process are completely different from those of a classical glycosylation and will be discussed later. In this case the stereochemistry at the anomeric position was dependent on the stereochemistry of the double bond and on the electrophile used.

Scheme 2. Reagents: i) LDA, THF, -78 to 0°C; ii) KH, THF, 40"C, 75% Z/E 2: 1; iii) Hg(I1) salts, PhSeCl or NIS; iv) NaBH4, Bu3SnH or LiAlH4 respectively.

9.1 En01 Ethers

199

BnO ?Me COOMe

BnO v

f- l o R =

CHO

i,ii BnO

COOMe

9

yo OMe 6

I

i,iii BnO

BnO

BnO

16

BnO

COOMe

I

14

BnO

COOMe

17 R = HgCl 19R=H

IZOR=H

Scheme 3. Reagents: i) THF, NaH, O T , 90%, Z/E 2: 1 from 7, 85%, Z/E 2 : 3 from 14; ii) CF3COOH/H20 9 : 1, 0 "C, 80%; iii) MeONa in MeOH, 95%; iv) Hg(OCOCF3)2,THF, 0 "C then KCI, 80%; v) Ph3SnH, anhydrous AcONa, toluene, 90%.

Sinay and Paquet cleverly applied this approach to the synthesis of sialic acid [2] and Kdo [3] glycosides (Scheme 3). The enol ether precursors were prepared in good yield by a Horner-Emmons reaction between the properly protected aldehydo sugar derivative 7 or 14 and the sugar containing alkoxyphosphono acetate 8, and the E/ Z mixtures obtained were easily separated by chromatography. With respect to the previous procedure, the double bond in the starting enol ether was trisubstituted. In the case of sialic acid disaccharides, the mercuriocyclization of the E diastereisomer

200

9 Other Methods of Glycosylation

9 gave the a-linked mercurioderivative 11 with complete stereo- and regioselectivity. Likewise, cyclization of the Z isomer 8 afforded exclusively the p-linked mercurioderivative 10. The remarkable stereoselectivity of the reaction has been rationalized in terms of the attack of the electrophile from the less hindered side of the double bond controlled by the adjacent stereocenter and by the chair conformation in the cyclization transition state. Treatment of the mercurioderivatives with triphenyltin hydride and sodium acetate gave the p- and a-sialic acid disaccharides 12 and 13. When the procedure was applied to the Kdo enol ether precursors 15 and 16, the cyclization was again highly stereoselective affording, after demercuriation, the a-linked Kdo disaccharide 19 from the Z isomer and the p-linked Kdo disaccharide 20 from the E isomer. The reaction was also applied to a phosphonoacetate linked to the 3 position of a properly protected N-acetylgalactosamine. In this case the Horner-Emmons reaction was completely stereoselective giving exclusively the E isomer, from which the P-linked Kdo disaccharide was easily obtained in good yield. The above procedures allow access to 2-deoxyglycosidesand oligosaccharides. In an analogous way such an approach could be used for the synthesis of 2-hydroxy glycosides provided that the enol ether is allowed to react with an electrophilic oxygen. Russo et al. developed this idea [4]through the epoxidation of hydroxy enol ethers followed by ring closure of the intermediate epoxides to give the desired 2-hydroxy glycosides. It is worthy of mention that these compounds, having a single free hydroxyl group, allow further manipulation at position 2. The starting enol ethers E- and 2-3were prepared according to Mukaiyama [ 11, through a Horner-Wittig reaction of alkoxymethylphosphine oxides with 2,3,5-tri0-benzyl-D-arabinose (Scheme 4) and ribose. After separation of the E/Z mixture, epoxidation with MCPBA of the of the hydroxy enol ethers, either in Z or E configuration of the double bond, gave directly the cyclized products 21-24 owing to the high reactivity of the intermediate enol ether epoxides, with a 6-endo-tet ring closure. As mentioned above, the stereochemical course of this process is controlled by factors which are completely different from those of a “classical” glycosylation reaction. In this case two new stereocenters are introduced, at the anomeric position and at C-2, and the stereochemical outcome of the process depends on the config-

OR

BnO

-

Bno&OR BnO

E-3

BBnOn

O 2 3

+

21 OH

e OR

B,,O

OR 23

Scheme 4. Reagents: i) MCPBA, CH2C12, Na2HP04.

+

BBnOn

I

22

OR

O

3

24 HoOR

9.1 En01 Ethers

201

uration of the double bond, on the stereoselectivity of the epoxidation reaction, and on the stereospecificity of the ring closure. While the epoxidation was not highly stereoselective, the ring closure was stereospecific in all cases studied except for the formation of mannosides with oxygenated aglycons. The same procedure was also applied to the synthesis of ketofuranosides [5] such as fructosides and psicosides, where the starting enol ethers contained an extra hydroxymethyl group linked to the double bond. This structural feature allowed the use, besides the MCPBA epoxidation, of the Sharpless epoxidation. Although the epoxidation reaction gave, in general, better results than in the former case, it is noteworthy that the stereoselectivity of the Sharpless epoxidation was somewhat unpredictable. This is probably due to the presence of a second hydroxyl group which can coordinate the titanium and, in some cases, counteract the normal course of the Sharpless epoxidation. 9.1.2 Exo-Enol Ethers

The synthesis of glycosides from exo-enol ethers through a iodoetherification reaction was initially proposed by Barrett [6].It is obvious that from this method only glycosides of ketosugars can be obtained. In this approach a hydroxy group of a sugar either at the anomeric carbon or at another position is initially converted into an ester with a series of carboxylic acids containing a silyloxy function. Different procedures were used for the esterification step at the anomeric position in order to obtain a- or p-anomeric esters with good stereoselectivity. The esters were methylenated using the Tebbe reagent, desilylated and iodocyclized to give reducing and nonreducing disaccharides in fair to good yield. It is noteworthy that, in the case of iodocyclization of anomeric vinyl ethers, the anomeric C-0 bond is not cleaved at the intermediate iodonium ion stage. The stereochemical outcome of the cyclization reaction is strongly dependent on the structure of the vinyl ether, and the stereoselectivity varies from modest to excellent. Barrett applied this procedure to the synthesis of sucrose (Scheme 5) [7]. 2,3,4,6Tetra- 0-benzyl-D-glucopyranose was converted into the a-glucopyranosyl ester 25 with excellent selectivity. In this case the methylenation using Tebbe reagent was unsuccessful. However, the Nozaki-Takai protocol gave the desired vinyl ether. After desilylation, the cyclization reaction gave only the (3-fructofuranosyl derivative 26 when the reaction was performed in the presence of silica. The sN2 displacement of the iodide proved to be difficult, as expected for a neopentylic position. The problem was solved using a radical-mediated substitution with TEMPO, and final deprotection afforded sucrose 29. A similar approach was used by Sinay (Scheme 6) for a new synthesis of Kdo glycosides and disaccharides [8]. The mannose derived hydroxyheptenitol 30 was converted into the exo-enol ether 31 through a multistep synthesis. Cyclization was achieved using potassium t-butoxide and iodine at low temperature to give the disaccharidic derivative 32 as the only product of the reaction. The transformation of the iodomethyl appendage into the carboxylic group required a difficult nucleo-

202

BnO B n

9 Other Methods of Glycosylation

OBn QSiPh;Bu

Bn007/yvoB -

O

25

q

0

...

i,u

OBn

111

26

Bnd

27

: \OBn

28

Bn6

Acb

bAc

OBn

HV

OH

Scheme 5. Reagents: i) TiC14, TMEDA, Zn, CH2Br2, THF (Nozaki-Takai protocol), 68%; ii) TBAF on silica, THF, then tBuOK, 12, THF, 25 "C, 65%; iii) TEMPO, Bu3SnH, benzene, hv, 80%; iv) Na, NH3, THF then Ac20, pyridine, 90%; v) MeONa, MeOH, 96%.

philic displacement with cesium acetate in HMPA at 140 "C followed by a standard sequence of deprotection, oxidation and esterification with diazomethane to afford the final disaccharide 33. 9.1.3 Endo-Glycals Glycals have been extensively used for the synthesis of 2-deoxyglycosides and oligosaccharides since Lemieux introduced their reaction with simple alcohols in the presence of iodine and silver salts [9]. Interesting results were obtained using elec-

COOMe

30

BnO BnO 31

BnoOMe

Scheme 6. Reagents: i) PivC1, pyridine, DMAP, 98%; ii) 9-BBN, THF, then H202, NaOH, 71%; iii) RuC13/NaI04 CC14, CH3CN, H20, 92%; iv) Methyl 2,3,4-tri-O-benzyl-a-~-glucopyranoside, DCC, DMAP, 78%; v) Tebbe reagent, 81%; vi) LiAlH4, THF, O"C, 93%; vii) tBuOK, then 12, THF, 92%; viii) CsOAc, HMPA, 130 "C, 68%; ix) MeONa in MeOH, rt, 98%; x) (COC1)2/DMSO then Et3N; NaC102, H202, CHjCN; xi) CHzN2 Et20, 68%.

9.1 En01 Ethers

BnO BnO

203

Bz0 34

35 31 39 R=H

Scheme 7. Reagents: i) IDCP, CH2C12,4A ms, 58% of 36, 79% of 38; ii) Ph-iSnH, AIBN, benzene, 94%.

trophiles such as IDCP, again described by Lemieux, and more recently exploited by Danishefsky [lo], who applied the concept of armed-disarmed glycosylation (Scheme 7). In this manner it was possible to use the benzyl protected glycal34 as a donor, and the benzoyl protected glycal35 as an acceptor, in the initial glycosylation reaction. The product still contains a glycal function which can be submitted for a second glycosylation reaction to finally give compound 39. Other groups have proposed the use of NIS and NBS as electrophiles. For example, Tatsuta introduced [ l l ] and applied the NBS mediated glycosylation to the synthesis of natural 2deoxyglycoside containing compounds [ 121. Thiem [ 131 developed the NIS-mediated glycosylation, which was applied by many groups to the synthesis of various natural products containing oligosaccharidic chains [ 141. Thiem has improved the method for the synthesis of disaccharide side chains of antracyclines by the transformation of the alcohol acceptor into the corresponding tin-alkoxide to enhance its reactivity, as in the glycosylation of 41 (Scheme 8) 1151. All these 2-deoxy-2-haloderivatives can be reductively dehalogenated to give the desired 2-deoxyglycosides. Sinay introduced the use of phenylselenyl chloride as an electrophile [ 161, which was exploited by Barrett in the avermectin a-disaccharide synthesis [ 171. Ogawa

HOIt..

/ 0

i ____)

40

42 R=I

41 43 R=H

AcO

Scheme 8. Reagents: i) (Bu3Sn)z0,benzene then NIS, CH3CN 65%; ii) BqSnH, AIBN, 76%.

204

9 Other Methods o j Glycosylation (OBn

BnO& BllO

t

34

ratio

Scheme 9. Reagents: i) (PhS)2, PhSC1, SbC15, CH2Cl2 -6O"C, 70°h

described [ 181 the addition of phenylsulfenate esters to glycals in the presence of TMSOTf while Franck performed the reaction using phenylbis (pheny1thio)sulfonium salts, in which the alcohol acceptor was again presented as the stannyl ether (see e.g. the glycosylation of 34 with 44, Scheme 9) [19]. The addition product was reduced to the 2-deoxyglycosides by using either Raney-Ni or a tin hydride. In these glycosylation reactions the use of NIS, NBS or phenylselenyl chloride afforded mainly the a-glycosides, whereas other electrophiles gave the pglycosides as the main products with moderate selectivity. Franck studied quite extensively the face-selectivity of his reaction and found that it depends on the arylthio substituents, on the nucleophile and on the glycal structure. Such an approach was also applied to the synthesis of glycofuranosides starting from furanoid glycals. Some examples of activation of glycals using protons as electrophiles are also described. PTSA [20], CSA [21], triphenylphosphine hydrobromide [22], and AG50 WX2 resin [23] were used as promoters allowing the direct formation of 2-deoxyglycosides with predominant a-anomeric configuration. For example, Tatsuta effected the glycosylation of a glycal in the presence of CSA for the synthesis of the antibiotic elaiophylin [21]. The main drawback of this procedure is the possible occurrence of a competitive Ferrier rearrangement. Glycals were also transformed into the corresponding epoxides and used as glycosy1 donors, as will be described in another chapter. 9.1.4 Exo-Glycals

The NIS-mediated glycosylation introduced by Thiem, was applied not only to endo-glycals, but also to exo-glycals for the preparation of glycosides of ketosugars. It was initially studied [24] on the exo-glycal of leucrose 48 (Scheme 10) which was obtained in a classical way by reductive elimination on the glycosyl bromide 47 with zinc dust. Treatment of 48 with NIS in the presence of the sugar acceptor 49 gave the trisaccharide 50 in about 50% yield as a single anomer. Similarly, van Boom obtained [25] a iodomethyl disaccharide by treatment of the exo-glycal derived from tetra- 0-benzybglucopyranose with IDCP in the presence of a protected mannose acceptor.

205

9.1 End Ethers

i

OBz 41

BzO

OAc

Br 48

49

50

OAc

Scheme 10. Reagents: i) Zn, CH3COONa, CH3COOH/H20 1 : I, -10"C, 65%; ii) NIS; 3 A ms, CH3CN, rt, 47%.

Attempts to apply an analogous approach to the synthesis of Kdo disaccharides were made by Sinay by reaction of an exo-glycal obtained from 2,3,5,6-di-O-isopropylidene-D-mannose with iodine and potassium t-butoxide, but after a positive result with methanol, sugar acceptors did not give the expected disaccharides [8]. Russo [26] developed the use of exo-glycals such as 51 and 54 for the synthesis of glycosides of ketosugars by epoxidation with dimethyldioxirane followed by a Lewis acid-mediated glycosylation (Scheme 11). In this way glycosides and disaccharides of fructose 56 and gluco-heptulose 53 were obtained. The stereoselectivity of the reaction was dependent on the diastereomeric ratio of the epoxides in the case of simple alcohol acceptors, while less reactive sugar alcohols gave moderate selectivity for fructose glycosides and good to excellent a-selectivity in the case of pyranosyl derivatives. The method was further developed for the synthesis of Kdo disaccharides (Scheme 12) [27]. The exo-glycal57, previously synthesized by Sinay [8], was epoxidized with dimethyldioxirane and glycosylated with different alcohol and sugar acceptors to give the hydroxymethyl derivatives such as 59. The advantage of this approach with respect to the NIS mediated glycosylation is the direct introduction of the oxygenated function at C-1 of the Kdo moiety thus avoiding the cumbersome substi-

OBn

9

ABn

55 a/!3 11:

Scheme 11. Reagents: i) DMD, CH2C12, 0°C; ii) ZnCl2 'OEt2, CH2C12, -78"C, (55-90%)

206

xF--

9 Other Methods of Glycosylation

OH

11

*

5 8 "

57

Scheme 12. Reagents: i) DMD, CH2C12, -78 "C; ii) ZnCl2 . OEt,, CHzC12, -78 "C, 82%).

tution of the iodide. The reaction occurs with excellent a-selectivity and its stereochemical outcome has been rationalized by means of semiempirical calculations. 9.1.5 Vinyl Glycosides This method may be seen as a more classical approach as vinyl ethers can act as leaving groups in glycosylation reactions (Scheme 13a). The method was again introduced by Sinay [28] and derived from the observation that mercury(I1) chloride induces cyclization of 1-propenyl glycosides of 2-acetamido sugars to give oxazoline derivatives. Some initial positive results using 1-propenyl glycosides promoted the

a)

BBnO n 0

G o BnO

T

+

BBnO n 0

i

BnoOCH3

60

b,

Ho% BnO

BBnO n O BnO G o Brio $

62

61

G

o

H

-

+

N3

63

BBnO

n N3

65

64

OBn

Brioo

Pia 5:1

BnoOCH3

~ p/a 6.611 o BnoOCH3

OBn

BnO

66

61

Scheme 13. Reagents: i) BF3. OEt,, CH3CN, -25°C; ii) TMSOTf, CH3CN, -25"C, 75% of 76 and 92% of 79.

~

9.1 En01 Ethers

207

development of isopropenyl glycosides as glycosyl donors. Thus, treatment of the anomeric acetates of tetra-0-benzylglucopyranose with the Tebbe reagent gave the desired isopropenyl glucoside 60. The donor was activated both with TMSOTf and BF3 . OEt2. The glycosylation reactions were performed both in acetonitrile and in dichloromethane. The first solvent gave the better yield and p-selectivity, as expected, owing to the formation of transient a-nitrilium species. The rationalization of the process in terms of an intermediate mixed acetal formation led to the hypothesis that the glycosylation should also occur by reverting the reacting partners, namely using a vinyl derivative of the acceptor with a 1-hydroxy sugar. This reversed approach was successfully applied to the glycosylation of 2-azido derivative 63, for which the isopropenyl glycoside itself is not available as the Tebbe reagent is not compatible with the azido group (Scheme 13b). Finally the same paper describes the use, as donors, of isopropenyl glycosyl carbonates, easily obtained from hemiacetal derivatives with the commercially available isopropenyl chloroformate (Scheme 13c). Strictly speaking these donors are anomeric carbonates but are included here as their efficiency as glycosyl donors derives from the presence of the double bond. The glycosylation reaction of donors such as 66 with various acceptors occurred smoothly in the presence of TMSOTf to give disaccharides in good to excellent yield. Isopropenyl glycosides were also studied by Chenault [29] with particular attention to mechanistic aspects (Scheme 14). The stereoselective synthesis of pivaloyl protected isopropenyl glycosides was achieved either using Tebbe reagent or by reaction of glycosyl halides with bis(acetony1)mercury. The glycosylation reaction (e.g. of 69 with 37) was promoted with various electrophiles the best results being obtained with Tf20, TMSOTf and NIS/TfOH. The use of IDCP gave electrophilic addition to the double bond with subsequent formation of acetals such as 71. Chenault found that there are interesting influences of the reaction conditions on the reactivity of the isopropenyl glycosides towards transglycosylation or electrophilic addition on the double bond. (OPiv

(OPiv

69

Scheme 14. Reagents:

NIS, TfOH, CH3CN, 70'31, ii) IDCP, CH3CN or CH2C12, 53%

208

9 Other Methods of Glycosylation

AcO

Br 12

-Ai:030n i

BnO BnO+

AcO

iv

13

74

Bno$&oF+ OBn

B;:oGonv

BnO active

BnO

6'

latent

BnO

BnO

l5

op-

latent

& $-on.

alp 1:20

BnO BnO BnO 71

Scheme 15. Reagents: i) 3-Buten-2-01, Hg(CN)Z,HgBrz, CH3CN, 70%; ii) MeONa in MeOH, 90%, iii) NaH, BnBr, DMF, 820/0; iv) (Ph3P)3RhCl/nBuLi,THF, 92%; v) TMSOTf, CH3CN, 78%.

Boons developed the use of vinyl glycosides as donors by modifying the leaving group and exploiting the concept of latent-active glycosylation strategy (Scheme 15) [30]. In his approach a sugar was protected at the anomeric position as 3-buten-2-yl glycoside (72+ 74). The isomerization of the double bond with Wilkinson's catalyst in the presence of a base afforded 2-butene-2-yl glycosides such as 75, which can undergo a Lewis acid catalyzed glycosylation reaction. The activation of the butenyl group with TMSOTf promoted the glycosylation reaction giving disaccharides in good yield. Such interesting procedures exploit the easy transformation of a protecting group at the anomeric position into an efficient leaving group. The use of an allyl glycoside (e.g. 76) as acceptor allows the reiteration of the procedure. Moreover, the stability of the allyl glycoside allows quite a simple and flexible protecting group strategy. This latent-active glycosylation strategy was also applied by Boons to the synthesis of libraries of both linear [31] and branched trisaccharides [32]. The main drawback of the procedure is the preparation of the starting glycosides [ 301: These are obtained from acetobromosugars and 3-buten-2-01 using HgBr2/ HgCN2 as promoters, a procedure quite laborious and problematic from a health and environmental point of view. Boons recently tried to improve the preparation of substituted allyl glycosides by enzymatic means, using almond 0-glycosidase [33]. Another type of vinyl glycoside was used by Schmidt for glycosylation reactions [34].The glycosyl donor was obtained by treatment of a sugar hemiacetal with ethyl phenyl propiolate and sodium hydride. Glycosylation reactions were performed using TMSOTf as a promoter and demonstrated that such vinyl ethers at the anomeric position are not very good leaving groups. Surprisingly, Takeda et al. proposed very recently the same method [35] without mentioning Schmidt's paper. The only difference is the procedure used for the preparation of the vinyl glycosides which are obtained by addition of tetra-0-benzylD-glucopyranose to some alkynyl esters and ketones in the presence of tributylphosphine. The glycosylation reaction is carried out in the presence of molecular sieves and requires 10 equivalents of TMSOTf, thus confirming the relatively low reactivity of such donors.

9.2 I-Hydroxy Sugars

...

111

209

B n O & o q BnO

Me0

MeoOCH3

81

82

Scheme 16. Reagents: i) Nysted reagent, THF, CH2C12, 75%); ii) TfOH, CHZC12, -2O”C, 67%; iii) Raney-Ni, yield very much batch-dependent (50-90%).

Finally, some examples of the use of bicyclic systems in which the anomeric vinyl ether is partly or completely contained in a ring are present in the literature [36]. Franck described the use of cycloaddition products obtained from glycals as donors (e.g. 78, Scheme 16). Compound 78 exhibited low reactivity, so it was modified by introduction of a second double bond using the Nysted reagent. The vinyl glycoside 79 so obtained revealed excellent glycosyl donor properties when activated with Lewis or protic acids, the best results being obtained with TfOH, to give fair to good yields of P-linked products. Raney-Ni desulfurization afforded the 2-deoxy-Pglycosides. The difference in reactivity between 78 and 79 introduces the possibility of using these compounds as latent-active donors [ 3 7 ] . Hetch et al. used cyclic ketene acetals for glycosylation reactions [38]. The donor is obtained from a glycosyl bromide by treatment with silver oxide and Hunig’s base and is activated with CSA in the presence of the acceptor to give an orthoester. TMSOTf catalyzed rearrangement of such an ortoesther gave the desired disaccharide.

9.2 l-Hydroxy Sugars The use of l-hydroxy sugars for the formation of glycosidic bonds would be one of the most fascinating glycosylation methods, owing to the simplicity and the stability of the “donor”. Of course, the OH group itself is not a good leaving group and needs to be activated. The simplest way for the activation of the OH group is the use of an acidic promoter in order to have a water molecule as leaving group. In this case, better results are obtained in the presence of a water scavenger. Other methods involve the transformation of the anomeric hydroxy group into an intermediate glycosyl donor, which is used without isolation in glycosylation reactions.

210

9 Other Methods of Glycosylation

\ OTBDPS

+ +

i

Ho H AllO O%

OTBDPS

OH 83

,O 84

AllO H O &

*'

Scheme 17. Reagents: i) pyridinium p-toluenesulfonate, CH2C12, azeotropic dehydration, 73%.

Finally, the activation at the anomeric position through the Mitsunobu reaction will be considered. Besides the 1-hydroxy sugars, the l-O-silylated glycosides used as glycosyl donors will also be included in this section. The classification into the following sub-sections is in some senses questionable, but may help the reader to find a selected topic more easily. 9.2.1 Acidic Activation Apart from the obvious problems given by the water formed during the reaction, the direct transformation of l-hydroxy sugars into glycosides suffers of two other main problems: Their poor reactivity as donors and the strong tendency to selfcoupling to give trehalose like structures. Such problems are easily overcome in the formation of simple glycosides when the alcohol acceptors can be used in large excess (i.e. they are sufficiently volatile and cheap). After Fischer proposed the acid catalyzed formation of simple glycosides [39], some other methods appeared in literature [40]. Wessel [41] used the triflic acid for the formation of simple glycosides using the alcohol as solvent. The method was improved by Fraser-Reid [42] for the preparation of pentenyl glycosides, easily obtained using DMSO as solvent, TfOH and only 3 equivalents of alcohol. Glycosylation with sugar acceptors was performed on the Kdo precursor 83 as donor (Scheme 17), using pyridinium p-toluenesulfonate as acid catalyst and removing water either with 4 A ms or by azeotropic dehydration [43]. Glucuronic acid glycosides have been obtained by treatment of the methyl 2,3,4tri-O-acetyl-D-glucuronate with alcohols and phenols in the presence of TMSOTf [44]. According to the reactivity of the alcohol, the reaction is believed to proceed via a SN2 like reaction on a silylated hydroxy group to give a-glycosides, or by formation of an oxonium ion to produce the P-anomers owing to the participation of the acetate. The reactivity of N-acetyl-D-glucosamine and galactosamine in hydrogen fluoride solutions was also studied [45]. More recently Mukaiyama developed a method for the glycosylation of 2,3,5-triO-benzyl-D-ribose using a catalytic amount of trityl salts as promoters, in the pres-

9.2 I-Hydroxy Sugars

BnO&o, BnO

f

B

$

o

G

Brio OMe 87

86

2 11

pla ,6,84

BnBnO O&%O+ Brio OMe

88

Scheme 18. Reagents: i) 5 mol%,TrB(CsF5)4, benzene/toluene 10: 1, Drierite, O T , 83%.

ence of Drierite as a dehydrating agent [46]. Without any additive the p-anomer was formed preferentially, while the presence of lithium salts reversed the stereoselectivity leading to a large predominance of the a-anomer. The same procedure was also applied to the synthesis of 2-deoxy-~-glucopyranosides88 (Scheme 18)

PI.

A very interesting possibility has arisen in recent years, namely the use of solid acidic catalysts both as a promoter and as a dehydrating agent. Such an approach, when applicable, has many advantages: The procedure is simple and environmental friendly, the catalysts are usually quite cheap and reusable. The main limitation to be noted from literature data, is that these catalysts are efficient in glycosylation of donors which form the oxonium ion quite easily, such as deoxysugars. Montmorillonite K-10 was used for glycosylation reactions with simple alcohols and sugar acceptors [48]. In particular, olivosides were obtained using this Lewis acid solid catalyst. Another recent example of dehydrative glycosylation of 1-hydroxy sugars was performed [49] using an heteropoly acid, H4SiW12040 (Scheme 19). Different glycosyl donors and acceptors were used, with the following results: The glycosylation of deoxy sugars requires small amounts of promoter, whereas a larger amount is needed for benzylated hexoses; the formation of simple glycosides is described for all the donors, while the preparation of disaccharides, such as 90, is performed only on deoxy sugars; the glycosylation reaction is usually quite stereoselective giving mainly the a-anomer. Although at an initial stage, such promoters deserve attention because they can also be used for the activation of less reactive glycosyl donors. Acidic Activation With Additional Reagents

Several other attempts to activate 1-hydroxy sugars with cationic activators or Lewis acid catalysts have appeared in literature, in which a second active species is generated in situ by additional reagents. Inanaga et al. used catalytic amounts of Yb(OTf)3 and methoxyacetic acid as promoters to obtain glycosides of perbenzylated glucose 91 and ribose in excellent

&

BnO BnO

OH

86

+

8'

N3

HO

89 N3

p/a 10:90

90

Scheme 19. Reagents: i) heteropoly acid (H4SiW12040) 10 Wt%, CH3CN, 2 5 T , 82%.

212

9 Other Methods of Glycosylation

BnO%oH BnO

BBnOn OBn

91

O

q

-

67 Brio OMe

i

BBnO n o &OBn o ~ o M ~

2515

68

Scheme 20. Reagents: i) Yb(OTf)s, CH30CH2COOH 10 mol% each, CH2C12, ms 4A, reflux, 99%.

yield (Scheme 20) [50]. In the same conditions it is also possible to obtain thioglycosides. Although the role of methoxyacetic acid is not clear, no reaction took place when it was absent. The use of trimethylsilyl derivatives in the presence of various Lewis acids is also described. Susaki has published [51] a glycosylation reaction promoted with TMSCl and Zn(0Tf)Z. In a similar way, Mukaiyama described [ 521 successful catalytic glycosylations using (TMS)20 in combination with four different Lewis acids, again studying the effect of LiC104 on the stereoselectivity of the reaction. In a further development the same group studied the use of Sn(0Tf)z with various trimethylsilyl derivatives [ 531. 9.2.2 Dehydrative Glycosylation

Starting from observations from different fields of organic synthesis, Mukaiyama introduced the use of dehydrating reagents for the activation of the anomeric position of 1-hydroxy sugars. In a first example [54]a diphosphonium salt, generated by reaction of a trialkyl- or a triarylphosphine oxide and triflic anhydride allowed the synthesis of 1,2-cis-ribofuranosides when used in combination with a base (Hunig's base),1!4 ms and an appropriate acceptor. Similar results were obtained when the phosphine oxide was substituted with an oxotitanium derivative [ 551 in conditions similar to those previously described. A third paper [56] introduced the use of a different metal derivative in order to improve the already good yields and selectivities on the glycosylation of 2,3,5-tri-O-benzylribose92. Excellent results were obtained with diphenyltin sulfide and triflic anhydride in the presence of LiC104, with CsF as an acid scavenger, using alcohols or trimethylsilyl ethers as acceptors (Scheme 21). It is worthy of note that most of the tin reagent is recovered at the end of the reaction.

Scheme 21. Reagents: i) diphenyltin sulfide, Tf20, CsF, CH2C12, 0°C; ii) -23"C, 99%.

1% 1

6-

RO RO

OH

1

OR 91R=Bn 94R=Bz

- OTf

RO

OR

9.2 I-Hydroxy Sugars

RROO

&

o GBnO OR

213

Brio OMe

96a R = Bn, 85%, alp 6040 %b R = Bz, 72%, p only

Scheme 22. Reagents: i) diphenylsulfoxide, Tf20, 2-chloropyridine, toluene/CHzClz 3 : 1, -78123 "C.

More recently, Gin et a]. developed [57] the glycosylation with 1-hydroxy sugars using diphenyl sulfoxide and triflic anhydride to give glycosides and disaccharides in good yields. Benzyl or benzoyl protected glycosyl donors (91 and 94, Scheme 22) reacted equally well, and the procedure even allowed the N-glycosylation of an amide. In the Presence of the Acceptor From the Beginning

In spite of their efficiency, the above procedures involve the prior activation of the anomeric position of the donor, followed by the addition of the acceptor. The selective activation of the donor in the presence of the acceptor would allow the presence of both reagents from the beginning of the reaction. Koto et al. introduced [ 5 8 ] a glycosylation reaction in which a mixture of both the donor and the acceptor, together with CoBr2, is treated with methanesulfonic acid to give the glycoside or disaccharide through the in situ formation of a glycosyl bromide. The procedure was further modified, in order to obtain a better control of the stereoselectivity, with Et4NC104 [59] or using TMSBr instead of methanesulfonic acid [60]. The same group also introduced the activation of the anomeric position of 1-hydroxy sugars with p-nitrobenzenesulfonyl chloride, silver triflate and Et3N, probably through the formation of an anomeric sulfonate intermediate [61]. From this procedure glycosides and disaccharides were obtained with predominant pconfiguration. The addition of N,N-dimethylacetamide reverted stereoselection to predominant formation of the a-anomer of 68 (Scheme 23) [62]. A simplified procedure, described by Szeja [63], involves in situ formation of glycosyl sulfonates under phase-transfer conditions followed by alcoholysis to give mainly a-glycosides. Recently Koto et al. proposed [64] another dehydrative glycosylation in which DAST is used to activate the donor in the presence of the acceptor, Sn(OTf)Z,

BBnOn O g o H OBn 91

+

BBnOn 61

O q Brio OMe

alp 13:2?

Scheme 23. Reagents: i) NsCl, AgOTf, dimethylacetamide, Et3N, CHzCIz, -40 "C, 88%.

214

9 Other Methods of Glycosylation

B n o S O H BnO

1

95

"Me,

B n*;BnO C -) Brio OMe ma

alp 82:18

Scheme 24. Reagents: i) (CCl3CO)20, TMSC104, ms 5A,EtzO, rt, 91%.

Bu4NC104, and Et3N. The reaction probably proceeds through the formation of a glycosyl fluoride, which is further converted into an anomeric triflate and eventually to the glycoside. Finally, an even more recent paper [65] described a novel dehydrative glycosidation using trifluoro- or trichloroacetic anhydride in the presence of TMSC104 and 5 A ms with preferential formation of the a-anomer of 96a (Scheme 24). The anhydride is believed to form the anomeric ester with the essential contribution of the 5 A ms as acid scavengers. The ester is then activated by the TMSC104 to give glycosides such as 96. 9.2.3 Mitsunobu Glycosylation

A particular method for the anomeric activation of 1-hydroxy sugars is the Mitsunobu reaction which has mainly found application for the synthesis of aryl glycosides. Following from an earlier publication [66], Garegg described [67] the use of this reaction for synthesis of aryl 8-D-mannopyranosides. Lubineau studied [68] the influence of the pK, of the acid or phenol in the formation of aryl and acylglycosides derived from three pesticides. Mitsunobu reaction was employed by Ogawa in the synthesis of the antibiotic hygromycin A [69]. A more extensive study was made by Roush in a project devoted to the synthesis of analogs of aureolic acid family [70], which contain oligosaccharidic chains constitutes of 2,6-dideoxypyranosides. It was observed that the glycosylation reaction proceeds through an SN2 mechanism, as the stereochemistry of the products reflect the anomeric configuration of the starting hemiacetal. Moreover, the reaction was also performed using 2-thiophenyl derivatives as donors (Scheme 25). The preferential formation of 8-glycosides, such as 99, seems to be due to the effect of 2-thiophenyl substituent on the anomeric ratio of 97 and not to a neighboring group-participation. In a recent paper, Dondoni et al. obtained [711

Scheme 25. Reagents: i) 2-naphthol, Ph3P, DEAD, toluene, O T , 74%.

9.2 1-Hydroxy Sugars

AcO 100

OSiMe,

OMe

BzO

OAc

101

i h -

215

AcO

OBz

OAc lo'

BzO

OMe OBz

Scheme 26. Reagents: i) TMSOTf, CH2C12: -50&-30"C, 76%.

glycosylated calix[4]arenes under Mitsunobu conditions. Mitsunobu reaction was also performed between glucuronic acid and phenolic chromium tricarbonyl complexes [72]. The phenol complexation is used in order to increase its acidity. Glycals ' of were shown to react under Mitsunobu conditions through an S N ~displacement the allylic hydroxy group [73]. Anomeric 0-acylation was exploited by Amos Smith within the synthesis [74] of phyllanthoside, an antineoplastic glycoside containing a (3-glycosyl ester linkage. Mitsunobu reaction has rarely been used for the synthesis of alkyl glycosides owing to the lower acidity of an alcoholic hydroxy group. An old example describes the formation of alkyl glycosides and disaccharides using, besides triphenylphosphine and DEAD, a mercuric salt [75].Descotes et al. studied the reactivity of 1,3,4,6-tetra-O-benzoyl-~-fructofuranose in Mitsunobu conditions with different nucleophiles including alcohols [ 761. Finally, Nicolaou exploited [77] the Mitsunobu reaction for the formation of an 0-glycosyl hydroxylamine in the synthesis of calicheamicin oligosaccharides. 9.2.4 1-0-Silyl Glycosides

The use of 1-0-silylglycosides as glycosyl donor was introduced by Tietze for the synthesis of aryl glycosides [78].Reaction of trimethylsilyl glycosides, protected as ester or benzyl ethers, with phenyltrimethylsilyl ethers in the presence of TMSOTf gave good yields of aryl glycosides. The method was developed by Glaudemans (Scheme 26) [79],in a version in which the hydroxy group of the acceptor was protected as a t-butyldiphenylsilyl ether (e.g. 101) and the 1-0-TMS derivative or galactopyranose 100 used as a donor, in the of 2,3,4,6-tetra-O-acetyl-~-glucopresence of TMSOTf. In this way it was possible to avoid the prior deprotection of the acceptor. Simple glycosides were obtained from the reaction between 1-0-TMS glycosides and alcohols using BF3 . OEt2 as promoter [go]. Kolar et al. used silylated glycosyl donors derived from 2,6-dideoxysugars for the synthesis of anthracycline derivatives [811. 2-Deoxy- and 2,6-dideoxy silylglycosides were also studied by Priebe et al. [82], particularly for their conversion into other synthetically useful anomeric derivatives. Starting from the observation of a transient anomeric silyl derivative in a previously studied glycosylation reaction on 1-hydroxy sugars [56], Mukaiyama applied a similar procedure (a catalytic amount of a Lewis acid, diphenyltin sulfide) to 1-0-silyl derivatives of furanoses and pyranoses. Thus 1,2-

2 16

9 Other Methods of Glycosylution

trans- or 1,243 ribofuranosides, and 1,2-& glucopyranosides were obtained using the appropriate reaction conditions [83]. Finally, a very recent work described the synthesis of spacer-linked neodisaccharides of L-daunosamine [ 841. In a glycosylation reaction on a monosilylated 1,4-butanediol the main product obtained was the monoglycosylated unprotected diol. This may suggest either a desilylation due to the acidic conditions or a higher reactivity of the silylated oxygen of the acceptor.

9.3 Esters and Related Derivatives 9.3.1 Esters

Among the various glycosyl derivatives, glycosyl esters are the most readily accessible and frequently serve as useful precursors for other glycosyl donors such as glycosyl halides. Despite the impressive development of new and efficient glycosylation methods, described in other chapters of this book, glycosyl esters still offer a good alternative as glycosyl donors, especially when more classical procedures fail. In this section, we will focus mainly on the 1-0-acyl donors introduced in the last decade, even if some older methods are mentioned. For a more detailed description of the most classical methods we refer to ref. [86] and references cited therein. The first example of 1-0-acyl sugars as glycosyl donors dates back to 1933, when Helferich described the stereoselective syntheses of both a and fl glucosides using zinc chloride or p-toluenesulfonic acid as a catalyst and 1,2,3,4,6-penta-O-acetyl-BD-glucose as a donor [85]. Since then, several applications of 1-0-acyl donors, namely 1-0-acetyl, l-O-(2’-substituted)-acetyland 1-0-benzoyl sugars, have appeared in the literature [86]. Furthermore, many different Lewis acids have been used as effective promoters in the glycosylation reactions, including tin tetrachloride [87], ferric chloride [88], BF3 . OEtz [89] and TMSOTf [90]. Mukaiyama made an extensive contribution to this area. First, he showed that 2,3,4,6-tetra-0-benzyl-l-O-bromoacetyl-fl-~-glucopyranose could be stereoselectively converted into a-glucosides with primary and secondary alcohols using trityl perchlorate as an activator [91]. This method was further developed using different promoters and applied to the synthesis of 0-glycosides and disaccharides from 1-0acyl donors and 0-trimethylsilylated nucleophiles (Scheme 27).

cat.

PO

PO

X=H, halogen, -OR

Scheme 27. Glycosylations from 1 -0-acyl donors and 0-trimethylsilylated acceptors.

9.3 Esters and Related Derivatives

r (OBn

103

1

I

L

I

,OBn B BnOn

B BnO n o S O A c BiO

217

O

G

BnO 105

ClO,

J

TMSOR

(OBn

TMSOAc

TMSCIO,

Scheme 28. General mechanism proposed for glycosylation reactions promoted by MCI,-AgC104 systems.

Thus, D-ribofuranosides with high a-stereoselectivity were obtained with the combined use of a catalytic amount of tin tetrachloride and tin(I1) triflate and a stoichiometric amount of lithium perchlorate using 1-0-acetyl-2,3,5-tri-O-benzylp-D-ribofuranose as a glycosyl donor and 0-trimethylsilylated alcohols or monosaccharides as acceptors [92]. Moreover, highly stereoselective syntheses of 1,2-cis glucosides from 1-0-acetyl glucose having a non participating group have been achieved using a combination of a Lewis acid and silver perchlorate; several Lewis acids corresponding to the general formula MCl, have been screened, tin tetrachloride and gallium trichloride ~ being the most effective [93]. The reaction was assumed to proceed via the S N type mechanism and the species such as 104 were postulated to be the active catalyst, generated from MCl, and silver perchlorate (Scheme 28). The activation of the anomeric acetoxy group affords the intermediate oxonium ion 105 stabilised by the perchlorate ion. This large counter anion blocks the p-face of the anomeric centre; silyl alkoxide attack from the a-face predominantly produces the a-glucoside along with trimethylsilyl perchlorate and MCln-lOAc, from which the catalyst is regenerated. Further, the effect of the leaving group on the stereoselectivity of this glycosylation reactions was explored by screening various l -0-acyl-2,3,4,6-tetra0-benzyl-D-glucopyranoses with tin tetrachloride and silver perchlorate as the catalytic system. The 2-(2-methoxyethoxy)acetoxygroup gave the best results. Using this glycosyl donor, either c1 or p glucosides were obtained with high stereoselectivity [94], depending on the catalytic system. Mainly a-glucosides were formed employing tin tetrachloride and silver perchlorate in diethyl ether at -5 "C (a/p ratios ranging from 96/4 to 98/2), whereas P-glucosides were achieved with tetra-

218

9 Other Methods of Glycosylution

chlorosilane-silver perchlorate in acetonitrile at - 10 "C (a/p ratios ranging from 9/91 to 1/99). It is worthy of note that the stereoselectivities of the above glycosylations were demonstrated to be independent on the a/p ratio of the starting substrate, confirming the formation of the oxocarbenium cation intermediate and therefore the S N type ~ mechanism. Mukaiyama's protocol was successfully applied to the synthesis of other pglycopyranosides and a or p furanosides. 1-0-Acetyl-2,3,4,6-tetra-0-pivaloyl-~-~-glucopyranose afforded in high yields 0glucosides and disaccharides using methyltrichlorosilane and silver perchlorate as the catalytic system [95].2-Amino-2-deoxy-~-~-glucoand galactopyranosides, which are still challenging tasks in carbohydrate chemistry, have been prepared in high yields from the corresponding 1-0-acetyl donors either using trichlorotin triflate or tin(I1) triflate as a catalyst, the latter being more advantageous for practical reasons [96]. The p configuration of the newly formed glycosidic linkages was ensured by exploiting the participation of N-2,2,2-trichloroethoxycarbonyl(Troc) as a neighbouring group. 2,3,5-Tri-O-benzyl-l-O-iodoacetyl-~-ribofuranose (alp = 1/5) gave the best results as a glycosyl donor in the synthesis of both a and p ribofuranosides: the combined use of silver perchlorate and lithium perchlorate afforded ribofuranosides with high a stereoselectivity. On the other hand, P-ribosides were obtained with different catalytic systems, such as the combination of silver perchlorate and diphenyltin sulfide. The milder tin(I1) chloride also gave high yields of p-ribosides, when used in combination with 2 mol% of tetrachlorosilane [97]. More recently, Reynolds described the syntheses of D-arabinofuranosides, Dgalactofuranosides and L-rhamnopyranosides viu tin tetrachloride mediated glycosylation of the corresponding l -0-acetyl donors [98]. Finally, the combination of dimethyl dichlorosilane and silver perchlorate was proved to be efficient in the a-sialylation of the 6-OH of a P-D-galactopyranosyl acceptor, using a 2-0-acetylated sialyl donor: the disaccharide product, a useful building block for the synthesis of sialyloligosaccharides, was obtained in 76% yield in propionitrile at -28 "C (alp = 83/17) [99]. As stated above, the 1-0-acetyl donors may still be an effective alternative to classical methods, as recently witnessed by the synthesis of a kedarcidin chromophore analogue (Scheme 29). In a key step of the synthesis, the alcohol 106 was

Kedarosamine

106

Scheme 29. Reagents: i) BF3 . OEtz, toluene, 0 "C, 80%.

107

9.3 Esters and Related Derivatives

219

AcO OAc

OAc 108

109

110

Scheme 30. Reagents: i) Lewis acid, CH2C12, -1O"C, 76-950/0; ii) Bu3SnH, AIBN, toluene, 1 lO"C, 80-95%.

glycosylated with the N-protected 1-0-acetyl kedarosamine and afforded the uglycoside 107 in 80% yield. It is worthy of note that the use of either the glycosyl imidate or the phenyl thioglycoside of the kedarosamine led to lower yields of the product [ 1001. An efficient synthesis of P-glycosides of 2,6-dideoxy sugars, occurring in many biologically relevant antibiotics such as digitoxin, venturicidins and olivomycin, was achieved using acetyl 3,4-di-O-acetyl-2,6-anhydro-2-thio-~-~-altropyranoside 108 as a glycosyl donor (Scheme 30). Several Lewis acids promoters and solvents were examined. The best results, in terms of yields and stereocontrol, were obtained when the glycosylations were carried out in dichloromethane in the presence of TMSOTf (1.1 eq.) at -10°C. The 2,6-anhydro-2-thio bridge of the obtained p-glycosides can be easily converted into the corresponding 2,6-dideoxy derivatives either by standard hydrogenolysis using Raney-Ni as a catalyst or by the treatment with tributyltin hydride and AIBN in toluene at 110 "C [loll. Lanthanoid(II1) triflates catalyzed glycosylations with 1-O-methoxyacetyl sugars as donors have been described. After a preliminary screening of many lanthanoid triflates, ytterbium triflate, Yb(OTf)3, was selected as the catalyst and acetonitrile as the reaction solvent. By this method, a variety of glycosides including disaccharides and thioglycosides were prepared in good to excellent yields. The stereoselectivities are sometimes modest, but they could be improved by increasing the reaction temperature to 53 "C. Under these conditions, the thermodynamically more stable p-isomer was obtained as the major product [102]. The Nicholas reaction [lo31 has been successfully exploited for a novel glycosylation reaction based on the internal delivery concept (Scheme 3 la). The suitably O-protected glycopyranoside derivative 111 bearing the cobalt-complexed 4-alkoxy6-phenyl-5-hexynoate residue at the anomeric position affords the corresponding propynyl cation 112 under Lewis acid catalysis. The intermediate 112 collapses spontaneously to form the oxonium ion delivering the cobalt-complexed y-lactone 113. The oxocarbenium cation is then intramolecularly captured by its counter alkoxy anion leading to glycoside 114. The rather laborious preparation of the starting substrates for the investigation of this glycosylation reaction is described in Scheme 3 1b. Various glycosyl "donors" of the gluco, galacto and manno series were synthesized, either O-benzylated or 67 was O-benzoylated, whereas only methyl 2,3,4-tri-O-benzyl-cr-~-glucopyranoside used as nucleophile.

220

9 Other Methods of Glycosylation

Ph b)

TBDPSO

BnO

115 61

117

(OBn

pp. (CO)co

k$co)3

0

118

alp 9191

113

Scheme 31. Reagents: i) CO~(CO)S, Et20, rt; ii) 67,BF3 . OEt2, CH2C12, 0°C; iii) CAN, MeOH, rt, overall yield 74%; iv) TBAF, THF, rt, 92%; v) PDC, DMF, rt; vi) 91, DCC, DMAP, HOBt, Et3N, CH2C12, rt; vii) Co2(CO)8, Et20, rt; viii) TMSOTf, EtCN, -65 "C, 740/0.

As an illustrative example, compound 117 was synthesized from 115 and 116. The treatment of 117 with Coz(C0)~gave the corresponding cobalt complexes, which afforded the disaccharide 118 with high p-stereoselectivity. The y-lactone 113 was also recovered in high yields [ 104). The p-nitrobenzoate was successfully employed as a leaving group in the synthesis of the biologically important disaccharide H, 2-O-(a-~-fucopyranosyl)-~-galactopyranose. Although the p-nitrobenzoate is not an excellent leaving group, in the case of the highly reactive 2,3,4-tri-O-benzyl-t-fucosyl derivative (a/p = 5/1) it was efficiently activated using BF3 . OEt2 in the presence of 1,3,4,6-tetra-O-acetyl-a-~galactopyranose as a glycosyl acceptor. The disaccharide H was obtained in yields

9.3 Esters and Related Derivatives

221

OBn-p-CI p-CBnO p-CBnO p-ClBnO 119

HO 0

BnO OMe

87

p-CBnO p-CBnO 120

alp 6194

Scheme 32. Reagents: i) TrB(C6F5)d 10 mol%, BuCN-CH2C12 2/1, Drierite, -23 "C, 97's.

ranging from 70Y0to 79%, and with a a/p ratio of 11 : 1, from which crystallization gave pure a anomer in 40-50% yield [ 1051. 9.3.2 Sugar Carbonates and Derivatives Glycosyl phenyl carbonates as glycosyl donors have been reported by Mukaiyama et al.: after activation in the presence of an acceptor, the leaving group decomposes into carbon dioxide and phenol. First, efficient a-sialylation of a 6-O-unprotected galactosyl acceptor was achieved with a 2-O-phenyl carbonate derivative of Nacetylneuraminic acid, under dimethyl dichlorosilane-silver perchlorate catalysis [99]. Glucosyl phenyl carbonate 119 was activated with a catalytic amount of trityl tetrakis(pentafluoropheny1)borate to provide (1 +6), (1'3) and (1+4)-linked disaccharides in very good yields and with excellent p-stereoselectivity (Scheme 32). Interestingly, the reaction can be performed on alkyl thioglycoside as acceptors, giving access to one-pot sequential syntheses of trisaccharides upon the activation of the alkyl thio group of the disaccharide product [ 1061. An attractive strategy for the formation of glycosidic bonds is based on bridging donor and acceptor. The decarboxylative glycosylation method, developed by Ikegami and co-workers [107], is an example. Donor and acceptor are connected through a carbonate bridge and the glycosidic bond is obtained by internal removal of carbon dioxide promoted by a Lewis acid (Scheme 33a). The mixed carbonates were prepared as a mixture of anomers by l-O-acylation of pyranoses (glucose and galactose) with activated alkyl carbonates of alcohols or monosaccharides (glucose and galactose). The decarboxylative glycosylations were best promoted by stoichiometric amounts of trialkylsilyl triflates (TMSOTf or TBDMSOTf) and the choice of the solvent was crucial. The highest yields were obtained using toluene and mesitylene. The stereoselectivities were only modest when a non participating group was present, furthermore, the anomeric ratio of the starting materials was not reflected in the products, as deduced starting from anomerically pure carbonates. Although an intramolecular mechanism was initially claimed, Schmidt et al. [ 1081 demonstrated with competition experiments that the acid promoted decarboxylative glycosylation follows an intermolecular course. Good a-stereoselectivities were obtained applying a catalytic version of the decarboxylative glycosylation, using the combinations of tin tetrachloride-silver perchlorate or CpzHfCl&ver perchlorate as promoters and EtzO as a solvent [ 1091. On the other hand, very high p-stereoselectivities were reported in the decarboxylative glycosylations of O-acyl protected

222

9 Other Methods of Glycosylation

BnO

123 121

Scheme 33. Reagents: i) K2C03, toluene, 88%; ii) TMSOTf, CH2C12, rt, 89%, only 0.

sugar carbonates. Using 0-acyl protected donors (i.e. 94), selective synthesis of mixed P-carbonates (122) was achieved from 0-succinimidyl-activated carbonates of various acceptors, mainly in the presence of K2CO3 (Scheme 33b) [ 1lo]. Ley and co-workers reported that various carbohydrate derivatives react readily via the C-1 hydroxy group with carbonyldiimidazole or thiocarbonyldiimidazole in ether (EtlO or THF) to give the 1-imidazolylcarbonyl-or thiocarbonyl- glycosides 124 and 125, respectively (Scheme 34). These compounds were reacted with alcohols and glycosyl acceptors in the presence of zinc bromide or silver perchlorate affording the corresponding glycosides in good yields. When the reactions were carried out in diethyl ether as a solvent, a glycosides were obtained with good stereoselectivity, whereas in dichloromethane the quantity of the anomers was increased, suggesting that a SN1 type process is operating [ 1111.

124

0

Scheme 34. 1 -1midazolylcarbonyl- (1 24), 1 -imidazolylthiocarbonyl- (125) and 2-pyridincarboxyl(126) glycosides.

9.3 Esters and Related Derivatives

223

The authors suggest that the displacement is facilitated by chelation of the metal atom to the carbonyl oxygen (or thiocarbonyl sulfur) and the N-3 of the imidazolyl group. This can be related to the remote activation concept originally introduced by Hanessian [ 1121. The same strategy, i.e. the introduction of a bidentate ligand as a leaving group, was exploited using glycosyl2-pyridinecarboxylates 126 as glycosyl donors (Scheme 34). These compounds were easily prepared by esterification of the anomeric hydroxy group of the sugar with picolinoyl chloride in the presence of triethylamine. Mild Lewis acids, such as tin(I1) triflate and copper(I1) triflate, were found to be more effective than other reagents, such as BF3 . OEtz and TMSOTf, in promoting the activation of the 2-pyridinecarboxy group. Interestingly, in preliminary experiments carried out both with alcohols and monosaccharides as nucleophiles, the stereochemical outcome of the glycosylation reactions was observed to be dependent on the choice of the metal salt and the solvent [ 1 131. This procedure was further applied to more glycosyl donors, namely mannopyranose, 2-azido-2-deoxy-glucopyranose and 2-deoxy sugar derivatives [ 1141, demonstrating that the stereochemistry of the glycosylation reactions was influenced not only by the promoter and the solvent, but also by the structure of the donor. Glycosyl 2-thiopyridylcarbonates have been described as effective glycosyl donors by Hanessian et al. However, this leaving group has been used mainly for the formation of N-glycosidic linkages in the synthesis of 1,2-cis and 1,2-trans nucleosides [115]. A new glycosylation method based on the mild activation of 0-glycosyl N-ally1 carbamates with soft electrophiles has been described by Kunz and co-workers [ 1161. This activation is similar to that of 0-glycosyl4-pentenoyl esters reported by Fraser-Reid [ 1171, with formation of an oxazolidinone instead of a lactone as the leaving group at the anomeric center. Anomeric N-ally1 carbamates were readily obtained from anomerically unprotected carbohydrates by reaction with ally1 isocyanate. Among the electrophiles used for the activation, the methyl bis-methylthiosulfonium hexachloroantimonate (TMTSB) was found to be the most effective in promoting the glycosylations. Various alcohols and O-unprotected sugars were used as acceptors, and the corresponding glycosides were obtained in good yields but with low stereoselectivity, except in cases where neighbouring-group participation occurred. The authors applied this procedure to the synthesis of the mucine core trisaccharide 131 showing that many different protecting groups, such as alkyl and silyl ethers, esters and acetals, remain unaffected under the mild conditions required for the TMTSB-promoted activation of 0-glycosyl N-ally1 carbamates (Scheme 35) [ 1 181. 9.3.3 Orthoesters and Oxazolines The first appearance of 1,2-0-orthoesters as glycosyl donors in glycosylation reactions was in 1971. Kochetkov et al. reported a teut-butyl orthoester as a donor upon activation with 2,6-dimethylpyridinium perchlorate [ 1 191 and extensively

224

9 Other Methods of Glycosylation Ph

I

127

A c AcO

C

Y

128

&

oms+ yo

&

AcO

N3

129

NPht

130

129

... AcO

0

oms AcO AcO 131

Scheme 35. Reagents: i) TMTSB, CH2C12, -15°C + -5"C, 67%; ii) HS(CH&$H, CH2C12, 83%; iii) TMTSB, CH2C12, -29°C + -1O"C, 48%.

PTSA,

explored this methodology introducing the 1,2-0-(1-cyanoethylidene) group, whose activation occurred under milder conditions. Other orthoester derivatives, such as 1,2-0-[1-(p-methylpheny1)thiolethylideneand 1,2-0-[I-"-( 1-phenylethylidene)amino]oxyl]-2,2-dimethylpropylidene, were also described [861. A recent useful application of the 1,2-0-(1-cyanoethylidene) glycosyl donors was found in the highly regioselective glycosylations of primary-secondary ditrityl ethers acceptors. Quite surprisingly, secondary trityl ethers were essentially more reactive than the primary ones, using cyanoethylidene derivatives of mannose and galactose as donors. All glycosylations gave the 6-0-trityl derivatives of 1-2 and 1-4 linked disaccharides in good yields and very high regioselectivity, even in the case of such a poorly reactive acceptor as the 0-4 of galactose. This unusual regioselectivity was attributed to the formation of an early transition state, where electronic factors, i. e. electron density at oxygen, rather than steric factors will become determinant [ 1201. There are two main advantages in using 1,2-0-orthoesters as glycosyl donors: -

-

the orthoester ring directs the nucleophilic attack to the back side, leading to 1,2trans glycosides; the nucleophilic opening of the orthoester ring (for example, with a simple alcohol) allows the contemporary and orthogonal protection of the C-1 and C-2 functionalities of the sugar.

During the last decade sugar orthoesters have been widely employed. However, their use has been more as protecting groups rather than as glycosyl donors in oligosaccharide synthesis. The new achievements concerning glycosylation reactions with sugar orthoesters exploit their chemistry in a quite different way. Recently, Kong and co-workers reported a new methodology for highly regio-

225

9.3 Esters and Related Derivatives

(OAC

/OH

(OAc AcO AcO Br 72

+

AcO

Ho%

BnO

AcO

-

Me (OH

OMe

y:o%

132

OMe

OMe

ii

7

E

133: R = H 1 3 4 R = Ac

AcO

TMSOTf

OMe 135:R=H 136R=Ac

Scheme 36. Reagents: i) AgOTf, 2,4-lutidine, CH2C12, rt, 95%; ii) AczO, pyridine, qu.; iii) TMSOTf, CH2C12, O"C, 78%.

and stereoselective synthesis of oligosaccharides. The procedure is based on the preparation of sugar-sugar orthoesters by coupling acetobromosugars with unprotected or partially protected sugar acceptors in high yields and excellent selectivity. Subsequent rearrangement in the presence of catalytic quantities of TMSOTf yields the oligosaccharide [121]. In a typical example, acceptor 132 was regioselectively converted into orthoester 133 from acetobromoglucose 72 using silver triflate in the presence of 2,4-lutidine (Scheme 36). Rearrangement of 133 with a catalytic amount of TMSOTf selectively afforded disaccharide 135 in 78% yield. The same compound was obtained in 80% yield when 4-0-acetylation of 133 ( 4 134) was performed prior to the rearrangement. These results are mainly due to steric factors (Scheme 36). The less hindered hydroxy group attacks the acyloxonium carbon of A generating orthoester 133 which, during the following rearrangement, is fragmented again into A and the trimethylsilylated acceptor: finally, a regio- and stereoselective backside attack on the C-1 of A gives disaccharide 135. Many other examples reported by the authors demonstrate the excellent regioselectivity in the orthoester formation. This new method was widely explored and applied to the synthesis of complex branched oligosaccharides. First, a variety of acetobromosugars and glycosyl acceptors were examined in terms of their ability to afford regioselective orthoester formation. It was found that both the regio and the stereoselectivity were dependent on the acceptor rather than the donor structure, according to the mechanism depicted in Scheme 36. Indeed, acetobromo derivatives of glucose, galactose,

226

9 Other Methods of Glycosylation

mannose, rhamnose, lactose and maltose invariably led to highly regioselective formation of the orthoesters with various acceptors [121], [122], [123]. On the other hand, the behaviour of the glycosyl acceptors under the orthoester formation conditions can be summarized as follows: n,6-unprotected acceptors (n = 2-4) always afford 1+6 linked oligosaccharides; selective 3-0-glycosylations were achieved with 2,3- and 3,4-unprotected sugars (glucose, mannose); - the rhamnose behaviour is an exception: Very high 3-regioselectivity was observed both with 3,4- and 2,3,4-unprotected sugars, whereas the 2,3-diol gave good 2-regioselectivity, further demonstrating that orthoester formation is very sensitive to steric factors [ 1221. -

The mannose-type acceptors gave particularly interesting results: The 3,4,6unprotected derivatives can be directly converted into the corresponding 3,6diorthoester with very high regioselectivity. Thus, using acetobromomannose 137 as a donor, the biologically important 3,6-branched mannotriose 141 was obtained in 88% yield by rearrangement of the diorthoester 140 (Scheme 37). When acetobromolactose was employed as a donor, a 3,6-branched pentasaccharide diorthoester was formed (88%)) the rearrangement of which afforded the 3,6-branched pentasaccharide in 740/0yield [124], [125]. Likewise, both 2,3,4- and 3,4-unprotected mannosyl acceptors gave 3-0-glycosylation regioselectively, by rearrangement of the corresponding orthoesters, leading to various very important mannose-based building blocks [ 1251. The versatility of this methodology was demonstrated by its application to the synthesis of biologically relevant branched oligosaccharides, such as the phytoalexin elicitor hexasaccharide [ 1241, [ 1261, [ 1271, a trisaccharide part of the Escherichia coli K26 antigen [ 1221 and the trisaccharide repeating unit of the 0 2 antigen occurring in the lipopolisaccharide of Stenotrophomonas maltophilia [ 1281. A related approach to the stereoselective synthesis of glycosidic linkages is based on the reductive cleavage of cyclic orthoesters generated by two sugar moieties. A C O y OAC ACOAAcOc

OAC O

q

+

Br 137

138

ii

139: R = H

141: R = Ac

140: R = Ac

Scheme 37. Reagents: i) AgOTf, 2,4-lutidine, CH2C12, rt, 91%; ii) Ac20, pyridine, qu.; iii) TMSOTf, CH2C12, -30 "C, 88%.

9.3 Esters and Reluted Derivatives

227

i

OMe

R, = H, R4= OBn R2=OBn, R, =OBn, % = H

142: R, = OBn, R, = H,

143: R, = H ,

OMe 144: 82% 145 74%

Scheme 38. Reagents: i) LiAlH4, AlC13, CH>ClZ/Et20, rt.

Such cyclic orthoesters are available from the corresponding sugar lactones through reaction with sugar diols in the presence of methoxy trimethylsilane and TMSOTf [ 1291. Using this procedure, 4,6-O-glucopyranosylidene, galactopyranosylidene and mannopyranosylidene acetals were synthesized. For the cleavage of the orthoesters, the combination LiAlH4-AlC13 was found to be the most effective reducing agent. The treatment of glycopyranosylidene acetals afforded the p-(1-+4)-linkeddisaccharides, with high regio- and stereoselectivity [ 1301. It is worthy of note that only the most sterically congested p-(1-4) glycosides were obtained, and not one of the other possible isomers, i.e. the p-(1-6) glycosides andfor a-glycosides, were detected in the reaction mixture, even in the case of the mannose-type donor. This high selectivity was attributed to the predominant coordination of the less hindered oxygen atom followed by the axial attack of a hydride. The procedure was further extended to the synthesis of glycosyl-galactosides (Scheme 38). The corresponding glycopyranosylidenes 142 and 143 were reductively cleaved with LiAlH4-AlC13 giving the highly congested p-( 1-4)-glycosyl-galactosides 144 and 145 with excellent regio- and stereoselectivity [131]. The mannose 1,2,6-orthoester 147 was recently reported as a potential glycosylating agent; after protection of the remaining hydroxy groups, the ring opening of the orthoester with nucleophiles could be achieved with 2-OH and 6-OH groups being orthogonally protected (Scheme 39). Compound 147 was prepared by mannose 1,2-0rthoester 146 using tert-butyldiphenylsilyl chloride and imidazole in DMF or, better, pyridinium triflate in pyridine. O-Acetylation or O-benzylation of 147 gave orthoesters 148a,b. The glycosylating properties of 148b were then examined: BF3 . OEtz was found to be the best promoter, and glucoside 149 was chosen as an exemplificative acceptor (Scheme 39). Disaccharide 150 was obtained in good yields. Interestingly, when an excess of orthoester was used, the partial formation of the trisaccharide due to the glycosylation of the 6’-hydroxy group took place [ 1321. Sugar tricyclic orthoesters, such as glucose 1,2,4-0rthoester derivatives, were used as monomers in polymerization reactions to afford cellooligosaccharides with a degree of polymerization up to 20. The 1,2,4-0rthopivalate was found to be the best monomer, both in terms of regio- and stereoregularity of the polymer produced, and it was successfully employed in the first chemical synthesis of cellulose [133].

228

9 Other Methods of Glycosylation

HO

HO HO

RO

146

147

148a:R = Ac 148b:R = En

-

+ En0

BnO 149

yo Bn0-&&/ooct BnO

148b

En0 150

Scheme 39. Reagents: i) Py . TfOH, pyridine, 89%; ii) AczO, pyridine, qu. or BnCI, raH, DMF, qu.; iii) TMSOTf, ClCH2CHzC1, 0 "C, 68%.

The electronic and steric effects of diverse orthoester groups (orthopropionate, orthoacetate and orthobenzoate) on the ring-opening polymerization of glucose 1,2,4-0rthoester were then explored. Although these compounds gave stereoregular polysaccharides, having only P-glucosidic bonds, the polysaccharides were not regioregular, consisting of a mixture of P-(1+4) and P-(1+2) units. These results were largely attributed to steric rather than electronic effects [ 1341. The 1,2-O-orthoesters derivatives of D-mannose and D-galactose were used as glycosyl donors to incorporate a-mannoside and P-galactoside branches at C-6 of chitin and its N-deacetylated form chitosan [ 1351 employing a N-phthaloyl 6-0-trimethylsilylated chitosan derivative as an acceptor. Sugar oxazolines are obtained from 2-acylammido-2-deoxy glycosyl donors through deprotonation of the oxazolinium ion intermediate 151 (Scheme 40). They are quite stable and can be used as glycosyl donors in 1,2-truns glycosylation reactions. In acidic media, the oxazoline 152 generates the oxazolinium ion, which can react with a suitable acceptor to provide the corresponding P-glycoside 153.

151

R

Y 153

Scheme 40. Glycosylation via oxazolinium ion intermediate.

9.3 Esters and Related Derivatives

229

Due to the strongly acidic conditions usually required for their activation, the oxazoline method has not found broad application in the field of oligosaccharide synthesis. They are incompatible with acid-sensitive protecting groups and with poorly nucleophilic acceptors. Their employment seems to be restricted to reactive acceptors. Despite these severe limitations, some important examples of the application of the oxazoline method have been reported [ 1361. Recently, Martin et al. described the synthesis of a- and P-u-glucosamine and galactosamine 1-phosphates using the more reactive trifluoromethyl oxazolines and dibenzyl phosphate [ 1371. was used as a monomer The methyl oxazoline of 3,6-di-O-benzyl-~-glucosamine for the synthesis of a dibenzylchitin-type polysaccharide. The polymerization under 1O-camphorsulfonic acid catalysis provided the polyaminosaccharide [ 1381. was used as a glycosyl The methyl oxazoline of 3,4,6-tri-O-acetyl-~-glucosamine donor in the glycosylation of ally1 6-O-pivaloyl-~-~-galactopyranoside: the reaction took 3-4 days in the presence of BF3 . OEtl and afforded the (3-(1+3)-disaccharide as the only regioisomer in 45-500/0 yield [ 1391. Finally, the unprotected methyl oxazoline of D-glucosamine was used as a substrate for the glycosidase chitinase in the synthesis of the N,N’-diacetylchitobiose, the repeating unit of chitin, via an enzymatic glycosylation [ 1401. 9.3.4 Phosphorus and Sulfur Derivatives Various glycosyl donors incorporating a variety of phosphorus-containing leaving groups have been developed and used for the stereoselective construction of a- and p- glycosidic linkages, with or without a neighbouring participating group [ 861. Besides the classical glycosyl phosphates and phosphites, the use of glycosyl diphenylphosphineimidates [141] and glycosyl phosphorodiamidimidothioates [142] as glycosyl donors have been described. New achievements in this area include both pentavalent and trivalent phosphorous-containing leaving groups. Glycosyl phosphoroamidates, phosphorimidates and phosphoramidimidates come into the first category. prepared from Shelf-stable glycosyl N,N,N’,N’-tetramethylphosphoroamidates, both O-benzyl and O-benzoyl-protected D-glucopyranose and D-galactopyranose, were coupled with a variety of primary and secondary glycoside alcohols in propionitrile in the presence of TMSOTf at -78 “C. In all cases, the chemical yields were very high as well as the p-stereoselectivities, demonstrating the considerable versatility of this glycosylation method [ 1431. Recent reports detail how this methodology has been extended. The high stability of glycosyl N,N,N’,N’-tetramethylphosphoroamidates allowed the reactivity to be modulated either by varying the nature of the protecting groups (benzyls or benzoyls) or the glycosylation conditions (Lewis acid and reaction temperature). This feature was exploited in a novel application of the “armed-disarmed” principle introduced by Fraser-Reid [ 1441. Thus, the “armed” donor 154 was coupled with the “disarmed” acceptor 155 under the aforementioned conditions to provide disaccharide 156 in 86% yield ( a :(3 = 5:95). Switching the protective groups from

230

9 Other Methods of Glycosylation

alp 5/95

BnO RO 154

155

156: R = Bz

ii

(OBn

158

0

oP@Me2)

157: R = Bn

BzO 0PWez)z

Scheme 41. Reagents: i) TMSOTf, EtCN, -78”C, 86%; ii) NaOMe, MeOH, then BnBr, NaH, DMF, 89% overall yield; iii) 155, TMSOTf, EtCN, -78 “C, 83%.

benzoyls to benzyls, the “armed” donor 157 was obtained, the glycosylation of which with acceptor 155 gave trisaccharide 158 in 83% yield (a: = 13:87, Scheme 41) [145]. On the other hand, the “armed-disarmed’’ concept was applied to glycosylation reactions between glycosyl N,N,N’,N‘-tetramethylphosphoroamidates and glycosides bearing other phosphorous-containing leaving groups, such as glycosyl diphenylphosphinimidates, phosphites and phosphorodiamidimidothioates, on each occasion taking advantage of their well documented reactivity differences [ 1451. An illustrative example is showed in Scheme 42. Fully benzylated glycosyl phosphites, such as 159, are known to be readily activated with BF3 . OEt, even at -78 “C giving very high p-selectivities, whereas the corresponding tetramethylphosphoroamidates were unreactive at below - 10 “C in the same conditions. The “armed” donor 159 was therefore coupled with

BnO

alp 3197

BnO

160

BnO

Bn 157 159

161

OMe

67

Scheme 42. Reagents: i) BF3 OEtl, CH2C12, -7O”C, 66%, only -78”C, 88%.

p; ii) 67, TMSOTf, EtCN,

9.3 Esters and Related Derioatives

-T&

T&OP(OEtJ2

-,

NF’h

II

0 OEt P-OEt

23 1

promoter ROH

(OMe

alp 1/30 0-P-OEt Me0

Me0 162

I

OEt

+

BnO I63

BnO BnO

164

Scheme 43. Reagents: i) PhN3, benzene, 50°C; ii) TMSOTf, EtCN, -78”C, 7 5 %

the “disarmed” acceptor 160 at -70°C giving disaccharide 157 in 66%1 yield (only p-anomer). This was subsequently activated with TMSOTf at -78°C in the presence of the alcohol 67 to afford the trisaccharide 161 (88%) a : p = 3 :97). Glycosyl N-phenyl diethylphosphorimidates were prepared by the treatment of the corresponding glycosyl phosphites with phenyl azide and used without further purification in glycosylation reactions with primary and secondary glycosyl acceptors. The chemical yields were generally good, whereas the stereoselectivities ranged from fair to only moderate. Scheme 43 shows an overview of the procedure and a selected example [ 1461. Interestingly, the authors observed a general a-selectivity using lutidinium ptoluensulfonate/tetrabutylammonium iodide as promoters and P-selectivity in the presence of TMSOTf. Moreover, the highest selectivities (a or p ) were obtained for the primary acceptor alcohols, whereas the ratios drop with secondary acceptors. The treatment of glycopyranoses with ethoxy bis(diisopropy1amino)phosphine in the presence of diisopropylammonium tetrazolide afforded the corresponding glycosy1 phosphoramidites 166 which, by reaction with phenyl azide, gave the glycosyl phosphoramidimidates 167. When they were used as glycosyl donors with various acceptors, the corresponding disaccharides were obtained, but only with modest pstereoselectivities. The best results were achieved with primary alcohols in propionitrile, in the presence of TMSOTf or BF3 . OEt2 (Scheme 44)[ 1471. The glycosyl phosphoramidites themselves were explored as glycosyl donors. First, 2-deoxyglycosidic linkages were synthesized. Excellent a-stereoselectivities were achieved using 2-deoxyglycosyl N,N-diisopropylphosphoramiditesas glycosyl donors, even in the case of hindered alcohol acceptors [148]. Compounds 169 were then coupled with selected primary and secondary glycosyl acceptors to give disaccharides with high p : a ratios, either using TMSOTf or BF3 . OEt2 as promoters. Propionitrile was found to be the best solvent for achieving high selectivities as well as good chemical yields (Scheme 45) [149]. Glycosyl diphenylphosphinyl donors were recently explored for the synthesis of D-ribofuranosides: the couplings were conducted with 1.5 eq. of TMSOTf at low temperature using simple alcohols or a-aminoacids as acceptors. Although only

232

9 Other Methods of Glycosylation

Me0 Me0

\

Me0

-

NPf, 165

(OMe

0-P-OEt

I

Me0

NPr',

166

...

167

ci/p 1/20

111

.___)

/OH

67

BnO

OMe

168

Scheme 44. Reagents: i) EtOP(NPr2')2, diisopropylammonium tetrazolide, CH2C12, rt; ii) PhN3, benzene, 45 "C; iii) 67,TMSOTf, EtCN, -78 "C, 73%.

modest stereoselectivities were observed, the glycosides were obtained in excellent yields. However, when a glucosyl acceptor was tested, the formation of a complex mixture of products was observed [ 1501. Glycosyl phosphorodithioates 171 (Scheme 46), prepared by action of 0,Odialkylphosphorodithioic acid on the corresponding glycals, were employed as glycosy1 donors in the synthesis of 2-deoxydisaccharides. The glycosylations of primary and secondary acceptors in the presence of silver fluoride proceeded sluggishly (from 7 to 12 days) and the a-stereoselectivities were somewhat low, except in a few cases [151]. This procedure was extended by the preparation of the related glycosyl phosphorothioates 172 and phosphoroselenoates 173 (Scheme 46). They were used as donors in the glycosylation of Diclofenac, a non-steroidal agent with potent antiinflammatory, analgesic and antipyretic activity, in order to produce new Diclofenacderived prodrugs. The corresponding glycosyl esters were obtained in good yields (72-84%) and with excellent P-selectivity, using silver carbonate as a promoter [ 1521. Among the sulfur containing leaving groups, glycosyl sulfonates have been described 11.531, but they were not further developed in recent years, although glycosyl triflates have been claimed as reactive intermediates in many glycosylation reaction 11541. Only one example of the use of glycosyl sulfates as glycosyl donors has been reported [ 1551. They were easily prepared from gluco- and mannopyranoses and em-

0-P

BnO BnO 169

BnO

9/91

BnO 95

Scheme 45. Reagents: i) BF3 . OEtz, EtCN, -78 "C, 92%.

170

References

171

172

233

173

Scheme 46. Glycosyl phosphorodithioates (171), glycosyl phosphorothioates (172) and glycosyl phosphoroselenoates (173).

ployed in the glycosylation of simple alcohols as well as primary and secondary sugar acceptors, providing the corresponding glycosides in fair to good yields, but with low stereoselectivities. Finally, sugar 1,2-cyclic sulfites have been investigated as glycosyl donors. Preliminary studies exploited these compounds for the introduction of reactive nucleophiles, such as azide or benzoate, on the anomeric position [156]. Effective b-0glycosylations with sugar cyclic sulfites were then reported, under catalysis of lanthanide(II1) triflates. However, the reactions required high temperatures (from 80 to 100 "C),limiting their application only to simple alcohols [ 1571. References I. a) K. Suzuki, T. Mukaiyama, Chem. Lett., 1982, 683-686; b) K. Suzuki, T. Mukaiyama, Chem. Lett., 1982, 1525-1528. 2. a) J. M. Beau, R. Schauer, J. Haverkamp, L. Dorland, J. F. G. Vliegenthart, P. Sinay, Curbohydr. Rex, 1980, 82, 125-129; b) F. Paquet, P. Sinay, Tetrahedron Lett., 1984, 25, 30713074. 3. F. Paquet, P. Sinay, J. Am. Chem. Soc., 1984, 106, 8313-8315. 4. a) F. Nicotra, L. Panza, F. Ronchetti, G. Russo, L. Toma, Tetrahedron Lett., 1985, 26, 807808; b) F. Nicotra, L. Panza, F. Ronchetti, G. Russo, L. Toma, J. Chem. SOC.,Perkin Trans. 1987, I , 1319-1324. 5. A. Boschetti, L. Panzd, F. Ronchetti, G. Russo, J. Chem. Soc., Perkin Trans. 1988, 1 , 33533357. 6. A. G. M. Barrett, B. C. B. Bezuidenhoudt, A. F. Gasiecki, A. R. Howell, M. A. Russell, J. Am. Chem. Soc., 1989, I l l , 1392-1396. 7. A. G. M. Barrett, B. C. B. Bezuidenhoudt, L. M. Melcher, J. Org. Chem., 1990; 55, 51965197. 8. A. Haudrechy, P. Sinay J. Org. Chem., 1992, 57, 4142-4151. 9. R. U. Lemieux, S. Levine, Can. J. Chem., 1%4,42, 1473-1480. 10. R. W. Friesen, S. J. Danishefsky, J. Am. Chem. SOC.,1989, 111, 6656-6660. 1 1 . K. Tatsutd, K. Fujimoto, M. Kinoshita, S. Umezawa, Carbohydr. Rex, 1977, 54, 85-104. 12. a) K. Tatsuta, T. Yamauchi, M. Kinoshita, Bull. Chem. Soc. Jpn. 1978, 51, 3035-3038; b) K. Tatsuta, Y . Amemiya, Y. Kanemura, H. Takahashi, M. Kinoshita, Tetrahedron Lett., 1982, 23, 3375-3378; c) K. Toshima, K. Tatsuta, M. Kinoshita, Tetrahedron Lett., 1986, 27, 47414741. 13. a) J. Thiem, H. Karl, J. Schwenter, Synthesis, 1978, 696-699; b) J. Thiem, H. Karl, Tetrahedron Lett., 1978, 4999-5002. 14. a) J. Thiem, S. Kopper, Tetrahedron, 1990, 46, 113-138; b) D. Horton, W. Priebe, M. Snaitzman, Carbohydr. Res., 1989, 187, 149-153; c) B. Abbaci, J. C. Florent, C. Monneret, J. Chem. Soc., Chem. Commun., 1989, 1896-1897; d) S. J. Danishefsky, H. G. Selnick, D. M. Annistead, F. E. Wincott, J. Am. Chem. Soc., 1987, 109, 8119-8120.

234 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25.

9 Other Methods of Glycosylation

J. Thiem, W. Klafie, J. Org. Chem., 1989, 54, 2006-2007. G. Jaurand, J. M. Beau, P. Sinay, J. Chem. Soc., Chem. Commun., 1981, 572-573. A. G. M. Barrett, T. A. Miller, Tetrahedron Lett., 1988,29, 1873-1874. Y. Ito, T. Ogawa, Tetrahedron Lett., 1987, 28, 2723-2726. G. Grewal, N. Kaila, R. W. Franck, J. Org. Chem., 1992, 57, 2084-2092. C. J. Tu, D. Lednicer, J. Org. Chem., 1987, 52, 5624-5627. See e.g. K. Toshima, K. Tatsuta, M. Kinoshita, Bull. Chem. Soc. Jpn., 1988, 61, 2369-2381. V. Bolitt, C. Mioskowski, S. G. Lee, J. R. Falck, J. Org. Chem., 1990, 55, 5812-5813. S. Sabesan, S. Neira, J. Org. Chem., 1991, 56, 5468-5472. J. Thiem, M. Kleeberg, Carbohydr. Rex, 1990, 205, 333-345. D. Noort, G. H. Veeneman, G. J. P. H. Boons, G. A. van der Marel, G. J. Mulder, J. H. van Boom, Synlett, 1990,205-207. 26. F. Nicotra, L. Panza, G. Russo, Tetrahedron Lett., 1991, 32, 4035-4038. 27. L. Lay, F. Nicotra, L. Panza, G. Russo, G. Sello, J. Carbohydr. Chem., 1998,17, 1269-1281. 28. A. Marra, J. Esnault, A. Veyrieres, P. Sinay, J. Am. Chem. Soc., 1992, 114, 6354-6360. 29. a) H. K. Chenault, A. Castro, L. F. Chafin, J. Yang, J. Org. Chem., 1996, 61, 5024-5031; b) H. K. Chenault, L. F. Chafin, J. Org. Chem., 1998,63, 833-840. 30. a) G. J. Boons, S. Isles, Tetrahedron Lett., 1994, 35, 3593-3596; b) G. J. Boons, S. Isles, J. Org. Chem., 1996, 61, 4262-427 1. 31. G. J. Boons, B. Heskamp, F. Hout, Angew. Chem. Int. Ed. Engl., 1996,35, 2845-2847. 32. M. Johnson, C. Arles, G. J. Boons, Tetrahedron Lett., 1998, 39, 9801-9804. Perkin Trans. 1997, I , 3357-3360. 33. R. R. Gibson, R. P. Dickinson, G. J. Boons, J. Chem. SOC., 34. Y. D. Vankar, P. S. Vankar, M. Behrendt, R. R. Schmidt, Tetrahedron, 1991,47, 9985-9992. 35. Y. Osa, K. Takeda, T. Sato, E. Kaji, Y. Mizuno, H. Takayanagi, Tetrahedron Lett., 1999, 40, I53 1-1 534. 36. C. H. Marzabadi, R. W. Franck, J. Chem. Soc., Chem. Commun., 1996, 2651L2652. 37. R. W. Franck, C. H. Marzabadi, J. Org. Chem., 1998,63, 2197-2208. 38. M. L. Sznaidman, S. C. Johnson, C. Crasto, S. M. Hetch, J. Org. Chem., 1995, 60, 39423943. See also the preparation of disaccharides by cycloaddition to sugar cyclic ketene acetals: S. C. Johnson, C. Crasto, S. M. Hetch, J. Chem. Soc., Chem. Commun., 1998, 1019-1020. 39. E. Fischer, Chem. Ber., 1893, 2400-2412. 40. See e.g.: a) A. A. Pavia, J. M. Rocheville, S. N. Ung-Chhun, Carbohydr. Res., 1980, 79, 7989; b) A. A. Pavia, S. N. Ung-Chhun, Can. J. Chem., 1981, 59, 482-489 (synthesis of trehaloses); c) A. Lubineau, J. C. Fischer, Synth. Commun., 1991,21, 815-818 (glycoside synthesis). 41. H. P. Wessel, J. Carbohydr. Chem., 1988, 7, 263-269. 42. P. Konraddson, C. Roberts, B. Fraser-Reid, Reel. Trav. Chim. Pays-Bas, 1991, 110, 23-24. 43. S. Horito, M. Tada, H. Hashimoto, Chem. Lett., 1991, 117-120. 44. B. Fischer, A. Nudelman, M. Ruse, J. Herzig, H. E. Gottlieb, J. Org. Chem., 1984, 49, 49884993. 45. J. Defaye, A. Gadelle, C. Pedersen, Carbohydr. Rex, 1989, 186, 177-188. 46. a) H. Uchiro, T. Mukaiyama, Chem. Lett., 1996, 79-80; b) H. Uchiro, T. Mukaiyama, Chem. Lett., 1996, 271-272 47. K. Takeuchi, S. Higuchi, T. Mukaiyama, Chem. Lett., 1997, 969-970. 48. a) K. Tomooka, Y. Nakamura, T. Nakdi, Synlett, 1995, 321-322; b) T. Jyojima, N. Miyamoto, Y. Ogawa, S. Matsumura, K. Toshima, Tetrahedron Lett., 1999, 40, 5023-5026. 49. K. Toshima, H. Nagai, S. Matsumura, Synlett, 1999, 1420-1422. 50. J. Inanaga, Y. Yokoyama, T. Hanamoto, J. Chem. Soc., Chem. Commun., 1993, 1090-1091. 51. H. Susaki, Chem. Pharm. Bull., 1994, 42, 1917-1918. 52. T. Mukaiyama, K. Matsubara, M. Hora, Synthesis, 1994, 1368-1373. 53. H. Uchiro, K. Miyazaki, T. Mukaiyama, Chem. Lett., 1997, 403-404. 54. T. Mukaiyama, S. Suda, Chem. Lett., 1990, 1143-1146. 55. S. Suda, T. Mukaiyama, Chem. Lett., 1991, 431-434. 56. T. Mukaiyama, K. Matsubara, S. Suda, Chem. Lett., 1991, 981-984. 51. B. A. Garcia, J. L. Poole, D. Y. Gin, J. Am. Chem. Soc., 1997,119, 7597-7598. 58. S. Koto, N. Morishima, S. Zen, Chem. Lett., 1976, 1109-1110. 59. S. Koto, N. Morishima, S. Zen, Bull. Chem. Soc. Jpn., 1982, 55, 1543-1547.

References

235

60. S. Koto, N. Morishima, C. Kusuhara, S. Sekido, T. Yoshida, S. Zen, Bull. Chem. Soc. Jpn., 1982,55,2995-2999. 61. S. Koto, T. Sato, N. Morishima, S. Zen, Bull. Chem. Soc. Jpn., 1980, 53, 1761-1762. 62. a) N. Morishima, S. Koto, S. Zen, Chem. Lett., 1982, 1039-1040; b) S. Koto, N. Morishima, M. Owa, S. Zen, Carhohydr. Res., 1984, 130, 73-83. 63. W. Szeja, Synthesis, 1988, 223-224. 64. M. Hirooka, S. Koto, Bull. Chem. Soc. Jpn., 1998, 71, 2893-2902. 65. M. Wakao, Y. Nakai, K. Fukase, S. Kusumoto, Chem. Lett., 1999, 27-28. 66. G. Grynkiewicz, Carbohydr. Rex, 1977, 53, CI 1-C12. 67. a) P. J. Garegg, T. Iversen, T. Norberg, Carbohydr. Res., 1979, 73, 313-314; b) K. Akerfeldt, P. J. Garegg, T. Iversen, Acta Chem. Scand. B, 1979, 33, 467-468. 68. A. Lubineau, E. Meyer, P. Place, Carbohydr. Rex, 1992, 228, 191-203. 69. a) N. Chida, M. Ohtsuka, S. Ogawa, Chem. Lett., 1988, 969-972. 70. a) W. R. Roush, X. F. Lin, J. Org. Chem., 1991,56,5740-5742; b) W. R. Roush, X. F. Lin, J. Am. Chem. Soc, 1995,117, 2236--2250. 71. A. Dondoni, A. Marra, M. C. Scherrmann, A. Casnati, F. Sansone, R. Ungaro, Chem. Eur. J., 1997, 3, 1774-1782. 72. G. T. Badmdn, D. V. S. Green, M. Voyle, J. Oryanomet. Chem., 1990, 388, 117-121. 73. A. Sobti, G. A. Sulikowski, Tetrahedron Lett.: 1994, 35, 3661-3664. 74. a) A. B. Smith, 111, K. J. Hale, R. A. Rivero, Tetrahedron Lett., 1986, 27, 5813-5816; A. B. Smith, 111, R. A. Rivero, K. J. Hale, H. A. Vaccaro, J. Am. Chem. Soc, 1991, 113, 2092-2112. 75. W. A. Szarek, H. C. Jarpell, J. K. N. Jones, Curhohydr. Res., 1977, 57, C13-CI6. 76. A. Bouali, G. Descotes, D. F. Ewing, A. Grouiller, J. Lefkidou, A. D. Lespinasse, G. Mackenzie, J. Curbohydr. Chem., 1992, 11, 159-169. 77. K. C. Nicolaou, R. D. Groneberg, J. Am. Chem. Soc, 1990, 112, 4085-4086. 78. L. F. Tietze, R. Fischer, H. J. Guder, Tetrahedron Lett., 1982, 23, 4661L4664. 79. E. M. Nashed, C. P. J. Glaudemans, J. Org. Chem., 1989, 54, 6116-6118. 80. D. X. Qiu, F. Y. Wang, M. S. Cai, Syntli. Coinmun., 1989, 19, 3453-3456. 81. a) C. Kolar, G. Kneissel, Angew. Chern. Int. Ed. Engl., 1990, 29, 809-810; b) C. Kolar, G. Kneissel, U. Knodler, K. Dehmel, Curhohydr. Rex, 1991, 209, 89-100. 82. W. Priebe, G . Grynkiewicz, N. Neamati, Tetruhrdron Lett., 1991, 32, 2079-2082. 83. T. Mukaiyama, K. Matsubara, Chem. Lett., 1992, 1041- 1044. 84. A. Kirschning, G. Chen, Tetrahedron Lett., 1999, 40, 4665~-4668. 85. B. Helferich, E. Shimitz-Hillebrecht, Chem. Ber., 1933, 66, 378--383. 86. K. Toshima, K. Tatsuta, Chem. Rev.,1993, 93, 1503-1531 87. a) R. U. Lemieux, W. P. Shyluk, Can. J. Chem., 1953, 31, 528-535; b) S. Hanessian, Y. Banoub, Curhohydr. Res., 1977, 59, 261-267. 88. a) M. Kiso, L. Anderson, Carhohydr. Res., 1979, 72, C15-CI7; b) L. M. Lerner, Carhohydr. Rex 1990,207, 138-141. 89. J. Dahmen, T. Frejd, G. Magnusson, G. Noori, Carbohydr. Res., 1983, 114, 328-330. 90. T. Ogawa, K. Beppu, S. Nakabayashi, Carbohydr. Res., 1981, 93, C6--C9. 91. T. Mukaiyama, S. Kobayashi, S. Shoda, Chem. Lett., 1984, 907-910. 92. T. Mukaiyama, T. Shinipuku, T. Takashima, S. Kobayashi, Chem. Lett., 1989, 145-148. 93. a) T. Mukaiyama, T. Takashima, M. Katsurada, H. Aizawa, Chem. Lett., 1991, 533-536; b) T. Mukaiyama, M. Katsurada, T. Takashima, Chem. Lett., 1991, 985-988. 94. K. Matsubara, T. Sasaki, T. Mukaiyama, Chem. Lett., 1993, 1373-1376. 95. T. Mukaiyama, M. Katsurada, Y. Iida, Chem. Lett., 1992, 2105-2108. 96. T. Mukaiyama, K. Matsubara, Chem. Lett., 1992, 1755-1758. 97. a) T. Mukaiyama, N. Shimomura, Chem. Lett., 1993, 781-784; b) N. Shimomura, M. Saitoh, T. Mukaiyama, Chem. Lett., 1994, 433-436. 98. A. K. Pathak, Y. A. El-Kattan, N. Bansal, J. A. Maddry, R. C. Reynolds, Tetrahedron Lett., 1998,39, 1497-1500. 99. T. Mukaiyama, T. Sasaki, E. Iwashita, K. Matsubara, Clzem. Lett., 1995, 455--456. 100. T. Vuljanic, J. Kihlberg, P. Somfai, J. Org. Chem., 1998, 63, 279--286. 101. K. Toshima, Y . Nozaki, S. Mukaiyama, K. Tatsuta, Tetrahedron Lett., 1992, 33, 1491-1494. 102. Y. Inanaga, Y. Yokoyama, T. Hanamoto, Tetrahedron Lett., 1993,34, 2791-2794.

236 103. 104. 105. 106.

9 Other Methods of Glycosylation

K. M. Nicholas, Acc. Chem. Res., 1987, 20, 207-214. C. Mukai, T. Itoh, M. Hanaoka, Tetrahedron Lett., 1997, 38, 4595-4598. F. Nicotra, L. Panza, A. Romano, G. Russo, J. Curbohydr. Chem., 1992, 11, 397-399. a) T. Mukaiyama, K. Miyazaki, H. Uchiro, Chem. Lett., 1998, 635-636; b) T. Mukaiyama, Y. Wakiyama, K. Miyazaki, K. Takeuchi, Chem. Lett., 1999, 933-934. 107. T. Iimori, T. Shibazaki, S. Ikegami, Tetruhedron Lett., 1996, 37, 2267-2270. 108. G. Scheffler,R. R. Schmidt, Tetrahedron Lett., 1997,38, 2943-2946. 109. T. Iimori, I. Azumaya, T. Shibazaki, S. Ikegami, Heterocycles, 1997, 46, 221-224. 110. I. Azumaya, T. Niwa, M. Kotani, T. Iimori, S. Ikegami, Tetrahedron Lett., 1999, 40, 46834686. 111. a) M. J. Ford, S. V. Ley, Synlett, 1990, 255-256; b) M. J. Ford, J. G. Knight, S. V. Ley, S. Vile, Synlett, 1990, 331-332. 112. S. Hanessian, C. Bacquet, N. Lehong, Curbohydr. Res., 1980, 80, C17-C22. 113. K. Koide, M. Ohno, S. Kobayashi, Tetrahedron Lett., 1991, 32, 7065-7068. 114. H. Furukawa, K. Koide, K. Takao, S. Kobayashi, Chem. Pharm. Bull., 1998,46, 1244-1247. 115. a) S. Hanessian, J. J. Conde, B. Lou, Tetrahedron Lett., 1995, 36, 5865-5868; b) S. Hanessian, J. J. Conde, H. H. Khai, B. Lou, Tetrahedron, 1996, 52, 10827-10834. 116. H. Kunz, J. Zimmer, Tetrahedron Lett., 1993, 34, 2907-2910. 117. J. C. Lopez, B. Fraser-Reid, J. Chem. Soc., Chem. Comrnun., 1991, 159-161. 118. H. Herzner, J. Eberling, M. Schultz, J. Zimmer, H. Kunz, J . Carbohydr. Chem., 1998, 17, 759-776. 119. N. K. Kochetkov, A. F. Bochkov, T. A. Sokolovskaya, V. J. Snyatkova, Curbohydr. Rex, 1971,16, 17-27. 120. Y. E. Tsvetkov, P. I. Kitov, L. V. Backinowsky, N. K. Kochetkov, Tetrahedron Lett., 1993, 34, 7977-7980. 121. W. Wang, F. Kong, J. Org. Chem., 1998,63, 5744-5745. 122. Y. Du, F. Kong, J. Carbohydr. Chem., 1999,18,655-666. 123. W. Wang, F. Kong, J. Curbohydr. Chem., 1999, 18, 451-460. 124. W. Wang, F. Kong, J. Org. Chem., 1999,64, 5091-5095. 125. W. Wang. F. Kong, Angew. Chem. Int. Ed. Engl., 1999,38, 1247-1250. 126. W. Wang, F. Kong, Tetrahedron Lett., 1998, 39, 1937-1940. 127. W. Wang, F. Kong, Curbohydr. Rex, 1999, 315, 117-127. 128. W. Wang, F. Kong, J. Curboydr. Chem., 1999,18, 263-273. 129. T. Iimori, H. Ohtake, S. Ikegami, Tetrahedron Lett., 1997,38, 3413-3414. 130. T. Iimori, H. Ohtake, S. Ikegami, Tetrahedron Lett., 1997,38, 3415-3418. 131. H. Ohtake, T. Iimori, S. Ikegami, Synlett, 1998, 1420-1422. 132. S. Hiranuma, 0. Kanie, C.-H. Wong, Tetrahedron Lett., 1999, 40, 6423-6426. 133. F. Nakatsubo, H. Kamitakahara, M. Hori, J. Am. Chem. Soc., 1996,118, 1677-1681. 134. M. Hori, H. Kamitakahara, F. Nakatsubo, Macromolecules, 1997, 30, 2891-2896. 135. a) K. Kurita, M. Kobayashi, T. Munakata, S. Ishii, S. Nishimura, Chem. Lett., 1994, 20632066; b) K. Kurita, H. Akao, M. Kobayashi, T. Mori, Y. Nishiyama, Polymer Bull., 1997,39, 543-549; c) K. Kurita, K. Shimada, Y. Nishiyama, M. Shimojoh, S. Nishimura, Mucromolecules, 1998, 31, 4764-4769. 136. a) L. Lay, L. Panza, G. Russo, D. Colombo, F. Ronchetti, E. Adobati, S. Canevari, Helu. Chim. Actu, 1995, 78, 533-538; b) D. Colombo, L. Panza, F. Ronchetti, Curbohydr. Rex, 1995,276, 437-441; c) J. Banoub, P. Boullanger, D. Lafont, Chem. Rev., 1992, 92, 1167-1 195 and references cited therein. 137. P. Brusca, 0. R. Martin, Tetrahedron Lett., 1998, 39, 8101-8104. 138. a) J. Kadokawa, Y. Watanabe, M. Karasu, H. Tagaya, K. Chiba, Mucromol. Rapid Commun., 1996,17, 367-372; b) J. Kadokawa, S. Kasai, Y. Watanabe, M. Karasu, H. Tagaya, K. Chiba, Macromolecules, 1997, 30, 8212-8217. 139. J. Xia, C. F. Piskorz, R. D. Locke, E. V. Chandrasekaran, J. L. Alderfer, K. L. Matta, Bioorg. Med. Chem. Lett., 1999, 9, 2941L2946. 140. S. Kobayashi, T. Kiyosada, S. Shoda, Tetrahedron Lett., 1997, 38, 21 11-21 12. 141. S. Hashimoto, T. Honda, S. Ikegami, Tetruhedron Lett., 1991, 32, 1653-1654. 142. S. Hashimoto, T. Honda, S. Ikegami, Tetrahedron Lett., 1990, 31, 4769-4772.

References

237

143. S. Hashimoto, Y. Yanagiya, T. Honda, H. Harada, S. Ikegami, Tetrahedron Lett., 1992, 33, 3523-3526. 144. D. R. Mootoo, P. Konradsson, U. Udodong, B. Fraser- Reid, J. Am. Chem. Soc., 1988, 110, 5583-5584. 145. S. Hashimoto, H. Sakamoto, T. Honda, S. Ikegami, Tetrahedron Lett., 1997,38, 5181-5184. 146. S. Pan, H. Li, F. Hong, B. Yu, K. Zhao, Tetrahedron Lett., 1997, 38, 6139-6142. 147. M.-J. Chen, K. Ravindran, D. Landry, K. Zhao, Heterocycles, 1997, 45, 1247- 1250. 148. H. Li, M.-J. Chen, K. Zhao, Tetrahedron Lett., 1997, 38, 6143-6144. 149. D. Niu, M.-J. Chen, H. Li, K. Zhao, Heterocycles, 1998, 48, 21-24. 150. G. Singh, I. Tranoy, Carbohydr. Lett., 1998, 3, 79 -84. 151. H. Bielawska, M. Michalska, J. Carbohydr. Chem., 1991, 10, 107-112. 152. J. Borowiecka, Liebigs Ann., 1997, 2147-21 50. 153. R. Eby, C. Schuerch, Carbohydr. ReJ., 1982, 102, 131-138. 154. D. Crich, W. Cai, J. Ory. Chem., 1999, 64, 4926-4930. 155. L. Cipolla, L. Lay, F. Nicotra, L. Panza, G. Russo, Tetrahedron Lett., 1994,35, 8669-8670. 156. A. Meslouti, D. Beaupere, G. Demailly, R. Uzan, Tetrahedron Lett., 1994, 35, 3913-3916. 157. W. J. Sanders, L. L. Kiessling, Tetrahedron Lett., 1994, 35, 7335-7338.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

10 Polymer-Supported Synthesis of Oligosaccharides Jiri J. Krepinsky and Stephen P. Douglas

10.1 Introduction Among the variety of biological activities of oligosaccharides, antiinfective effects are the reason some oligosaccharides are close to a practical use [ 11, provided that the oligosaccharides can be cleanly made on large scale. Glycosylations by enzymic (cf. Part 3 of this Volume) and chemical methods compete to meet this challenge. Classical chemical methods of oligosaccharide synthesis invariably require extensive purification of the desired reaction products, mostly by chromatography, which makes chemical syntheses of oligosaccharides very laborious and expensive. Similar problems encountered in the syntheses of oligopeptides and oligonucleotides were dealt with successfully using polymer-supported synthetic and purification techniques. A polymer support facilitates quick and efficient purification of products assuming that these reactions proceed to completion. Although some indications suggest that polymer supports may influence the glycosylation reaction itself [2], the primary purpose of the polymer support is rapid purification after every step in the reaction sequence leading to the formation of an oligosaccharide [ 3 ] . Therefore, polymer supports will not make the chemistry of oligosaccharide synthesis any simpler which is unfortunate since this chemistry is complex and many details important for the synthesis remain unknown despite the rapid progress in this area during recent years documented, for instance, in this Volume. However, because of the significantly simpler purification, synthetic sequences become enormously faster. Furthermore, for syntheses of combinatorial oligosaccharide libraries, polymer support is essential since it allows purifications of the desired intermediates or products as groups, which is difficult to accomplish otherwise. Four factors combine to make oligosaccharide synthesis challenging: i) several hydroxyl groups of similar reactivity are present in monosaccharides, from which oligosaccharides are generally constructed and which almost invariably require laborious protection-deprotection schemes; ii) difficulty to form anomerically pure glycosidic linkages; iii) acid lability of the glycosidic linkages (hemi-acetals); iv)

240

10 Polymer-Supported Synthesis of Oligosaccharides

incompletely known intramolecular interactions. The most common glycosylation technique utilizes a nucleophilic substitution at the anomeric carbon with a hydroxyl group (“acceptor”) of the aglycon. In order to prevent other hydroxyls from taking part in this reaction, they must be protected [4] (and subsequently deprotected) by methods compatible with the labile glycosidic linkages [ 5 ] . As is common for acetal formation reactions, glycosylations are often both incomplete and accompanied by numerous side-reactions, so that chromatography is usually needed to clean desired products after each synthetic step. Therefore, the synthesis of an oligosaccharide in solution is extremely time consuming, costly, and environmentally unfriendly in part because of the large volumes of solvents used. Despite the ability of this methodology (c.f. this Volume) to provide almost any oligosacaccharide [6], if the chromatographic purification is the bottleneck step in the chemical oligosaccharide syntheses, removal of the requirement for chromatography may make chemical synthesis of oligosaccharides significantly more practical. Methods employing enzymes have been developed to overcome these difficulties (cf. Part 3 of this Volume). Glycosyltransferases [7] and glycosidases working in reverse [S], or combined enzyme and chemical strategies [9, 101 have been examined. Syntheses employing glycosyl transferases using soluble polymer supports, such as water-soluble polyvinyl alcohol [ 1I] and polyacrylamide [ 121, to which a monosaccharide was bound via a photolabile (O-nitrobenzyl) linker, also established that a spacer between the polymer support and the oligosaccharide must be of certain length so that the reactions can occur [13]. Despite considerable advantages, such as not needing a hydroxyl protection-deprotection scheme, enzymic syntheses have their own problems. One problem is the requirement of glycosyl transferases to employ relatively unstable and difficult to prepare intermediates. Furthermore, transferases cannot as yet be used to synthesize oligosaccharides unknown in Nature.

10.2 General Reflections Polymer supported designs of oligosaccharide synthesis began to be explored thirty years ago encouraged by the solution of analogous problems in the syntheses of oligonucleotides and oligopeptides by polymer-supported strategies. The spectacular success of solid-state syntheses of oligopeptides and oligonucleotides on polystyrene resins and controlled-pore glass is due not only to the polymer support but also to specifically designed chemical procedures to be performed on solid state support. The main benefit of procedures involving solid state reactants, i.e. the purification of desired intermediates from by-products, excess reagents and products of their decomposition by simple washing, is combined with prompt quenching of the reaction and fast work-up, eliminating overexposure of reaction products to excess reagents and by-products. The first polymer-supported oligosaccharide synthesis by Schuerch’s group employed the Koenigs-Knorr reaction and its variants, using an immobilized glycosyl acceptor on an organic solid polymer (Scheme 1) [ 141. It was followed by inves-

10.2 General Reflections

+

J

t

P

& O Q ! ! j

P

24 1

242

10 Polymer-Supported Synthesis of Oligosaccharides

tigations of an inorganic support, namely porous glass, zirconized and silanized [ 151. The results were not satisfactory; among the problems were decreased reaction rate and lower yields compared to solution strategies, incomplete coupling, and poor anomeric specificity. These problems seemed to be attributed to the solid state rather than its incompatibility with certain aspects of Koenigs-Knorr chemistry. Unfortunately, these difficulties stymied further progress in polymer-supported glycosylations, although the method was not entirely abandoned by van Boom [16, 171 and others [ 181, providing various oligosaccharides using Merrifield cross-linked polystyrene and controlled-pore glass. Although it was suggested that the use of a soluble polymer could improve the results [19], it took another twenty years until the first report in 1991 from Krepinsky’s group on successful1 applications of polyethyleneglycol as soluble support was published (Scheme 1) [20]. Since both Koenigs-Knorr (promoted by silver triflate) and trichloroacetimidate chemistry gave comparable results on polyethylene glycol, it seemed to confirm the belief that it was the solid state support, i.e. two phase condition, which failed glycosylations [20]. That this conclusion was erroneous was demonstrated when the glycal method of glycosylation developed by Danishefsky’s group 1211, was exploited by that group for the synthesis of oligosaccharides using immobilized glycosyl donors on polystyrene support 1221. This scenario is reported to provide excellent anomeric specificity although an exception was noticed [23, 241, which was subsequently remedied by a modification of the procedure 1251. Oligosaccharide chains can be extended (i) from the reducing end to the nonreducing terminus (using polymer-supported acceptors), as well as (ii) from the nonreducing end to the reducing terminus (using polymer-supported donors) (Scheme 2). Both strategies have serious problems due to incomplete glycosylations and/or donor decomposition. Assuming that a small portion (realistically perhaps 3-5%) of polymer-supported monosaccharide M1 acceptor fails to be glycosylated to a disaccharide D1, the disaccharide D1 in the subsequent step is extended to a trisaccharide T1, but a portion of the previously unglycosylated monosaccharide M 1 gives disaccharide D2. These second round glycosylations are again incomplete. After the third glycosylation the following compounds will be polymer-bound in addition to the desired tetrasaccharide: the monosaccharide M 1, disaccharides D 1, D2, and D3, and trisaccharides T1 and T2. After each additional glycosylation the mixture becomes even more complicated, yet all impurities originating from the donors are absent. Thus the desired product must be separated from the truncated sequences (= failures), after detachment from the polymer support. To alleviate the situation, excess glycosylation reactants are recommended to be applied in portions (Scheme 2), as is practiced to obtain “difficult sequences” in polymer-supported oligopeptide syntheses. Incomplete anomeric specificity causes further complications. Starting the synthesis from the nonreducing terminus, i.e. using a polymer-bound donor presents the same problem of truncated sequences. Since donors are usually quite reactive, a portion of a donor may be transformed into molecules which do not react anymore with acceptors and since they are fastened to the polymer, they are carried through the synthetic sequence (Scheme 2) and must be removed by chromatography at the end of the synthesis. This will remain problematic unless

10.2 General Reflections

k

\

4

I

4

0

\

243

244

10 Polymer-Supported Synthesis of Oligosaccharides

Glycal Assembly

0

2 Self-policingFailure

ZnClDHF Due to their relatively higher polarity the "failures" are easily separated after the synthesis by simple Chromatography.

Scheme 3. Two-phase polymer supported oligosaccharide synthesis using galactal donor attached

to an insoluble polymer.

stable donors are employed, which react specifically with acceptors. One solution has been developed by Danishefsky's group (cf. Schemes 3 and 4) [26]. Failures are self-policing in that they are easily removed at the end of the synthesis by simple chromatography owing to their highly polar nature [26]. An ingenious solution of this problem was developed by Ogawa's group (Scheme 5 ) [27]. The monosaccharide added last to the sequence is equipped with a hydrophobic group, such as 2-(trimethylsilyl)ethyl, which after detachment from the polymer permits isolation of the sequence containing the hydrophobic group by reversephase chromatography (investigated by Hindsgaul's group) [28], thus separating it from most failure sequences. As already mentioned, solid-state oligopeptide and oligonucleotide syntheses required the development of synthetic methods compatible with the solid state processes. This should be equally or more indispensable for oligosaccharide synthesis since syntheses of oligosaccharides are an even more daunting task than the syntheses of oligonucleotides or oligopeptides. This is actually happening, since in spite of this complexity, the progress in designing glycosylation methods is steadily accelerating.

10.2 General Reflections

245

Glycal Assembly

B

> opiv

0BnO 1

MeOTUDTBP

OPiv

BnO

OBn

I

=Bn*

--

BnO

,o& BnO

OH

opiv

Scheme 4. Two-phase polymer supported oligosaccharide synthesis using on-support transformation of glucal into thioglycoside donor attached to insoluble polymer.

+

E

" L & Y

Repeat

-

r

Y

Non-tagged impurities easily removed by chromatography Scheme 5. Polymer supported oligosaccharide synthesis utilizing hydrophobic handle attached in the final stage of the synthetic sequence. The handle allows to separate most of failure sequences accumulated during the synthesis.

246

10 Polymer-Supported Synthesis of Oligosuccharides

10.3 Polymer Supports Automated two phase polymer-supported syntheses of oligopeptides and oligonucleotides provide their products in micromolar to millimolar quantities because of the combination of loading capacity (the loading capacity is a number of anchoring sites in millimoles per gram of a support) of solid polymers (cross-linked polystyrene, controlled-pore glass, Tentagel, Argogel) and the limit of the support volume (mass) practically usable. Aside from mechanical stability of the support, the most important consideration is the chemical stability of both the product and the supporting system (i.e. the polymer and the tether). Since prolonged contacts with the reagents may induce decomposition of the polymer-bound product and the supporting system itself, the reactions should be as fast as possible. Given a small scale of the synthesis, the reagents can be used in a huge excess (e.g. hundredfold) to ensure rapid reactions. However, to synthesize larger quantities of shorter sequences it became necessary to employ polymeric supports which can be practicably handled in substantially larger volumes than the solid polymers, such as polyethyleneglycol o-monomethylether (MPEG) of m.w. = 5,000-6,000 [29, 301, albeit the loading capacity of MPEG at approximately 0.2 mmol/g is not much different from loading capacities of polystyrene-based supports. MPEG is soluble in most solvents including water and thus the polymer-bound synthon can be dissolved under many reaction conditions. After completion of the reaction an ether such as methyl t-butyl ether is poured into the reaction mixture and the polymer-bound product precipitates because MPEG is insoluble in ethers. The remaining components of the reaction mixture can be removed from the desired MPEG-bound product by simple filtration (Scheme 1). One of the original arguments for using a soluble polymer support for oligosaccharide syntheses was based on the necessity to obtain oligosaccharides in larger quantities because biological effects are only observed in higher concentrations. Furthermore, it was assumed that glycosylations on soluble supports are more easily controlled than those immobilized on insoluble matrices, and that all reactants in solution would allow reaction kinetics and anomericity control similar to those observed in solution chemistry [19, 20, 311. Fortunately these arguments did not prevent solid supports from being investigated. Thus it was shown in quick succession that almost any solid support used previously for oligopeptide and oligonucleotide syntheses could be employed for oligosaccharide syntheses (Table 1): polystyrene crosslinked with divinylbenzene (1-2%) [21] usually derived from “Merrifield and Wang resins” [32] or Rink resin [33], polystyrene with grafted polyethylene glycol (TentaGel [34], Argogel [35]), and controlled-pore glass (CPG) [36], as well as novel supports such as polystyrene crown [37], in combination with a suitable chemistry and a tether between the polymer and oligosaccharide chain. Now it appears that there are no significant differences between glycosylations performed in solution without a polymer support, with a soluble polymer support, or using a solid polymer support [38], although some variations were observed 12). It should be noted, of course, that insoluble catalysts are much more likely to be

10.4 One-Phase Systems (Syntheses in Solution)

247

Table 1. Polymer supports showing the functionality to which linkers are attached. Supports for one-phase (solution) syntheses are marked with an asterisk (*); the remainder are supports for twophase syntheses. Except more heavily crosslinked macroporous polystyrene, all supports for two phase syntheses are 1-2% cross-linked polystyrene-divinylbenzene.The first four resins (Merrifield, Wang, Rink, aminomethyl) have loading capacity 0.1-0.3 mmol/g, and swell in solvents such as dichloromethane, methanol, dimethylformamide, and pyridine. Macroporous polystyrene has loading capacity 0.3-0.6 mmol/g, and maintains constant pore size in most solvents, including water. In both Tentagel and Argogel, swelling in organic solvents, polyethylene glycol is grafted on 1-2% cross-linked polystyrene-divinylbenzene with loading capacity 0.3 mmol/g. Argogel is claimed to be more chemically stable. Soluble Polvstvrene* Aminornew

Polvethvlene Glvcol*

Merrfield

\, TentaGelTM

X= OH, NHz, others

effective in association with a soluble support, which can be considered a protecting group, as exemplified by the removal of boronate diester from 4,6-positions of galactose by Amberlite IRA-743 [39].

10.4 One-Phase Systems (Syntheses in Solution) In the synthesis of oligosaccharides, a soluble polymer is bound to the first member of the oligosaccharide chain by a linker permitting removal of the polymer from the

248

I 0 Polymer-Supported Synthesis of Oligosaccharides

completed oligosaccharide, after the completion of the synthesis. Most physical properties of such synthons, are dictated by the polymer and they are not much different from those of the polymer, unless oligosaccharides on the polymer become very large. The polymer can be linked to either the acceptor or the donor. Polyethyleneglycol o-monomethylether (MPEG) and linear polystyrene have been investigated, although the latter has been used only infrequently (Table 1). Polyethyleneglycol = ,-monomethylether (MPEG)

MPEG of different average molecular weights is commercially available [40]. MPEG of average m.w. = 5,000 are most frequently used, but m.w. = 2,000 and m.w. = 12,000 were used as well. Depending on size, MPEG forms rigid rod-like structures stabilized by intramolecular interactions [41]. It is soluble in most apolar solvents (benzene, toluene, methylene chloride, chloroform, acetonitrile, acetone) as well as in water, and insoluble in hexane, diethylether and t-butylmethyl ether, and cold ethanol. It can be dissolved in hot tetrahydrofurane and ethanol, therefore it is commonly purified by reprecipitation from ethanol. Otherwise it is purified by precipitation with an ether from its solutions in the organic solvents cited above. Although MPEG is strongly hygroscopic, the adsorbed water can be almost completely removed by azeotropic distillation, e.g. with toluene or benzene; wet MPEG precipitates with ether only reluctantly. t-Butylmethyl ether is safer to use for the precipitation than diethylether since its boiling and flash points are higher [42]. Recovery of MPEG derivatives is usually no less than 95%. This recovery is substantially lower when MPEG is complexed with metal cations (e.g. mercury, chromium, cerium). Therefore many common metal-containing reagents cannot be used with substrates containing MPEG. MPEG also precipitates less easily and is more difficult to filter if it carries higher molecular weight synthons. Precipitations with alcohols (e.g. methanol) give similar results. To improve precipitation properties of MPEG when the synthesized oligosaccharide extends beyond five monosaccharides and its protected form approaches m.w. 5,000, a suitable remedy is to use MPEG of average mw = 12,000 [43]. Also, shorter MPEG of average mw = 2,000 was used in some cases, and instead of the precipitation a filtration through a silica gel column was used for purification [44]. Although MPEG is stable enough under many reaction conditions that it can be considered a protecting group, some conditions cause its partial degradation [29]; for instance pyridinium salts peel off one or more ethylene glycol units [45]. Furthermore, MPEG cannot be located directly on the anomeric carbon. Treatment of an MPEGyl glycoside, e.g. 2,3,4,6-tetra-O-acetyI-~-D-galactoside with a glycosylating agent such as acetobromomannose causes a glycosyl exchange to take place is formed, and soluble glycolyl and MPEGyl 2,3,4,6-tetra-O-acetyl-a-D-mannoside or (bis)glycolyl 2,3,4,6-tetra-O-acetyl-p-D-galactoside can be isolated from the mother liquors [46] after the MPEG-bound saccharide is precipitated (Scheme 6). The solubility of MPEG in organic solvents allows monitoring of the progress of the synthesis by high field 'HNMR spectroscopy using the signal of its single OCH3 group (6 = 3.380 ppm) as internal standard. Since the glycol peak of MPEG at

10.4 One-Phase Systems (Syntheses in Solution)

249

AcO-7

Ace% AcO

OAc

OCH~C€IZ(OCH~CH~),,OCH~+

AcoA%AcO X

1 AcoA* AcO

Promoter

+

Scheme 6. An example of instability of polyethylene glycol directly attached to the anomeric carbon.

6 = 3.640 ppm is very strong, it must be suppressed by irradiation for 3.0 s to increase the dynamic range and obtain a good quality spectrum. Despite this, sufficient structural information to identify the product can be usually obtained from its 'HNMR spectrum. Linear Polystyrene As with MPEG, linear polystyrene is soluble in benzene, toluene, chloroform, and insoluble in alcohols (methanol) and ethers (e.g. 2-methoxyethyl ether). The first monosaccharide was linked via thioglucoside to chloromethylated polystyrene (prepared from the polymer using chloromethyl methyl ether in the presence of stannic chloride) in benzene solution, and subsequent glycosylations were also performed in benzene. Polymer bound disacharides were isolated by precipitation using 2-methoxyethyl ether or methanol. Although the original report seemed unpromising [ 18b], an application of recent methods of oligosaccharide synthesis is desirable since linear polystyrene should be stable under conditions incompatible with the use of MPEG. Linear polystyrene to which a protected monosaccharide was bound directly was also prepared by copolymerization of styrene with four different derivatives of protected glucose substituted on C-6 with 4-vinylbenzoate, and some reactions have been investigated [ 19a].

250

10 Polymer-Supported Synthesis of Oligosaccharides

10.4.1 Linkers First members of growing oligosaccharide chains usually are not bound to polymeric supports directly but through a linker. Two major reasons for this arrangement are that a too short span between the carbohydrate and the support may create interactions detrimental to the course of the synthesis (however, some reactions may be improved [2]). Thus the tether maintains a safe distance between the support and the growing oligosaccharide chain. The major reason for inserting a tether is that it allows for an easy release of the carbohydrate from the polymer support. Usually the direct bond between the saccharide and the support would be stronger than the glycosidic bonds between two monosaccharides making impossible the removal of the oligosaccharides from the supports. Should the bond between a saccharide and the polymer involve the anomeric carbon, this bond often would be too labile to tolerate all reactions required by the synthetic sequence without a partial decomposition of this bond. Several linkers were examined for one phase systems (Table 2). The choice of a linker is determined by the reactions employed in the reaction sequence and on the form of the final oligosaccharide, which further depends on the eventual use of the oligosaccharide and the possible utilization of the linker or a part of it. As in any organic chemical synthetic scheme, reagents and protecting groups must act in harmony; soluble polymers-linkers can be regarded as special protecting groups. The following examples will illustrate such criteria. Succinoyl Diester The extensive use of succinoyl diester in polymer-supported solution syntheses of oligopeptides and oligonucleotides invited its application in polymer-supported synthesis of oligosaccharides. Since it is essential that both the protective groups and the linker are stable during the synthetic sequence, it is obvious that the baselability of ester linkages prohibits temporary protection with ester groups to be removed before the completion of the synthesis. Particularly unstable are ester groups located on anomeric carbons. Moreover, suitably located succinoyl-MPEG groups are prone to migrations in acidic environment, for instance 2-O-SuMPEG+ 3-0-SuMPEG in acceptor P-mannosides with OH-3 unprotected [47]. Despite these restrictions the acceptor-bound succinoyl-MPEG [48] was found to permit glycosylations with glycosyl trichloroacetimidates promoted by boron trifluoride, triflic anhydride, and trimethylsilyl or triethylsilyl triflates, with glycosyl halogenates promoted by silver triflate, and with thioglycosides promoted by the iodonium ion. It is compatible with ester long-term protection, with allylic, benzylic type (including benzylidene-type acetals), and certain silicon-based protecting groups, and deprotection with hydrogenolysis. To synthesize an oligosaccharide, MPEG is bound to a hydroxyl of the first monosaccharide through the succinoyl diester linker (Scheme 1). Next, another hydroxyl of this monosaccharide is deprotected, glycosylated, temporary protection on the second monosaccharide is removed and the above process is repeated until the required oligosaccharide is synthesized. Although incomplete glycosylation can

0

-S-Polystyrene

-0CHz \ /-CHzO-PEG

CH&BnOH

3) Acetolytic PEG Removal with Sc(OTO3/AczO

Removal

2) Hydrogenolytic PEG-DOX

1) Reductive PEG Removal

'Abbreviations. Donors: A= Trichloroacetimidate; B= Bromide; C= Sulfoxide; D= ThioalkyVaryl; E= Fluoride

7) Thio

6 ) DOX

Removal

!) Hydrogenolysisfor Complete

Removal

18b

55

54

54

27

0

) NaOMdMeOH for PEG

5 ) Benzylphenol

53

TEA/DCM

44

Dioxirane oxidation then NaOMmeOH

4) Fluorene

2,209 49, 51,52,91

Reference

Base-catalysed (DBU; NaOMe)

Removal

44

Donor(s) Used'

Hydrogenolysis

-0CHZ \ 1-O(CH&COO-PEG

-OCO(CHz)zCO-PEG

Structure

3) Phenylacetamide

2) Thioethanol

1) Succinoyl

Linker

~

Table 2. Major linkers used in one-phase (solution) oligosaccharide synthesis showing chemical structure of the linker, glycosylating agent. and removal methods used.

252

10 Polymer-Supported Synthesis of Oligosaccharides

be minimized by using glycosylating agents in excess, it is probably more efficient to add an excess of the glycosylation agent in several portions. The synthesis of the methyl heptaglucoside 1 is an excellent example of a regioselective glycosylation of the primary hydroxyl in 4,6-diol using succinoylated MPEG and iodonium promoted glycosylations with thioglycosides [49]. After the glycosylation, the unreacted OH-4 was capped by acetylation. A comparison with a classical solution synthesis [50]confirmed, as expected, that the MPEG-supported synthesis was significantly faster. Another example of a synthesis using MPEG-Subound acceptor are syntheses of heparan sulphate-like oligomers up to a dodecamer [51]. MPEGSu- can be also linked to a donor as shown in the synthesis of an oligomer of N-acetylneuraminic acid [521. Succinoyl diester has been further modified by linking to 9-hydroxymethylfluorene (Table 2) and using amino MPEG (i.e. MPEG terminating with amino group). Increasing sensitivity of the linker towards bases, this modification made it possible to distinguish between base-labile long-term protecting groups such as pivaloyl on the oligosaccharide while removing it from the polymer-linker with a mild base [53]. Dioxyxylyl Diether (DOX) The base-lability of succinoyl diester linker severely limits the selection of protecting groups available for an oligosaccharide synthesis, so a more versatile tether was required. A diether bond of a,a'-DiOxyXylyl diether, - O C H Z C ~ H ~ C H ~(DOX) O-, [54] is not limited by restriction of the succinoyl linker: (i) bound via a hydroxyl or as an 0-glycoside, it is stable under many reaction conditions including glycosylation; (ii) it is easily removable from a finished oligosaccharide by hydrogenolysis under mild acidic condition; (iii) using suitable hydrogenolytic conditions MPEG only is removed and DOX is converted into 4-methylbenzyl protecting group; (iv) ,using scandiumtictriflate/aceticanhydride[551 DOX can be converted into 4(acetoxymethy1)benzyl protecting group by removal of MPEG only (Scheme 7); (v) it is easily accessible from a,a'-dichloro-p-xylene. The MPEG-DOX grouping is readily synthesized from MPEG-OH and c1,a'dichloro-p-xylene by Williamson ether synthesis yielding MPEG-DOX-CI, and, after hydrolysis, the alcohol MPEG-DOX-OH [ 541. MPEG-DOX-OH can be

10.4 One-Phase Systems (Syntheses in Solutiorz)

Carbohydrate Portion

253

QiQxy Xylyl Portion u MetbylPEG Portion

Carbohydrate-OCH

HJH+

AczO/DCM Carbohydrale-OH

I?$?

CH,C,

,CCH,

Carbohydrate-OCH,

Scheme 7. An example of a versatile linker that can be transformed during or at the end of a synthetic sequence into other synthetically useful forms.

glycosylated to attach a saccharide at the anomeric position (an impossible connection through succinoyl tether [20, 461) to create an MPEG-supported acceptor. In a practical example (Scheme l), the MPEG-DOX-OH is glycosylated by 2-0acety1-3,4,6-tri-O-benzyl-a-D-mannopyranosyl trichloroacetimidate. The 2-0-acetyl group as a short-term protecting group can be easily deacetylated to a hydroxyl acceptor for the next glycosylation; it also serves for steric control of glycosylation via neighbouring group participation. Repetition of the glycosylation and hydrolytic steps gave the expected protected MPEG-DOX-bound pentasaccharide. Standard deprotection including the removal from the polymer support yielded the desired pentamannopyranoside [Manp(al-2)]4Manp(2). Target oligosaccharides are often contaminated with impurities originating in deprotection steps which usually can be removed by chromatography of peracetylated oligosaccharides. Peracety-

254

10 Polymer-Supported Synthesis of Oligosaccharides

lated oligosaccharides are commonly more suitable for final purification by chromatography than deprotected compounds; after deacetylation pure oligosaccharides are obtained readily [56]. The attachment of MPEG-DOX to a carbohydrate hydroxyl other than the anomeric one is achieved by the reaction of such hydroxyl with MPEG-DOX-Cl by Williamson ether synthesis [541. Syntheses of oligosaccharides on MPEG, tethered with dioxyxylyl (DOX) and related 4-oxybenzyloxy diethers, can be performed even more speedily using orthogonal glycosylation [27] and multicomponent carbohydrate [ 571 coupling strategies, which are described in detail elsewhere in this Volume. 10.4.2 Chemistry Investigations In addition to its use as polymer support for oligosaccharide syntheses, MPEGDOX-OH established itself as a useful aide for investigations of problems associated with glycosylations. This is possible because electrophiles competing for nucleophilic acceptors with glycosylating agents can be readily identified after precipitating products of reaction between such electrophiles and MPEG-DOX-OH. These competing reactions complicate results of glycosylations including those polymer supported; therefore it is urgent to suppress them as much as possible. Better understanding of the competing reactions makes this objective realizable. Thus it was established that the trans (p) anomeric specificity of some 2-acetamido glycosylating agents does not require oxazoline intermediacy, and that formation of oxazolines represents a side reaction. It was further found that oxazoline cannot be an obligatory intermediate since MPEG decomposes under the conditions required

10.5 Two-Phase Systems (Syntheses on Solid Supports)

I

BnTBnO

OBnOBn

O

A

’h3q

TBSO

TBSO

SMe

Ph-

255

BnO SMe

OBnOBn

OH

Scheme 8. “Reverse” use of a polymer support for the synthesis of P-mannoside linkage. The synthetic helper arm (MPEG) is removed from the desired product by simple precipitation.

by oxazoline intermediacy [58]. The origins of acyl transfer from 2-acyloxy glycosylating agents to acceptors [59a] have been studied in combination with calculations based on density functional theory [59b] in order to eliminate this very common side reaction. Novel reactants in glycosylation reactions can be efficiently evaluated using MPEG-DOX-OH as illustrated by the investigation of dimethylboron triflate as a promoter [60]. The formation of the P-mannoside linkage by intramolecular aglycon delivery [61] was modified to be performed on MPEG support: in the last step of the synthesis the j3-mannoside remains in solution after precipitation of MPEG-bound byproducts (Scheme 8) [62].

10.5 Two-Phase Systems (Syntheses on Solid Supports) Polymers used as a support for oligosaccharide synthesis should be commercially available or easily modifiable, for instance by introduction of spacers and linkers. The supports should be both mechanically and chemically stable, and have reasonable loading capacity. Three main types of solid support for oligosaccharide syntheses were essentially “borrowed” from oligonucleotide and oligopeptide solid state syntheses (cf. Table 1): (i) functionalized controlled pore glass (CPG); (ii) functionalized polystyrene cross-linked with divinylbenzene (e.g.: Merrifield, Wang, Rink resins); and (iii) crosslinked polystyrene resin with grafted functionalized polyethylene chains (TentaGel, Argogel). The last two types of supports (ii and iii) were compared as to their reaction kinetics. It was concluded that, contrary to the

256

I0 Polymer-Supported Synthesis of Oligosaccharides

common belief, reactions on PEG-containing beads do not proceed faster and that the kinetics depends on the reaction rather than on the support [63]. Compared to one-phase syntheses, monitoring progress of the synthesis of twophase syntheses presents a problem because of the insolubility of the support. Usually the product is detached from a portion of the polymer-bound material and analyzed by convential techniques, such as TLC, NMR spectroscopy and mass spectrometry. Monitoring of the progress of the synthesis without the loss of material by high resolution magic angle spinning NMR spectroscopy of solids is an important development [641. Controlled Pore Glass This has become the support of choice for oligonucleotide syntheses. It is mechanically quite sturdy, does not change its volume in different solvents, and it withstands fairly strong chemical conditions to which it is exposed for a short time (for instance, in an oligonucleotide synthesizer. However, it becomes unstable on prolonged exposure to bases and acids [65] and has a very low loading capacity. Cross-Linked Polystyrene This support was well tested for oligopeptide syntheses, and this led to its adoption for solid state synthesis of oligosaccharides. It has to be preswollen before the synthesis, and it also changes its volume during the synthesis as the synthesized chain becomes more voluminous. It is acid and base resistant, and can be used with linkers cleavable by photolysis, oxidation, fluorolysis, and several other conditions (Table 3). However, it does not swell in some solvents, for instance diethylether. Macroporous highly cross-linked poly(styrene-co-divinylbenzene) resins [661, such as Argopore [67],maintain a constant volume in organic solvents, including protic solvents, and even water. Polyethyleneglycol Grafts on Cross-Linked Polystyrene While the above mentioned supports are rigid, functionalized polyethyleneglycol chains on polystyrene in TentaGel [68] and Argogel [69] are highly mobile and can be considered to be in “solution” [70]. TentaGel (Table 1) is used extensively for oligonucleotide and oligopeptide syntheses; because of the polyethyleneglycol nature of the support, somewhat diminished stability towards acids would be expected. Because of the polystyrene matrix, Tentage1 should be swollen before use; swelling of its beads was determined to be higher or comparable to 2% cross-linked polystyrene-divinylbenzene beads [71]. It has been observed that the purity of oligonucleotides synthesized on TentaGel is significantly better than after synthesis on Teflon or CPG supports [65]. It is not known if this applies also to the synthesis of oligosaccharides. Mechanically TentaGel is more delicate than crosslinked polystyrene supports. Properties of Argogel are similar to those of TentaGel but it is claimed that its chemical stability is significantly improved (cf. Table 1 for its chemical structure), and also because of its double loading capacity.

~~

Linker

*

-

0 1 0 CPG 0,

O2N

= TG = PS(C)

,0(CH&O-PS

-NHCO(CH~)SCONH-AP

-0-CH2-p

CH3

-0(CHZ)~-S~-(CH~)&ONHCH~-PS

I I

CH3

-So-OCH2CONH-TG

A

-OCO(CH2)2CONHCH2COO-TG

-O-CO(CH2)2CO-NH-PS

-O-CO(CH2)2CO-NH-TG

-0-CO(CH2)2CONH-CPG

\

-S(CH2)3Si-

/

/

Structure

D. H. Rich, S. K. Gurwa, J. Am Chem. SOC.,1975,97,1575.

F

b. Rich's Linker

9) a. Pentenyl

I) Nitrobenzyl

r) p-Acylaminobenzyl

butylcarbonamide

i ) Ethoxydimethylsilyl-

acetamide

); QMercaptophenol-

I)CMercaptophenol

3) Sucanoyl-glycine

1) Succinoyl

1) Mercaptopropyl

~

Donor(s) Used

34,36b

32% Aq. Ammonia

NIA

hv /THF

DDQ/DCM, H20

NIA

Hg(OCOCF3)2 DCM, H 2 0

NaOMe/MeOH

37

32€

67

32j

S4,86

32a

82

36b

36b, 77

32%Aq. Ammonia

32%Aq. Ammonia

36a

~

Reference

NBS in THF/MeOH with DTBP

Removal

Table 3. Major linkers used in two-phase (solid state) oligosaccharide synthesis showing chemical structure of the linker, glycosylating agent, and removal methods used.

-S(CH&O-PS(M)

A

32b, 32i

3%

DMTSB-DIPEA DCM/MeOH NBSlDTBP

32e

32e

21

TrBFflCM

NaOMelMeOH

TBAFl HOAc

24,25b

22

32d

1)hvmHF for Partial Removal 2) PhSSiMeflnI2 for Conversion to Thioglycoside TBAF/ HOAc TBAM HOAc

32b

73

Reference

NaI, then NaOAc or N d

AcOH/THF/HzO

Removal

'Abbreviations. Donors: A= Trichloroacetimidate; B=Thioalkyl/aryl; C= Sulfoxide; D= n-Pentenyl; EFluoride; F= Glycal Polymers: CPG= Controlled-Pore Glass; TG= TentaGelR; PS(M)= Merrifield Resin; PS(C)= Polystyrene-Graft Crown (Cbiron Mimotopes, San Diego, CA, 92121); AP= ArgoPoreR

b. Mercaptopropanol

A

1s) a. Mercaptothiopropanol

A

F

F

F

Donor(s) Used

A

-S(CH&-S-PS(M)

R= Ph, 'Pr

R=pb R= 'fi

OZN

-TPS R

I

R

-S02(CH&-PS(M)

Et

I

-Si-(CH2)4-PS

I

Et

Structure

b. p-Oxybenzyl

14) a. p-Alkoxybenzyl

13) Diaryl(a1kyl)silyl

11) Sulfonyl

10) Diethylsilyl

Linker

Table 3 (continued)

10.5 Two-Phase Systems (Syntheses on Solid Supports)

259

10.5.1 Linkers

As in one-phase polymer supported synthesis, linkers must not interfere with the synthetic procedure, must be completely compatible with all synthetic steps, and must facilitate easy removability under conditions which harm neither the oligosaccharide nor the support. Many linkers have been adopted from other branches of solid state chemistry, others have been designed de now, and many more will be designed according to the requirements of a particular synthesis. It should be noted, of course, that a more general practical adoption of polymer-supported synthetic methodologies depends to a certain extent also on the general employability of a linker. Linkers utilized in the two-phase synthesis of oligosaccharides are surnmarized in Table 3. Dialkyl- or Diaryl-Silyl In some cases the solid support, i.e. polystyrene, has to be functionalized first, and the novel functional group represents the linker. The case of silicon-based linkers is a fitting example since it is easily removable by fluoride anions (e.g. tetrabutylammonium fluoride) and yet stable under most conditions utilized in oligosaccharide synthetic strategies. Cross-linked polystyrene-divinylbenzeneis first randomly lithiated [72],followed by reaction with diphenylsilyldichloride to create polystyrene-bound silyl chloride [24,271 which can be linked to a carbohydrate hydroxyl. The relative instability of the diphenylsilyl group to acidic conditions led to the replacement of the phenyl group with the isopropyl group [2Sb], using diisopropylsilyldichloride. Diethylsilyl functionality indirectly linked to the polymer [ 731 was investigated as a linker in a series of glycosylations under different conditions. Another promising silicon-based linker is S-(2-oxyethyldimethylsilyl)valeroyl[32JI which, in analogy with the cleavage of (trimethylsily1)ethyl glycosides [74],is transformed into an anomeric ester, e.g. propionate, by treatment with BF3.EtzO and propionic anhydride (other acyl anhydrides most likely can be used as well). From this glycosylating agents such as trichloroacetimidate can be subsequently prepared. The amidic bond between the valeric acid of the linker and the amino group of the functionalized polymer is sufficiently stable under most conditions used in oligosaccharide synthesis. Thioglycoside Linkers Thioglycosidic bonds are admirably stable under most conditions used for oligosaccharide synthesis and yet can be cleaved with many reagents, such as mercuric trifluoroacetate, N-bromosuccinimide with 1,6-di-t-butylpyridine, dimethylmethylthiosulfonium triflate or tetrafluoroborate, or oxidatively using dioxirane followed by NaOMe in methanol. Therefore cross-linked polystyrene utilizing thioglycoside [32a], or sulfide-based linkers are commonly used strategies [32h, i]. In fact, they were among the first linkers used in the two-phase oligosaccharide synthesis. 1,4Thiophenol, 1,3-propanedithiol, 1,3-thiopropanol, 2-(4-thiophenyloxy)acetamide have also been linked to the functionalized polystyrene support.

260

10 Polymer-Supported Synthesis of Oligosaccharides

1

THF J.

Reducing Oligosaccharide

ProtectedGlycoside

Scheme 9. A photochemically removable linker (top) can be combined with an ester funcionality (bottom) to increase versatility of the tether.

Linkers Cleavable by Photolysis The o-nitrobenzyl protection group has been used in oligosaccharide syntheses infrequently [14e], since it requires a long period (15-20 hours) of irradiation in a photochemical (e.g. Rayonet) reactor at 350 nm. The photoredox process can be substantially accelerated by the presence of a methoxy group in p-position to the nitrogroup [75]. However, the scope of the reaction is limited to furnishing upon photolysis a reducing saccharide when the tether is hooked to the anomeric carbon. The versatility of this type of linker, first used for an oligosaccharide synthesis by Zehavi in 1973, was recently expanded by Nicolaou's group (Scheme 9). In this novel form of the linker the nitrobenzyl moiety was connected to the anomeric carbon via 4-hydroxy benzoyl ester; therefore photolysis results in removal of the polymer producing a glycoside of p-hydroxy benzoyl ester. Alternatively, treatment with trimethylsilyl thiophenol/zinc iodide converts the complete linker assembly into a thioglycoside, an excellent glycosylating reagent.

10.6 Examples of Syntheses It would be of interest to evaluate different polymer-supported synthetic strategies for the preparation of the same oligosaccharides. To date, two oligosaccharides have been synthesized using both the one-phase system on MPEG and the two-

10.7 Combinatorial Libraries

261

phase system using crosslinked polystyrene. One example is the synthesis of the heptasaccharide 1. One method [32c] utilized phenolic polystyrene and the photolabile 5-hydroxy-2-nitro-benzyl alcohol tether, while the other [491 employed MPEG and the succinoyl linker. Because of the different synthetic designs a meaningful comparison of the two supports, such as to the advantages of one support over the other, is not possible. The second example is the synthesis of the pentamannoside [Manp(al-2)I4Manp (2). Mannose oligosaccharides are superbly suited to these studies since a-anomers are formed exclusively under appropriate glycosylation conditions. The one-phase (solution) system used MPEG and the DOX linker [54] and the two-phase system (solid state) utilized crosslinked polystyrene and a thioglycoside linker [32h]. To monitor reactions, the one-phase strategy took advantage of the support’s solubility to use NMR spectroscopy; the two-phase strategy employed cleavage of a small sample from the solid support, and analyzed the sample by either NMR spectroscopy or mass spectrometry (MALDI-TOF) [32h]. The results of these pentamannoside syntheses are not fully comparable since the perbenzylated pentamannoside obtained by the two-phase method was not deprotected. As noted before, the use of cross-linked polystyrene support for syntheses of oligosaccharides by the glycal strategy is discussed elsewhere in this Volume. In Ogawa’s group the p-alkoxybenzylether linker, used in one phase polymer supported synthesis, was utilized in combination with Wang’s resin to synthesize polylactoseamines of various lengths [ 32eI. Finally, polymer-supported synthesis has been used to synthesize uncommon monosaccharides [761. Controlled pore glass has been used to synthesize an a1-2 trimannoside [36a], several disaccharides [36b], and glycosylated oligonucleotides [77]. The examples confirm that CPG-supported oligosaccharide synthesis is feasible, although its limited loading capacity and scale-up may eventually become a restricting factor. TentaGel-N, linked via the succinoyl linker to a monosaccharide acceptor was utilized using trichloroacetimidate chemistry to prepare disaccharides [34] and trisaccharides [78]. It was observed that TentaGel is slightly unstable under the acidic conditions required by trichloroacetimidate chemistry [78], most likely due to the acid sensitivity of PEG [79]. No similar sensitivity was observed in similar applications employing Argogel [ 781.

10.7 Combinatorial Libraries Polymer-supported strategies of oligosaccharide syntheses are ideally suited to preparations of combinatorial oligosaccharide libraries since purification of all products of a particular synthetic step is done at the same time and it would not be expected that individual library members would be lost during the purification process. Both one-phase and two-phase polymer supported strategies have been examined, and a review was published recently [go]. More recent articles that were not discussed in this review include: syntheses of disaccharide [81] and trisaccharide 1821 libraries leading to the discovery of novel antibacterial agents [83], and prepa-

262

I0 Polymer-Supported Synthesis of Oligosaccharides

ration of libraries on TentaGel beads as polyvalent molecules [84]. Often carbohydrates (including oligosaccharides) form a scaffold for a library [85],and the variations in carbohydrate structure may have an effect on biological activity of the constructs in the library [86].

10.8 Capping Capping, or protection of the unreacted hydroxyl to prevent it from reacting in the next condensation step, is a very important operation in oligonucleotide automatic synthetic protocols. In oligosaccharide synthesis capping seems to be much less important, although sometimes unreacted hydroxyls are acetylated to prevent their glycosylation in the next step [49]. However, it seems more useful to repeat the glycosylation step, in analogy with syntheses of “difficult sequences” in oligopeptide synthesis, rather than to use capping. It should be noted that important variants of polymer supported synthesis of oligosacharides, such as the “self-policing” glycal strategy (Scheme 3 ) or strategies involving a hydrophobic handle (Scheme 5), do not require capping. The last mentioned strategy eventually may become more generally adopted.

10.9 Concluding Remarks The polymer-supported synthesis of protected oligosaccharides has attracted considerable attention of organic chemists during the last decade of the 20th century. Full reviews dedicated to the polymer supported techniques of oligosaccharide synthesis were published recently [26, 871, significant parts of reviews on oligosaccharide synthesis have been reserved for polymer-supported methods [5g, 88, 891 and parts of reviews on polymer-supported methodology have been dedicated to oligosaccharide syntheses [ 31. As well, a review on combinatorial carbohydratebased libraries was recently published [90]. It appears to be a suitable methodology for the synthesis of oligosaccharides in quantities required for biomedical applications; for instance, fragments of the capsular polysaccharide of Huemophilus inflenzae were synthesized both by one-phase [91] and two-phase [92] polymer supported methods. It is clear that the methodology, while not perfect, will become more widely used, as better and more efficient reaction conditions are discovered and the essential components become less expensive. References 1. D. Zopf, S. Roth, The Lancet, 1996,347, 1017-1021. 2. 0.-T. h u n g , S. P. Douglas, D. M. Whitfield, H. Y. S. Pang, J. J. Krepinsky, New J. Chem., 1994, 18, 349-363.

References

263

3. D. J. Gravert, K. D. Janda, a) Chem. Revs., 1997, 97, 489-509, b) Current Opin. Chem. Biol., 1997, I , 107-113. 4. For recent reviews, see e.g. a) K. Jarowicki, P. Kocienski, J. Chem. Soc. Perkin Trans. I , 1999, 1589-1615; b) 0. Seitz, Angew. Chem. Int. Ed. Engl. 1998,37, 3109-3111. 5. For reviews, see e.g. H. Paulsen, Angew. Chem. Int. Ed.Eng1. a) 1982,21, 155-224; b) 1990, 29, 823; c) H. Paulsen, Revs Chem. Soc. 1984, 13, 15-45; d) R. R. Schmidt, Angew. Chem. Int. Ed. Engl. 1986, 25, 212; e) P. Sinay, Pure Appl. Chem. 1991, 63, 519-528; f ) K. Toshima and K. Tatsuta, Chem. Revs., 1993, 93, 1503-1531; g) R. R. Schmidt, J. C. Castro-Palomino, 0. Retz, Pure Appl. Chem. 1999, 71, 729-744; h) R. R. Schmidt, Pure Appl. Chem. 1998, 70, 397-402. 6. E. g. Modern Methods in Carbohydrate Synthesis (Eds.: S. H. Khan, R. A. O’Neill), Harwood Academic Publishers, Amsterdam, 1996. 7. For instance: a) K. J. Kaur, G. Alton, 0. Hindsgad, Carbohydr. Res., 1991,210, 145; b) C.-H. Wong, Y. Ichikawa, T. Krach et al. J. Am. Chem. Soc., 1991, 113, 8137; c) S. Roth, US Pat. 5,180,674, Jan 19, 1993; d) Y. Ichikawa et al. J. Am. Chem. Soc., 1992, 114, 9283; (e) for a review, see Y. Ichikawa, G. C. Look, C.-H. Wong, Anal. Biochem., 1992,202, 215. 8. a) J. Thiem, B. Sauerbrei, Angew. Chem. Int. Ed. Engl., 1991, 30, 1503-1505; b) B. Sauerbrei, J. Thiem, Tetrahedron Lett., 1992, 33, 201-204. 9. For instance, G. C. Look, Y. Ichikawa, G.-J. Shen, P.-W. Cheng, C.-H. Wong, J. Org. Chem., 1993, 58,4326-4330. 1994, 116, 1135-1136. 10. M. Schuster, P. Wang, J. C. Paulson, C.-H. Wong, J. Am. Chem. SOC., 11. a) U. Zehavi, M. Herchman, Carbohydr. Rex, 1984, 128, 160-164; b) U. Zehavi, S. Sadeh, M. Herchman, Carbohydr. Res., 1983, 124, 23-34; c) U. Zehavi, M. Herchman, Curbohydr. Res., 1986, 151, 371-378. 12. T. Wiemann, N. Taubken, U. Zehavi, J. Thiem, Carbohydr. Res., 1994,257, C1LC6. 13. S.-I. Nishimura, K. Matsuoka, Y. C. Lee, Tetrahedron Lett., 1994, 35, 5657-5660. 14. a) J. M. J. Frechet, in Polymer-Supported Reactions in Organic Synthesis (P. Hodge, D. C. Sherrington, Eds.), Wiley, Chichester, 1980, pp. 293 and 407; b) J. M. J. Frechet, C. Schuerch, J. Am. Chem. Soc., 1971, 93, 492-496; c) Carbohydr. Res., 1972, 22, 399-412; d) N. K. Mathur, C. K. Narang, R. E. Williams, Polymers as Aids in Organic Chemistry, Academic Press, New York, 1980, Chapter 6; e) U. Zehavi, Adv. Carbohydr. Chem. Biochem., 1988, 46, 179-204; f ) J. M. J. Frechet, Tetrahedron 1981, 37, 663-683. 15. R. Eby, C. Schuerch, Carbohydr. Res., 1975, 39, 151-155. 16. a) P. Westerduin, G. H. Veeneman, Y. Pennings, G. A. van der Marel, J. H. van Boom, Tetrahedron Lett., 1987,28, 1557-1560; b) G. H. Veeneman, S. Notermans, R. M. J. Liskamp, G. A. van der Marel, J. H. van Boom, Tetrahedron Lett., 1987,28, 6695-6698. 17. G. H. Veeneman, H. F. Brugghe, H. van den Elst, J. H. van Boom, Carbohydr. Rex, 1990,195, C1-C4 (after p. 308). 18. a) U. Zehavi, A. Patchornik, J. Am. Chem. Soc., 1973, 95, 5673-5677; b) S.-H. L. Chiu, L. Anderson, Carbohydr. Rex, 1976, 50, 227-238; c) G. Excoffier, D. Gagnaire, J.-P. Utille, M. Vignon, Tetrahedron, 1975, 31, 549-553; d) for a review see; A. Malik, H. Bauer, J. Tschakert, W. Voelter, Chemiker Ztg., 1990, 114, 371-375. 19. a) R. D. Guthrie, A. D. Jenkins, J. Stehlicek, J. Chem. Soc. ( C ) , 1971, 2690-2696; b) R. D. Guthrie, A. D. Jenkins, G. A. F. Roberts, J. Chem. Soc. Perkin Trans. I, 1973, 2414-2417. 20. S. P. Douglas, D. M. Whitfield, J. J. Krepinsky, J. Am. Chem. Soc., 1991, 113, 5095-5097. 21. For example, J. T. Randolph, K. F. McClure, S. J. Danishefsky, J. Amer. Chem. Soc., 1995, 117, 5712-5719. 22. S. J. Danishefsky, K. F. McClure, J. T. Randolph, R. B. Ruggeri, Science, 1993,260, 1307-1309. 23. C. M. Timmers, G. A. van der Marel, J. H. van Boom, Red. Trav. Chim Pays-Bas, 1993, 112, 609-6 10. 24. J. T. Randolph, S. J. Danishefsky, Angew. Chem. Znt. Ed. Engl., 1994, 33, 1470-1473. 25. a) P. H. Seeberger, M. Eckhardt, C. E. Gutteridge, S. J. Danishefsky, J. Am. Chem. Soc, 1997, 119, 10064-10072; b) C. Zheng, P. H. Seeberger, S. J. Danishefsky, J. Org. Chem., 1998, 63, 1126-1 130. 26. P. H. Seeberger, S. J. Danishefsky, Acc Chem. Res., 1998, 31, 685-695. 27. Y. Ito, 0. Kanie, T. Ogawa, Angew. Chem. Int. Ed. Engl., 1996,35, 2510-2512. 28. a) M. M. Palcic, L. D. Heeze, M. Pierce, 0. Hindsgaul, Glycoconjugate J., 1988,5,49-63; b) P. Stangier, M. M. Palcic, D. R. Bundle, Curbohydr. Res., 1995, 267, 153-159.

264

10 Polymer-Supported Synthesis of Oligosaccharides

29. a) J. M. Harris, J. Macromol. Sci.-Rev. Macromol. Chem. Phys., 1985, C25, 325-373; b) J. M. Harris in Poly(Ethy1ene Glycol) Chemistry: Biotechnical and Biomedical Applications (J. M. Harris, Ed.) Plenum Press, New York 1992, pp. 2. 30. a) E. Bayer, M. Mutter, The Peptides (E. Gross, J. Meienhofer, Eds.), Academic Press, New York, 1980, 2, 286; b) G. M. Bonora, C. L. Scremin, F. P. Colonna, A. Garbesi, Nucl. Acids Res., 1990, 18, 3155-3159; c) G. M. Bonora, G. Biancotto, M . Maffini, C. L. Scremin, Nucl. Acids Res., 1993, 21, 1213-1217. 31. J. J. Krepinsky, S. P. Douglas, D. M. Whitfield, in Methods in Enzymology, 242, Neoglycoconjugates, Pt. A: Synthesis (Y. C. Lee, R. T. Lee, Eds.), Academic Press, San Diego, CA. 1994, pp. 280-293. 32. a) L. Yan, C. M. Taylor, R. Goodnow, Jr., D. Kahne, J. Am. Chem. Soc., 1994, 116, 69536954; b) J. A. Hunt, W. R. Roush, J. Am. Chem. Soc., 1996, 118, 9998-9999; c) K. C. Nicolaou, N. Winsinger, J. Pastor, F. DeRoose, J. Am. Chem. Soc., 1997, 119, 449-450; d) K. C. Nicolaou, N. Watanabe, J. Li, J. Pastor, N. Winssinger, Angew. Chem. Znt. Ed. Engl., 1998,37, 1559-1561; e) H. Shimizu, Y. Ito, 0. Kanie, T. Ogawa, Bioorg. Med. Chem. Lett., 1996, 6, 2841-2846; f ) K. Fukase, K. Egusa, Y. Nakai, S. Kusumoto, Mol. Divers., 1996,2, 182-188; g) J. Rademann, R. R. Schmidt, Tetrahedron Lett., 1996, 37, 3989-3990; h) J. Rademann, R. R. Schmidt, J. Urg. Chem., 1997, 62, 3650-3653; i) J. Rademann, A. Geyer, R. R. Schmidt, Angew. Chem. Znt. Ed. Engl., 1998, 37, 1241-1245; j) D. Weigelt, G. Magnusson, Tetrahedron Lett., 1998, 39, 2839-2842. 33. L. Szabo, B. L. Smith, K. D. McReynolds, A. L. Parrill, E. R. Morris, J. Gervay: J. Urg. Chem., 1998, 63, 1074-1078. This series of sialooligomers (up to the octamer) are linked by (1+ 5 ) amidic bonds between two units of neuraminic acid and not by (2+) glycosidic bonds. 34. M. Adinolfi, G . Barone, L. De Napoli, A. Iadonisi, G. Piccialli, Tetrahedron Lett., 1996, 37, 5007-5010. 35. S. P. Douglas, J. J. Krepinsky, to be published. 36. a) A. Heckel, E. Moss, K.-H. Jung, J. Rademann, R. R. Schmidt, Synlett, 1998, 171-173; b) M. Adinolfi, G. Barone, L. De Napoli, A. Iadonisi, G. Piccialli, Tetrahedron Lett., 1998, 39, 1953-1956. 37. R. Rodebaugh, S. Joshi, B. Fraser-Reid, H. M. Geysen, J. Org. Chem., 1997, 62, 5660-5661. 38. S. Manabe, Y. Ito, T. Ogawa, Synlett, 1998, 628-630. 39. G. G. Cross, D.M. Whitfield, Synlett, 1998, 487-488. 40. Suitable MPEG mw = 5,000 was obtained from Fluka AG, Buchs, Switzerland. 41. a) Technical Bulletin: Polyglycols Hoechst. Polyethylene Glycols: Properties and Applications. Hoechst, Frankfurt (1983); b) J. Dale, Zsr. J. Chem., 1980, 20, 3. 42. t-Butylmethyl ether of sufficient quality and degree of dryness for this purpose is available inexpensively in bulk quantities from A R C 0 (Atlantic Richfield) Chemicals. 43. S. P. Douglas, D. M. Whitfield, J. J. Krepinsky, unpublished observations. 44. L. Jiang, R. C. Hartley, T.-H. Chan, J. Chem. Soc. Chem. Commun., 1996, 2193-2194. 45. S. P. Douglas, D. M. Whitfield, H. Y. S. Pang, J. J. Krepinsky, unpublished observations. 46. a) D. M. Whitfield, S. P. Douglas, J. J. Krepinsky, Tetrahedron Lett., 1992, 33, 6795-6798; b) S. P. Douglas, H. Y. S. Pang, J. J. Krepinsky, unpublished results. 47. M. Y. Meah, D. M. Whitfield, S. P. Douglas, J. J. Krepinsky, unpublished results. 48. J. J. Krepinsky, in Modern Methods in Carbohydrate Synthesis (S. H. Khan, R. A. O’Neill, Eds.), Hanvood Academic Publishers, 1996, pp. 194-224. 49. R. Verduyn, P. A. M. van der Klein, M. Douwes, G. A. van der Marel, and J. H. van Boom, Recl. Trav. Chim. Pays-Bas, 1993, 112,464-466. 50. R. Verduyn, M. Douwes, P. A. M. van der Klein, E. M. Mosinger, G. A. van der Marel, J. H. van Boom, Tetrahedron, 1993, 49, 7301-7311. 51. C. M. Dreef-Tromp, H. A. M. Willems, P. Westerduin, P. van Veelen, C. A. A. van Boeckel, Bioorg. Med. Chem. Lett., 1997, 7, 1175-1180. 52. L. 0. Kononov, Y. Ito, T. Ogawa, Tetrahedron Lett., 1997, 38, 1599-1602. 53. Y. Wang, H. Zhang, W. Voelter, Chem. Lett., 1995, 213-274. 54. S. P. Douglas, D. M. Whitfield, J. J. Krepinsky, J. Am. Chem. Soc., 1995, 117, 2116-2117. 55. S. Mehta, D. M. Whitfield, Tetrahedron Lett., 1998, 39, 5907-5910. 56. D. M. Whitfield, H. Y. S. Pang, J. P. Carver, J. J. Krepinsky, Can. J. Chem., 1990,68,942-952.

References

265

57. S. Manabe, Y. Ito, T. Ogawa, Molecules Online, 1998,2, 40-45. 58. G. Hodosi, J. J. Krepinsky, Synlett, 1996, 159-162. 59. a) R. U. Lemieux, Chemistry in Canada, 1964, 16, 14; b) A. Berces, I. Chrzczanowicz, A. Obuchowska, D. M. Whitfield, J. J. Krepinsky, J. Am. Chem. Soc., 1999, manuscript to be submitted, and references therein. 60. Z . G . Wang, S. P. Douglas, J. J. Krepinsky, Tetrahedron Lett., 1996, 37, 6985-6988. 61. F. Barresi, 0. Hindsgaul, a) J. Am. Chem. Soc., 1991, 113, 9376; b) Can. J. Chem., 1994, 72, 1447- 1465. 62. Y. Ito, T. Ogawa, J. Am. Chem. Soc., 1997, 119, 5562-5566. 63. W. B. Li, B. Yan, J. Org. Chem., 1998, 63, 4092-4097. 64. P. H. Seeberger, X. Beebe, G. D. Suckenick, S. Pochapsky, S. J. Danishefsky, Angew. Chem. Int. Ed. Engl., 1997, 36, 491-493. 65. B. P. Zhao, G. B. Panigrahi, P. D. Sadowski, J. J. Krepinsky, Tetrahedron Lett., 1996, 37, 3093-3096. 66. M. Hori, D. J. Gravert, P. Wentworth, Jr., K. D. Janda, Biorg. Med. Chem. Lett., 1998, 8, 2363-2368. 67. K. Fukase, Y. Nakai, K. Egusa, J. A. Porco, Jr., S. Kusumoto, Synlett, 1999, 1074-1078. 68. Rapp Polymere, GmbH, 7072 Tubingen, Germany. 69. Argonaut Technologies, 887 Industrial Blvd., Suite G, San Carlos, CA 94070, USA. 70. a) E. Bayer, Angew. Chem. Znt. Ed. Engl., 1991, 30, 113-129; b) E. Bayer, W. Rapp in Poly (Ethylene Glycol) Chemistry: Biotechnical and Biomedical Applications (J. M. Harris, Ed.) Plenum Press, New York 1992, pp. 325-345. 71. M. E. Wilson, K. Paech, W.-J. Zhou, M. J. Kurth, J. Org. Chem., 1998, 63, 5094-5099. 72. T.-H. Chan, W.-Q. Huang, J. Chem. Soc. Chem. Commun., 1985,909-911. 73. T. Doi, M. Sugiki, H. Yamada, T. Takahashi, Tetrahedron Lett., 1999, 40, 2141-2144. 74. K. Jansson, T. Noori, G. J. Magnusson, J. Org. Chem., 1990,55, 3181-3185. 75. H. Venkatesan, M. M. Greenberg, J. Org. Chem., 1996, 61, 525-529. 76. S. Kobayashi, T. Wakabayashi, M. Yasuda, J. Org. Chem., 1998, 63, 4868-4869. 77. M. Adinolfi, G. Barone, L. De Napoli, L. Guariniello, A. Iddonisi, G. Piccialli, Tetrahedron Lett., 1999, 40, 2607-2610. 78. S. P. Douglas, J. J. Krepinsky, unpublished observations. 79. A. A. Patel, I. Chrzczanowicz, J. J. Krepinsky; manuscript in preparation. 80. Z.-G. Wang, 0. Hindsgaul, Adv. Exptl. Med. Biol., 1998, 435, 219-236. 81. D. J. Silva, H. Wang, N. M. Allanson, R. K. Jain, M. J. Sofia, J. Org. Chem., 1999, 64, 59265929. 82. T. Zhu, G.-J. Boons, Angew. Chem. Int. Ed. Engl., 1998, 37, 1898-1900. 83. M. J. Sofia, N. Allanson, N. T. Hatzenbuhler, R. Jain et al, J. Med. Chem., 1999, 42, 31933198. 84. R. Liang, J. Loebach, N. Horan, M. Ge, C. Thompson, L. Yan, D. Kahne, Proc. Natl. Acad. Sci. USA, 1997, 94, 10554-10559. 85. a) T. Wunberg, C. Kallus, T. Opatz, S. Henke, W. Schmidt, H. Kunz, Angew. Chem. Int. Ed. Engl., 1998,37,2503-2505; b) M. J. Sofia, R. Hunter, T. Y. Chan, A. Vaughan, R. Dulina, H. Wang, D. Gange, J. Org. Chem., 1998, 63, 2802-2803. 86. R. Liang, L. Yan, J. Loebach, M. Ge, Y. Uozumi, et al, Science, 1996,274, 1520-1522. 87. U. Zehavi, Reactive & Functional Polymers, 1999, 41, 59-68. 88. a) H. M. I. Osborn, T. H. Khan, Tetrahedron, 1999,55, 1807-1850; b) Y. Ito, S. Manabe, Curr. Opin. Chem. Biol., 1998,2, 701-708. 89. a) G.-J. Boons, Tetrahedron, 1996, 52, 1095-1121; b) G.-J. Boons, Contemp. Org. Synth., 1996, 3, 173-200; c) D. M. Whitfield, S. P. Douglas, Glycoconj. J., 1996, 13, 5-17. 90. M. J. Sofia, Molec. Diversity, 1998, 3, 75-94. 91. A. A. Kandil, N. Chan, P. Chong, and M. Klein, Synlett, 1992, 555-557. 92. S. Nilsson, M. Bengtsson, T. Norberg, J. Carhohydr. Chem., 1992, 11, 265-285.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

11 Glycopeptide Synthesis in Solution and on the Solid Phase Hovst Kunz and Michael Schultz

11.1 Introduction Glycopeptides are defined as fragments of glycoproteins consisting of saccharide units which are covalently linked to a peptide backbone. The structural diversity emerging from the combination of amino acid and carbohydrate functionality is immense [l-51. Table 1 summarizes the known types of natural linkage types between amino acid side-chains and the first carbohydrate residue. The most abundant carbohydrates are highlighted and the corresponding structural details are depicted. In nature, especially in eucaryotic organisms, most proteins are glycosylated. This observation suggests that such a sophisticated result of evolution should be accompanied by increased functional complexity [6]. In fact, the last decade has revealed numerous fascinating insights into glycoprotein biology. In the course of this introduction both the typical biological role of such a complex glycoprotein and the characteristic preparative problems one faces will be demonstrated by the example of P-selectin glycoprotein ligand-1 (PSGL-1) [7]. Organisms defend themselves against inflammation by deploying white blood cells (leukocytes) which infiltrate the inflamed tissue [8-121. One of the first signals for patrolling leukocytes to migrate from the blood vessel into neighboring tissue is presented on the endothelium of the vessel. This signal is the expression of the glycoprotein receptor P-selectin. Initial weak adhesion to the ligand PSGL-1 located on the surface of leukocytes to this selectin leads to retarded leukocyte movement (‘rolling’). Several subsequent adhesion processes mediated by other recognition molecules finally lead to migration of the leukocytes through the endothelium. In chronic diseases such as rheumatoid arthritis or reperfusion injury a promising treatment could involve inhibition of the aforementioned selectin-ligand interaction [ 111. In this context, the selectin ligand itself is the first lead structure. Biological investigations have revealed the critical binding segment of PSGL-1 to be a sulfated oligopeptide located in the N-terminus and carrying an oligosaccharide terminated with sialyl Lewis’ on a defined threonine residue (Scheme 1) [7, 13, 141.

268

11 Glycopeptide Synthesis in Solution and on the Solid Phase

Table 1. Linkage regions in natural glycoproteins. Amino acid

Carbohydrate

Asparagine

N-Acetylgalactosamine N-Acetylglucosamine D-Glucose L-Rhamnose

Hydroxylysine

D-Galactose

Hydroxyproline

L-Arabinose D-Galactose

Serine

Galactose Glucose Xylose Fucose

Threonine

Linkage region

I HO

HNJ 0

+O+

Fucose

HO

Serine or Threonine

N-Acetylgalactosamine N-Acetylglucosamine Mannose

AcHN

R

R=H, CH3

Tyrosine

Glucose Galactose

,O

O

11.1 Introduction

s0:so:-

I /

so?-

269

HO

I

Q A T E Y E Y L D Y D F L P E T E P I

I

kHo..0H

' T O H

OH

COH

AcHN

OH

1

Sialyl Lewis' (%ex)

1 n=2-6 Scheme 1.

If PSGL-1 is not sulfated, the molecule is no longer recognized by P-selectin but it still serves as a ligand for another adhesion molecule, E-selectin [15]. Because PSGL-1 is a glycosylated protein the over-expression of this factor in a suitable cell-line is not a possible method for the synthesis of an effective compound. Such an approach would not deliver the sulfated and glycosylated form. On the other hand, PSGL-1 cannot be isolated from natural sources as a homogeneous compound in a suitable quantity for medicinal purposes. At this point recent developments of glycopeptide synthesis justify closer retrosynthetic analysis of compound 1, although in terms of size and complexity 1 constitutes quite a challenging target. Thus, the first step is simplification of the structure in such a way that biological activity and experimental feasibility are both taken into account. Biological results suggest that mainly the terminal tetrasaccharide (sialyl Lewis") is required for recognition. Furthermore, only one of the sulfated tyrosine residues is considered essential for binding. Thus, the minimum target molecule is represented by a sulfated heptapeptide carrying sialyl Lewis" (2) (Scheme 2) [ 161. Further retrosynthetic analysis of this structure splits the target into the corresponding amino acid building blocks required for solid-phase synthesis (Scheme 2). N- and side-chain-protected amino acid derivatives are commercially available except for the glycosylated threonine. The glycosylated key building block can be substituted by the more readily available glycosylated asparagine derivative which is constructed from suitably protected monosaccharides 3-6 and an aspartic acid moiety ready for glycosylation. For the realization of such a synthetic approach a detailed knowledge of glycosylation chemistry and protecting group strategy becomes important [4, 17, 181. Fundamental methods in both fields relevant to the synthesis of 2 shall therefore be outlined in the following paragraphs. 11.1.1 Which Protecting Groups are Suitable for Carbohydrates (Table 2)?

For 0-glycosylation, at least, the answer to this question depends mainly on the desired stereochemistry of the products. Ether type hydroxy-protecting groups which do not exert a neighboring group-effect, e.g. benzyl (Bzl), in the 2-position lead almost exclusively to glycosides with axial linkages (usually a ) to the used glycosyl acceptor if respective reaction conditions are applied (see below) [ 191. The

270

I 1 Glycopeptide Synthesis in Solution and on the Solid Phase H-Tvr(S032-)-Asp-Phe-Leu-Pro-Glu-Asn-OH ,OH

/

Fmoc-Amino Acids-OH and Fmoc-Asn(SLeX)-OH ======3 SLeX-N3

Scheme 2.

target molecule 2 contains two a-linked saccharides and, indeed, the fucose donor 4 is an ideal example of the rule of preferred cis-glycosylation employing benzylprotected sugars. The formation of the a-sialoside, however, cannot be influenced by an assisting group neighboring the anomeric center. Thus, acyl type protection of hydroxy groups, which is readily introduced into the valuable neuraminic acid, can be applied after protecting the carboxy group as the methyl ester (6). The remaining donors galactose 5 and glucosamine 3 should be equipped with a combination of ether- and acyl-type protection. This pattern ensures the orthogonal stability of protecting groups during the synthetic protocol. Benzyl ethers can be removed by hydrogenation and ester type groups are cleaved by use of catalytic amounts of base (table 2).

Table 2. Common protecting groups in carbohydrate chemistry. Protecting group

Introduction

Removal ~~

CsHsCH20CH3COO-COOCH3

NaH/benzyl bromide Pyridine/acetic anhydride H+/methanol

H2/Pd-C cat. NaOMe/MeOH dil. NaOH

11.1 Introduction

27 1

11.1.2 Which Glycosylation Methods are Useful for the Formation of Glycopeptides?

Again the range of methods useful for the formation of the different types of linkage will be illustrated using compound 2 as an example [4, 18, 20-231. Formation of Asparagine N-Glycosides The asparagine-glucosamine linkage is the most common type of linkage found in natural glycoproteins. Because asparagine cannot be used directly as a glycosyl acceptor the linkage is usually formed by coupling suitable glycosylamine (saccharide) and aspartic acid derivatives. The amino function of the saccharide is accessible either from unprotected saccharides and ammonium hydrogen carbonate [24] or via the hydrogenation of an intermediate azide which is prepared, for instance, from glycosyl halides in phase-transfer-catalyzed reactions with sodium azide [25]. Coupling of the glycosylamine to aspartic acid can be accomplished by use of established peptide-coupling reagents such as isobutoxyisobutyldihydroquinoline (IIDQ, Scheme 3) or TBTU [ 171. a-Fucosylation

A very efficient introduction of a-fucosyl moieties, e.g. into the 3- or 4-position of a glucosamine acceptor, is possible via the in situ anomerization procedure developed by Lemieux and coworkers [26]. Tri-0-benzylated fucosyl bromide 7 is a useful glycosyl donor in these cases [27]. Activation of this donor by catalytic amounts of Br- results in the intermediate formation of the reactive P-halide (Scheme 4).

4 HZ Ra-Ni

Acoq Aliquat

AcHN CI NaN3

N H ~IIDQJDIPEA

N)

AcHN

AcHN

T-

Fmoc-Asp(0H)OtBu

Scheme 3.

Scheme 4.

TNH

L~

272

I I Glycopeptide Synthesis in Solution and on the Solid Phase

OBzl 5

OBzl 8

Scheme 5.

The P anomer spontaneously reacts with nucleophiles to give the desired afucoside. On the other hand, so-called disarmed donors such as Ac3FucBr carrying electron-withdrawing protecting groups do not react under these conditions. Formation of the P-Lactosamine Linkage

P-Galactosylation of the 4-hydroxy group of glucosamine is difficult because of the significantly reduced nucleophilicity of this acceptor group in the vicinity of protected or glycosylated neighboring hydroxy functions, as was shown in earlier investigations. The powerful trichloroacetimidate methodology [28, 291 is most recommendable to achieve the desired conversion (Scheme 5). Trichloroacetimidate donors are easily prepared from the corresponding 1-OH precursors and are stable towards basic conditions [30]. a-Sialylation

N-Acetylneuraminic acid contains no C-3 substituent but carries a carboxy group at position 2. Both factors complicate the stereoselective and efficient formation of the desired a-2,3-sialoside in example 2 [31]. A neighboring group at C-3, required for stereoselective synthesis, can be introduced temporarily (SPh, SePh) and removed after completion of the sialylation [32]. This strategy, however, requires additional steps and results in reduced overall yield. Recently, thioalkyl/aryl sialosides and Ssialosyl xanthates have been used to achieve regio- and stereoselective formation of a-glycosides of N-acetyl neuraminic acid derivatives (Scheme 6) [33-351. The 2-SR sialosides can be activated by use of N-iodosuccinimide (NIS) and traces of TfOH or with alkyl sulfenyl triflate [33, 36, 371. Under these conditions the desired a-product 10 can be obtained in yields above 50% without the need for purification by HPLC. The regioselectivity of the NeuNAc-SR donor for the 3-OH group of the galactose residue of the acceptor 9 is particularly advantageous [38].

11.1.3 Glycopeptides Containing Particularly Sensitive Linkages

Because of the particular sensitivity of glycopeptides, precautions must be taken during the assembly of the target molecule, in particular during the deblocking reactions.

273

11.1 Introduction AcO

Ac0w0cH3

L

SEt

AcHN 6

AcO

NIS/TfOH

N3 c

AcO

OBzl

10

NHAc

OBzl

NHAc

9

Scheme 6.

Acid Sensitivity In general, the 0-glycosidic linkages between saccharide residues and also between saccharides and amino acids are sensitive towards acids. For instance, the a-fucosidic linkage is only compatible with the trifluoroacetic acid-promoted cleavage of side-chain protection or anchoring groups if acyl protection is used (Scheme 7). In the presence of benzylated or methoxy benzylated (OMpm) carbohydrate moieties, treatment with trifluoroacetic acid can lead to extensive destruction of the corresponding saccharide linkage [39, 401. Base Sensitivity If strongly basic conditions (pH > 11) are used, the a-CH linkages of amino acids in peptides and, in particular, in peptide esters are likely to epimerize (Scheme 8). In

OBzl 1 OBzl

-0BA

A

C*o--

Peptide

I

NHAc

COOtBu

1

1

Peptide

I 0

NHAc

COOtBu

TFA TFA

Complete loss of the fucose residue

Scheme 7.

4

nNH

AcO

1

Selective cleavage of the rert-butyl ester

274

I 1 Glycopeptide Synthesis in Solution and on the Solid Phase

addition, with serine/threonine glycosides the removal of the a proton can result in complete elimination of the glycan portion by an E 1cB mechanism:

Scheme 8.

The risk of side-reactions is particularly increased if ester protecting groups (e.g. the methyl ester in 10) are cleaved using, for example, dilute sodium hydroxide. The progress of these deblocking reactions should, therefore, always be carefully monitored [41]. Fortunately, the basic conditions required for the removal of the Fmoc group (morpholine, short treatment with piperidine) and the 0-acetyl groups (catalyst NaOMe, pH 8.5) are so mild that undesired reactions can be minimized. For this reason the N-Fmoc and the 0-acetyl groups are well established as standard protecting tools in glycopeptide synthesis [ 171.

11.2 Synthesis of Glycopeptides in Solution Despite the successful development of a routine methodology for the solid-phase synthesis of peptides and glycopeptides (see below) a synthesis in solution is still the method of choice, in particular if glycopeptides with demanding saccharide components are to be synthesized. Short sequences with a relatively high content of complex saccharide residues can be prepared efficiently in solution avoiding the loss of valuable building blocks required in excess in syntheses on the solid phase. The development of new coupling reagents or protecting groups, moreover, often requires an initial verification in solution. Particularly difficult sequences are advantageously prepared in solution by coupling preformed fragments [42]. Finally, synthesis of glycopeptides in solution is required, not only if a few milligrams of the product are required for a first attempt at biological evaluation (and claiming) but also if preparative amounts of a glycopeptide with valuable biological properties are to be obtained. 11.2.1 0-Glycopeptides

Glycopeptides Carrying N-Acetylgalactosamine (Tn-Antigen)

N-Acetylgalactosamine (Tn-antigen) is usually found a-glycosidically linked to serine or threonine. The Ser/Thr(GalNAc) residue also represents the key building

11.2 Synthesis of Glycopeptides in Solution

215

block of the important glycoprotein family of the mucins [6, 43, 441. Mucins are highly glycosylated proteins produced by a wide variety of epithelial and endothelial cells. Despite their normal cell-protecting function, the appearance of mucin molecules with diminished carbohydrate chains has also been correlated with a negative prognosis in the development of tumors. In particular, Tn (aGalNAc), T (PGal-1,3-aGalNAc), STn (aNeuNAc-2,6-aGalNAc) and the regioisomeric ST [ PGal-I,3-(aNeuNA~-2,6)-aGalNAc or aNeuNAc-2,3-PGal-l,3-aGalNAc] antigens have been described as tumor-associated antigens. [8, 45, 461. The synthesis of partial structures of these biologically relevant cell surface components has, therefore, become a major target in glycopeptide synthesis. Some results of these efforts have already been applied in the fields of immunology (vaccination) and diagnosis

WI.

MUCl is the mucin which has been most thoroughly investigated [48]. It consists of numerous glycosylated tandem repeats of 20 amino acids. A bisglycosylated decapeptide of this repeating unit was synthesized in solution by use of enzymatic protecting group methodology [49]. The key building blocks for this synthesis were the Fmoc-protected Ser/Thr-aGalNAc residues which were accessible in eight steps starting from galactose via the azidonitrate 11 [50]and the thioethyl glycoside 12 [51] (Scheme 9). Because the azido group in the 2-position does not exert a neighboring-group effect, a mixture of a- and P-glycoside with prevailing a-anomer is obtained by reaction with the Fmoc-protected serine heptyl ester after activation with a soft electrophile (DMTST) [52].Subsequently, the C-terminal heptyl ester in 13 was cleaved in a lipase-catalyzed reaction [53] to give building block 14. Coupling of 14 with preformed glycosylated dipeptide 15 using water-soluble carbodiimide/HOBt gave the bisglycosylated tripeptide which was converted into the N-acetyl galactosamine derivative by use of thioacetic acid. After N-terminal enzymatic cleavage of the C-terminal heptyl ester and coupling with tripeptide 16 (+17), the Fmoc-protected N-terminus was removed and the amino component condensed with tetrapeptide 18 to give the bisglycosylated decapeptide (Scheme 10).

I ) AczOiHC104 2) PBr, 3) Zn/CuS04NaOAc AcOWH20 4) ceric ammonium AcO nitrateNaN3

Gal

-

1 ) EtOCSSK AcO ONo2 2) NaVacetone

OAC

AcO

11

AcO @SE' 12

Fmoc-Ser(OH)-OHep DMTST

\

FmocHN---r-COOHep

Aco+0213 AcO

Scheme 9.

OAc

276

I1 Glycopeptide Synthesis in Solution and on the Solid Phase 1) EDC/HOBt

FmocHN-cH-cooH II 13

*

7H2

pH7,3I0C

go

AcO OAc OAC 14

H-TF-Ala-OHep

I

aAc3GalN3

15

-

aAc3GalNHAc

I

FmocSer-Thr-Ala-Pro-Pro-Ala-OHep

I

2) CH3C(O)SH 3) LipaseM 4) EDCRIOBt aAc3Ga NHAc H-Pro-Pro-Ala-OHep 16

l7

\

1)Morpholine/Ac20 2)Ac-Pro-Ala-Pro-Gly-OH 18 3)Lipase N 4)NaOMe/MeOH GalNHAc

AcPro-Ma-Pro-Gly-Ser-Thr-Ala-Pro-Pro-Ala-OH GalNHAc

19

Scheme 10.

The target molecule 19 was obtained by lipase-catalyzed heptyl ester cleavage and employing sodium methanolate for the removal of the 0-acetyl groups. In summary, this result showed a set of three base-labile protecting groups (Fmoc, OAc, OHep) to be removable selectively and thus useful for the synthesis of complex glycopeptidic structures. A conformational investigation of the target compound 19 was conducted by use of state-of-the-art NMR techniques [49]. The results indicate that carbohydrate residues can influence the conformation of the peptide backbone. Further studies on lipase-catalyzed ester hydrolysis revealed that more hydrophilic structures such as the methoxyethoxy ethyl ester, were generally better substrates for the enzymes, in particular when the peptides and glycopeptides were sterically hindered (Scheme 11) [54, 551. A different approach to obtaining potential Tn antigen-containing vaccines is to assemble dimeric and trimeric structures from Ser(aGa1NAc) building blocks, which are equipped with an aminobutyric acid spacer molecule: Ac-Ser (aGalNAc)-Ser (aGalNAc)-Ser (aGalNAc)-CO-NH-( CH2) X O O H The resulting dimeric antigen was coupled to a lipopeptide and was successfully used to stimulate antibody (IgM and IgG) production in mice [ 5 6 ] . Glycopeptides Carrying the T-Antigen (Gal-GalNAc)

The key steps in the formation of the T antigen linked to serine or threonine usually involve azidonitrate 11 which is converted into the methyl glycoside, then deacety-

lipase

Aloc-Ser(GalNAc)-Val-Phe/O~O~O\

Scheme 11.

Aloc-Ser(GalNAc)-Val-Phe

277

11.2 Synthesis of Glycopeptides in Solution Ph

I

AcO

&q OAc

\n

2) 1) AcOH NIS/TfOH

I+

.+

Hok&

3) AczO/pyridine 4)ACZ"'H2SU4 * 5 ) TiBr4

N3

20

&;% OAc

21

N3

Br

Scheme 12.

lated and regioselectively benzoylated at the 4- and 6-OH groups [57]. This acceptor molecule can be efficiently and stereoselectively galactosylated in position 3 to give the desired disaccharide. Conversion into a suitable donor can be achieved after acetolysis of the anomeric position using TiBr4. Important disadvantages of glycosyl halides as donor molecules are their instability and the need for silver or mercury salts as activating agents. In this context an improvement was achieved by introduction of thioalkyl/aryl donors which can be stored and easily activated by means of a soft electrophile [58]. A combination of thioglycoside methodology with pentenyl glycoside activation [59] led to thiopentenyl glycosides as glycosyl donors (e.g 20) [60]. Attack of iodonium reagents at the double bond and subsequent exo-trig cyclization forms a thiotetrahydrofuran as the leaving group, which can be substituted by the glycosyl acceptor. After conversion of the intermediate methyl glycoside into a bromo glycoside 21, glycosylation of the Fmoc-protected threonine phenacyl (Pac) ester, reduction of the azido group and reductive elimination of the Pac ester [61], a building block for convergent glycopeptide fragment condensations was obtained (Scheme 12). Use of phenacyl protection avoids the application of harsh conditions for the liberation of the C-terminal carboxy group (e.g. tBu/CF3COOH). Coupling of T antigen derivative 22 with pentapeptide 23 gave a fully protected fragment of the MUCl repeating unit. Removal of all remaining protecting groups was achieved in high overall yield by standard deblocking procedures. ROESY NMR analysis suggests that glycosylation of the threonine induces a cis peptide bond between Thr and Arg in 24 (Scheme 13) Fmoc-Thr-OH 1) IIDQ

H-Arg(Pmc)-Pro-Ala-Pro-Gly-OtBu 23

AcO&O& I ) Fmoc-Thr-OPac AgOTf 2)CHxCOSH f

AcA coAc 22

\

2) morpholine 3) AcZOlpyridine 4) TFA 5 ) NaOMeMeOH

Ac-Thr-Arg-Pro-Ala-Pro-GI y-OH 21

H

O OH

Scheme 13.

OH

e

O g 24 OH OH

218

I1 Glycopeptide Synthesis in Solution and on the Solid Phase

AcO

1) dimethyldioxirane 2) 6-O-triisopropylsilyIgalactal/ZnC12 3)AC20, EtjN, DMAP

-

1) TBAF

R=TIPS

26

3) NaN3lCAN 4) PhSWDEA 5 ) CCI&N/KzCO,

AcO


gNH \

2) TMSOTO27 _, ....-3TO27

OAc

0

OAc

N3

CCI3

28

27

Scheme 14.

Glycopeptides Carrying the Sialyl T Antigen (NeuAca2,6[Gal~l,3]GalNAc) The synthesis of a nonapeptide carrying three sialyl T residues was recently reported [62]. As for the preparation of Tn and T antigens, glycal and azidonitrate intermediates play key roles throughout the synthetic strategy. The glycal 25 was subjected to epoxidation. Activation of the epoxide with ZnC12 and reaction with a 6-0-silyl protected galactal gave disaccharide 26. After fluoride-mediated removal of the TIPS group, the 6-OH groups were sialylated by use of sialyl phosphite 27 [63, 641 and activated with TMSOTf. The resulting protected sialyl T glycal was converted into the azidonitrate which was subsequently hydrolyzed. The resulting hemiacetal can be transformed into the trichloroacetimidate 28 by treatment with K2C03 and trichloroacetonitrile (Scheme 14). Upon activation with BF30Et2 the P-trichloroacetimidate 28 gave high yields of predominantly a-glycosylated Fmoc-protected serine benzyl ester. Subsequent to the reduction of the 2-azido groups and the deprotection of the carboxy group the serine and threonine building blocks 29 carrying the sialyl T antigen were introduced into a convergent peptide synthesis. The peptide couplings were achieved with IIDQ, and Fmoc removal was performed with potassium fluoride in the presence of 18-crown-6 [65]. Finally, complete deprotection by standard procedures gave the mucin-type target 30 (Scheme 15). To date, the immunoreactivity of this glycopeptide has not been reported. A similar strategy also using glycals as the crucial carbohydrate building blocks was used in the successful synthesis of glycopeptides containing regioisomeric sialyl (2, 3) T antigen carbohydrate side-chains [66]. An alternative sequence of reactions also based on an early introduction of the sialic acid was applied for the construction of a sialyl T antigen serine/threonine building blocks [67].

279

11.2 Synthesis of Glycopeptides in Solution 28

1 ) BF30Etz/Fmoc-Se/Thr-OBzl 2) CH3COSH 3) Hz/lO% Pd-C AcO

t

a2,6NeuNAc(p I ,3Gal-aGalNAc)

-

'

I

AcPro-Ile-Val-Ser-Th-Ser-Asp-Pro-Val-OH a2,6NeuNAc(p 1,3Gal-aGalNAc)

I

30

I

Scheme 15.

Glycopeptides Carrying 0-GlcNAc The existence of natural glycoproteins containing 0-linked N-acetylglucosamine was unknown until 1984 [68]. Since then, numerous investigations have focused on this ubiquitous structural motif [69]. Recently, it was found that the 0-GlcNAc moiety might be connected with important phenomena such as Alzheimer's disease. The development of methods for a straightforward and efficient approach towards 0-GlcNAc-containing peptides followed evaluation of the biological background. The major synthetic problem is the formation of the b-configured GlcNAc-Ser/Thr linkage. Although the 2-N-acetamido group has the desired neighboring effect, the intermediate cation has a high tendency to form the rather unreactive oxazolidine ring (Scheme 16). To overcome these drawbacks the trichloroethyloxycarbonyl (Tcoc) group was used for temporary protection of the 2-amino group [70]. In combination with the anomeric thioethyl group an efficient donor for the construction of p-GlcNAc-Ser/ Thr building blocks was provided [71].The reactivity of 31 was even high enough to enable the glycosylation of dipeptides in acceptable yields. In particular, the sensitive tert-butyl ester as the C-terminal protection remained unaffected during both the mild activation procedure and the subsequent conversion of the Tcoc group into the acetamide (Scheme 17). With 32 in hand, the orthogonal stability of the tert-butyl and the Z protecting groups facilitates extension of the peptide chain in the N-terminal direction.

Nu-

Scheme 16.

+

""yo

Nu-

280

A AcOc

11 Glycopeptide Synthesis in Solution and on the Solid Phase

ONH G S E31 ~

1)DMTST/ZSerAlaOtBu 1) H2/Pd-C 2) AcOWZn 2) Z-Ala 3)Ac20/pyridine EDCBIEA 2-Ser-Ala-OtBu

-

I

PAC~GICNAC32 0~OCH*CCI,

NHAc

Scheme 17.

Other glucosamine donors based on the N-Tcoc and N-tetrachlorophthaloyl protecting groups in combination with the anomeric trichloroacetimidate also enabled other efficient syntheses of GlcNAc-Ser/Thr building blocks [72]. Despite these achievements the 0-GlcNAc-containing building blocks for the solution-phase synthesis of a bisglycosylated and phosphorylated hexapeptide from RNA polymerase I1 were obtained by the activation of the oxazoline 34 in the presence of urethane-protected serine 33 (Scheme 18) [73]. The unusual N-terminal protection used herein facilitates the subsequent deblocking of the N-terminus by penicillin G acylase under exceptionally mild conditions. For assembly of the target peptide the GlcNAc-Ser derivative 35 had to be modified: elongation at the C-terminus led to glycodipeptide 36 and introduction of the tert-butyl ester and liberation of the terminal a-amino function gave building block 37. After coupling of both GlcNAc compounds (-38)) N-terminal deprotection, condensation with a suitable Pro-Thr moiety, and another N-terminal deprotection, a phosphorylated serine residue was introduced at the N-terminus (-39). Simultaneous removal of the amino and phosphate protecting groups using Pd(0) ally1 transfer followed by standard deblocking procedures gave target 40: 11.2.2 N-Glycopeptides N-Glycopeptides Carrying Natural Saccharide Side-Chains Natural N-glycoproteins usually carry a central (‘core’) pentasaccharide 41 linked to aspartic acid (Scheme 19). The Asn-GlcNAc unit is, therefore, an ideal model compound for methodological investigations towards the preparation of N-glycopeptides. That is, it served for the development of C-terminal protecting groups cleavable by lipases [ 5 5 ] . This class of enzyme was chosen because it does not cleave peptide or other linkages present in glycopeptides. For the synthesis of 0-glycopeptides, heptyl esters as lipase substrates had already been successfully used (see above) [54].In this context it was found that the hydrophobicity of heptyl esters is not a prerequisite for a sufficient conversion. Instead, rates of hydrolysis by lipases were distinctly higher for esters derived from ethylene glycol.

281

11.2 Synthesis of Glycopeptides in Solution

OH

BF3-OEtz

1)

PhAcOZ-Ser-OH

iPrN F O t B u iprm HSer-OtBu

2) penicilline G acylaseMHS03

I

c

A & % !AcO

AcHN

PAc~GIcNAc

35

31 1 ) EDC/HOB~-Pro-OAll 2) Pd(O)/morpholine

EDC/HOBt

PhAcOZ-Ser-Pro-OH

I

PAc~GIcNAc 36

Ph AcOZ-Ser-Pro-Ser-0th

I

PAc~G~cNAc

OPO(OH),

PGlcNAc

\

38

OPO(OAlI),

PAc3GlcNAc

I

I

Aloc-Ser-Pro-Thr-Ser-Pro-Ser-OtBu PGlcNAc

40

Scheme 18.

Mana 1,6

\

Manal,3 /

Man-Pl,4-GlcNAc-P1,4-GlcNAc-P-Asn

\

/ " Z e " :41

Scheme 19.

I

PAc~GIcNAc

39

282

-

I 1 Glycopeptide Synthesis in Solution and on the Solid Phase

PeptdieyO-o,n-O ,. ,

Peptide O

v

OMEE

K 0

O 0’-

OME

42

Scheme 20.

COOMe Tcoc ,

H 43

1

l)AcjGlcNAc-NHZ45 IIDQ Tcoc-Asn-Ser-Ala-OH 2)Lipase N Lipase A6 Tcoc , I N COOMe m-Ser-Ala-OMe PAc3GlcNAc H IIDQ OME 44 4)Lipase A6 46

Scheme 21.

The methoxyethyl (ME) esters (e.g. 42, Scheme 20) combined high substrate acceptance with optimum purification properties and high solubility in aqueous solution. As a model target in the area of N-glycopeptides glycotetrapeptide 46 was synthesized. Starting from the bis-ME aspartic acid ester 43 the p-ester was selectively cleaved in a lipase-catalyzed reaction. Condensation of compound 44 with AqGlcNAc-NH2 45 gave the fully protected asparagine derivative. Upon treatment with a different type of lipase, the a-carboxy group could be deblocked and condensed with a dipeptide ME ester. After elongation at the N-terminus the resulting ME ester again could be cleaved by use of a lipase (+46, Scheme 21). In summary, the selectivity and mildness of the lipase-catalyzed reactions was established as a tool applicable to multifunctional and sensitive glycopeptide targets in particular. Variants of the ‘core’ 41 shown in Scheme 19 with increased structural complexity were found, which can be related to pathogenic phenomena. As a basis for further biological investigations, the complete N-glycan of the so-called complex type carrying a pentapeptide was recently synthesized using a chemoenzymatic strategy (Scheme 22) [741. This molecule contains several preparative challenges that warrant more detailed description. The synthetic strategy applied can be summarized as:

I

xxx

NeuSNAca-2,6-Galp-l,4-GIcNAc~-I ,2-Mana-l,

\

I

Man-P-l,4-GlcNAc-P 1,4ClcAc-P-Asn 2,6-Galp- 1,4-GlcNAc~-1,2-Mana-1,4

I

xxx I

Scheme 22.

283

11.2 Synthesis of Glycopeptides in Solution

1) 2) 3) 4)

assembly of heptasaccharide 47 [75]; introduction of aspartic acid at the anomeric center; selective deprotection and extension of the peptide chain; and sequential enzymatic transfer of galactose and neuraminic acid on to the nonreducing ends of the glycan chain

The synthesis started from chitobiosyl azide 48 which is available in 11 synthetic steps. The azido group herein is essential for the chosen strategy because it: 1) remains stable during the applied reaction conditions, and 2) enables the generation of the anomeric amino function required for introduction of aspartic acid [76]. Next, a P-mannoside linkage must be established and two major problems must be overcome because commonly used neighboring groups, and the anomeric effect lead to the formation of the undesired a-mannosides. During the past 20 years several efficient solutions to this key problem have been provided [22].The strategy used here is based on the epimerization of the 2-hydroxy group subsequent to the formation of a p-glucosidic linkage (Scheme 23) [77, 781. To this end, a phenylcarbamoyl group at 3-OH and a leaving group at 2-OH (OTf) were introduced (49). An intramolecular s N 2 reaction (DMF, 40-50 "C) resulted in

NPth

\

OBzl

AcO

+

48

&

o+

6"

OAc

49

1) BFtOEt,

2) K2603kIeOH 3) dimethoxytoluene/H+

phqo

4) 5 ) TfiOlpyridine DMFipyridine

0HO a B

6 ) AcOWdioxane 7) NaOMdMeOH

O

e

:

z

An-, UDLl

&

NPth N3 50

I

\ OAc

47

Scheme 23.

g

1) A-trichloroacetimidate BF30Et2 2) AcOW80"C 3) B-trichloroacetimidate BF30Et2 I

NPth OBzl

284

11 Glycopeptide Synthesis in Solution and on the Solid Phase

the formation of the 2,3-cis carbonate which upon treatment with NaOMe/MeOH gave the trisaccharide acceptor 50. Advantageously, the 2”,3”-diol 50 was now selectively glycosylated with the disaccharide donor A at position 3”. After removal of the 4”,6/’-benzylidene group a second selective glycosylation with B gave heptasaccharide 47 in high yield. This glycan was now subjected to a manipulation of protecting groups followed by the conversion of the azido group into the P-configured amino function, by use of propanedithiol, and condensation with a fully protected aspartic acid derivative. All remaining protecting groups were subsequently removed by Pd-catalyzed hydrogenation and the Fmoc group was introduced at the terminal amino function. The efficiency and selectivity of modern peptide-coupling reagents thereafter facilitated the extension of the peptide chain at C- and N-terminal functional groups in the presence of the totally deblocked heptasaccharide side-chain. All remaining protecting groups were removed before the final enzyme-catalyzed transfer of galactose and neuraminic acid on to both non-reducing ends of the glycan chains (Scheme 24). During these transformations the addition of alkaline phosphatase was essential to prevent product inhibition by UDP [79, 801. These final steps yielded the target undecasaccharide glycopentapeptide 51 in 91% yield for the last two steps. It is obvious why the authors of the aforementioned synthesis have preferred peptide assembly in solution to the solid-phase approach:

1) the synthetic effort required for peptide assembly is small compared with the construction of the saccharide chain

YOOHHo

I

s,, I

Ser

I

Ser I Asn

I

Phe

HO

Scheme 24.

51

11.2 Synthesis o j Glycopeptides in Solution

285

2) on the solid phase, a molar excess of building blocks must usually be used to ensure high coupling yields. For the saccharide-asparagine moiety this would have entailed tedious re-isolation of this valuable compound. In general, therefore, glycopeptides with a comparably short peptide sequence and complex and valuable carbohydrate parts can be efficiently prepared in solution. This is also true for N-glycopeptides carrying Lewis antigen-type side-chains. As an alternative to the chemical and chemoenzymatic routes, N-glycopeptides can be obtained by a convergent strategy [ 8 1] using glycans isolated from natural Nglycoproteins or oligosaccharides and presynthesized peptides 1821. The unprotected saccharide having the reducing terminal N-acetylglucosamine is converted into a glycosylamine by use of excess ammonium hydrogen carbonate [24] and then coupled with a peptide [82]. By this strategy, large oligosaccharide side-chains can be introduced in the molecules. This is also true for the construction of N-glycopeptides from N-acetylglucosamine-containing glycopeptides and extension of their glycan portion by trans glycosylation-catalyzed saccharide transfer from N-glycoproteins [ 831. Despite the possible microheterogeneity of the natural glycoprotein glycans, the limitation of these reactions is that only analytical amounts (<1 pmol) are obtained. N-Glycopeptides with Lewis-Type Saccharide Side-Chains

The biological importance of Lewis-type saccharide structures, and typical methods for their chemical preparation, have already been outlined in the introduction. In particular, the preparation of peptide backbones carrying a high density of Lewis type antigens is of increasing interest. Such clustered structures are often more immunoreactive than the corresponding monovalent antigens [84]. This observation was the motivation for a chemoenzymatic synthesis of the trivalent N-glycopeptide 52 (Scheme 25) [85]. Selectively protected GlcNAc-Asn units were prepared as suitable precursors and conjugated via an amino acid of variable length. The trivalent glycopeptide was converted into the corresponding tris-sialyl LewisX glycoconjugates by stepwise elongation with galactosyltransferase/UDP-galactose, sialyltransferase/CMP-sialic acid, and fucosyltransferase/GDP-fucose. By this approach, 20 mg of target compound 52 were obtained and used for E-selectin binding assays. An inhibition four times stronger than that of the monovalent SLe"-Asn residue was achieved. Without the restrictions of enzymatic reactions, such as the strict specificity of glycosyltransferases and the limited availability of enzymes and nucleotide substrates, a purely chemical approach has its advantages in generating a broader structural variety and larger product quantities. For this purpose, the SLeXiaazides are ideal precursors (see Scheme 6). They can be: 1) produced on a gram scale; 2) easily converted into the corresponding P-glycosylamino derivatives and 3 ) efficiently coupled to form stable amide linkages.

286

I1 Glycopeptide Synthesis in Solution and on the Solid Phase 0

HO

0

HO

J

HO

J

Scheme 25.

In this manner, SLe" amine 53 was combined with a tetrapeptide partial sequence from fibrinogen required for binding to the integrin receptor GpIIb [29]. For this conjugation the C-terminal alanine of RGDA derivative 54 served the as coupling position (Scheme 26). The fully deprotected SLeX-RGDA glycopeptide 55 inhibited P-selectin binding with high affinity (1Cso 25 pmol), but had no affinity for E-selectin. Structurally even more sophisticated is the trivalent cycloheptapeptide 57 (861. The synthesis of this complex molecule involved the preformation of the cyclic peptide 56 containing three aspartic acid residues (Scheme 27). After removal of the tert-butyl side-chain protecting groups simultaneous condensation with three equivalents of SLeX-lactone amine 53 using HATU/HOAT/DIEA gave the cycloheptapeptide carrying three SLeX side-chains. This was fully deprotected in two further steps. The resulting trivalent antigen proved to inhibit the binding of human tumor cells (HL60) to E-selectin in the low millimolar range (IC50 0.35 mM), but in contrast with 55 had no affinity for P-selectin.

11.3 Glycopeptide Synthesis on the Solid Phase In contrast with the construction of glycopeptides in solution, automated assembly on solid supports promises rapid access to the desired target compounds [87-891.

11.3 Glycopeptide Synthesis on the Solid Phase

53

OH

287

I OBzl

nn-1

OH

NH

I

Scheme 26.

This only holds true, however, if certain structural and methodological prerequisites are fulfilled: 1) glycosylated amino acids must be provided in sufficient quantity as N-Fmocprotected compounds containing a free a-carboxylic function and 2) reaction conditions and polymer-anchoring systems applied during the solidphase procedure and the cleavage and deprotection steps must be adapted to the sensitive glycosylated compounds. Throughout the following section, solutions of these problems will be presented in connection with recently published solid-phase 0- and N-glycopeptide syntheses. 11.3.1 O-Glycopeptides Glycopeptides Carrying N-Acetylgalactosamine (Tn-Antigen)

The biological relevance of a-GalNAc-carrying glycopeptides has been outlined above. For the preparation of a glycopeptide from the homophilic recognition domain of Epithelial Cudherin 1 (E-CAD, 58, Scheme 28) the particular importance of the anchoring structure and of adapted coupling conditions can be demonstrated [901.

288

I 1 Glycopeptide Synthesis in Solution and on the Solid Phase

51

Scheme 27.

58

Scheme 28.

11.3 Glycopeptide Synthesis on the Solid Phase

289

58

Scheme 29.

The depicted target efficiently was constructed with the help of the allylic anchoring principle and a newly developed coupling reagent. To this end, the starting amino acid (Glu) was linked to the polymeric support (Tentagel@)via the Hycron anchor comprising the internal standard p-alanine, a flexible and polar oligo(ethy1ene glycol) spacer, and a 1,4-dihydroxybut-2-ene unit (59, Scheme 29). Subsequent initial attempts to synthesize the E-CAD partial structure with the help of TBTU [91] as coupling reagent gave only low yields although TBTU had already proved efficient in glycopeptide synthesis. Obviously, this sequence is prone to back-folding, and particularly efficient activation of the corresponding amino acids building blocks had to be achieved. This problem was finally overcome by the development of a new coupling reagent: PfPyU ( N ,N , N ’ , N’-bis(tetramethylene)-O-pentafluorophenyluronium hexafluorophosphate) 60 [92]. Compared with TBTU, activation with PfPyU in model reactions resulted in eightfold increased reaction rates. This observation was confirmed during the E-CAD synthesis. First, Fmoc-AlaVal-OH was coupled to the starting glutamic acid. This step was advisable to prevent diketopiperazine formation on the stage of the resin-bound dipeptide. All coupling steps were conducted using Pf PyU and were followed by morpholine-

290

-

11 Glycopeptide Synthesis in Solution and on the Solid Phase "Rink"

BzO Fmoc-He-Ser-Gly-Ile-GI

OH

NH

2) TFA/H20

Fmoc-Ile-Ser-Gly-lle-Gly-NH2

I

aBz3GalN3

Scheme 30.

mediated Fmoc removal. Finally, the fully assembled glycopeptide was detached from the resin by Pd(0)-catalyzed ally1 transfer to N-methylaniline and gave the glycododecapeptide 58a in an overall yield of 55%. After removing all acid labile protecting groups, the 0-acetyl and Fmoc groups were cleaved by Zemplen transesterification. The high overall yield and purity of 0-glycododecapeptide 58a demonstrated the efficiency of the Hycron/Pf PyU methodology. The approach described above employed conventional insertion of a preformed GalNAc-Ser building block into the peptide chain to be synthesized. A method presented recently performs solid-phase glycosylation subsequent to peptide assembly [93]. Obviously, with this strategy the construction of glycopeptide libraries containing different glycan side-chains becomes feasible. The improvement achieved by the authors concerned the nature of the polymer support. A polyoxyethylenepolyoxypropylene (POEPOP) resin with balanced polarity and hydrophobicity was developed. In combination with an acid-labile RINK linker [94] this system served for the synthesis of pentapeptide 62 containing one serine residue (Scheme 30). Moreover, the properties of the POEPOP resin enabled the solid-phase glycosylation of the serine hydroxy group using (among others) perbenzoylated 2-azidogalactosyl donor 61. The target Tn glycopeptide 63 was obtained in high yield after detachment from the resin. 0-Glycopeptides Carrying the T Antigen (Gal-GalNAc)

Similar to the above mentioned on-resin glycosylation of hydroxy amino acids, the solid-phase elongation of saccharide chains linked to peptides has also been described. With the help of this method the preparation of T antigen carrying MUC2

11.3 Glycopeptide Sjmthesis on the Solid Phase

291

Ac-Pro-Thr(tBu)-Thr-Thr(tBu)-Pro-ile-Ser(tBu)-Thr(tBu) -RINK-

BIo%0Yc13 BzO 64

NH

Ho+

G-" Ph

Ac-Pro-Thr-Thr-Thr-Pro-Ile-Ser-Thr-NH2

66

HO OH

Scheme 31.

fragments was achieved [95]. To this end, a suitably protected a-azidogalactosyl threonine unit was used in a solid-phase glycopeptide synthesis (Scheme 31). In the resulting glycooctapeptide 65 the unsubstituted 3-OH group of azidogalactose served as an acceptor for a solid-phase P-galactosylation using galactosyl trichloroacetimidate 64 [30]. After optimization of the reaction conditions and reagent purity, the disaccharide octapeptide 66 was obtained in 67% yield after cleavage from the resin with TFA and complete deprotection. It should again be noted that the resin employed had a decisive influence. 0-Glycopeptides Carrying the Sialyl Tn Antigen (NeuNAc-a2,6-GalNAc)

The sialyl Tn antigen is considered to be one of the most important tumor-associated carbohydrate antigens [46, 961. Because of its immunoreactivity the application of clustered STn conjugates as cancer vaccines is already being investigated in clinical studies [47]. Because it is assumed that both the peptide and carbohydrate parts of the corresponding conjugates are important for a tumor-selective cellular and humoral immune response [97], the synthesis and application of tumor related glycopeptides are other major areas currently being investigated. Only recently has the successful incorporation of a suitable STn-Ser/Thr building block into a solidphase glycopeptide synthesis been achieved [98]. The key element of the strategy used here consisted in the construction of the disaccharide threonine unit 70. This was achieved starting from the known GalNAc-Thr moiety already carrying the required Fmoc and the tert-butyl ester protecting group (Scheme 32). Despite the base-sensitive Fmoc group, the removal of the 0-acetyl groups was selectively accomplished by careful treatment with NaOMe in MeOH at pH 8.2.

292

I1 Glycopeptide Synthesis in Solution and on the Solid Phase

YOAc

COOtBu HNFmoc

I ) AC20 1 p p d ~ n e 88% , 2) TFA,anaole, quant

69

B

OAc

OAc

70

Scheme 32.

The resulting trio1 acceptor 68 was then sialylated by use of sialyl xanthate 67 [34] and methylsulfenyl triflate/AgOTf [37] as the promoter. At -62 "C the reaction proceeded almost diastereoselectively and regioselectively and the glycosylated threonine derivative 69 was obtained in 32% yield after purification. The desired building block 70 for solid-phase synthesis was provided after O-acetylation and cleavage of the tert-butyl ester group. Glycosyl threonine 70 was subsequently incorporated into a partial sequence of the MUCl repeating unit known to be tumorrelevant [99]. To this end, the starting amino acid proline attached to the Hycron anchor (see above) was amino-deprotected and coupled with Boc-Ala-OH. Changing to the Boc methodology at this stage prevented diketopiperazine formation on the resin-linked dipeptide strongly favored for an Ala-Pro sequence (Scheme 33). This intermittent change of protecting groups was possible because of the stability of the anchor towards acids and bases. Further chain extension and Pd (0)-catalyzed detachment from the resin gave the protected glycoundecapeptide. After acidolytic removal of the side-chain protecting groups the O-acetyl groups and the methyl ester were cleaved by carefully optimized treatment with dilute aqueous NaOH in MeOH and aqueous NaOH (pH 11.5) (-71). At pH < 11.5 no methyl ester cleavage was achieved whereas above this pH epimerization and b-elimination of the carbohydrate occurred. The attention paid to the preparation of STn and other sialic acid-containing glycopeptides of biological importance became apparent with the appearance of another publication reporting the successful synthesis of a STn-carrying decaheptapeptide from HIV at almost the same time [loo].

11.3 Glycopeptide Synthesis on the Solid Phase

293

aAcgNeu(OMe)-(2-6)-aAc2GalNac

I

Ac-Ala-Pro-Pr~Ala-His(Trt)-Gly-Val-Thr-Ser(tBu)-Ala-Pro-HYCRON-~Ala-AMPS

I)[Pd(PPh3)4], DMSO I DMF, rnorpholine 2) TFA, anisole, EtSMe 3) NaOH, MeOH 4) 5 mM NaOH

H

*’HO HO

71

Scheme 33.

0-Glycopeptides Carrying the 2,3-Sialyl T Antigen

Another biologically important saccharide found in nature and resulting from incomplete glycosylation (see Tn, T, STn) is the sialyl T antigen [ 1011. Only recently, a 27 amino acid peptide carrying the 0-glycosidically linked a2,3NeuNAcGalpl, 3GalNAc structure was synthesized [102]. Again, a glycosylated serine had to be prepared before the solid-phase procedure (Scheme 34). Starting with di-0-benzyl galactose residue 72 sialylation of the 3-OH was achieved by use of the perbenzylated sialic acid donor 73 [103]. This donor was temporarily equipped with a thiophenyl residue in the 3 position to ensure high stereoselectivity in the mercury salt-promoted a-sialylation. Upon desulfurization and treatment with DBU a disaccharide was obtained wherein the lactone serves as an internal protecting group. After oxidative cleavage of the anomeric methoxyphenyl group and conversion into the trichloroacetimidate 74, the fully protected trisaccharide serine was obtained by BF3 .OEt2-mediated reaction of 74 with TBDMS-protected azidogalactosyl serine residue 75. Protecting group manipulation and carboxy deprotection using Pd(O)/N-methylaniline gave building block 76 which was used for the solid-phase assembly of the peptide employing dicyclohexylcarbodiimide/HOBt as coupling reagent and an acid-labile anchoring group.

294 HO HO

I 1 Glycopeptide Synthesis in Solution and on the Solid Pliase OBzl

Bzlo

BzlO

72

+

BzlO

W AcHN C

’

OMe

O

BzlO

O

BSPh z

I

73

I ) HgO)

2) Ph3SnH. AIBN 3) DBWTHF 4) CAN 5 ) CCI3CNDBU I ) BF30Etz

76

75

Fmoc-Ser-OAll

NeuNAca2,3Gal!31,3GalNAca 1

I

H-Thr-Val-Val-Gln-Pro-Ser-Val-Gly-Ala-~a-Ala-Gl~~o-Val-Va~-Pro-Pro-Cys-Pro-Gl~~g-Ile-~g-His-Phe-Lys-Val-OH

77

Scheme 34.

Cleavage from the resin was followed by an unusual debenzylation step which was performed using TMSOTf and TFA [104]. By this procedure the hydrogenolytic cleavage of benzyl ethers-often found difficult in such complex molecules-was avoided. Finally, opening of the lactone ring with NaHC03/D20 and treatment with 1,4-dithiothreitol to reduce the undesired disulfide bridges gave the target glycoheptacosapeptide 77. 0-Glycopeptides Carrying 0-GlcNAc Side-Chains

The biological importance of 0-GlcNAc-carrying peptides has already been mentioned. Again, a prerequisite for the solid-phase approach is the availability of Fmoc-Ser/Thr(GlcNAc)-OH building blocks. Doubtlessly, the most rapid synthesis of these derivatives was achieved by Lewis acid-promoted activation of peracetylated glucosamine in the presence of Fmoc-Ser or Fmoc-Thr. Including the preparation of the starting compounds only three steps gave the desired glycosylated amino acid. In this instance the sugar oxazoline is the reactive intermediate [ 1051. A similar conversion can be performed starting from the corresponding N-Tcoc derivative [ 1061. The number of protecting-group manipulations can also be reduced by use of

11.3 Glycopeptide Synthesis on the Solid Phase

ACO&O&OQ; AcO

H *t

HNAC

R

O

F

R=Y CH3

/

295

OH

H-Ala-Val-Ser-Thr-GIu-Pro-Phe-Gly-Arg-NH~ I

I

F 78

79

HO HO HNAC

Scheme 35.

Fmoc-Ser/Thr pentafluorophenyl esters [ 1071. The latter remain stable during the glycosylation process and so suitable solid-phase building blocks 78 are obtained after the generation of the N-acetamido residue. Solid-phase synthesis with Pfp esters was performed on a poly(ethy1ene glycol)-derived resin equipped with an acidlabile linker [108]. By following a standard protocol for the coupling reactions, detachment from the resin, and deprotection, the bisglycosylated nonapeptide 79 was obtained and used for detailed NMR investigations (Scheme 35). A chemoenzymatic approach to O-GlcNAc peptides with less general applicability has recently been reported [ 1091. O-Glycopeptides Carrying O-Linked Fucose

Nothing was known about the appearance of O-fucosylated proteins until the early 90s. Then, the uncommon Fuc-Ser/Thr motif was found in the growth factor domains of several coagulation and fibrinolytic proteins [ 1101. The structure also occurs in a small serine protease inhibitor isolated from Locustu migratoriu (83; Scheme 36) [ 1 I I]. In the latter both the anomeric configuration of the fucose and its biological function remain unclear. To evaluate the influence of the single fucose moiety on the conformation and activity of this protein, the fucosylated and unfucosylated form were synthesized [ 1121. The required a-fucosylated threonine building block (which was assumed to be the more likely a isomer) was prepared in four steps from a suitable Mpm etherprotected 1-thioalkyl fucose 80 and the Fmoc/tBu-protected threonine derivative 79 [71]. Activation of the donor 80 was achieved with DMTST and resulted in the formation of the a-configured glycoside 81. Because of the acid-sensitivity of the afucosidic linkage, 2,6-di-tert-butylpyridine had to be added to the reaction mixture. The Mpm ethers were removed by oxidation with ceric ammonium nitrate followed by O-acetylation and cleavage of the tert-butyl ester with formic acid. The intermittent exchange of the hydroxy protecting groups was essential to enhance the acid stability of the fucoside before liberation of the terminal carboxy group and final peptide deprotection [40]. By use of the building block 82, the solid-phase synthesis was performed by means of an acid-sensitive resin and BOP [ 1 131 as coupling reagent. Release from the solid support and side-chain deprotection was achieved by

296

I1 Glycopeptide Synthesis in Solution and on the Solid Phase

-

Fmoc-Thr(0H)-OH tBuOWDCC CUCl Fmoc-Thr(0H)-OEJu

MpqFucSEt 80

Fmoc/'?OtBu

1 DMTST 2,643 -ferfbutylpyridine

@O ' MOMpm pm MpmO

1) CAN 2) AczO/pyridine

3, HCooH 81

-

Fmoc-Thr(aAc,Fuc)-OH 82

Fucose

I

HOOC -

Scheme 36.

treatment with TFA. Finally, disulfide formation and 0-acetyl removal were performed simultaneously in diethylmethylamine in the presence of air. Careful HPLC purification gave the monofucosylated peptide 83 containing 36 amino acids and three disulfide bridges. A straightforward synthesis towards the fucosyl threonine block 82 was performed by direct activation of peracetylated fucose with BF3.0Et2 in the presence of Fmoc-Thr-OH [114]. Small amounts of 0-configured impurities had to be removed from the crude product by HPLC, however. A more laborious procedure for the construction of building block 82 and its incorporation into a undecapeptide has also been reported [ 1 151. 0-Glycopeptides Carrying a Sialyl Lewis Antigen Structure The first synthesis of a glycopeptide carrying 0-linked SLe" was recently described 11161. The major achievement therein consisted of the enzymatic assembly of the SLe" tetrasaccharide on a suitable solid support. This approach promises a rapid and variable construction of complex glycopeptides. The starting point was the preparation of the corresponding O-GlcNAc-containing octapeptide 85 linked to controlled-pore glass (CPG) via the aforementioned Hycron linker [ 1171 by use of standard protecting groups and coupling conditions. Because the ally1 ester unit herein proved unstable under the typical conditions used for the deacetylation of sugar side-chains, a building block with unprotected sugar

11.3 Glycopeptide Synthesis on the Solid Phase

%

OAc

AcO AcO

1) BF30Et*/Z-Thr 2) NaOMehleOH 3) Pd-C/MeOH 4) Fmoc-OSu

Fmoc-Thr-OH

NHAc

84

" - h o

/ w

NHAc

--

Ac-Lys-Pro-Pro-Asn-Thr-Thr-Ser-Ala-HYCRON-CPG ,H0&0,

297

85

Ac-Lys-Pro-Pro-Asn-Thr-Thr-Ser-Ala-HYCRON-CPG

H6p NHAc

NHAc

\

HO

Scheme 37.

hydroxy groups (84) had to be provided according to the pathway depicted in Scheme 37. After removal of the tert-butyl and trityl side-chain-protecting groups with TFA, galactosylation, sialylation, and fucosylation were successively performed using the corresponding glycosyltransferasesand donor substrates. Finally, the advantages of mild and reliable detachment from the solid support by use of the Hycron/Pd(O) system was demonstrated with this complex and sensitive structure (86).

11.3.2 N-Glycopeptides Compared with O-glycopeptides, access to N-glycopeptides on the solid phase is more readily possible via two different pathways [42]-conventional sequential incorporation of preformed and suitably protected building blocks into the solidphase synthesis or assembly of the peptide chain, selective removal of the aspartic acid side-chain protecting groups, and formation of the amide linkage between glycosyl amines and the peptide backbone [81] on the solid phase. The simplicity of amide formation and access to a great variety of glycosyl amines enables the use of the second method for the synthesis of N-glycopeptides whereas only single examples of solid-phase O-glycosylation have been reported in the synthesis of O-glycopeptides.

298

11 Glycopeptide Synthesis in Solution and on the Solid Phase H-Q(GkNAc)YGGFL-OH H-Q(Fuc)YGGFL-OH H-Q(GlcNAc(p 1,~GICNAC)YGGFL-OH H-Q(GalP 1,4Glc)YGGFL-OH H-Q(GalP 1,3(Fuca 1,4)GlcNAcP1,3GalB1,3Glc)YGGFL-OH H-Q(Gal6SO3)YGGFL-OH

f

H-N(Ga1)YGGFL-OH H-N(Glca 1,4Glcal,4Glca 1,4Glca 1,4Glca 1,4Glcla I,4Glc)YGGFL-OH H-N(GlcN2,6(S0332)YGGFL-OH H-N(Ga16SOi)YGGFL-OH H-N(GaL4)YGGFL-OH

Fmoc-E(OAIl)Y(tBu)GGFL-SASRIN

’/

’

Fmoc-D(OAII)Y(tBu)GGFL-SASIUN

-

Fmoc-Y(tBu)GE(OAII)GFL-SASRIN

H-Y GO(GlcNAcEFL-OH H-YGQ(Gal)GFL-OH H-YGQ(Glca 1,4Glc)GFL-OH

Fmoc-Y(tBu)GGFLE(OAII)-SASRM

H-QGQ(GIc!~I,~GIc)GFL-OH

\

Fmoc-Y(tBu)GGFLD(OAlI)-SASRIN \

H-QGY(GICNAC~SO~~)GFL-OH

\ H-YGGFLQ(MmNAc)-OH H-Y GGFLQ(Glc)-OH H-YGGFLQ(GalP 1,4Man)-OH H-YGGFLQ(Glca 1,4Glca 1,4Glca 1,4Glca 1,4Glca 1,4Glc1a 1,4Glc)-OH H-Y GGFLN(G1cNAc)-OH H-YGGFLN(Ga1NAc)-OH H-YGGFLN(Lm)-OH

Scheme 38.

The Construction of N-Glycopeptide Libraries on the Solid Phase The development of efficient methodology for on-resin N-glycopeptide formation has been described as a starting point toward the preparation of glycopeptide libraries (Scheme 38) [ 1181. Employing an acid-labile anchoring system (Sasrin resin [ 1191) five different hexapeptides containing either glutamic or aspartic acid allyl ester residues were synthesized. Pd(0)-catalyzed selective cleavage of the allyl ester and subsequent formation of the corresponding pentafluorophenyl esters enabled conversion into N-glycopeptides by use of unprotected glycosyl amines and a mixture of DIEA and HOBt. Detachment from the resin and removal of tert-butyl side-chain protecting groups were achieved by treatment with TFA. The combination of mono- and oligoglycosidic amines with five peptides resulted in 23 different N-glycopeptides (see Scheme 38) including interesting products with sulfated glycosyl residues [ 1201. Problems were encountered with fucosylated glycosides-this 0-glycosidic linkage was cleaved even by dilute TFA. Another difficulty that prevents general applicability is the tendency of activated aspartic acid derivatives to rearrange to aspartimides. A solution to this problem was provided by appropriate partial protection of

11.3 Glycopeptide Synthesis on the Solid Phase

299

the peptide backbone. With the help of the N-2-acetoxy-4-methoxybenzyl (AcHmb) group on the alanine residue, aspartimide formation throughout the convergent solid-phase synthesis of a chitobiosylated hexapeptide was suppressed [ 1211. The AcHmb group can be removed in two steps-hydrazine-mediated deacetylation and subsequent treatment with TFA. Sequential N-Glycopeptide Synthesis on the Solid Phase with Oligosaccharides from Natural Sources The use of oligosaccharide residues from natural sources in solid-phase peptide synthesis requires two steps, isolation in sufficient quantity and conversion into suitable building blocks. Whereas conversion of the reducing end of very complex glyco-structures into glycosyl amines has already been achieved [24, 1221, the primary step of detaching sufficient quantities of N-linked oligosaccharides from their natural precursors still requires simplification. Very recently a modified hydrazinolysis procedure has enabled the isolation of di- and trianntenary structures from fetuin and high-mannose type compounds from ribonuclease B on a 100-500-mg scale (Scheme 39) [123]. Upon treatment with aqueous ammonium bicarbonate the corresponding p-configured glycosylamines were obtained and coupled to activated Fmoc-Asp-OtBu. After tert-butyl removal the resulting building blocks were successfully used in solid-phase synthesis. Compounds such as the undecaglycodecapeptide 87 were obtained.

Val

I I Thr I Ile

OH

I<

OH O 'H 0,

OH AcHN

OH

flnu 5

P " LOH O H

Scheme 39.

87

300

I1 Glycopeptide Synthesis in Solution and on the Solid Phase

Certainly, this methodology holds great potential for particular saccharide sidechains that can be isolated from natural sources and are not readily accessible via chemical synthesis.

11.4 Conclusion The synthetic methods reviewed in this chapter and in earlier survey articles now make available glycopeptides of considerable complexity. These methods have been provided by the development of distinctly chemo-, regio-, and stereoselective chemical reactions supplemented by efficient techniques adopted from biochemistry and biotechnology. The step-by-step assembly of glycopeptides in solution is still the method of choice if larger quantities of the target compounds ( 2 1 mmol) are required. The synthesis of glycopeptides containing important but complex saccharide residues remains a major task of glycopeptide chemistry. The introduction of sialic acid residues-considered to be a limitation in the past-can now be achieved efficiently. Remarkable progress has also been made in the use of enzymatic methods. The latter have been shown to be of particular advantage for the assembly of carbohydrate side-chains in the context of linear or convergent synthetic strategies. In combination with improvements in solid-phase technology, e.g. adapted resins and coupling reagents, further progress in the preparation of biologically important glycopeptides can be expected. Abbreviations AgOTf All

Bn Boc BOP Bz Bzl CAN DCC DIEA DMAP DMF DMSO DMTST EEDQ

Fmoc

silver triflate ally1 benzyl t-butyloxycarbonyl benzotriazole-1-yl-oxy-tris(dimethylamin0)phosphoniumhexafluorophosphate benzoyl benzyl ceric ammonium nitrate N ,N' -dicyclohexylcarbodiimide

diisopropylethylamine dimethylaminopyridine dimethylformamide dimethylsulfoxide dimethylthiomethylsulfonium triflate 2-ethoxy-1-ethoxycarbonyl-1,2-dihydrochinoline 9-fluorenylmethoxycarbonyl

References

301

Fuc Gal GalNAc GlcNAc HATU

fucose galactose 2-acetamido-2-desox y-galactose 2-acetamido-2-desox y-glucose 2-( 1H-9-azobenzotriazole-1-yl)-l,l,3,3-tetramethyluronium hexafluorophosphate hydroxyazabenzotriazole HOAt hydroxybenzotriazole HOBt Sheptadecanoyl HYCRON (E)-17-hydroxy-4,7,1O,I3-tetraoxa-l IIDQ 1-isobutyloxycarbonyl-2-isobutyloxy1,2-dihydrochinoline lactose Lac mannose Man p-methoxyphenylmethy1 MPm N-acetyl neuraminic acid NeuNAc PEGA polyethylene glycol-dimethylacrylamide co-polymer pentafluorphenyl PfP PhAcOZ p-(phenylacetoxy)benzyloxycarbonyl POEPOP polyoxyethylene-polyoxypropylene copolymer rotating frame nuclear overhauser spectroscopy ROESY super acid sensitive resin SASRIN sialyl-Lewis-)< SLe" 2-(1H-benzotriazole-1-yl)- I, 1,3,3-tetramethyluroniumtetrafluoro-borate TBTU 'Bu tert-butyl Tcoc N-trichloroethoxycarbonyl TFA trifluoroacetic acid TMSOTf trimethylsilyl triflate benzyloxycarbonyl Z

References H. Lis, N. Sharon, Eur. J. Biochem., 1993, 218, I . A . Varki, Glycohiol., 1993, 3 , 97. R. A. Dwek, Chem. Rev., 1996, 96, 683. G. Arsequell, G. Valencia, Tetrahedron Asym., 1997, 8, 2839. B. K. Hayes, G. W. Hart, Curr. Opin. Struct. Biol., 1994, 4, 692. N. Jentoft, TZBS, 1990, 15, 291. D. Sako, K. M. Comess, K. M. Barone, R. T. Camphausen, D. A. Cumming, G. D. Shaw, Cell, 1995, 83, 323. 8. T. A. Springer, Cell, 1994, 76, 301. 9. M. L. Phillips, E. Nudelman, F. C. A. Gaeta, M. Perez, A. K. Singhal, S.4. Hakomori, J. C. Paulson, Science, 1990, 250, 1130. 10. M. Larkin, T. J. Ahern, M. S. Stoll, M. Shaffer, D. Sako, J. O'Brien, C.-T. Yuen, A. M. Lawson, R. A. Childs, K. M. Barone, P. R. Langer-Safer, A. Hasegawa, M. Kiso, G. R. Larsen, T. Feizi, J. Biol. Chem., 1992, 267, 13661. 11. E. E. Simanek, G. J. McGarvey, J. A. Jablonowski, C.-H. Wong, Chem. Rev, 1998, 98, 833. 12. H. Lis, N. Sharon, Chem. Rev., 1998, Y8, 637. 13. W.-j. Liu, V. Ramachandran, J. Kang, T. K. Kishimoto, R. D. Cummings, R. P. McEver, J. Biol. Chem., 1998, 273, 1078.

I. 2. 3. 4. 5. 6. 7.

302

11 Glycopeptide Synthesis in Solution and on the Solid Phase

14. P. P. Wilkins, R. P. McEver, R. D. Cummings, J. Biol. Chem., 1996, 271, 18732. 15. F. Li, H. P. Erickson, J. A. James, K. L. Moore, R. D. Cummings, R. P. McEver, J. Biol. Chem., 1996,271, 6342. 16. M. DeLuca, L. C. Dunlop, R. K. Andrews, J. V. Flannery, R. Ettling, D. A. Cumming, G. M. Veldman, M. C. Berndt, J. Biol. Chem., 1995, 270, 26734. 17. H. Kunz, M. Schultz, in D. G. Large, C. D. Warren (Eds.): Glycopeptides and Related Compounds, Marcel Dekker. Inc., New York 1997, p. 23. 18. H. G. Garg, K. v. d. Bruch, H. Kunz, Adu. Carbohydr. Chem. Biochem., 1994,50,277. 19. H. Paulsen, Angew. Chem. Int. Ed. Engl., 1982, 21, 155. 20. H. Kunz, Angew. Chem. Znt. Ed. Engl., 1987,26, 294. 21. K. Toshima, K. Tatsuta, Chem. Rev., 1993, 93, 1503. 22. 0. Lockhoff, in H. Hagemann, D. Klamann (Eds.): Methoden der Organischen Chemie, Vol. E14a/3, G. Thieme, Stuttgart 1992, p. 621. 23. R. R. Schmidt, Angew. Chem. Znt. Ed. Engl., 1986,25, 212. 24. L. M. Likhosherstov, 0. S. Novikova, V. A. Derevitskaja, N. K. Kochetkov, Carbohydr. Rex, 1986, 146, C1. 25. H. Kunz, H. Waldmann, Angew. Chem. Znt. Ed. Engl., 1985,24, 883. 26. R. U. Lemieux, J. I. Haymi, Can. J. Chem., 1965,43,2162. 27. H. Lonn, Carbohydr. Res., 1985, 139, 105. 28. R. R. Schmidt, J. Michel, Angew. Chem. Znt. Ed. Engl., 1980, 19, 731. 29. U. Sprengard, G. Kretzschmar, E. Bartnik, C. Huh, H. Kunz, Angew. Chem. Znt. Ed. Engl., 1995, 34, 990. 30. R. R. Schmidt, in H. Ogawa, A. Hasegawa, T. Suami (Eds.): Carbohydrates-Synthetic Methods and Applications in Medicinal Chemistry, Verlag Chemie, Weinheim 1992, p. 66. 31. R. R. Schmidt, ASC Symposium Series, 1994, 560, 276. 32. K. Okamoto, T. Goto, Tetrahedron, 1990, 46, 5835. 33. A. Hasegawa, H. Ohki, T. Nagahama, H. Ishida, M. Kiso, Carbohydr. Res., 1991,212, 277. 34. A. Marra, P. Sinay, Carbohydr. Res., 1989, 187, 35. 35. H. Lonn, K . Stenvall, Tetrahedron Lett., 1992, 33, 115. 36. T. Murase, H. Ishida, M. Kiso, A. Hasegawa, Carbohydr. Res., 1988, 184, C1. 37. F. Dasgupta, P. J. Garegg, Carbohydr. Rex, 1988, 177, C13. 38. W. Stahl, U. Sprengard, G. Kretzschmar, D. W. Schmidt, H. Kunz, J. Prakt. Chem., 1995, 337, 441. 39. H. Kunz, C. Unverzagt, Angewj. Chem. Znt. Ed. Engl., 1988, 27, 1697. 40. C. Unverzagt, H. Kunz, Bioorg. Med. Chem., 1994,2, 1189. 41. B. Liebe, H. Kunz, Helv. Chim. Actu, 1997, 80, 1473. 42. D. G. Large, I. J. Bradshaw, in D. G. Large, C. D. Warren (Eds.): Glycopeptides and Related Compounds, Marcel Dekker, Inc., New York 1997, p. 295. 43. P. L. Devine, I. F. C. McKenzie, BioEssays, 1992, 14, 619. 44. J. J. R. Gum, Biochem. SOC.Trans., 1995, 23, 795. 45. I. Brockhausen, J.-M. Yang, J. Burchell, J. T.-Papadimitriou, Eur. J. Biochem., 1995, 233, 607. 46. S. H. Itzkowitz, M. Yuan, C. K. Montgomery, T. Kjeldsen, H. K. Takahashi, W. L. Bigbee, Y. S. Kim, Cancer Rex, 1989, 49, 197. 47. R. R. Koganty, M. A. Reddish, B. M. Longenecker, in D. G. Large, C. D. Warren (Eds.): Glycopeptides and Related Compounds, Marcel Dekker, Inc., New York 1997, p. 707. 48. S. Gendler, J. Taylor-Papadimitriou, T. Duhig, J. Rothbard, J. Burchell, J. Biol. Chem., 1988, 263, 12820. 49. P. Braun, G. M. Davies, M. R. Price, P. M. Williams, S. J. B. Tendler, H. Kunz, Bioorg. Med Chem., 1998, 6, 1531. 50. R. U. Lemieux, R. M. Ratcliffe, Can. J. Chem., 1979, 57, 1244. 51. H. Paulsen, W. Rauwald, U. Weichert, Liebigs Ann. Chem., 1988, 75. 52. M. Ravenscroft, R. M. G. Roberts, J. G. Tillett, J. Chem. Soc., Perkin Trans., 1982, 1568. 53. A. Reidel, H. Waldmann, J. Prakt. Chem., 1993, 335, 109. 54. J. Eberling, P. Braun, D. Kowalczyk, M. Schultz, H. Kunz, J. Org. Chem., 1996, 61, 2683. 55. M. Gewehr, H. Kunz, Synthesis, 1997, 1499.

References

303

T. Toyokuni, S.4. Hakomori, A. K. Singhal, Bioorg. Med. Chem., 1994, 2, 1119. H. Paulsen, M. Paal, Carbohydr. Res., 1984, 135, 71. P. Fugedi, P. J. Garegg, H. Lonn, T. Norberg, Glycoconjugate J., 1987, 4, 97. D. R. Mootoo, P. Konradsson, B. Fraser-Reid, J. Am. Chem. Soc., 1989, 111, 8540. M. Leuck, H. Kunz, J. Prukt. Chern., 1997,339, 322. B. Liining, T. Norberg, J. Tejbrant, Glyeoconj. J., 1989, 6, 5. D. Sames, X.-T. Chen, S. J. Danishefsky, Nature, 1997, 389, 587. T. J. Martin, R. Brescello, A. Toepfer, R. R. Schmidt, Glycoeonj. J., 1993, 10, 16. M. M. Sim, H. Kondo, C. H. Wong, J. Am. Chem. Soc., 1993,115,2260. J. Liang, W.-R. Li, M. Joullie, Synth. Conzmun., 1994, 24, 187. J. B. Schwarz, S. D. Kuduk, X.-T. Chen, D. Sames, P. W. Glunz, S. J. Danishefsky, J. Am. Chem. SOC., 1999, 121, 2662. 67. S. Komba, M. Meldal, 0.Werdelin, T. Jensen, K. Bock, J. Chem. Soc., Perkin Trans. 1,1999, 415. 68. G. W. Hart, R. S. Haltiwanger, G. D. Holt, W. G. Kelly, Annu. Rev. Biochem., 1989, 58, 785. 69. G. W. Hart, D. M. Snow, Int. Rev. Cytol.; 1998, 181, 43. 70. J. Banoub, P. Boullanger, D. Lafont, Chem. Soc. Rev., 1992, 92, 1167. 71. M. Schultz, H. Kunz, Tetrahedron Asymmetry, 1993, 4, 1205. 72. U. K. Saha, R. R. Schmidt, J. Chem. Soc., Perkin Trans. I , 1997, 1855. 73. T. Pohl, H. Waldmann, J. Am. Chern. Soc., 1997, 119, 6702. 74. C. Unverzagt, Tetrahedron Lett., 1997, 38, 5627. 75. C. Unverzagt, Angew. Chem. Int. Ed. Engl., 1994: 33; 1102. 76. C. Unverzagt, H. Kunz, J. Prakt. Chem., 1992,334, 570. 77. H. Kunz, W. Giinther, Angew. Chern. Int. Ed. Engl., 1988, 29, 1068. 78. W. Gunther, H. Kunz, Curbohydr. Rex, 1992, 228, 217. 79. C. Uuverzagt, S. Kelm, J. C. Paulson, Carbohydu. Rex, 1994, 251, 285. 80. C. Unverzagt, H. Kunz, J. C. Paulson, J. Am. Chem. Soc., 1990, 112, 9308. 81. S. T. Cohen-Anisfeld, P. T. Lansbury, J. Am. Chem. Soc., 1993, 115, 10531. 82. S. Y. C. Wong, G. R. Guile, W. T. Rademacher, R. A. Dwek, Glycoconj. J., 1993, 10, 227. 83. M. Mizuno, K. Haneda, R. Iguchi, I. Muramotot, T. Kawakami, S. Aimoto, K. Yamamoto, T. Inazu, J. Am. Chem. Soc., 1999, 1999, 284. 84. T. Feizi, Curr. ,Opin. Stuuct. B i d , 1993, 3, 701. 85. G. Baisch, R. Ohrlein, Angew. Chem., Int. Ed. Engl., 1996, 35, 1812. 86. U. Sprengard, M. Schudok, W. Schmidt, G. Kretzschmar, H. Kunz, Angew. Chem., Int. Ed. Engl., 1996, 35, 321. 87. S. J. Danishefsky, J. Y. Roberge, in D. G. Large, C. D. Warren (Eds.): Glycopeptides and Reluted Compounds, Marcel Dekker, Inc., New York 1997, p. 245. 88. M. Meldal, Curr. Opin. Struct. Biol.,1994, 4, 710. 89. G. Barany, N. Kneib-Cordonier, D. G. Mullen, Int. J. Peptide Protein Res., 1987, 30, 705. 90. J. Habermann, H. Kunz, Tetrahedron Lett., 1998, 39, 265. 91. R. Knorr, A. Trzeciak, W. Bannworth, D. Gillessen, Tetrahedron Lett., 1989, 30, 1927. 92. J. Habermann, H. Kunz, J. Prukt. Chem., 1998, 340, 233. 93. A. Schleyer, M. Meldal, M. Renil, H. Paulsen, K. Bock, Angew. Chem., Int. Ed Engl., 1997, 36, 1976. 94. H. Rink, Tetrahedron Lett., 1987,28, 3787. 95. H. Paulsen, A. Schleyer, N. Mathieux, M. Meldal, K. Bock, J. Chem. Soc., Perkin Trans. I , 1997, 281. 96. S. H. Itzkowitz, E. J. Bloom, W. A. Kokal, G. Modin, S.4. Hakomori, Y. S. Kim, Cancer, 1990, 66, 1960. 97. H. Kunz, in C. H. Schneider (Ed.): Peptide3 in Immunology, J. Wiley and Sons, Ltd., ChiChester 1996, p. 10. 98. B. Liebe, H. Kunz, Helv. Chim. Acta, 1997, 80, 1473. 99. T. R. E. Stadie, W. Chai, A. M. Lawson, P. G. H. Byfield, F . G . Hanisch, Eur. J. Biochenz., 1995,229, 140. 100. M. Elofsson, L. A. Salvador, J. Kihlberg, Tetrahedron, 1997, 53, 369. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66.

304

11 Glycopeptide Synthesis in Solution and on the Solid Phase

101. J. U. Yang, J. C. Byrd, B. B. Siddiki, Y. S. Chung, M. Okumo, M. Sowa, Y. S. Kim, K. L. Matta, I. Brockhausen, Glycobiology, 1994, 4, 873. 102. Y. Nakahara, Y. Nakahara, Y. Ito, T. Ogawa, Tetrahedron Lett., 1997,38, 721 1. 103. Y. Nakahara, H. Iijima, T. Ogawa, Tetrahedron Lett., 1994, 35, 3321. 104. N. Fujii, A. Otaka, 0. Ikemura, K. Akaiji, J. Chem. Soc., Chem. Comm., 1987, 1987, 274. 105. G. Arsequcll, C. Krippner, R. A. Dwek, S. Y. C. Wong, J. Chem. Soc., Chem. Commun., 1994, 1994, 2383. 106. L. A. Salvador, M. Elofsson, J. Kihlberg, Tetrahedron, 1995, 51, 5643. 107. E. Meinjohanns, M. Meldal, K. Bock, Tetrahedron Lett., 1995, 36, 9205. 108. U. K. Saha, L. S. Griffith, J. Rademann, A. Gcyer, R. R. Schmidt, Carbohydr. Rex, 1997, 304, 21. 109. K. Witte, 0. Seitz, C.-H. Wong, J. Am. Chem. Soc, 1998, 120, 1979. 110. R. J. Harris, M. W. Spellman, Glycobiology, 1993, 3, 219. 111. N. Nakakura, H. Hietter, A. v. Dorssclear, B. Luu, Eur. J. Biochem., 1992, 204, 147. 112. H. Hietter, M. Schultz, H. Kunz, Synlett, 1995, 1219. 113. D. Nguyen, A. Heitz, B. Castro, J. Chem. Soc., Perkin Trans. I , 1987, 1915. 114. M. Elofsson, S. Roy, L. A. Salvador, J. Kihlberg, Tetrahedron Lett., 1996, 37, 7645. 115. S. Peters, T. L. Lowary, 0 . Hindsgaul, M. Meldal, K. Bock, J. Chem. Soc., Perkin Trans. I, 1995, 3017. 116. 0. Seitz, C.-H. Wong, J. Am. Chem. Soc, 1997, 119, 8766. 117. 0. Seitz, H. Kunz, Angew. Chem. Int. Ed. Engl., 1995, 34, 803. 118. D. Vetter, D. Tumelty, S. K. Singh, M. H. Gallop, Angew. Chem., Znt. Ed. Engl., 1995,34, 60. 119. M. Mergler, R. Tanner, J. Gostcli, P. Grogg, Tetrahedron Lett., 1988, 29, 4009. 120. C. Galustian, R. A. Childs, C. T. Yucn, A. Hasegawa, M. Kiso, A. Lubineau, G. Shaw, T. Feizi, Biochem., 1997, 36, 5260. 121. J. Offer, M. Quibell, T. Johnson, J. Chem. Soc., Perkin Trans. I , 1996, 175. 122. A. Lubineau, J. Auge, B. Drouillat, Carhohydr. Rex, 1995, 266, 211. 123. E. Meinjohanns, M. Meldal, H. Paulsen, R. A. Dwck, K. Bock, J. Chem. Soc., Perkin Trans. I, 1998, 549.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

12 Glycolipid Synthesis Hideharu Ishida

12.1 Introduction Glycolipids are classified into three groups, glycosphingolipids, glycoglycerolipids and others which include lipid A of lipopolysaccharide. Gangliosides, sialylated forms of glycosphingolipids, are most intensively studied because of their biological as well as chemical interest. As shown Table 1, gangliosides are further classified into several subclasses depend on the structures of oligosaccharide backbone. In the first section, the synthesis of ganglioside G D l a as a example of ganglioseries ganglioside is described, including the topics which are common even to other series gangliosides, for example preparation of building blocks and transformation of oligosaccharides into glycolipids. In the following section, the synthesis of polysialo ganglio-series gangliosides is described as the extension of this methodology.

12.2 Synthesis of Ganglio-Series Gangliosides In this section, synthesis of ganglioside G D l a as the example of a-series gangliosides which carry sialic acid at 0-3’ of lactose, is described. 12.2.1 Retrosynthetic Analysis of Ganglioside GDla

Retrosynthetic analysis of ganglioside G D l a (Scheme 1 [l]) shows two import concepts which can be applied for all of gangliosides in common. First, lipid part should be incorporated to the sialo-oligosaccharide after complete construction of the sugar chain. When the lipid is coupled with the glucose,

306

12 Glycolipid Synthesis

Table 1. Fundamental carbohydrate structure of gangliosides. hematoside-series Galp(l+I)Cer Galp(1 +4)GlcP( 1+1)Cer ganglio-series

GalNAc~(1+4)Gal~(l+4)Glc~(l+l)Cer Galp( 1+3)GalNAcP( 1+4)Galp( 1-+4)Glcp(1+1)Cer lacto- and neolacto-series [GalP(l+3)GlcNAc~(l+3)]nGal~(1+4)Glc~(l+l)Cer

[Gal~(l+4)GlcNAc~(l+3)]nGal~(l-+4)Glc~(l+l)Cer globo- and isoglobo-series

GalNAc~(l+3)Gala(l+4)Gal~(l+4)Glc~(l+l)Cer GalNAc~(l+3)Gala(l+3)Gal~(l+4)Glc~(l+l)Cer

which is the reducing terminal of oligosaccharide, at the first stage of the synthetic route, the resulting glucosyl ceramide has too low reactivity to serve as a glycosyl acceptor for further elongation of oligosaccharide chain, because mainly of the steric hindrance caused by the lipid. Second, sialic acid should be introduced to the oligosaccharide in the form of building block, e.g. sialylgalactose or sialyllactose. Because of the structural feature, sialic acid affords the corresponding glycoside in good yields only on the coupling with reactive acceptors. When reacted with less reactive acceptors, like oligosaccharides, sialic acid is easily transformed into the 2,3-dehydro derivative which is the major by-product of the reaction and lower the yields. The detail of this matter is discussed separately in this book. It is also of importance to note that the stereochemistry of the anomeric center was used to be determined by the chemical shifts of NMR of H-3ax, H-3eq, H-4 by the empirical rule [2]. This means that the stereochemistry of a novel sialoside synthesized by a coupling of sialic acid and complex sugar chain can not be determined absolutely by NMR. This problem can be solved by employing the building block, the stereochemistry of which has already been determined. 12.2.2 Preparation of Sialylgalactose Donor as Building Block

The sialylgalactose 1 [2] synthesized by the established procedure can be readily transformed into the corresponding glycosyl donor as the building block which is required to synthesize gangliosides in a tailor-made manner (Scheme 2 [3]). Compound 1 was converted by (1) benzoylation of the remaining hydroxyls, followed by either (2) selective transformation of 2-(trimethylsily1)ethyl (SE) group at the reducing end into the corresponding p-acetate by treatment [4] with boron trifluoride etherate-acetic anhydride and (3) introduction of the methylthio group by treatment [51 with (methy1thio)trimethylsilane and boron trifluoride, or (4) chemo-

12.2 Synthesis of' Ganglio-Series Gangliosides

301

,."* OH

AcHN HO "'OH HO

I OBZ

HO-Cl3H2,

HN

x-

C17H35

or AcO

OAc

AcO

OBz

"OAc AcO

HO+CI3H2, N3

SMe

AcO

OAc AcO

I .

n

__

< - s S M e NPhth AcO AcO

"'OAc

Scheme 1.

selective deprotection of SE group by treatment [4]with trifluoroacetic acid and (5) activation as trichloroacetimidate [6] to the desired donors, respectively. The use of resulting donors excludes any possibility of the P-sialyl glycoside formation. The same procedure can be applied for the conversion of disialylgalactose to the corresponding donor (see 3-(2)).

308

12 Glycolipid Synthesis ACO

OAc

OSE

AcHN AcO

’1

AAcOc O U 0 + OAc C02Me OSE

AcHN AcO

2)s 3)

BzO

OBz

2

1 AcO A c

OAc C02Me O U O *

SMe

AcHN

BzO

OBz

3

4 1) Bz,O, DMAP, Pyr., 2) Ac20, BF3*OEt2,3) TMSSMe, TMSOTf, 4) TFA, 5 ) CCISCN, DBU

Scheme 2.

12.2.3 Construction of Oligosaccharide The most important synthetic problems in the systematic synthesis of a-series ganglio-series gangliosides is a formation of GM2 structure, GalNAcb1-3(NeuAca23)Galbl-4Glc, synthesis of which has been found to be difficult because of its compact and lipid structure [7]. As shown in Fig. 1, retrosynthetic analysis of GM2 gives two synthetic strategies; (A): GA2 + NeuAc and (B): GM3 GalNAc. Because introduction of NeuAc into the isolated hydroyxl at 0-3’ of GA2 trisaccharide was found to be difficult by the conventional procedures reported, the strategy B is recommended for use.

+

A GalNAcP1, 4 GalPl+4Glc 3

’

NueAca2

GM2

GalNAcp,

P

4 GalPl+4Glc

NeuAc

+

GalNAc

GA2

\ B NueAca2’

Gal01-4Glc 3 GM3

Figure 1.

+

309

12.2 Synthesis of Gunglio-Series Ganyliosides A

OAc

C

O

AcHN

S

OSE

0 En0 AcO

~

&ty

OAc

~

SMe NPhth

5 A

~

-k

BnO

OBn

HO

~

6 C

AcHN

O

~

0 BnO OBn

~

~

~

,OSE OBn

NPhth

7

BzO 4cHN

+

OBz SMe

ACO""' AcO

C02Me

OBz

OAc

3

OAc

9

1) NIS-TfOH, 2) Lil, 3) N2H4*H20,4) Ac20, Pyr., 5) CH2N,,

6)80% aq. AcOH, 7) NIS-TfOH Scheme 3.

For the systematic synthesis of ganglio-series gangliosides, sialyllactose derivative 5 [8] and galactosamine derivative 6 [8] were selected as the glycosyl acceptor and the glycosyl acceptor, respectively (Scheme 3). The glycosylation of 5 with 6 in the presence of NIS-TfOH as the glycosyl promoter [9]gave the desired P-glycoside 3 in 68% yield. Methyl ester of the sialic acid was cleaved by treatment with LiI [ 101 to give the carboxyl group free derivative to present the side-product formation during

310

12 Glycolipid Synthesis

the treatment with hydrazine monohydrate to cleave the phthaloyl group. Schmidt et al. [ 1 I] employed the N-trichloroethoxycarbonyl group in place of phthaloyl group to protect the galactosamine donor to skip this tedious steps. After conversion of phthalimide into acetamide, isopropylidene group was removed to give the tetrasaccharide acceptor 8. The glycosylation of 8 with 3 in the presence of dimethyl(methy1thio)sulfonium triflate [ 121 gave the expected hexasaccharide in 66% yield.

12.2.4 Transformation of Oligosaccharide into Glycolipid Transformation of oligosaccharides into glycolipids can be accomplished be following our standard methods (Scheme 4). First, benzyl group used in oligosaccharide must be replaced with acetyl group, because final deprotection must be accomplished without affecting the double bond of ceramide. Hydrogenolysis over Pd/C, followed by treatment with Ac2O in pyridine led to replacement of all 0benzyl groups by 0-acetyl groups. Second, SE group at the anomeric position is removed by treatment with trifluoroacetic acid in CH2C12 (41 in a quantitative yield. When the anomeric position was protected with acetyl group, converted form benzyl group as described abve, treatment with hydrazine monoacetate (N2H4.AcOH) [ 131 cleaves the anomeric 0-acetyl group with high regioselectivity. And the following treatment with trichloroacetonitrile in the presence of 1,8-diazabicycl0[5.4.0]undec7-ene (DBU) 1141 afforded the hexasaccharide donor. The resulting glycosyl donor was then coupled with ceramide [15] or azidosphingosine [ 16, 171 which is a synthetic precursor of ceramide. Although the former strategy can shorten the synthetic route, the yields of coupling are rather low. Ogawa et al. [ 161 and Schmidt et al. [ 101 were successful in increasing the yield by employing the pivaloyl group for the protection of hydroxyl at the C-2 of the reducing terminal glucose, in place of acetyl group which causes the formation of the corresponding orthoester to decrease the coupling yield. The latter strategy, the coupling with azidosphingosine, is now popularly employed because of the good yield for coupling and the versatility to accept a series of fatty acids as components of ceramide, although the resulting glycosyl azidosphigosine nessitates the further transformation, reduction of azido to amine and introduction of fatty acids. The reduction was performed mainly by two procedures, treatment with H2S in aqueous pyridine [I81 or treatment with triphenylphoshine [19].When the hydroxyl at C-3 of azidosphingosine is protected with benzoyl group, both of the procedures often cause the migration of benzoyl from hydroxyl to amine formed. This problem can be solved by having the temperature during the reduction or by employing tertbutyldiphenylsilyl group [ 191 for the protection in place of benzoyl group. Finally, the complete deprotection of the acyl group, e.g. acetyl, benzoyl or pivaloyl group, followed by the saponification affords the desired gangliosides. Scheme 4 illustrates our standard methods to transform oligosaccharides into the corresponding gangliosides. When the silyl protection was employed, treatment [20] with tert-butylanmonium fluoride was required before the deacylation.

12.3 Synthesis of Polysiulo Gunglio-Series Gangliosides

3 11

OSE AcHN

OBn OAc

9 1) H2, Pd(OH)2I MeOH 2) AczO, Pyr.

3) CFjCOzH I CHzCl2 4) CCIICN, DBU I CH2CIz

TMSOTf I CH2C12

5) H O & C , ~ H ~ ~ OTBDPS 6) PPh3,HzO I C6Hs

7) stearic acid, WSC I CH2C12 8)l.OM TBAF I CH&N

9) MeONa I MeOH, 0.2M KOH V

OH

\ no-

\

OH

L..

C02H

GD1 a Scheme 4.

12.3 Synthesis of Polysialo Ganglio-Series Gangliosides Polysialo gangliosides are common in nature, having a(2+8)glycosidic linkage between two sialyl residues. In this section, the synthesis of ganglioside G Q l b as an example of polysialo ganglio-series gangliosides is described in comparison with that of GDla.

3 12

12 Glycolipid Synthesis

u1 \

W u)

?

12.3 Synthesis of Polysiulo Gunglio-Series Gangliosides

3 13

12.3.1 Retrosynthetic Analysis of GQlb An efficient method for obtaining disialyl glycoside is the use of dimeric sialic acid donor, which can be readily prepared from naturally available colominic acid. As shown in Scheme 5 [21], ganglioside GQlb can be synthesized in almost same procedure as that of G D l a by employing dimeric sialic acid in place of monomeric sialic acid, except for the construction of GD2 which is the key intermediate for the synthesis of GQlb. It is of worth to note that the dimeric sialic acid donor is enough reactive to glycosylate the isolated hydroxyl at 0-3' of GA2, which can not be glycosylated with the monomeric sialic acid. This surprising result made the synthetic route of GQlb short and practical.

12.3.2 Preparation of Building Block The disialylgalactose obtained by the glycosylation of the galactose acceptor with the dimeric sialic acid donor, was converted into the glycosyl donor as the same procedure as described for the sialylgalactose donor Scheme 6 [22].

11

Bz20, DMAP, Pyr., 2) Ac20, BF3*OEt2,3) TMSSMe, TMSOTf Scheme 6.

3 14

12 Glycolipid Synthesis OBn

AcO 4 0 A c

,OBn

co

+ OBn OBn

12 13

Aco O ,Ac

co

14

/

AcO o,Ac

co

/

11

3,

15 1) NIS-TfOH, 2) 80% aq. AcOH, 3) DMTST

Scheme 7.

12.3.3 Construction of Oligosaccharide

GD2 pentasaccharide acceptor 14 was selected as the key compound to achieve the systematic synthesis of polysialo ganglio-series gangliosides (Scheme 7). Glycosylation of GA2 acceptor 12 with a-sialyl-(2+8)-sialic acid donor 13 in the presence of NIS-TfOH in acetonitrile at -25 "C, afforded the desired pentasaccharide in 50% and no p-isomer was isolated. De-0-isopropylidenation of the pentasaccharide with aqueous 80% acetic acid afforded the acceptor 14 in 91% yield. It is noteworthy that the key intermediate 14 was prepared by one step, de-0-isopropylidenation, after the construction of GD2 pentasaccharide, in contrast to the synthesis of G D 1a. Glycosylation of 14 with sialyl a(2-8)sialyl galactose donor 11 by use of

315

12.4 Conclusion

+

Glc-Cer GaCGlc-Cer

GaCGlc-Cer Ned5Ac

GalNAc-Gal-Glc-Cer

GA2

+

+

Gal-GalNAc-GaCGlc-Cer

+

+ +

Gal-GalNAc-GaCGlc-Cer Ned5Ac

J.

NeyCAc Gal-GalNAc-Gal-Glc-Cer Ned5Ac GDclx I I

I' I/

GM2

+

GaCGalNAc-GaCGlc-Cer N e e Neu5Ac GDlb

GM1

f Gal-GalNAc-Gal-Glc-Cer Ned5Ac Ne@ GTlb Neu5Ac

GDla

I

I

f

GaCGalNAc-GaCGlc-Cer Ne&Ac NedsAc I

,

1

I I

I'

Gal-GalNAc-Gal-Glc-Cer Nei5Ac Ned5Ac Neu5Ac

GT1a

[a-series]

+

Gal-GalNAc-Gal-Glc-Cer Ned5Ac Ned5Ac Ned5Ac Ned5Ac

,

I

I

GT1act

Gal-Glc-Cer Nel(5Ac Neu5Ac GD3

GalNAc-GaCGlc-Cer Ne@ Neu5Ac GD2

Gal-GalNAc-Gal-Glc-Cer NedSAc

GM l b

I I I

GM3

GalNAc-Gal-Glc-Cer Ned5Ac

GA1

I I I I

~

I

GQ1b

[b-series)

NeukAc

GQlbcx Figure 2.

DMTST in CHZCl2 gave the octasaccharide 15 in 52% yield, which was transformed into the final compound by our standard methods described in the previous section.

12.4 Conclusion In this chapter, the methodology to synthesize a series of a-series as well as b-series ganglio-series gangliosides is described. By combining the two procedures, all of

3 16

12 Glycolipid Synthesis

ganglio-series (Fig. 2), GM2 [8, 91, GM1 [l, 91, G D l a [l],G T l a [23], GD3 [22, 241, GD2 [25], G D l b [Zl], G T l b [21], and G Q l b [21]. In addition, a series of novel ganglio-series gangliosides, GTlacl [27] and GQlbcl [28] (also see, Fig. 2) was successfully synthesized as the application of the procedures. Other series of gangliosides, sialyl Le" [29], sialyl dimeric Le" [30, 311, and sialyl Lea [32] of lacto- and neolacto-series gangliosides, and sialyl globopentasaccharide (SSEA-4) [33, 341 and sialyl isoglobopentaosyl ceramide [ 351 of globo- and isoglobo-series gangliosides in a tailor-made manner, respectively.

References 1. A. Hasegawa, H. K. Ishida, T. Nagahama, M. Kiso, J. Curbohydr. Chem. 1993,12, 703-718. 2. T. Murase, A. Kameyama, K. P. R. Kartha, H. Ishida, M. Kiso, A. Hasegawa, J. Curbohydr. Chem. 1989, 8, 265-283. 3. A. Kameyama, H. Ishida, M. Kiso, A. Hasegawa, Curbohydr. Res. 1990, 200, 269-285. 4. a) K. Jansson, T. Frejd, 3. Kihlberg, G. Magnusson, Tetrahedron Lett. 1986, 27, 753-756; b) K. Jansson, S. Ahlfors, T. Frejd, J. Kihlberg, G. Magnusson, J. Dahmen, G. Noori, K. Stenvall, J. Org. Chem. 1988, 53, 5629-5647. 5. V. Pozsgay, H. J. Jennings, Tetrahedron Lett. 1987,28, 1375-1376. 6. a) R. R. Schmidt, J. Michel, Angew. Chem. Int. Ed. Engl. 1980, 19, 731; b) R. R . Schmidt, J. Michel, M. Roos, Liebigs Ann. Chem. 1984, 1343. 7. Y.-T. Li, S.-C. Li, A. Hasegawa, H. Ishida, M. Kiso, A. Bernardi, P. Brocca, L. Raimondi, S. Sonnino, J. Biol. Chem. 1999,274, 10014-10018. 8. A. Hasegawa, T. Nagahama, H. Ohki, M. Kiso, J. Carbohydr. Chem. 1992, 11, 699-714. 9. a) P. Konradsson, U. E. Udodong, B. Fraser-Reid, Tetrahedron Lett. 1990, 31, 4313-4316; b) G. H. Veeneman, S. H. van Leeuwen, J. H. van Boom, Tetrahedron Lett. 1990, 31, 13311334. 10. M. Sugimoto, M. Numata, K. Koike, Y. Nakahara, T. Ogawa, Curbohydr. Rex 1986, 156, cl. 11. J. C. Castro-Palomino, G. Ritter, S. R. Fortanato, S. Reinhardt, L. J. Old. R. R. Schmidt, Angew. Chem. Int. Ed. Engl. 1997, 36, 1998-2001. 12. P. Fiigedi, P. J. Garegg, Curbohydr. Res. 1986, 149, c9-cI2. 13. G. Excoffer, D. Gagnaire, J. P. Utille, Curbohydr. Res. 1975, 39, 308-373. 14. R. R. Schmidt, J. Michel, Angew. Chem. Znt. Ed. Engl. 1980, 19, 731. 15. K. Koike, Y. Nakahara, T. Ogawa, Glycoconjugate J. 1986, 1, 107-109. 16. Y. Ito, M. Kiso, A. Hasegawa, J. Curbohydr. Chem. 1989,8,285-294. 17. R. R. Schmidt, P. Zimmermann, Angew. Chem. Znt. Ed. Engl. 1986,25, 725-726. 18. T. Adachi, Y. Yamada, 1. Inoue, M. Saneyoshi, Synthesis 1977, 45-46. 19. M. Mori, Y. Ito, T. Ogawa, Curbohydr. Res. 1990. 195, 199-224. 20. T. Ehara, A. Kameyama, Y. Yamada, H. Ishida, M. Kiso, A. Hasegawa, Curbohydr. Res. 1996.281.237-252. 21. H.-K. Ishida, H. Ishida, M. Kiso, A. Hasegawa, Tetrahedron: Asymmetry. 1994, 5, 24932512. 22. H.-K. Ishida, Y. Ohta, Y. Tsukadd, M. Kiso, A. Hasegawa, Carbohydr. Rex 1993,246, 75-88. 23. H.-K. Ishida, H. Ando, H. Ito, H. Ishida, M. Kiso, A. Hasegawa, J. Curbohydr. Chem. 1997, 16, 413-428. 24. Y. Ito, M. Numata, M. Sugimoto, T. Ogawa, J. Am. Chem. SOC.1989, 111, 8508. 25. Y. Matsuzaki, S. Nunomura, Y. Ito, M. Sugimoto, Y. Nakahara, T. Ogawa, Carbohydr. Res. 1993,242, cl. 26. H.-K. Ishida, Y. Ohta, Y. Tsukada, Y. Isogai, H. Ishida, M. Kiso, A. Hasegawa, Curbohydr. R ~ s 1994,252, . 283-290. 27. H. Ito, H. Ishida, M. Kiso, A. Hasegawa, Curbohydr. Res. 1997, 304, 187-190. 28. K. Hotta, H. Ishida, M. Kiso, A. Hasegawa, J. Curbohydr. Chem. 1995, 14, 491-506.

References

3 17

29. A. Kameyama, H. Ishida, M. Kiso, A. Hasegawa, J. Curbohydr. Chern. 1991, 10, 549-560. 30. A. Kameyama, T. Ehara, Y. Yamada, H. Ishida, M. Kiso, A. Hasegawa, J. Curbohydr. Chem. 1995, 14, 507-523. 31. G. Hammel, R. R. Schmidt, Tetrahedron Lett. 1997, 38, 1173-1 176. 32. A. Kameyama, H. Ishida, M. Kiso, A. Hasegawa, J. Curbohydr. Chem. 1994, 13, 641-654. 33. H. Ishida, R. Miyawaki, M. Kiso, A. Hasegawa, J. Curbohydr. Chern. 1996, 15, 163-182. 34. J. M. Lassaletta, K. Carlsson, P. J. Garreg, R. R. Schmidt, J. Urg. Chem. 1996,61, 6873-6880. 35. H. Ishida, R. Miyawaki, M. Kiso, A. Hasegawa, Carhohydr. Res. 1996, 284, 179-190.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

13 Stereoselective Synthesis of P-Mannosides Vince Pozsguy

13.1 Introduction P-D-Mannopyranosyl residues are frequently found in oligosaccharides and glycoconjugates in living systems. This unit is omnipresent in N-linked glycoproteins that contain a pentasaccharide core (1) in which the P-linked mannose residue can carry additional sugars [I] (Scheme 1). P-Mannopyranose moieties are often part of the 0-specific polysaccharides (OSPs) of Gram-negative bacteria [2-8] and of bacterial capsular and em-polysaccharides [6, 9-15]. P-Linked mannopyranose is also a common residue in plant polysaccharides [ 161. Further examples of the occurrence of the P-mannosidic linkage are the glycosphingolipids in wheat flour [ 171 and in the freshwater bivalve Hyriopsis schlegelii [ 181. Interestingly, this residue is absent from mammalian glycolipids [19]. Cell-wall phosphomannans of yeasts contain linear chains of up to fourteen P-1,2-linked mannose residues [20, 211. P-Mannopyranose units also occur in secreted small molecules. Examples include the phospholipase inhibitor caloporoside [22] and free oligosaccharides detected in a patient with mannosidosis [23] and isolated from ripening tomatoes [24]. The related P-linked 2-acetamido-2deoxy-D-mannopyranose also frequently occurs in OSPs [ 8, 251 and in capsular polysaccharides [26-3 11. The corresponding mannuronic acid is present in the enterobacterial common antigen [32] and in other bacterial polysaccharides [6, 331. The occurrence of the 0-mannosyl unit in many biological systems creates a biological rationale for its synthesis. Whereas the simple methyl [34-381 and phenyl PD-mannopyranoside [ 391 have been known for many years, the synthesis of complex oligosaccharides containing this unit is still a challenge for two reasons [40-441. First, the anomeric effect facilitates the formation of the 1,2-truns-mannopyranosyl (a) linkage thus it disfavors the P-linkage and, second, the 1,2-cis arrangement of the equatorial aglycone and the axial functionality at C-2 in P-mannopyranosides harbors repulsive steric effects [40]. Glycosyl donors with a participating group at 0 - 2 and halide-ion catalytic conditions promote the exclusive formation of (Y man-

320

13 Stereoselective Synthesis ojP-Mannosides

a-D-Man

1: fa

p-D-Man-(1 -+4)-p-D-GlcNAc-( 14 4 )--pD-GlcNAc-(1+Asn

1

a-D-Man

Scheme 1.

nosides. Once formed, the P-mannosidic linkage might be prone to anomerization under acidic conditions adding another factor to be considered in synthetic sequences involving this linkage [45, 461. Whereas anomeric pairs of galacto- and H-2 coupling constants, glucopyranosides can easily be differentiated by their 3JH-1. this is not so for mannopyranosides. Because of the gauche relationship between the H-1 and H-2 protons in either anomer, the 3JH-1,H-2 couplings (- 1-3 Hz) are not suitable for diagnostic purposes [47]. Reliable information can be gained from in, c-1 heteronuclear coupling traresidual NOE data [48] and from the one-bond, constants [49]. The chemical shift of the H-5 proton (- 3.3-3.4 ppm) is also of diagnostic value [50, 511. The specific rotation can also be used for configurational assignment, if both anomers are available [51]. This chapter reviews methods for the construction of P-mannosidic linkages. Together with the more popular protocols, less widespread and emerging approaches are also covered in the hope they will stimulate further progress.

'&-,

13.2 Chemical Methods 13.2.1 Glycosylation with Mannosyl Donors Mannosylation using Insoluble Promoters Silver cations on the surface of insoluble catalysts anchor mannosyl bromides from the a-side [52]. As a consequence, highly reactive alcohols can attack the anomeric center from the P-side in an S N type ~ reaction, thus leading to P-mannosidic linkages. The first application of this principle was reported by Perlin [53] in 1961 who condensed the cyclic carbonate-protected mannosyl bromide [541 2 and the reactive aglycone 3 under silver oxide activation to afford the first synthetic disaccharide (4) that contains a P-mannopyranosidic linkage (Scheme 2). Because of the strain imposed on the pyranose ring by the fused carbonate moiety in 2, the formation of a planar oxocarbonium ion from it is disfavored and an S N ~ type anomeric displacement predominates [ 551 when strongly nucleophilic primary [56]and secondary alcohols [57-591 are used as the aglycones. Condensations with moderate nucleophiles proceed in lower yields and with reduced P-selectivity. Interestingly, AgzO-assisted reaction of the carbonate 2 with weak nucleophiles gives, exclusively, a-mannosides [ 60, 6 11.

321

13.2 Chemical Methods FHzOAc

&

AcO

Br

CHzOH

HOH& 1 . Ag20.1,

2.Ba(OWzb -t

AcO OAc

2

3

Scheme 2.

+

H

O

OBn

c1 5

~

Ag2C03 O CH2CI2 Yield: 64%

~

E@&&o, Bn

6

7

Scheme 3.

The use of silver carbonate as the promoter has been successful for p-selective condensation between the 0-benzylated donor 5 and the reactive, non-carbohydrate primary alcohol 6 to give 7 (Scheme 3) [62] but it is less practical in other combinations [62-661. Because activation by Koenigs-Knorr promoters produces an equimolar amount of water, loss of the donor as a result of hydrolysis is difficult to prevent even in the presence of drying agents [52]. High yields have been achieved with Paulsen's silver silicate activator precipitated on aluminum oxide that can also act as a water scavenger 1641. A threshold level of acceptor reactivity is necessary for satisfactory p-stereoselectivity-whereas coupling of mannosyl bromide 8 with the anhydro derivative 9 proceeds in an excellent yield and high p-selectivity [67] to afford 10, mannosylation of the less nucleophilic disaccharide 11 with the same donor affords the trisaccharide 12 as a 1 : 1 a/P anomeric mixture (Scheme 4) [68].

-

10 R r -.

8

+ $ G e ONPhth B - & $CH20Bn J ) B n

-

AAll

c

~

&

o

~

o

B

n

All CH20Bn

NPhth

11

Scheme 4.

Yield: 76%

0 CHzOBn

12. cr/D - I

Bn NPhth

13 Stereoselective Synthesis of p-Mannosides

322

1

Br 13

-55 to -30°C Yield: 87%

14

Scheme 5.

+ Br 15

0~~ pphth 16

Ag silicate CH,CI, -15 to25 OC Yield: 22%

Bn

17

OMP NPhth

Scheme 6.

A more predictable approach to this problem has become possible by van Boeckel’s recognition that an electron-withdrawing, acyl substituent at 0-4 of mannosyl donors promotes P-selectivity whereas such substituents at 0 - 3 and 0-6 reduce it [52]. Highly reactive donor-acceptor combinations seem exempt from this rule [69, 701-silver silicate-promoted condensation of the mannosyl donor 13 with n-octanol affords the P-linked mannoside 14 in excellent yield even without an acyl substituent at 0-4 (Scheme 5) [69]. Paulsen’s activator has been used with success to synthesize a large number of P-mannosides related to N-glycoproteins [61, 62, 64, 67-79], glycolipids [ 80-831, and other natural products 1451. Among other insoluble promoters proposed for P-mannoside synthesis van Boeckel’sporous silver silicate [84] has emerged as a most efficient activator [85, 861. This catalyst even promotes 0-selective mannosylation by donor 15 at 0-4 of the disaccharide 16 to afford the 4,6-branched trisaccharide 17 that corresponds to an Asn-linked core structure (Scheme 6) 1861. Garegg’s silver zeolite [87] is also a powerful promoter [82, 88-90] and has been reported to afford increased yields and P-selectivity over silver silicate 182, 911. Other activators, including siluer imiduzolute [92], silver sulicylate [93], and thallium zeolite [88] have also been investigated. The Sulfonate Approach

The leaving group properties of sulfonate groups were first exploited in p-mannoside synthesis by Perlin [94, 951. In the first step of this approach mannofuranose 18 is treated with Tf20 to afford the anomeric triflate 19. This is converted in situ to the bromide 20 by Bu4NBr. Subsequent addition of MeOH led to a 2: 3 mixture of the methyl a- and P-mannosides 21 without heavy metal promoters (Scheme 7). Schuerch has synthesized O-benzyl-protected mannosyl brosylate, tosylate, tresylate, and triflate derivatives 23 from the corresponding chloride 22 and the respective silver sulfonate [96]. On exposure to MeOH, the sulfonates 23 give a nearly 1 : 1 mixture of methyl a- and P-mannosides 24 (Scheme 8).

13.2 Chemical Methods r

323

1

Scheme 7.

Scheme 8.

Scheme 9.

Improved P-selectivity has been obtained by using a mesyl [97-991 or a benzylsulfonyl group [I001 at 0 - 2 of the donor. Such groups are non-participating and by their electron-withdrawing capacity they disfavor anomeric oxocarbonium ion formation, thus favoring direct substitution. For example, reaction of the mesyl derivative 25 with cyclohexanol affords the p-mannoside 26 in 75% yield with excellent p-selectivity (Scheme 9) [99]. Condensation of the benzylsulfonyl-protected donor 27 with the moderately nucleophilic alcohol 28 has led to mannobiose 29 in 49% yield, in a 1 :2 alp ratio (Scheme 10) [loo]. In these studies Schuerch has shown that to achieve optimum pselectivity complete formation of the mannosyl sulfonate is essential before addition of the aglycone component [ 1001. A major development in the sulfonate approach was Crich's serendipitous observation that Kahne-activation of the mannosyl sulfoxide 30 with Tf20 in the presence of DTBMP, at -78 "C in CHzC12, followed by exposure of the intermedi-

z q n +f : Bn CF,CH2S020 27

Scheme 10.

!

M&N

HO

OSO2Bn

< )' ::

Yield:49%

Bn

OMe

28

R = p-BrC&

2 9 . a / p = 1:2

OMe

324

13 Stereoselective Synthesis ojp-Mannosides

phTq

Tf20, DTBML CH?CI?

phTw +

Bn

I

CF3S020

30

u

thenodC

AcOAMe

A20

31

I

OMe

rll

32

Yield: 95%

33, p/a > 25

Scheme 11.

ate triflate 31 to the alcohol 32 led to the disaccharide 33 in an excellent yield and very high P-selectivity (Scheme 11) [51].The a anomeric stereochemistry of the triflate 31 has been demonstrated by low-temperature NMR experiments [101, 1021. Hindered secondary and tertiary alcohols are also suitable acceptors [51, 1031071. Mannosyl triflates can also be prepared from thiomannosides and mannosyl bromides by reaction with PhSOTf, or silver triflate, respectively. As in Schuerch’s method [ 1001, the order of the addition of the reactants is critical for the success. The intermediate triflates should be prepared first before addition of the alcohol to prevent attack by the latter at the initially formed oxocarbonium ion. Additional requirements are the presence in the mannosyl donors of non-participating (benzyl) groups at 0 - 2 and 3 and a benzylidene-acetal moiety at 0 - 4 and 0-6. Crich’s method is likely to enjoy widespread popularity because of its simple experimental protocol combined with high yields and excellent P-stereoselectivity.

Intramolecular Mannosylation The previous examples demonstrate that intermolecular mannosylations proceed with high P-selectivity only with donors that are structurally disposed to direct sN2type substitution at the anomeric carbon and disfavor the formation of oxocarbonium ion intermediates. Because of the large number of variables, the steric outcome of intermolecular mannosylations is predictable only in closely related systems. Stork has proposed a conceptually new, intramolecular mannosylation protocol that has initially been implemented in the first stereocontrolled synthesis of a p-Cmannoside [ 1081. The starting compound is the phenylseleno mannoside 34 in which a 2-phenylethynyl group is anchored to 0 - 2 through a silicon tether. Radicalinitiated removal of the anomeric phenylseleno group (+35) in concert with cyclization of the phenylethynyl group on to the anomeric radical provides 36 from which fluoride-mediated detachment of the silicon connector affords 0-C-mannoside 37 (Scheme 12).

34

Scheme 12.

35

36

37 EK= 10.1

325

13.2 Chemical Methods

Me\

I.BuLi. THF, , S i ,C I 2. Me2SiC12

ye Bn SPh Yield: quant.

Yield: quant.

38

39

OMe

SPh 41

40

MCPBA -78 10 20 "c

-25 to 0 'C

OMe

S(0)Ph Yield: 73%

42

43

Scheme 13.

The principle of intramolecular C-mannosylation has been extended to the synthesis of 0-linked P-mannosides [ 1091. In the first step, a chlorodimethylsilyl moiety is introduced to the primary hydroxyl group of the 0-benzylated aglycone 38 to afford compound 39. Next, chlorosilane 39 is connected to the thiomannoside 40 to give the silaketal intermediate 41. Subsequent conversion to the sulfoxide 42 followed by Kahne-activation [ 1 101 affords the P-linked disaccharide 43 stereospecifically (Scheme 13). This version of the temporary connection method is not suitable for mannosylation of secondary alcohols because of the failure to fashion a silaketal tether between HO-2 of a thiomannoside and a weak nucleophile. More recently, Stork has demonstrated that the silaketal 46 involving a secondary alcohol can be prepared in a one-pot process by using equimolar amounts of the mannosyl sulfoxide 44 instead of the parent thiomannoside, the sugar 45 to be tethered and dichlorodimethylsilane [ l l l ] . Treatment of 46 with Tf20 affords the P(1-3) linked disaccharide 47 in a high yield (Scheme 14). A problem of the Stork method is signaled by a serendipitous finding-that Kahne-activation of the tethered sulfoxide 48 gives mostly the 1,6- (49) instead of the targeted 1,4-linked disaccharide [ 11 11 (Scheme 15). This observation indicates that the silicon-tethered oxygen is not always the most nucleophilic for reaction with the activated anomeric carbon. Hindsgaul has proposed the isopropylidene acetal as an alternative, reactive tether between 0 - 2 of the mannosyl donor and the linkage oxygen of the aglycone B n M Tf20, CH2CIz

OMe -100 10 20°C* %:& BnO nB

Bn

BnO

OMe

S(0)Ph

Yield 88%

44

Scheme 14.

45

B nO OMc

Yield 82% 46 S(O)Ph

41

326

13 Stereoselective Synthesis of p-Mannosides

Scheme 15.

[ 112, 1 131. In the first phase the vinyl ether 51 is prepared from the acetate 50 by methylene-transfer from the Tebbe-reagent. Next, the thioglycoside 51 and the aglycone 52 are connected by acid-catalyzed addition of the latter to the vinyl ether moiety to afford the acetal-tethered intermediate 53. In the final phase, NIS/TfOHinduced anomeric activation proceeds in concert with the attack of the tethered oxygen at the anomeric carbon to afford the P-linked disaccharide 54 (Scheme 16) [114]. Whereas both primary and secondary carbohydrate alcohols are suitable for the linking, the yield is very sensitive to structure and to experimental conditions. A further limitation is the acid-sensitivity of the acetal-tether; this can lead to hydrolysis and transacetalization [ 1141. Ogawa has introduced an interresidual 4-methoxybenzylidene-acetal moiety as the stereocontroling element for intramolecular P-mannosylation [ 1 151. This tether can be fashioned by oxidation of the 4-methoxybenzyl group at 0 - 2 of the mannosy1 fluoride 55 with DDQ to give an intermediate (56) that is stabilized by the aglycone alcohol 57 to form the mixed acetal 58 as a single stereoisomer (Scheme 17) [ 1161. Interestingly, oxidative coupling of the fluoride 59 and the methoxybenzyl ether 60 furnished 58 as a 3 :2 mixture of diastereoisomers [ 1161. The configuration of the acetal carbon atom is dictated by kinetic control. Anomeric activation of the acetal58 afforded the P-mannosyl disaccharide 61 in a stereospecific process that is rather insensitive to the configuration of the acetal (Scheme 17) [116]. The 4-methoxybenzyl-assisted P-mannosylation is also suitable for the blockwise synthesis of higher-membered oligosaccharides [ 117, 1181. For example, DDQ-

Cp2TiCHzClAlMez

SEI

Yield: 90% ,OBn

54

Scheme 16.

Bn'O I OMe

NIS. TfOH

p-TsOH

THF. C6H5CH,

+

DTBMP

OMe

S El 51

~

Bn

52

Yield: 57%

SE1

53

Yield: 77'0

13.2 Chemical Methods

327

1 DBMP

55

MeCN

%aYn B"

Bn

OUII

OBn

Yield: 67% 61

b

59

Bn

OBII

60

Scheme 17.

Scheme 18.

induced attachment of the branched trimannoside thioglycoside 62 to the chitobiose acceptor 63 followed by MeOTf-mediated anomeric activation afforded the pentasaccharide 64 in 42% overall yield (Scheme 18) [ 1181. This compound corresponds to the core oligosaccharide of Asn-linked glycoproteins and represents the most complex setting for which stereospecific pmannosylation has been achieved. The methoxybenzylidene-acetal procedure has also been implemented on a polymer support [ 1 191. In a related approach, studied by Ziegler, the precursor is a 'prearranged glycoside' (e.g. 65) constructed by connecting the mannosyl donor and the acceptor alcohol by means of a persisting succinyl bridge [120, 1211. Activation of 65 by NIS/ TMSOTf gives a 2: 1 mixture of the a- and the P-linked mannosyl disaccharide 66 (Scheme 19) [120]. Anomeric selectivity in this approach is sensitive to the activation procedure and also depends on the glycosyl acceptor component [122]. In contrast with the intramolecular mannosylation methods discussed above, in Ziegler's method the hydroxyl group to be mannosylated is unprotected and the tether remains intact in the glycosylation step.

Other Mannosyl Donor-Based Methods Starting from the epoxide 67 Kochetkov prepared the mannosyl thiocyanate 68 [123] which reacts with the trityl ether 69 in the presence of trityl perchlorate to afford the disaccharide 70 with an a/p ratio of 2 : 1 (Scheme 20) [ 1241.

328

13 Stereoselective Synthesis of P-Mannosides

66, a/p = 2:l

65

Bzo'OBn

Scheme 19.

1. NH4SCN, 18-crown-6 C6HS.acetone, -20 'C 2. CCI3COBr,CSH5N -20 OC Yield: 40%

Bn

CHf.212 SCN

Bn

OMe Yield: 58%

67

68

70, a/p = 21

69

Scheme 20,

Inazu's procedure [125] uses the mannosyl phosphinothioate 72 as the mannosyl donor; it can be prepared from the hemiacetal71 with Me2P(S)C1under basic conditions. Condensation of 72 with the alcohol 73 in the presence of iodine and trityl perchlorate affords the mannosyl disaccharide 74 in 45% yield, with an a/P ratio of 1 : 3 (Scheme 21). Tatsuta has examined the glycosyl donor properties of thiomannosides under NIS/TfOH activation and found that the best P-selectivity can be obtained with the bulky donor 75 that in the presence of the alcohol 76 has afforded the mannoside 77 as a 3 : 1 mixture of the P and a anomers (Scheme 22) [22]. Hashimoto found that low-temperature activation of mannosyl chlorides with silver triflate in the presence of alcohols leads to varying ratios of a- and P-linked mannosides [45, 126, 1271. A study of the factors controlling stereoselectivity in silver triflate-mediated mannosylation has been published [ 1281. Activation of 0benzyl-protected mannosyl fluorides by a combination of Sn(OTf)2 and La(C104)3 has been reported by Shibasaki to afford mixtures of a- and P-mannosides [129]. The large proportions of a-mannosides formed in the homogeneous reactions outlined in this section indicate the involvement of oxocarbonium-ions enabling the anomeric effect to dictate the steric outcome. Because the stereoselectivity of these approaches is crucially dependent on the reacting partners and conditions their scope remains limited.

I . BuLi

OH THF. -30 "C 71

Scheme 21.

Yield: 75%

BnO Bn

-z w

2. Me2P(S)CI

Bn

OP(S)Me2+

72

73

13.2 Chemical Methods OBn

+

gBn

H

o OBn

BnO

329

OTBDPS

NIS,TfOH ~

____C

Bn

OBn QBn

CHzCl2,40 O C Yield: 88%

BnO 15

16

77,a/p = I :3

Scheme 22.

13.2.2 Epimerization of P-Glucopyranosides at C-2 The Oxidation-Reduction Approach Extending Theander's finding [ 1301 that reduction of hexopyranosiduloses proceeds with high stereoselectivity, Lindberg has developed a powerful approach to Pmannosidic linkages [ 13I]. In the first phase, the glucopyranosyl donor 78 that carries a selectively removable, participating group at 0 - 2 is condensed with the aglycone to give the P-glucoside 79. Subsequently, the HO-2 group is deprotected (i 80). Glucal 1,2-epoxides can also be precursors to such intermediates [ 1321. In the next step C-2 is oxidized to give uloside 81. In the final phase, the uloside is reduced to afford the p-manno product 82 together with usually
Direct Inversion Intermolecular nucleophiles The first one-step conversion of P-glucosides to P-mannosides has been reported by Miljkovic [ 1501. In his approach a mesyloxy group at C-2 of glucoside 83 is displaced by a benzoyl group to afford the mannoside 84 (Scheme 24). In a related reaction 2-deoxy-2-fluoro-~-mannoside86 has been prepared [ 15I] by fluoride displacement of the 2-triflyloxy group in 85 (Scheme 25).

78

Scheme 23.

13 Stereoselective Synthesis of p-Munnosides

330 % o T & h +p

BzOK. DMF

Me

OMe

7

MsO

OMe

Yield: 62%

83

84

Scheme 24.

phTow BudNF

BZ

OMe

TfO

p BzOh TOMes

EiGzT Yield: 3540%

86

85

Scheme 25.

These substitutions proceed in moderate yields because of repulsive electrostatic interaction between the incoming nucleophile and the lone pair of the ring oxygen atom [I521 and non-bonded steric interactions in the transition state [150]. Extension of the repertoire of leaving groups has alleviated these problems and made the direct displacement a powerful approach to P-mannoside synthesis. David's variation uses an imidazolylsulfonyloxy group as the leaving group at C2 and Bu4NOBz as the nucleophile [153]. In a similar strategy, a C-2 triflyloxy group is displaced by a combination of CH3C02Cs and 18-crown-6 under sonication to give P-mannosides [ 1541. An analogous method reported by Furstner employs BmNOAc as the nucleophilic reagent in conjunction with ultrasound [ 155, 1561. David has also reported the conversion of double-triflylated P-galactosides to P-mannosides [157, 1581. According to this procedure, Tf0-4 of P-galactoside 87 is displaced by BuNOBz at room temperature to give the gluco compound 88. In the second phase heating at 100 "C causes displacement of Tf0-2 to afford mannoside 89 (Scheme 26). A variation of this approach uses CH3COzCs and CF3C02Cs in combination with 18-crown-6 [ 159, 1601. An improved protocol for the P-galacto--,p-manno conversion and identification of by-products has been reported by Sinay [161]. Intramolecular nucleophiles

Kunz has taken advantage of an entropically favored, intramolecular substitution reaction [ 1621 to construct P-mannosides by configurational inversion at C-2 of P-

Bz

OMe

All

TfO 87

Scheme 26.

25

OC

TfO

sn

Bu~NOBZ * OMe C6H3Me I00 O C Yield: 64%

89

13.2 Chemical Methods

'yo Ac%

ROH AgOTf AcO

C6%

OR 1 , K2C03. MeOH

7

Z.C&CH(OMe): 3 Tf20 Yldd ca 75%

Ac&NHPh -3OOC Yield: 65-808

NHPh

331

90

91

PhTO OO& o l 75 "C Yield: 65-808 93

NHPh 92

OK

PhT& i&, 94

Scheme 27.

glucosides [152, 1631. The first phase is assembly of a P-glucopyranosyl donor (90) with a urethane-type group at C-3. This donor is then condensed with the aglyconic component to form the P-glucopyranoside 91. Subsequently, a benzylidene-acetal moiety is installed at 0 - 4 and 0 - 6 and HO-2 is triflylated (+ 92). In the final stages, thermally induced intramolecular attack of the carbamoyl oxygen at C-2 affords the P-mannoside 93 with small amounts of the iminocarbonate 94 (Scheme 27). The acetal moiety is essential for the success of this approach. Using this procedure, Unverzagt has synthesized the core-region trisaccharide P-D-Manp-1,4-P-DGlcpNAc-l,4-P-~-GlcpNAcin multigram quantities [ 1641. C-2 epimerization by intramolecular substitution has also been achieved with the 3-0-benzoyl analog of the 0-glucoside 92 [165]. 13.2.3 The 2-Ulosyl Donor Method Introduced by Lichtenthaler [42, 1661, this method utilizes the limited tendency of 2-glycosulosyl a-bromides to form oxocarbonium ions upon anomeric activation, thus favoring an S N type ~ reaction at C-I. In the first phase the arabino-hexulosyl bromide 96 is prepared by exposure of the readily available acyloxyglycal 95 to NBS/EtOH [48, 166-1711. Next, the bromide 96 is activated by an insoluble silver promoter in the presence of the aglycone. Because the 2-0x0 function not only supports the sN2 mechanism by its electron-withdrawing capacity but also is a nonparticipating group, the product is almost exclusively the P-glycoside 97. The use of soluble promoters with bromide 96 or with the corresponding pentenyl and thioglycosides yields substantial amounts of a-linked products [42, 1671. In the final step, the 0x0 group is reduced as in Lindberg's method to afford the p-mannoside 98 (Scheme 28) [167]. The low anomeric reactivity of 2-ulosyl bromides can be a limitation when moderately nucleophilic alcohols are used as aglycones. Attempts

Yield: 92% 95

Scheme 28.

Yield: 93%

Yicld:Yi% 96

97

98

332

13 Stereoselective Synthesis of P-Mannosides

Scheme 29.

to overcome t h s problem include the use of ulosyl iodides and 0-benzyl protected ulosyl donors [48, 169, 1721 in combination with van Boeckel's silver silicate promoter [84]. 13.2.4 Anomeric 0-alkylation

Alkylation of 1-0-Metal Complexes Schmidt has developed a technically simple route to p-manno furanosides and pyranosides in which the sodium or potassium salt of mannose hemiacetals is reacted with strong alkylating agents derived from the aglyconic component [ 173-1781. For example, treatment of the sodium complex of mannofuranose 99 with the triflate 100 gives the corresponding mannofuranoside 101 as a 1 :2 ratio of the ct and the P anomers (Scheme 29) [ 1751. It is believed that intramolecular complexation between the alkali metal and the sugar oxygens locks the anomeric oxygen in the P position before the mannosyloxy anion attacks the aglycone. The Stannylene Acetal Method Five-membered cyclic dibutylstannylene acetals formed on vicinal cis-axialequatorial pairs of hydroxyl groups selectively enhance the nucleophilicity of the equatorial oxygen in 0-alkylation reactions [179, 1801. On the basis of this rule Schuerch surmised that alkylation of the tin complex 103, having the anomeric oxygen locked in the equatorial position, should lead to P-mannosides [181]. This assumption was proved by treating the mannose diol 102 with BuzSnO (+ 103) followed by in situ exposure to alkyl halides whereby P-mannopyranosides 104 were formed stereospecifically (Scheme 30) [ 1811.

B Bn :

102

Scheme 30.

103

Yield: 94-100%

~

o

= M e Or A" 104

R

13.2 Chemicul Methods

%fNH

333

Yield: 82%

Bn

Scheme 31.

107

108

OMe Yield: 67%

109

Scheme 32.

This strategy [99, 1821 has been used by Nicolaou for stereospecific synthesis of trehalose analogs [50]. The procedure involves reaction of the complex 103 with the a-mannosyl donor 105 under TMSOTf activation to furnish the 0,a-linked mannosy1 disaccharide 106 without the a,a-connected stereoisomer (Scheme 3 1). In another application of Schuerch's concept, 0-triflyl derivatives of monosaccharides are used as electrophiles [183]. For example, reaction of the 4-0-triflyl guluctoside 108 with a sixfold molar excess of the partially protected dibutylstannylene complex 107 in DMF affords the p-1,4-linked mannosyl-glucoside 109 in 67% yield. The scope of this protocol is limited by configurational inversion at the linkage carbon and solubility problems (Scheme 32) [ 1831. 13.2.5 Miscellaneous Methods Radical Inversion of the Anomeric Chirality of a-D-Mannopyranosides This approach was first reported by Kahne in 1988. He found that hydrogen atom transfer to hexopyranosyl radicals having an alkoxy group at C-1 occurs, preferentially, from the a-side giving rise to P-glycosides with high stereoselectivity [ 1841. For example, the anomeric radical 111 generated by photolysis of the mannose hemithio orthoester 110 is converted under reducing conditions to the mannoside 112 with an impressive 18 : 1 ratio of the and a anomers (Scheme 33).

zqsMe

Bu3SnH, AIBN C6H5Me

Bn

bMe 110

Scheme 33.

30 "C Yield: 85% 111

112,cdp=l:18

334

13 Stereoselective Synthesis of p-Mannosides

Scheme 34.

In Curran’s method 11851 the first step is homolytic cleavage of the Br-C bond in the radical translocating group in 113 to produce the aryl radical 114. Subsequent intramolecular 1,6-hydrogen transfer from the anomeric carbon translocates the radical to C-1 (-115). In the final phase the radical is quenched from the a side by intermolecular hydrogen transfer from Bu3 SnH to afford the P-mannoside 116 (Scheme 34). A major problem with this approach is epimerization at C-2 through 1,5-hydrogen abstraction producing the a-yluco isomer 117. Another competing reaction is the intermolecular replacement of bromine by hydrogen in the radical translocating group. Additional approaches capitalizing on stereospecific hydrogen translocation to the anomeric radical have been described by Crich 1186, 1871. Despite excellent stereoselectivity, the radical approach is “rendered impractical by the circuitous methods needed for generation of the key anomeric radical” [51]. Reductive Cleavage of Cyclic Orthoesters Inspired by Liptak’s regioselective transformations of hexopyranose 4,6-benzylidene acetals by hydrogenolysis [ 1881, Ikegami has developed a new entry to glycosidesynthesis 11891. In the first phase the mannopyranosylidene acetal 120 is prepared from lactone 118 and the diol 119. In the subsequent step the interresidual spiro orthoester is reduced with LiAlH4/AlCl3 to produce the P-linked disaccharide 121 regio- and stereospecifically (Scheme 35). This method is limited to the preparation of P(1-4) linkages.

De novo Syntheses Tietze has synthesized ethyl P-D-mannopyranoside from non-carbohydrate precursors [ 1901. The procedure involves [4 + 21 hetero Diels-Alder cycloaddition of the enol ether 122 and the chiral oxazolidinone 123 under promotion by MeZAlC1 to give the dihydropyran 124 in very high diastereoselectivity. Reductive cleavage of

118

Scheme 35.

119

Yield: 77%

120

Yield 98%

121

13.2 Chemical Methods

''3

V

+

Ac

op N

A

o

1. Ac,O, NaOAc 2. BH3 . Me2S

/

Yield: 95%

U"O ~.

Yield: 85%

Ulld

123

I22

335

Bn6

Yield: 82%

Bn6

126

125

I24

Scheme 36.

the oxazolidinone moiety (+125) followed by hydroboration with oxidative workup has provided P-mannoside 126 (Scheme 36). In a similar approach, Boger has prepared enantiomeric mixtures of P-mannopyranosides using achiral precursors (Scheme 36) [191].

13.2.6 2-Acetamido-2-deoxy-~-~-mannopyranosides Two main directions have emerged to this group of compounds. In the first an azido group is employed as the precursor to the acetamido group [ 192-1941. In Paulsen's method the 2-azido-mannosyl donor 127 is used as the ManNAc synthon that is condensed with the aglycone component under silver silicate activation to afford the P-linked mannoside 128 (Scheme 37). In the alternative route [159, 195, 1961 a P-glucopyranoside is synthesized first, followed by azide-displacement at C-2 to provide the 2-azido-2-deoxy-Pmannopyranoside unit in a process as reviewed for the parent compound in Section 13.2.2. The second direction is based on 2-ulosyl intermediates. In one approach [ 197, 1981 an uloside e.y. 129 is synthesized first, as reviewed in Section 13.2.2, then converted to the oxime 130. Subsequent reduction and N-acetylation affords the Plinked N-acetyl-mannosaminyl disaccharide 131 (Scheme 38).

$,

25'C

Yield: 83%

127

128, cc/p = 2:3

Scheme 37.

E - d k 2 O H .

0 129

Scheme 38.

&;?*::

HCI 0CH20Bn Yield: 80%

Bn

LiAIH,,

2. AczO

Bn

'IHL

:-;&

CH20Bn

HO"

130

BnOoMe

Bn 0 CHzOBn 131

13 Stereoselective Synthesis of p- Mannosides

336

\

B z d h Br

OBz Yield: 74%

132

BzON Yield: 93%

133

134

135

Scheme 39.

In Lichtenthaler's method [33, 42, 199-2011 the oximinoglycosyl bromide 133 is used as the p-mannosamine precursor that can be prepared from the benzoylated hydroxyglycal 132. Condensation of 133 with MeOH using A g ~ C 0 3as the promoter has provided the p-linked glycoside 134 from which N-acetyl mannosaminide 135 can be obtained by BH3 reduction and N-acetylation [199] (Scheme 39). This approach is also suitable for glycosylating less reactive, secondary hydroxyl groups and for blockwise oligosaccharide synthesis [42]. 13.2.7 Aryl P-D-mannopyranosides P-Mannopyranosides with a chromo- or fluorogenic aromatic aglycone, e.g. a 4nitrophenyl, naphthyl, or 4-methylumbelliferyl group, are valuable enzyme substrates. The 4-nitrophenyl derivative is also employed as a mannosyl donor in glycosidase-catalyzed P-mannosylation. These compounds can be prepared in 4070% yields by reaction of the mannosyl bromide 2 [54] with sodium [202] or potassium phenolates [203, 2041. Garegg has synthesized aryl P-mannosides in 50-650/0 yields by condensation of dicyclohexylidene-mannopyranose with phenols in the presence of diethyl azodicarboxylate and triphenylphosphine [44, 2051. Epimerization of C-2 of aryl P-D-glucopyranosides by the Lindberg protocol (Section 13.2.2) has also been described [206]. 13.2.8 1-Thio-P-D-mannopyranosides 4-Nitrophenyl 1-thio-P-D-mannopyranoside has been synthesized by reaction of acetochloromannose [207] or the mannosyl bromide 2 [202] with the sodium salt of 4-nitrothiophenol. A similar reaction between sodium thiophenoxide and acetobromomannose in HMPT gives phenyl 1-thio-P-D-mannopyranoside quantitatively [208]. In a related reaction the sodium salt of 1-thio-(3-D-mannopyranosehas been alkylated with benzyl bromide as in Schmidt's approach (Section 13.2.4) to give benzyl P-thiomannoside in 86%) yield [209]. P-Linked 1-thiomannopyranosides can also be prepared by BF3-catalyzed anomerization of the readily available a-anomers [46]. The Lindberg protocol has also been used for the synthesis l-thioP-D-mannopyranosides [ 146, 1491. Aryl 1-thio-p-mannofuranosides have been syn-

13.3 Enzymatic Synthesis

136

137

331

138

Scheme 40.

thesized by reaction of di-isopropylidene-mannofuranose 99 and diary1 disulfides in the presence of triethylphosphine [210]. 13.2.9 P-D-Mannopyranosylamines The only N-glycoside that contains a P-linked mannopyranosyl residue was synthesized by palladium-catalyzed condensation of the amine 136 with the chloropurine 137 to afford the nucleoside antibiotic analog 138 together with the a-linked isomer in a 6: 1 ratio [211]. It is of interest to note the exclusive formation of the P amine 136 upon catalytic reduction of the corresponding azide, with no trace of the c1 anomer (Scheme 40).

13.3 Enzymatic Synthesis Although chemical methods have provided access to many oligosaccharides containing the p-mannopyranoside linkage, their overall efficiency is quite low because of numerous protecting group manipulations and glycosylation steps. As an alternative enzymatic methods have been receiving increasing attention. Both exomannosidases (glycosyl hydrolases) [47, 196, 2 12-2 161 and mannosyl transferases [21, 217, 2181 might be suitable biocatalysts for stereospecific p-mannoside synthesis. Although mannosiduse-catalyzed P-mannosylation (reverse hydrolysis) uses relatively readily available enzymes and mannosyl donors, e.g. p-nitrophenyl p-Dmannopyranoside or even mannose [216],overall yields are low [215].An additional difficulty is that when a saccharide is used as the acceptor alcohol the regioselectivity is not always predictable [47]. In the mannosyl transjerase approach a pmannosyl unit is transferred from GDP-mannose to the acceptor in a regio- and stereospecific manner. The high cost of GDP-mannose, the restrictions that might be imposed by substrate specificity, and instability and limited availability and of mannosyl transferases can be serious difficulties that require considerable investment before this method can be used for preparative-scale synthesis of 0-mannosides. A more detailed account of the enzymatic approach can be found in Chapters 23 to 30.

338

13 Stereoselective Synthesis of P-Mannosides

13.4 Conclusions Numerous developments in the last decade have rendered the construction of the problematic P-mannoside unit a less formidable challenge. Many years after its introduction, Paulsen’s silver silicate procedure is still often used for P-mannoside synthesis. Of the new methods Crich’s version of the sulfonate approach stands out because of its scope, operational simplicity, high overall yields, and P-selectivity. The more laborious inter- and intramolecular inversions of the C-2 chirality of P-glucosides are also powerful methods but require more steps that reduce the overall yield. Of the intramolecular glycosylation approaches Ogawa’s 4-methoxybenzylidene-assisted intramolecular glycosylation is a generally applicable method that is compatible with a variety of manipulations and is uniquely suitable for block-synthetic operations. Schuerch’s stannylene acetal approach has proved particularly useful for the stereoselective synthesis of trehalose analogs. Rapid advances in biotechnology are likely to open up efficient biocatalytic approaches to P-mannosyl oligosaccharides.

References 1. J. F. G. Vliegenthart and J. Montreuil, in Glycoproteins (Eds. J. Montreuil, J. F. G. Vliegen-

thart, H. Schachter), Elsevier, Amsterdam, 1995, Vol. 1, pp. 13-28. 2. D. R. Bundle, Topics. Curr. Chem., 1990, 154, 1-37. 3. C. Erbing, L. Kenne, B. Lindberg, S. Hammarstrom, Curbohydr. Res., 1978, 60, 400-403. 4. B. Lindberg, K. Leontein, U. Lindquist, S. B. Svensson, G. Wrangsell, A. Dell, M. Rogers, Curbohydr. Rex, 1988, 174, 313. 5. M. Betley, J. W. Redmond, N. H. Packer, D. Liu, P. Reeves, Curbohydr. Rex, 1997, 303, 313-318. 6. L. Kenne, B. Lindberg, in The Polysucchurides (Ed. G. 0. Aspinall) Academic Press, New York, 1983, Vol. 2, pp. 287-363. I . L. Galbraith, S. G. Wilkinson, Carbohydr. Rex, 1997, 303, 245-249. 8. U. Zahringer, H. Rettenmaier, H. Moll, S. N . Senchenkova, Y. A. Knirel, Curbohydr. Res., 1997,300, 143- 151. 9. V. Gloagen, L. Vebret, L. Hoffmann, H. Morvan, Y. Plancke, J.-M. Wieruszeski, G. Strecker, XVIII Int. Curbohydr. Symp., 1996, Milan, Italy, AP075. 10. D. Garozzo, G. Impallomeni, E. Spina, L. Sturiale, XVIII Int. Carbohydr. Symp., 1996, Milan, Italy, A0017. 11. P. L. Hackland, H. Parolis, L. A. S. Parolis, Curbohydr. Res., 1988, 172, 209-216. 12. Y.-M. Choy, G. G. S. Dutton, A. M. Zanlungo, Can. J. Chem., 1973,51, 1819-1825. 13. S. Kalelkar, J. Glushka, H. Van Halbeek, L. C. Morris, R. Cherniak, Curbohydr. Res., 1997, 299, 119-128. 14. D. Garazzo, G. Impallomeni, E. Spina, L. Sturiale, Curbohydr. Rex, 1998, 307, 113-124. 15. N. A. Paramanov, L. A. S. Parolis, H. Parolis, I. F. Boan, J. Anton, F. Rodriguez-Valera, Carbohydr. Res., 1998, 309, 89-94. 16. J. F. Kennedy, C. A. White, in Carbohydrate Chemistry (Ed. J. F. Kennedy), 1988, Clarendon Press, Oxford, pp. 220-226. 17. M. Ohnishi, S. Ito, Y. Fujino, Agric. Biol. Chem., 1985, 49, 3609-3611. 18. M. Sugita, Y. Nakano, T. Nose, 0. Itasaka, T. Hori, J. Biochem. (Tokyo), 1984, 95, 47-55. 19. N. Hada, T. Takeda, Y. Ogihara, Carbohydr. Rex, 1994, 258, 93-104. 20. P. A. Trinel, G. Lepage, T. Jouault, G. Strecker, D. Poulain, FEBS Let., 1997, 416, 203-206.

References

339

21. A. Suzuki, N. Shibata, M. Suzuki, F. Saitoh, H. Oyamada, H. Kobayashi, S. Suzuki, Y. Okawa, J. Biol. Chem., 1997,272, 16822-16828. 22. K. Tatsuta, S. Yasuda, Tetrahedron Let., 1996, 37, 2453-2456. 23. N. Sharon, H. Lis, H. in Glycosciences (Eds. H.-J. Gabius, S. Gabius) 1997, Chapman and Hall, London, pp. 133-162. 24. B. Priem, R. Gitti, C. A. Bush, Plant. Phys., 1993, 102, 445-458. 25. N. A. Kocharova, 0. V. Bystrova, S. A. Borisova, A. S. Shashkov, A. S, Y. A. Knirel, XVIII Znt. Curbohydr. Symp., 1996, Milan, Italy, AP008. 26. P. Branefors-Helander, L. Kenne, B. Lindberg, K. Petersson, P. Unger, Carbohydr. Res., 1981, 97,285-291. 27. F.-P. Tsui, R. Schneerson, R. A. Boykins, A. D. Karpas, W. Egan, Carbohydr. Rex, 1981, 97, 293-306. 28. P. Branefors-Helander, L. Kenne, B. Lindberg, K. Petersson, P. Unger, Curbohydr. Res., 1981, 88, 77-84. 29. F.-P. Tsui, R. Schneerson, W. Egan, Carbohydr. Rex, 1981, 88, 85-92. 30. C. Adlam, J. M. K. Knights, A. Mugridge, J. C. Lindon, P. R. W. Baker, J. E. Beesley, B. Spacey, G. R. Craig, L. K. Nagy, J. Gen. Microbiol., 1984, 130, 2415-2426. 31. H. J. Jennings, Curr. Top. Microbiol. Zmmunol., 1990, 150, 97-127. 32. C. Lugowski, E. Romanowska, L. Kenne, B. Lindberg, Curbohydr. Res., 1983,118, 173-181. 33. E. Kaji, Y. Osa, K. Takahashi, M. Hirooka, S. Zen, F. W. Lichtenthaler, Bull. Chem. SOC. Jpn., 1994, 67, 1130-1140. 34. J. K. Dale, J. Am. Chem. Soc., 1924,46, 1046-1051. 35. H. S. Isbell, H. L. Frush, J. Research Natl. Bur. Standards, 1940,24, 125-151. 36. H. G. Bott, W. N. Haworth, E. L. Hirst, J. Chem. SOC.,1930, 2653-2659. 37. B. Helferich, G. Duve, Chem. Ber., 1958, Y I , 1790-1793. 38. W. A. Bonner, J. Am. Chem. SOC.,1958,80, 3372-3379. 39. B. Helferich, S. Winkler, Chem. Ber., 1933, 66, 1556-1558. 40. H. Paulsen, Angew. Chem. Znt. Ed., 1982, 21, 155-173. 41. P. J. Garegg, Chemtracts, Org. Chem., 1992, 5, 389-393. 42. E. Kaji, F. W. Lichtenthaler, Trends. Glycosci. Glycotechnol., 1993, 5, 121-143. 43. F. Barresi, 0. Hindsgaul, in Modern methods in carbohydrate synthesis (Eds. S. H. Khan, R. O’Neill), Harwood Academic Publishers, Amsterdam, 1996, pp. 25 1-276. 44. B. V. McCleary, Methods Enzymol., 1988, 160, 515-518. 45. J. Tamura, S. Horito, H. Hashimoto, J. Yoshimura, Curbohydr. Rex, 1988, 174, 181-199. 46. D. T. Connolly, S. Roseman, Y. C. Lee, Curbohydr. Res., 1980, 87, 227-239. 47. N. Taubken, B. Sauerbrei, J. Thiem, J. Curbohydr. Chem., 1998,12, 651-667. 48. F. W. Lichtenthaler, T. Schneider-Adams, J. Org. Chem., 1994, 59, 6728-6734. 49. R. Kasai, M. Okihara, J. Asakawa, K. Mizutani, 0. Tanaka, Tetrahedron, 1979, 35, 14271432. 50. K. C. Nicolaou, F. L. van Delft, S. R. Conley, H. J. Mitchell, Z. Jin, R. M. Rodriguez, J. Am. Chem. SOC.,1997,119, 9057-9058. 51. D. Crich, S. Sun, Tetrahedron, 1998, 54, 8321-8348. 52. C. A. A. van Boeckel, T. Beetz, S. F. van Aelst, Tetruhedron, 1984, 40,4097-4107. 53. P. A. J. Gorin, A. S. Perlin, Can. J. Chem., 1961,39, 2474-2485. 54. T. J. Tewson, M. Soderlind, J. Curbohydr. Chem., 1985, 4, 529-543. 55. M. A. E. Shaban, I. E. Ary, D. A. Jeanloz, R. W. Jeanloz, Carbohydr. Rex, 1975,45, 105-1 14. 56. R. K. Jain, K. L. Matta, Curbohydr. Rex, 1996, 282, 101-111. 57. G. M. Bebault, G. G. S. Dutton, Curbohydr. Res., 1974, 37, 309-319. 58. V. I. Betaneli, M. V. Ovchinnikov, I. V. Backinowsky, N. K. Kochetkov, Curbohydr. Rex, 1980,84,211-224. 59. N. K. Kochetkov, V. I. Torgev, N. N. Malysheva, S. A. Shashkov, Tetrahedron, 1980, 36, 1099-1105. 60. M. A. E. Shaban, R. W. Jeanloz, Carbohydr. Rex, 1976, 52, 103-114. 61. M. A. E. Shaban, R. W. Jeanloz, Curbohydr. Rex, 1976,52, 115-127. 62. T. Sugawara, K. Irie, H. Iwasawa, T. Yoshikawa, S. Okuno, H. Watanabe, T. Kato, M. Shibukawa, Y. Ito, Curbohydr. Rex, 1992,230, 117-149.

340

13 Stereoselective Synthesis of P-Mannosides

H. Paulsen, 0. Lockhoff, Tetrahedron Let., 1978, 402774030, H. Paulsen, 0. Lockhoff, Chem. Ber., 1981, 114, 3102-3114. P. J. Garegg, T. Iversen, Curbohydr. Rex, 1979, 70, C13-C14. P. J. Garegg, T. Iversen, R. Johansson, Acta Chem. Scund., 1980, B34, 505-508. H. Paulsen, M. Heume, Z. Gyorgydeak, R. Lebuhn, Carbohydr. Rex, 1985, 144, 57-70. T. Ogawa, T. Kitajama, T. Nukada, Carbohydr. Rex, 1983, 123, C5-C7. H. Paulsen, M. Springer, F. Reek, E. Meinjohanns, I. Brockhausen, H. Schachter, Justus Liebigs Ann. Chem., 1995, 53-66. 70. P. Grice, S. V. Ley, J. Pietruszka, H. M. I. Osborn, H. W. M. Priepke, S. L. Warriner, Chem. Eur. J., 1997, 3, 431-440. 71. T. Ogawa, K. Katano, K. Sasajima, M. Matsui, Tetrahedron, 1981,37, 2779-2786. 72. H. Paulsen, R. Lebuhn, Angew. Chem. Int. Ed., 1982,21,926-927. 73. T. Ogawa, T. Kitajima, T. Nukada, Curbohydr. Res, 1983, 123, C8-Cll. 74. H. Paulsen, R. Lebuhn, Justus Liebigs Ann. Chem., 1983, 1047-1072. 75. H. Paulsen, R. Lebuhn, Carbohydr. Res., 1984, 130, 85-101. 76. T. Ogawa, M. Sugimoto, T. Kitajima, K. K. Sadozai, T. Nukada, Tetrahedron Let., 1986,27, 5739-5742. 77. F. Yamazaki, T. Nukada, Y. Ito, S. Sato, T. Ogawa, Tetrahedron Let., 1989, 30, 4417-4420. 78. H. Paulsen, M. Heume, H. Nurnberger, Curbohydr. Res., 1990,200, 127-166. 79. T. Nukada, T. Kitajima, Y. Nakahara, T. Ogawa, Curbohydr. Rex, 1992, 228, 157-170. 80. K. Koike, M. Mori, Y. Ito, Y. Nakahara, T. Ogawa, Agric. Biol. Chem., 1990,54,2931-2939. 81. M. Mori, Y. Ito, T. Ogawa, Carbohydr. Rex, 1990, 195, 199-224. 82. 0. Kanie, T. Takeda, N. Hada, Y. Ogihara, J. Curbohydr. Chem., 1991,10, 561-581. 83. K. Koike, M. Mori, Y. Ito, Y. Nakahara, T. Ogawa, Biosci. Biotech. Biochem., 1993,57,698702. 84. C. A. A. van Boeckel, T. Beetz, A. C. Kock-van Dalen, H. van Bekkum, Rec. Truo. Chim. Pu~s-Bus,1987, 106, 596-598. 85. Z.-W. Guo, Y. Nakahara, T. Ogawa, Bioorg. Med. Chem., 1997,5, 1917-1924. 86. Z.-W. Guo, Y. Itoh, Y. Nakahara, T. Ogawa, Carbohydr. Rex, 1998,306, 539-544. 87. P. J. Garegg, P. Ossowski, Actu Chem. Scand., 1983, B37, 249-251. 88. D. M. Whitfield, R. N. Shah, J. P. Carver, J. J. Krepinsky, Synth. Commun., 1985, 15, 737747. 89. T. Takeda, N. Hada, Y. Ogihara, Chem. Pharm. Bull., 1992,40, 1930-1933. 90. R. N. Shah, D. A. Cumming, A. A. Grey, J. P. Carver, J. Krepinsky, Curbohydr. Rex, 1986, 153, 155-161. 91. P. J. Garegg, C. Henrichson, T. Norberg, P. Ossowski, Carbohydr. Rex, 1983, 119, 95-100. 92. P. J. Garegg, R. Johansson, B. Samuelsson, Actu Chem. Scand., 1981, B35, 635-636. 93. G. Wulff, J. Wichelhaus, Chem. Ber., 1979, 112, 2847-2853. 94. J. Leroux, A. S. Perlin, Curbohydr. Rex, 1976, 47, C8-CIO. 95. J. Leroux, A. S. Perlin, Carbohydr. Rex, 1978, 67, 163-178. 96. E. S. Rachaman, R. Eby, C. Schuerch, Carbohydr. Rex, 1978,67, 147-161. 97. V. K. Srivastava, C. Schuerch, Carbohydr. Res., 1980, 79, C13-CI6. 98. V. K. Srivastava, C. Schuerch, J. Org. Chem., 1981, 46, 1121-1126. 99. E. S. El Ashry, C. Schuerch, Carbohydr. Rex, 1982, 105, 33-43. 100. L. F. Awad, E. S. H. El Ashry, C. Schuerch, Bull. Chem. SOC.Jpn., 1986,59, 158771592, 101. D. Crich, S. Sun, J. Am. Chem. Soc., 1997,119, 11217-11223. 102. D. Crich, W. L. Cai, J. Org. Chem., 1999,64,4926-4930. 103. D. Crich, S. Sun, J. Am. Chem. Soc., 1998, 120,435-436. 104. D. Crich, S. Sun, J. Org. Chem., 1997, 62, 1198-1199. 105. D. Crich, S. Sun, J. Org. Chem., 1996, 61, 4506-4507. 106. D. Crich, Z. Dai, Tetrahedron Let., 1998, 39, 1681-1684. 107. D. Crich, Z. M. Dai, Tetrahedron, 1999, 55, 1569-1580. 108. G. Stork, H. S. Suh, G. Kim, J. Am. Chem. Soc., 1991,113,7054-7056. 109. G. Stork, G. Kim, J. Am. Chem. Soc., 1992, 114, 1087-1088. 110. L. Yan, D. Kahne, J. Am. Chem. Soc., 1996, 118, 9239-9248. 111. G. Stork, J. J. La Clair, J. Am. Chem. Soc., 1996, 118, 247-248. 63. 64. 65. 66. 67. 68. 69.

References

341

F. Barresi, 0. Hindsgaul, J. Am. Chem. Soc., 1991, 113, 9376-9377. F. Barresi, 0. Hindsgaul, Synlett, 1992, 759-761. F. Barresi, 0. Hindsgaul, Can. J. Chem., 1994, 72, 1447-1465. Y. Ito, T. Ogawa, Angew. Chem. Int. Ed., 1994, 33, 1765-1767. M. Lergenmuller, T. Nukada, K. Kuramochi, A. Dan, T. Ogawa, Y. Ito, Eur. J. Org. Chem., 1999, 136771376, 117. A. Dan, Y . Ito, T. Ogawa, J. Org. Chem., 1995, 60, 4680-4681. 118. A. Dan, Y. Ito, T. Ogawa, Tetrahedron Let., 1995, 36, 7487-7490. 119. Y. Ito, T. Ogawa, J. Am. Chem. Soc., 1997, 119, 5562-5566. 120. T. Ziegler, G. Lemanski, A. Rakoczy, Tetrahedron Let., 1995,36, 8973-8976. 121. T. Ziegler, G. Lemanski, Eur. J. Org. Chem., 1998, 163-170. 122. T. Ziegler, G. Lemanski, Angew. Chem. Int. Ed., 1998, 37, 3129-3132. 123. N. K. Kochetkov, N. N. Malysheva, A. V. Demchenko, N. G. Kolotyrkina, E. M. Klimov, Bioorg. Chem. (Russia), 1991, 17, 1655-1659. 124. E. M. Klimov, A. V. Demchenko, N. N. Malysheva, N. K. Kochetkov, Bioorg. Chem. (Russia), 1991, 17, 1660-1664. 125. T. Yamanoi, K. Nakamura, H. Takeyama, K. Yanagihara, T. Inazu, Bull. Chem. Soc. Jpn., 1994,67, 1359-1366. 126. J. Yoshimura, S. Horito, J. Tamura, H. Hashimoto, Chem. Let., 1985, 1335-1338. 127. J. Tamura, S. Horito, J. Yoshimura, H. Hashimoto, Carbohydr. Res., 1990, 207, 153-165. 128. J. Tamura, R. Terayama, T. Shimadate, J. Carbohydr. Chem., 1993, 12,893-902. 129. W.-S. Kim, H. Sasai, M. Shibasaki, Tetrahedron Let., 1996, 37, 7797-7800. 130. 0. Theander, Acta Chem. Scand., 1958, 12, 1883-1885. 131. G . Ekborg, B. Lindberg, J. Lonngren, Acta Chem. Scand., 1972,26, 328773292, 132. K. K. C. Liu, S. J. Danishefsky, J. Org. Chem., 1994, 59, 1892-1894. 133. A. Liptak, J. Hajko, Unpublished results, 1998. 134. H. Boren, G. Ekborg, K. Eklind, P. J. Garegg, A. Pilotti, C. Swahn, Acta Chem. Scand., 1973, 27, 2639-2644. 135. N. K. Kochetkov, B. A. Dmitriev, 0. S. Chizhov, E. M. Klimov, N. N. Malysheva, A. Y. Chernyak, N. E. Bayramova, V. I. Torgov, Carbohydr. Res., 1974, 33, C5-C7. 136. N. K. Kochetkov, B. A. Dmitriev, N . N . Malysheva, A. Chernyak, E. M. Klimov, N . E. Bayramova, V. I. Torgov, Carbohydr. Rex, 1975, 45, 283-290. 137. G. Ekborg, J. Lonngren, S. Svensson, Acta Chem. Scand., 1975, B29, 1031-1035. 138. E. E. Lee, G. Keaveney, P. S. O’Colla, Carbohydr. Res., 1977, 59, 268-273. 139. C. D. Warren, C. Auge, M. L. Laver, S. Suzuki, D. Power, R. W. Jeanloz, Carbohydr. Rex, 1980, 82, 71-83. 140. C. Auge, C. D. Warren, R. W. Jeanloz, M. Kiso, L. Anderson, Carbohydr. Res., 1980,82, 8595. 141. G. Ekborg, C. P. J. Glaudemans, Carbohydr. Rex, 1984,129,287-292. 142. P. J. Garegg, C. Hallgren, J. Carbohydr. Chem., 1992, 11, 425-443. 143. H. H. Lee, L. N. Congson, D. M. Whitfield, L. R. Radics, J. J. Krepinsky, Can. J. Chem., 1992, 70, 2607-2617. 144. J . Kerekgyarto, J. G. M. van der Ven, J. Kamerling, A. Liptak, J. F. G. Vliegenthart, Carbohydr. Res., 1993, 238, 135-145. 145. S. H. Khan, K. L. Matta, Carbohydr. Res., 1995,278, 351-362. 146. P. H. Seeberger, P. F. Cirillo, S. Hu, X. Beebe, M. T. Bilodeau, S. J. Danishefsky, Enantiomer, 1996, I , 311-323. 147. S. Sarbanja, A. K. Misra, N. Roy, Synth. Commun., 1998,28, 2559-2570. 148. J. Kerekgyarto, J. P. Kamerling, J. B. Bouwstra, J. F. G. Vliegenthart, A. Liptak, Carbohydr. Res., 1989, 186, 51-62. 149. S. J. Danishefsky, S. Hu, P. F. Cirillo, M. Eckhardt, P. H. Seeberger, Chem. Eur. J., 1997, 3, 161771628, 150. M. Miljkovic, M. Gligorijevic, D. Glisin, J. Org. Chem., 1974, 39, 3223-3226. 151. A. Olesker, A. Dessinges, T. T. Thang, G. Lukacs, C. R. Acad. Sci. Paris Ser. 2, 1982, 295, 575-517. 152. W. Gunther, H. Kunz, Carbohydr. Rex, 1992,228, 217-241.

112. 113. 114. 115. 116.

342

13 Stereoselective Synthesis of /3-Mannosides

153. S. David, A. Malleron, C. Dini, Curbohydr. Rex, 1989, 188, 193-200. 154. I. Matsuo, M. Isomura, R. Walton, K. Ajisaka, Tetruhedron Let., 1996, 37, 8795-8798. 155. A. Fiirstner, I. Konetzki, J. Org. Chem., 1998, 63, 3072-3080. 156. A. Fiirstner, I. Konetzki, Tetrahedron Let., 1998, 39, 5721-5724. 157. S. David, A. Fernandez-Mayoralas, Carbohydr. Res., 1987, 165, clILc13. 158. J. Alais, S. David, Carbohydr. Rex, 1990, 201, 69-77. 159. K. Sato, A. Yoshimoto, Y. Tdkai, Bull. Chem. SOC.Jpn., 1997, 70, 885-890. 160. K. Sato, H. Seki, A. Yoshimoto, H. Nanaumi, Y. Takai, Y. Ishido, J. Curbohydr. Chem., 1998, 17, 703-727. 161. Y.-M. Zhang, J.-M. Mallet, P. Sinay, Carbohydr. Res., 1992,236, 73-88. 162. H.-W. Hagedorn, R. Brossmer, Helv. Chim. Actu, 1986, 69, 2127-2132. 163. H. Kunz, W. Giinther, Angew. Chem. Znt. Ed., 1988,27, 1086-1087. 164. C. Unverzagt, Angew. Chem. Int. Ed., 1994,33, 1102-1104. 165. I. A. Ivanova, A. V. Nikolaev, J. Chem. Soc. Perkin Trans 1, 1998, 3093-3099. 166. F. W. Lichtenthaler, E. Kaji, Justus Liebigs Ann. Chem., 1985, 1659-1668. 167. F. W. Lichtenthaler, E. Kaji, S. Weprek, J. Org. Chem., 1985, SO, 3505-3515. 168. F. W. Lichtenthaler, U. Klares, M. Lergenmiiller, S. Schwidetzky, Synthesis, 1992, 179-184. 169. F. W. Lichtenthaler, T. Schneider-Adams, S. Immel, J. Org. Chem., 1994, 59, 6735-6738. 170. F. W. Lichtenthaler, U. Klares, Z. Szunnai, B. Werner, Carbohydr. Rex, 1997,305, 293-303. 171. F. W. Lichtenthaler, T. W. Metz, Tetrahedron Let., 1997, 38, 5477-5480. 172. A. Fiirstner, I. Konetzki, Tetrahedron, 1996, 52, 15071-15078. 173. R. R. Schmidt, M. Reichrath, U. Moering, Tetrahedron Let., 1980,21, 3561-3564. 174. R. R. Schmidt, U. Moering, M. Reichrath, Chem. Ber., 1982, 115, 39-49. 175. A. Terjung, K.-H. Jung, R. R. Schmidt, Justus Liebigs Ann. Chem., 1996, 1313-1321. 176. A. Terjung, K.-H. Jung, R. R. Schmidt, Curbohydr. Rex, 1997,297, 229-242. 177. J . Tamura, R. R. Schmidt, J. Curbohydr. Chem., 1995,14, 895-911. 178. W. Klotz, R. R. Schmidt, Justus Liebigs Ann. Chem., 1993, 683-690. 179. S. David, S. Hanessian, Tetrahedron, 1985, 41, 643-663. 180. David, S. in Preparative carbohydrate chemistry, (Ed. S. Hanessian), Marcel Dekker, Inc., New York, 1997, pp. 69-83. 181. V. K. Srivastava, C. Schuerch, Tetrahedron Let., 1979, 3269-3272. 182. A. Dessinges, A. Olesker, G. Lukacs, T. T. Thang, Carbohydr. Res., 1984, 126, C6-C8. 183. G. Hodosi, P. Kovac, Curbohydr. Res., 1998, 308, 63-75. 184. D. Kahne, D. Yang, J. J. Lim, R. Miller, E. Paguaga, J. Am. Chem. Soc., 1988, 110, 87168717. 185. N. Yamazaki, E. Eichenberger, D. P. Curran, Tetrahedron Let., 1994, 35, 6623-6626. 186. D. Crich, S. Sun, J. Brunckova, J. Org. Chem., 1996, 61, 605-615. 187. D. Crich, J.-T. Hwang, H. Yuan, J. Org. Chem., 1996, 61, 6189-6198. 188. A. Liptak, I. Jodal, P. Nanasi, Curbohydr. Rex, 1975, 44, 1-11. 189. T. Iimori, H. Ohtake, S. Ikegami, Tetrahedron Let., 1997,38, 3415-3418. 190. L. F. Tietze, A. Montenbruck, C. Schneider, Synlett, 1994, 509-510. 191. D. L. Boger, K. D. Robarge, J. Org. Chem., 1988, 53, 5793-5796. 192. H. Paulsen, J. P. Lorentzen, W. Kutschker, Carbohydr. Res., 1985, 136, 153-176. 193. H. Paulsen, B. Helpap, J. P. Lorentzen, Carbohydr. Rex, 1988, 179, 173-197. 194. T. Sugawara, K. Igarashi, Carbohydr. Res., 1988, 172, 195-207. 195. B. Classon, P. J. Garegg, S. Oscarson, A.-K. Tiden, Carbohydr. Rex, 1991, 216, 187-196. 196. M. Scigelova, S. Singh, D. H. G. Crout, J. Chem. SOC.Perkin I , 1999, 777-782. 197. E. Micheli, F. Nicotra, L. Panza, F. Ronchetti, L. Toma, Curbohydr. Rex, 1985, 139, C1-C3. 198. R. Andersson, I. Gouda, 0. Larm, M. E. Riquelme, E. Scholander, Curbohydr. Rex, 1985, 142, 141-145. 199. E. Kaji, F. W. Lichtenthaler, Y. Osa, K. Takahashi, S. Zen, Bull. Chem. Soc. Jpn., 1995, 68, 2401-2408. 200. Y. Osa, E. Kaji, K. Takahashi, M. Hirooka, S. Zen, F. W. Lichtenthaler, Chem. Let., 1993, 1567-1570. 201. E. Kaji, F. W. Lichtenthaler, T. Nishino, A. Yamane, S. Zen, Bull. Chem. Soc. Jpn., 1988, 61, 1291-1297.

References 202. 203. 204. 205. 206. 207. 208. 209. 210. 21 1. 212. 213. 214. 215. 216. 217. 218.

343

K. L. Mattd, J. J. Barlow, Curhohydr. Rex, 1976, 48, 294-298. M.-C. Courtin-Duchateau, A. Veyrieres, Curhohydr. Res., 1978, 65, 23-33. H. P. Kleine, R. S. Sidhu, Curbohydr. Rex, 1988, 182, 307-312. K. Akerfeldt, P. J. Garegg, T. Iversen, Actu Chem. Scund., 1979, B33, 467-468. N. Kashimura, K. Kawaguchi, Ayric. Biol. Chern., 1976, 40; 1621-1624. M. Blanc-Muesser, J. Defaye, H. Driguez, Tetruhedron Let., 1976, 4307-4310. V. Pedretti, A. Veyrieres, P. Sinay, Tetruhedron, 1990, 46, 77-88. P. L. Durette, T. Y. Shen, Carbohydr. Res., 1980, 81, 261-274. A. Fiirstner, Justus Liehigs Ann. Chem., 1993, 1211-1217. N. Chida, T. Suzuki, S. Tanaka, I. Yamada, Tetrahedron Let., 1999, 40, 2573-2576. H. Itoh, Y. Kamiyama, Journal of Fermentution and Bioengineering, 1995, 80, 510-512. H. Fujimoto, M. Isomura, K. Ajisaka, Biosci. Biotech. Biochem., 1998, 61, 164-165. T. Usui, M. Suzuki, T. Sato, H. Kawagishi, K. Adachi, H. Sano, Glycocorzj. J., 1994, 11, 105110. S. Singh, M. Scigelova, D. H. G. Crout, J. Chrm. SOC.Chem. Commun., 1996, 993-994. N. Kildemark, K. Nilsson, XIX. Znt. Curhohydr. Symp., 1998, San Diego, CA, USA, BP035. G. M. Watt, L. Revers, M. C. Webberley, I. B. H. Wilson, S. L. Flitsch, Curhohydr. Rex, 1998,305, 533-541. L. Revers, R. M. Bill, I. B. H. Wilson, G. M. Watt, S. L. Flitsch, Biochim. Biophys. Actu, 1999, 1428, 88-98.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

14 Special Problems in Glycosylation Reactions: Sialidations Mukoto Kiso, Hideharu Ishida and Hiromi It0

14.1 Introduction Sialoglycoconjugates such as gangliosides and sialoglycoproteins have been recognized to play important roles in many biological processes [ 11. Sialooligosaccharide chains on cell surfaces are exposed as ligdnds to the external environment, and are capable of various biological functions, not only serving as receptors for hormones, viruses, and bacterial toxins, but also as mediators in cell growth, cell differentiation, cell adhesion, immune responses, and oncogenesis, etc. The most representative sialic acid, N-acetylneuraminic acid (NeuSAc, 1) is usually attached to 0 - 3 or 0-6 of galactose (Gal) or 0 - 6 of N-acetylgalactosamine (GalNAc) with an a(2-3) or a(2-6) linkage, and to 0-8 of another Neu5Ac with an w(2-8) linkage, giving diverse structures and functions. Because of this diversity systematic understanding of structure-function relationships of sialoglycoconjugates at the molecular level necessitates an efficient method for regio- and a-stereoselective glycoside synthesis of sialic acids.

14.2 Sialidation by the Koenigs-Knorr Method Several successful preparations of a-sialyl glycosides have been achieved mainly by use of the classic Koenigs-Knorr method employing N-acetylneuraminyl halides 2 as sialyl donors [2]. Although coupling of 2 (X = C1) with primary sugar hydroxyls, such as 3, gave a mixture of a-glycoside (4) and 0-glycoside ( 5 ) in good yields (Scheme 1) [3], the yield and stereoselectivity of the glycosides with hindered sugar hydroxyls were generally poor. For example, the sialidation of secondary sugar hydroxyls, such as 6, afforded an a,P mixture in low yields accompanied by a major by-product 7 (Scheme 2) [4], and the similar results were also obtained in couplings

346

14 Special Problems in Glycosylation Reactions: Siulidutions HO AcHN HO\W' W

C

0

2

H

~AcHN ~

HO

~

C

O

&

AcO

1 (Neu5Ac)

2

X = CI, Br etc

+

2 ( X = CI)

BnO

sieves 4 8, CHzC12

R

3 (R = OBn, N3)

AcouOAc

AcHN

"BnO " O0F o B n

AcO

OBn BnO

4 (a-glycoside) 36 % (R = OBn) 42 % (R = N3)

5 (p-glycoside) 48 % (R = OBn) 36 % (R = N3)

Scheme 1.

2 (X = CI)

t

HO

sieves 4 8, CH2C12

OBn 6

AcHN

OBn

AcO

C02Me

AcHN AcO

a (6 -12 %)

(3 (6 -19 %)

Scheme 2.

7

l

e

14.3 Siulidution Using an Auxiliary Group at C-3

347

of 2 (X = C1 or Br) with the lactose derivatives [5a, 5b]. Some improvements in the yield and a-stereoselectivity have been achieved by employing modified KoenigsKnorr conditions [2, 5c]. Particularly annoying is the competitive elimination which results from the deoxy center at C-3, giving the 2,3-dehydro derivative 7 as the major by-product. As critical problems, there are several disadvantages to the stereoselective synthesis of a-sialyl glycosides: 1) the anomeric center (C-2) of sialic acid is not only sterically hindered, but the reactivity is also electronically disfavored by the carboxylic acid function, which destabilizes the oxonium ion intermediate necessary for glycosidation; 2) the lack of a substituent at C-3 precludes the suitable neighboring participation leading to a-glycoside; and 3) the thermodynamically favored product is the P-glycoside. These combined factors are considered to disfavor formation of the desired aglycoside.

14.3 Sialidation Using an Auxiliary Group at C-3 The use of 2-halo-3-P-substituted sialic acid derivatives 8 as sialyl donors is an efficient means of overcoming the disadvantages in a-sialyl glycoside syntheses (Scheme 3) [2].Early work employing 8 (X = Br, Y = OH, P = Ac) showed that primary sugar hydroxyls react smoothly with this sialyl donor to give the desired

7

-

PO C02Me

AcHN

p0\\lts' AcHN

PO

PO

8

9

X = Halides Y = OH, SPh, SePh P = Ac, Bn

Scheme 3.

ROH

PO*"'

348

14 Special Problems in Glycosylation Reactions: Siulidutions

+

8

H O f l0O BnO e O B n

(X = CI or Br, Y = SPh, P = Bn)

HO

OBn

OBn 11

Hg(CN)2 HgBr2

Brio

sieves Molecular cc144A

B nBnOO AcHN

OBn

C02Me k

f

64 % (X = CI)

p

OBn OBn

OBn

HO

78 % (X = Br)

Brio &

o

12

OAc 7-

PhSCl CHcC12 30 C 2 days

AcOu' W L AcHN AcO

O

z M SPh

+

e

BnO* BnO

Brio OMe

8

AgOTf

(X = CI, Y = SPh, P = Ac)

40

13

Mokcular sieves 4 A

Na2HP04 A

c OAc O U

~

&

AcHN AcO

BnO BnO BnO OMe 14

Scheme 4a.

a-glycosides 9 with modest stereoselectivity. Nevertheless, the total yield and astereoselectivity of the glycosides with secondary sugar hydroxyls were generally poor. Many attempts to solve this problem were examined, and the thiophenyl auxiliary proved to be superior, as shown in Scheme 4a [6]. In fact, coupling of 8 (X = C1 or Br, Y = SPh, P = Bn) with a benzyl-protected lactose acceptor 11 gave excellent yields of the sialyl a(2-3')lactose derivative 12 [6a]. This procedure was immediately applied to the preparation of the sialyl a(2-8)sialic acid building blocks toward the synthesis of ganglioside GD3 and GQlb (Scheme 4b) [ 7 ] . The acetyl-protected sialyl donor 8 (X = C1, Y = SPh, P = Ac) was also examined for coupling with 13 to form a-sialoside 14 in 40% yield [6b]. Additional multiple steps for preparation of glycosyl donor 8 and removal of the C-3 substituent to arrive at the desired products 10 are, however, required (see Scheme 3), even though improvements in the yield and a-stereospecificity of the glycosides are achieved.

-

14.4 Siulidution Using 2-Tltioglycosidrs

BnO 8

(X = Br, Y = SPh, P = Bn)

OH

+

349

Hg(CN)2 HgBr2 C02Me

AcHN

Molecular

sieves 4A

cc14

BnO

(64%)

."W

C02Me

Bnvn C02Me

Scheme 4b.

14.4 Sialidation Using 2-Thioglycosides, Xanthates, or Phosphites of Sialic Acids in Acetonitrile 14.4.1 Thioglycosides Facile and highly regio- and a-stereoselective sialyl glycoside syntheses have been achieved [S, 91 by using the 2-thioglycosides of NeuSAc (15, 16) or 3-deoxy-w glycero-~-gaZacto-2-nonulosonicacid (Kdn, 17) as sialyl donors, and the partially protected sugar acceptors (18-26) (Figure l), in the presence of thiophilic promoters such as dimethyl(methy1thio)sulfonium triflate (DMTST) (promoter A) [9c, lo] or N-iodosuccinimide (N1S)-trifluoromethanesulfonic acid (TfOH) (promoter B) [ l l ] in acetonitrile (Table 1, Figure 2). The sialyl donors (15-17) were readily prepared [9, 12, 131 in three steps from NeuSAc or Kdn. Early work employing methyl 2-thioglycoside 15 as the glycosyl donor and DMTST as the promoter indicated [S, 91 that glycosylation in acetonitrile gives predominantly a-sialyl glycosides. Also, the use of lightly protected sugar acceptors (Figure I), in which several OH groups are unprotected, efficiently gave the desired sialyl glycosides a(2-3)- or a(2-6)-linked to the Gal residue in 50-70% yields, even in large-scale reactions [13]. Later, it was found that the use of phenyl 2-thioglycoside 16 [ 121 as the sialyl donor, promoted by NIS/TfOH, gives higher yields (Table 1, entries 3-5). Sialidation of trisaccharide acceptors (25, 26) with 16 afforded the corresponding sLeX (36) [14] and sLea (37) [15] derivatives in good yields (Table 1, entries 18 and 19). Reaction with reactive hydroxyls, such as OH-6 of 22 or OH-3 of 24, however, resulted in anomeric mixtures (entries 12-17). A

350

14 Special Problems in Glycosylation Reactions: Sialidations

Sialyl donors:

AcHN AcO 15R=Me 16R=Ph

AcO AcO 17

Sugar acceptors:

18 R’ = H, R2 = BZ 19 R’ = R2 = Bn

21 R = B z 22 R = Bn

23

24

J @ o ) O B n NHAc OBz OBn

&

HO Me

OBz

OBn

OBn

BnO OBn 25

26

Figure 1. Sialyl donors (15-17) and sugar acceptors (18-26).

similar tendency was observed also in the reaction with OH-6 of the GalNAc residue. Thus, the gangliotriose derivative 38 reacted with 16, promoted by NIS/ TMSOTf in acetonitrile, to afford 39a and 39p in 65% and 15% yield, respectively (Scheme 5). The use of NIS/TMSOTf, instead of NIS/TfOH, as the glycosyl promoter might give higher a-selectivity for reactive hydroxyls. Modification of the leaving group at C-2 in 2-thioglycosides of sialic acid has been found to modulate the reactivity as sialyl donors [ 161. The N-acetoxyacetamido (40a) [ 171, N-benzyloxyacetamido (40b) [ 181, and Ntrifluoroacetamido (42) [ 191 derivatives of 15 or 16 were also successfully coupled

14.4 Sialidation Using 2-Thioglycosides Products:

AcopOAc

R'

R2

R3

27

H

H

Bz

28

Bn

H

Bn

29

H

30

benzylidene

z:N2$$ b ; 33R=Bz 34R=Bn

OBz

31 R = NHAc 32 R = OAC

OSE

AcO

RO

OSE

AcO

CO2Me

OH

.

c

OAc o b

p

o C02Me

AcHN AcO

0 HO

o&

BnO

OBn

OSE OBn

35

OSE

OBn OBn

Figure 2. Structures of sialidation products.

351

352

14 Special Problems in Glycosylution Reactions: Siulidations

Table 1. DMTST (A)”-or NIS/TfOH (B)b-promoted sialidation of sugar acceptors (18-26) with methyl or phenyl 2-thioglycoside of sialic acid (15-17) in acetonitrile [Sc].

Entry

Donor

Acceptor

Promoter

Isolated yield (%)

Product

a 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

18 18 18 19 20 18 23 23 23 21 21 22 22 22 24 24 24 25 26

15 15 16 16 16 17 15 16 17 15 15 15 15 16 15 15 16 16 16

A B B B B B A B B A B A B B A B B B B

52 61 70 70 65 49 47 55 46 70 59 50 51 50 30 59 58 58 69

27 21 27 28 29 30 31 31 32 33 33 34 34 34 35 35 35 36 37

Reactions were performed at - 15 to -20 “ C . bReactions were performed at -30 to -40°C.

a

HO ,OH*

16 b

OBn

OBn

38

NIS / TMSOTf Molecular

sieves 3 A CH3CN -30‘C

AcHN

OSE OBn 39 a (65 %), p (15 %)

Scheme 5.

P 0 0 0 0 0

0 0 0 0 0 0 15 26 25 8 10 11 0 0

353

14.4 Sialidation Using 2-Thioglycosides

18

SPh

NIS I TMSOTf CH$N -35 c (54 %)

40a

0

41a

24

NH

'

BnO" C\

c o

SMe OAc

40b

NIS / TfOH CH&N -23"C

OAc

A

CQMe o

c

B n o Z C0y

OAc

Ho

(55 %)

~

~

20

c o

OAC

42

~

OBn

41b

NH

CF3-C:

o

BnO

OBn

Acow;: AcO

~

NIS I TfOH CH&N/CHzCI, -30 "C (61 %)

OAc

*

CF3-C;A NHc OAC 0 W 0

i

M

p OSE

0 \

Ph

43

Scheme 6.

with 18,24 or 20, promoted by NIS/TfOH in acetonitrile, providing the desired sialyl-u(2-3)Gal derivatives 41a,41b,and 43 in yields of 45, 55, and 61%, respectively (Scheme 6). The N-benzyloxycarbonyamino (Boc) derivative (44)of sialyl donor 15 reacted with the lactose derivatives 45 and 46,promoted by PhSeOTf/ TMSOTf in tetrahydrofuran, to give 47 (84%) and 48 (46%), u predominantly (Scheme 7) [20]. When this reaction was conducted in dichloroethane, the sialosides were preferred products [21]. An efficient method for obtaining oligosialyl glycosides is the use of dimeric or trimeric Neu5Ac glycosyl donors, which can be prepared from naturally available colominic acid [22]. The lactonized phenyl 2-thioglycosides of dimeric [23] or trimeric [24] sialic acids (49)were each coupled with sugar acceptors 18 and 19 by use of NIS/TfOH as the promoter in acetonitrile, as just described, giving 30-48% of the desired a-glycosides (50,51) selectively (Scheme 8) [9d]. The reactivity of monosialyl (15,16)and disialyl (52)donors for the highly hindered OH-3 of the internal Gal residue (53)was examined (Scheme 9) [25], providing di- and trisialooligosaccharides (54, 55), the key intermediates for the synthesis of u-series gangliosides GTlau [25] and GQlbu [26]. As summarized in Table 2, the lactonized disialyl donor (52)gave the best result (Table 2, entry 6; 44%, u only) promoted by NIS/TfOH, whereas the monosialyl donors (15,16) afforded 34% (entry 1) and 40% (entry 5) of the desired a-sialosides, respectively, promoted by DMTST or

o

s

E

354

-

14 Special Problems in Glycosylation Reactions: Sialidations OH

OBn

OBn

HO H o

c

OBn

45

PhSeOTf TMSOTf THF

1

g

;

*

OBn

OBn

OBn

46

AcO\\‘’ BocHN AcO

c

44

BocHN AcO OBn OBn

48 4 6 % ( a : f 3 = 8 : 1 )

47 8 4 % ( a : P = l l : l )

Scheme 7.

SPh

+

18or19

AcHN AcO

49 (n = 0,l)

N ISKfO H Molecular sieves 4 A (30-48 %)

OSE

50 R’ = H, R2 = BZ 51 R’ = R2 = Bn

Scheme 8.

14.4 Sialidation Using 2-Thioglycosides OAc AcO%

15116

AcHNAcO

2%

CO2Me

.O.

'OSE

AcNH

+

or

355

k1 '0

.t

53

O

G OBn

OBn

OAc A

c

O

k

C02Me

g

AcO * O AcO\\'''

SPh AcO

52

AcHN

Scheme 9.

Table 2. a-Glycosylation" of 53 with sialyl donors 15, 16, and 52 in acetonitrile. Entry

Donor

Promoter

Product

Isolated yield (%)

54 54

34 I

54 54 55

26 40 44

~~

1 2

15 15

DMTST NIS/TfOH

?

16

NT9ITMCnTf

4

16 16 52

NIS/TfOH NIS/TMSOTf NIS/TfOH

5

_I

6 a

All reactions were performed at

~

._I

-I-----.

- 15 "C.

9A --

in A"

356

14 Special Problems in Glycosylation Reactions: Siulidutions 60

Acow

1

RSOTf

Me

Act37 R

SR

C02Me

AcO

(4

I I

-

FH3

TMSOTf

70

+ a-sialosides

Figure 3. A proposed mechanism for a-predominant sialidation in acetonitrile.

NIS/TMSOTf. The yields and a-selectivity are quite appreciable, considering the bulkiness of the donor and the steric hindrance on the sialidation site. Taking into account all these results the critically important factors seem to be (1) the rate of generation of the oxocarbenium ion, (2) the preferred formation of the p-acetonitrilium ion intermediate, and (3) the nucleophilicity of sugar acceptors; this suggests the reaction mechanism [8c, 91 proposed in Figure 3. This highly practical sialidation method has been extensively applied [9] to the synthesis of numerous gangliosides, sialooligosaccharides, and their derivatives and analogs by using not only monosialyl (15-17, 40, 42 etc.), di- and trisialyl (49, 52) donors, but also building blocks such as 56-59 shown in Figure 4 (9; see also 'Glycolipids' in this book). 14.4.2 Xanthates and Phosphites

As alternative sialyl donors, the use of 2-xanthates 60 [27-291 or 2-phosphites 70 [30, 311 of NeuSAc has been examined (Schemes 10 and 11). Early work employing DMTST as the glycosyl promoter showed [27] that sialyl 2-xanthate 60 reacts with Gal acceptor 61 or 6 in acetonitrile to give predominantly

14.4 Siulidution Using 2-Thioglycosides AdqOAc

357

AmuoAc C02Me

C02Me

PO \op

OBz

57

56 AcO

4 0 A c

58

co

59

Figure 4. Building blocks for the systematic synthesis of sialooligosaccharides, gangliosides, and their derivatives and analogs. X = NHAc, NHCOCH~OAC,NHCOCF3, NHBoc, OAc, efc. Y = SMe, OC(= NH)CC13, etc. P = Ac, Bz, etc.

a-sialosides 63 or 64 (Scheme lo), similarly to the corresponding sialyl2-thioglycosides. Improvements in the yield and a-stereospecificity of the products were achieved [28] by using methylsulfenyl triflate [32] as the promoter and a lightly protected lactose acceptor 62 which corresponds to 23 in Figure 1 , to afford 65 in 63-82% yield. The reactive sugar hydroxyl, such as OH-6 of Gal or GalNAc, gave an a,p mixture, however, even at -70 "C [28]. Similar results were also obtained by using phenylsulfenyl triflate as promoter in acetonitrile (Scheme 11) [29]. Sialyl 2-phosphites 70 were also found to react with sugar acceptors (11, 13, 71) promoted by TMSOTf in acetonitrile (Scheme 12) [30, 311, giving a-sialosides (7274) predominantly. 14.4.3 Reaction Mechanism A possible reaction mechanism is shown in Figure 3. When the 2-thioglycosides (a) of sialic acids are specifically activated by the thiophilic promoter at low temperature, the cyclic oxocarbenium ion (d), necessary for almost all known glycosylation reactions, is formed by the initially produced intermediates (b) or (c), and subsequently, reacts with acetonitrile to generate the p-acetonitrilium ion (e), which then undergoes S Ndisplacement ~ at the anomeric center by sugar hydroxyls to give predominantly a-glycosides. The common oxocarbenium ion intermediate ( d ) is also thought to be formed by activation of 2-xanthate (60) or 2-phosphites (70), generating the p-acetonitrilium ion (e)which provides predominantly a-sialosides. It has been demonstrated that the a-D-glucopyranosyl acetonitrilium ions are stereospecifically generated from the corresponding oxocarbenium ions in dry ace-

358

14 Special Problems in Glycosylution Reactions: Siulidutions

BzO "-0~~

+

or

6

OH

AcHN

61 AcO 60

4 62

(A) DMTST CH$N -15

(6)AgOTf

c

t

MSB CH&N/CH
AcouOAc AcHN

C02Me

AcHN

A&

OMe

AcO

63 a (48 %), fi (16 %)

64 a (26%), p (4 %)

,OBz AcHN AcO

-

65 (63 82 Yo)

Scheme 10.

tonitrile [33-351. It seems most plausible, therefore, that stereospecific generation of the p-acetonitrilium ion of Neu5Ac from the oxocarbenium ion holds the key to the a-predominant formation of sialyl glycosides in acetonitrile medium as shown in Figure 3 . Thus, the reactive sugar hydroxyls have a chance to attack other reaction intermediates before the generation of the fi-acetonitrilium ion of NeuSAc is complete, giving an anomeric mixture. Coupling with highly hindered secondary alcohols, however, generally results in low yields (but high a-selectivity) of the desired sialyl glycosides with formation of 7 as a major by-product.

14.5 Furtlzev Solutions to the Problem

359

.OTBDMS

un

HO(

-0Bn

n

OBn

I OAc A AcH N

c

67

66

PhSCI/AgOTf/DTBP CH,CN/CH2CI,, -70 "C

,OTBDMS

C02Me O ~ 'OMe

AcO \OBn 68 74 % (a$ = 955)

or

35 (78%; a:p = 96:4)

Scheme 11.

14.5 Further Solutions to the Problem To solve the problem in sialidation (1) introduction of an auxiliary group at C-3 in sialyl donors, (2) specific activation of the anomeric carbon C-2 by use of thioglycosides, xanthates or phosphites in acetonitrile medium, and (3) the use of sterically less hindered sugar acceptors have been attempted and found to be highly effective for obtaining a-sialosides in a regio- and stereospecific manner. Combinations of these procedures have resulted in further improvements. 14.5.1 Combination of C-3 Auxiliary and Sterically less Hindered Sugar Acceptors Combination of C-3 auxiliary (SPh) in sialyl donor 8 and sterically less hindered sugar acceptors (75, 77) has been successfully applied to the synthesis of sialyl LeX [33] and ganglioside GM3 [34] to give the desired 76 and 78 in 63% and 72% yields, respectively (Scheme 13). This method has been extended to the synthesis of disialyl lactose derivative 81 by the coupling of 80 with 77 (Scheme 14) [35].

14 Special Problems in Glycosylation Reactions: Sialidutions

360

OAc

A AcHN C O w E : i e

+

11/13

or 0

AcO

Po

70 (R = Et, Bn)

Ph

71

TMSOTf Molecular sieves 3A

CH&N -40 "C

AcouoAc C02Me AcHN BnO

OBn BnO

OMe

72

73

70-80% (a:P= 4:l

- 5.6:l)

50-55% (aonly)

Ph

74 38 % (aonly)

Scheme 12.

14.5.2 Combination of C-3 Auxiliary and Specific Activation of the Anomeric Center C-2 Further improvements in yield and a-stereospecificity have been achieved [36, 371 by combining the C-3 auxiliary (SPh) and specific activation of the anomic carbon C-2 by use of 2-thioglycosides (82) in acetonitrile (Scheme 15). Thus the coupling of 82 with 24, promoted by (A) methylsulfenyl triflate [36] or (B) phenylsulfenyl triflate [37] in acetonitrile at -40°C, gave 83 in 71% and 83% yields, respectively. Even when coupling was performed with the sterically hindered acceptor 84, the desired a-sialoside 85 was obtained in 67% yield [36]. Coupling with 86, however, resulted in 28% yield.

14.5 Further Solutions to the Problem

8

(X = Br, Y = SPh, P = Bn)

i-

OBn

63%

1

75

Hg(CN)2kwr2 Molecular sieves 4A, CC4 40 “C, 2 days

AcHN

OBn

76

8 (X = CI, Y = SPh, P = Ac)

+ OPlV

72%

AgOTf Molecular sieves 3A,CH3CN 60 “C, 2 days

SEt

AcHN AcO OPIV

78

Scheme 13.

361

14 Special Problems in Glycosylution Reactions: Siulidations

362

AcO

I::!;

C02Me

AcHN

AgOTf Molecular sieves 4R Na2HP04 CHpC12/toluene

AcO

79

-*

AcHN

(X = CI, Y = 8SPh, P

77

AcHN AcO

80 (49 %)

AcHN

AgOTf SnC12 Molecular

SEt

AcHN

sieves 4R -AW300 Na2HP04 CH3CN

OPiV

81 (39 %)

Scheme 14. OAc AcHN A c

O

AcO%;i,i;e".OAc

CqMe S -

AcO

82

24

*

(A) MeSBdAgOTf (6)PhSCI/AgOTf CH+N -40

(R = Me, Et)

C02Me

AcHN AcO

OSE

HO

C

OBn

Brio

83 (71 %, 83 %)

OBn

84 85 (67 %)

AcHN ACo

MPMO

86 AcHN

BZO

87 (28%)

Scheme 15.

OBn

14.5 Further Solutions to the Problem

363

AcO'

Phd

90

8 8 X = CI 89 X = OP(OEt)2

(A) AgOTf, C2H4C12 (B) TMSOTf, CH3CN

AcO

AcHN A

c OAcO

C02Me G PhOBnO

AcHN AcO

OBn

Fs PhO HO

OBn

Brio

OBn

C02Me

AcHN AcO

92 (68%, 83 %)

91 (89 %, 88 %) Scheme 16.

Dramatic improvement in the yield of a-sialyl glycosides has been achieved [38] by using the 3-phenoxythiocarbonyloxy derivatives of the 2-chloride (88) and 2phosphite (89) as sialyl donors (Scheme 16). Even the chloride derivative 88 coupled smoothly with 11 or 90, promoted by AgOTf in dichloroethane, to afford 91 (89%) and 92 (68%) in high yields. Specific activation of C-2 by diethyl phosphite (89), promoted by TMSOTf in acetonitrile, improved the yield of 92 (83%).

14.5.3 Thioglycoside of N,N-Diacetylneuraminic Acid and Combination with C-3 Auxiliary A novel N,N-diacetyl sialyl donor 93, prepared by N-acetylation of 15, has been found to react with 18 more smoothly than 15, giving 94 in 72% yield (Scheme 17) [39]. This method was immediately combined with C-3 auxiliary [40]. In fact, a novel sialyl donor 95 was found to couple with 84 or 86 in higher yields than the corresponding Neu5Ac donor 82 (R = Et) shown in Scheme 15, affording 96 (83%) and 97 (44%).

14 Special Problems in Glycosylation Reactions: Siulidutions

364

18 __7

NIS/TfOH

#

AyN

Molecular 93 __

sieves 38, CH3CN -40 "C

94 (72%)

04

COzMe SEt

AcO

Molecular

I

sieves 3A CH3CN/CH&I, -45 "C

96 (83%)

97 (44 %)

Scheme 17.

References 1. (a) W. Reutler, E. Kottgen, C. Bauer, W. Gerok in Sialic Acids-Chemistry, Metabolism and Function, Cell Biology Monographs, Vol. 10 (Ed.: R. Schauer) Springer, New York, 1982, pp. 263-305; (b) S. Hakomori, J. Bid. Chem. 1990, 265, 18713-18716; (c) S. Kelm, A. Pelz, R. Schauer, M. T. Filbin, S. Tong, M.-E. de Bellard, R. L. Schnaar, J. A. Mahoney, A. Hartnell, P. Bradfield, P. R. Crocker, Curr. Biol. 1994, 4 , 965-972; (d) C. Mitsuoka, K. Ohmori, N. Kimura, A. Kanamori, S. Komba, H. Ishida, M. Kiso, R. Kannagi, Proc. Natl. Acad. Sci. USA 1999, 96, 1597-1602. 2. (a) K. Okamoto, T. Goto, Tetrahedron 1990, 46, 89-102; (b) M. P. DeNinno, Synthesis 1991, 583-593. 3. (a) H. Paulsen, H. Tietz, Carbohydr. Res. 1984, 125, 47-64; (b) H. Paulsen, U. von Deessen, H. Tietz, ibid. 1985, 137, 63-77. 4. (a) T. Ogawa, M. Sugimoto, Carbohydr. Res. 1985, 135, c5-c9; (b) M. Numata, M. Sugimoto, K. Koike, T. Ogawa, ibid. 1987, 163, 209-225. 5. (a) M. Sugimoto, T. Ogawa, Glycoconjugate J. 1985, 2, 5-9; (b) H. Paulsen, U. von Deessen, Carbohydr. Res. 1986, 146, 147-153; (c) S. J. Danishefsky, J. Gervay, J. M. Peterson, F. E. McDonald, K. Koseki, T. Oriyama, D. A. Griffith, J. Am. Chem. Soc. 1992,114, 8329-8331. 6. (a) Y. Ito, T. Ogawa, Tetrahedron Lett. 1988, 32, 3987-3990; (b) T. Kondo, H. Abe, T. Goto, Chemistry Lett. 1988, 1657-1660.

References

365

7. Y. Ito, S. Nunomura, S. Shibayama, T. Ogawa, J. Org. Chem. 1992,57, 1821-1831. 8. (a) 0. Kanie, M. Kiso, A. Hasegawa, J. Curbohydr. Chem. 1988, 7; 501-506; (b) T. Murase, H. Ishida, M. Kiso, A. Hasegawa, Curbohydr. Res. 1988, 184, cl-c4; (c) A. Hasegawa, T. Nagahama, H. Ohki, K. Hotta, H. Ishida, M. Kiso, J. Carbohydr. Chem. 1991, 10, 493-498. 9. (a) A. Hasegawa, M, Kiso in Carbohydrates-Synthetic Methods and Applications in Medicinal Chemistry (Eds.: H. Ogura, A. Hasegawa, T. Suami), Kodansha/VCH, Tokyofleinheim, 1992, pp. 243-266; (b) A. Hasegawa in Synthetic Oligosaccharides~lndispensableProbes f o r the Life Sciences (Ed.: P. Kovdc), ACS Symposium Ser. 560, ACS, Washington DC, 1994, pp. 184-197; (c) Hasegawa et al., Methods Enzymol. 1994, 242, 158-198; (d) A. Hasegawa in Modern Methods in Carbohydrate Synthesis (Eds.: S. H. Khan, R. A. O’Neill), Harwood Academic Publishers, The Netherlands, 1996, pp. 277-300; (e) A. Hasegawa, M. Kiso in Prepururive Carbohydrate Chemistry (Ed.: S. Hanessian), Marcel Dekker, Inc., 1997, pp. 357-379. 10. P. Fugedi, P. J. Garegg, Carbohydr. Rex 1986, 149, c9-cl2. 11. (a) G. H. Veeneman, S. H. van Leeuwen, J. H. van Boom, Tetrahedron Lett. 1990, 31, 13311334; (b) P. Konradsson, U. E. Udodong, B. Fraser-Reid, Tetrahedron Lett. 1990, 31, 43134316. 12. A. Marra, P. Sinay, Carbohydr. Res. 1989, 187, 35-42. 13. A. Hasegawa, H. Ohki, T. Nagahama, H. Ishida, M. Kiso, Carbohydr. Res. 1991, 212, 277281. 14. A. Hasegawa, K. Fushimi, H. Ishida, M. Kiso, J. Curbohydr. Chem. 1993, 12, 1203-1216. 15. Y. Makimura, H. Ishida, M. Kiso, A. Hasegawa, J. Curbohydr. Chem. 1996; 15, 1097-1 11 8. 16. R. Roy, F. 0. Andersson, M. Letellier, Tetrahedron Lett. 1992, 33, 6053-6056. 17. A. Hasegawa, A. Uchimura, H. Ishida, M. Kiso, Biosci. Biotech. Biochem. 1995, 59, 10911094. 18. T. Sugata, R. Higuchi, Tetrahedron Lett. 1996, 37, 2613-2614. 19. S. Komba, C. Galustian, H. Ishida, T. Feizi, R. Kannagi, M. Kiso, Angew. Chem. Int. Ed. Engl. 1999,111, 1131-1133. 20. S. Fujita, M. Numata, M. Sugimoto, K. Tomita, T. Ogawa, Carbohydr. Rex 1992, 228, 347370. 21. Y. Ito, T. Ogawa, Tetrahedron Lett. 1988, 29, 1061-1064. 22. R. Roy, R. A. Pon, Glycoconjugate J. 1990, 7, 3-12, 23. H.-K. Ishida, Y. Ohta, Y. Tsukada, M. Kiso, A. Hasegawa, Curbohydr. Res. 1993,246, 75-88. 24. (a) H.-K. Ishida, H. Ishida, M. Kiso, A. Hasegawa, J. Curbohydr. Chem. 1994, 13, 655-664; (b) H. Ando, H. Ishida, M. Kiso, A. Hasegawa, Curbohydr. Rex 1997, 300, 207-217. 25. H. Ito, H. Ishida, M. Kiso, A. Hasegawa, Carbohydr. Res. 1998, 306, 581-585. 26. K. Hotta, H. Ishida, M. Kiso, A. Hasegawa, J. Curbohydr. Chem. 1995, 14, 491-506. 27. A. Marra, P. Sinay, Curbohydr. Res. 1990, 195, 303-308. 28. H. Lonn, K. Stenvall, Tetrahedron Lett. 1992, 33, 115-1 16. 29. V. Matrichonok, G. M. Whitesides, J. Org. Chem. 1996, 61, 1702-1706. 30. T. J. Martin, R. R. Schmidt, Tetruhedron Lett. 1992,33, 6123-6126. 31. M. M. Sim, H. Kondo, C.-H. Wong, J. Am. Chem. SOC.1993,115, 2260- 2267. 32. F. Dasgupta, P. J. Garegg, Curbohydr. Res. 1988, 177, c13-c17. 33. K. C. Nicolaou, C. W. Hummel, N. J. Bockovich, C.-H. Wong, J. Chem. Soc. Chem. Commun. 1991, 870-872. 34. T. Tomoo, T. Kondo, H. Abe, S. Tsukamoto, M. Isobe, T. Goto, Carbohydr. Res. 1996, 284, 207-222. 35. T. Kondo, T. Tomoo, H. Abe, M. Isobe, T. Goto, J. Carbohydr. Chem. 1996, 15, 857-878. 36. T. Ercegovic, G. Magnusson, J. Org. Chem. 1995,60, 3378-3384. 37. V. Martichonok, G. M. Whitesides, Carbohydr. Res. 1997, 302, 123-129. 38. J. C. Castro-Palomino, Y. E. Tsvetkov, R. R. Schmidt, J. Am. Chem. Soc. 1998, 120, 54345440. 39. A. V. Demchenko, G.-J. Boons, Tetrahedron Lett. 1998,39, 3065-3068; Chem. Eur. J. 1999,5, 1278- 1283. 40. N. Hossian, G. Magnusson, Tetrahedron Lett. 1999, 40, 2217-2220.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

15 Special Problems in Glycosylation Reactions: 2-Deoxy Sugars Aluin Veyrihes

15.1 Introduction 2-Deoxy and 2,6-dideoxy glycosides are important structural units in many natural products including antitumor drugs (anthracyclines, aureolic acids, calicheamicin, esperamicin), antibiotics active against Gram-positive bacteria (erythromycins, orthosomycins), antibiotics inhibiting platelet aggregation (angucyclines), drugs used in the treatment of cardiac insufficiency (cardiac glycosides), antiparasitic agents (avermectins). Whereas the therapeutic effect of these drugs is usually mediated by their aglycone part, the glycosidic part is essential for biological activity by regulating the pharmacokinetic properties, not only transport, but also molecular recognition (interaction with nucleic acids, for example). The role of carbohydrates in biologically active natural products has been recently reviewed [ 11. Therefore, methods for the efficient and stereoselective construction of deoxyglycosidic linkages will have useful applications in medicinal and bioorganic chemistry by allowing to understand biological mechanisms and elaborate new and less toxic drugs. The lack of a stereodirecting neighboring group adjacent to the anomeric center makes 2-deoxyglycoside synthesis a particular challenge. Moreover, the absence of electron-withdrawing substituent at C-2 makes the glycosidic bond much more acid labile, giving rise to easy hydrolysis or anomerization. Glycals soon appeared as ideal precursors especially for the synthesis of 2-deoxy-a-glycosides which are thermodynamically more stable than the P-isomers. The most frequently employed strategy for accessing 2-deoxy-~-glycosidesuses temporary equatorial blocking groups at C-2 which can be removed after glycosylation. These groups should stabilize the glycosidic bond and provide the anchimeric assistance required for a good P-selectivity. The synthetic aspects of the chemical construction of 2-deoxyglycosides have been the subject of four recent reviews [2-51.

368

15 Special Problems in Glycosylation Reactions: 2-Deoxy Sugars

15.2 Electrophilic Additions to Glycals: Mechanistic Aspects and Applications to the Synthesis of 2-Deoxyglycosides Addition of an electrophile E+ to a glycal such as protected D-glucal 1 gives a positively charged species which can be viewed as an oxycarbenium ion (2 or 3) in resonance with a cyclic onium ion (4 or 5 respectively); the weight of each structure is highly dependent of the nature of the electrophile and the polarity of the solvent (Scheme 1). If the addition of an electrophile is made in the presence of an alcohol R'OH, one can easily conceive that the glycosylation stereoselectivity will be dictated by the face-selectivity of electrophilic attack, at least if a true onium ion is involved. Trans-diaxial opening of the cyclic ion will occur under assistance by a lone pair of the pyranose oxygen atom. An above-plane addition should lead to an a-D-manno pyranoside 6 (diaxial outcome); a below-plane addition should give a P-D-glucopyranoside, first as a boat conformer 7, then flipping to the usual chair conformer 8 (diequatorial outcome). If a true oxycarbenium ion is involved, one can expect that the stereoselectivity will be governed by the kinetic anomeric effect [6-81 leading to a predominant axial stereochemistry at C-1, the a/P ratio being of course modulated by steric effects (a 3-axial substituent can override the stereoelectronic effect). Oxycarbenium ions 2 and 3 lead to a-D-manno 6 and a-D-gluco 9 pyranosides respectively. Altogether, 2-deoxy-a-glycopyranosides can be predicted to be more easily attained after replacement of E by hydrogen.

RO

Scheme 1. Electrophile-promoted addition of an alcohol to protected D-glucal 1 .

-

15.2 Electrophilic Additions to Glycals Ph3P-HBr 0.05 e ROD 10

H

11

D

369

OR

12

Scheme 2. Ph3P-HBr promoted addition of a deuterated alcohol to 3,4,6-tri-0-acetyl-~-glucal.

15.2.1 Protonation of Glycals

The acid-catalyzed addition of an alcohol to an acetylated glycal appears to be the most direct method for the synthesis of a 2-deoxyhexopyranoside. However, the acid catalyst has to be carefully chosen in order to avoid the Ferrier allylic rearrangement. Bolitt et al. [9] have reported that triphenylphosphine hydrobromide allows mild and high-yield protonation, then subsequent glycosylation (Scheme 2). Glycals with all equatorial substituents, like 3,4,6-tri-O-acetyl-~-ghcal10, give mainly a-glycoside products. Sabesan and Neira [ 101 have disclosed another efficient procedure where a dehydrated strongly acid ion-exchange resin in acetonitrile is used; lithium bromide is added to generate traces of hydrogen bromide needed for the reaction. Both methods give good results only with primary alcohols and water. However, when TMS triflate-triethylamine is used [ 111 at low temperature instead of a protic acid, silylation of the secondary alcohol 14 generates in situ Et3NH+ TfO- which activates the glycal 13 and gives the 2’-deoxy-a-(1-4)-disaccharide 15 in excellent yield (Scheme 3 ) . Franck [12] has shown that the major axial stereochemistry at the anomeric center is not due to a trans addition. No strong bridging of the proton can be really expected and the oxycarbenium ion 11 should be the reactive intermediate (Scheme 2). The use of deuterated alcohols ROD with catalytic amounts of Ph3P-HBr showed that deuteron delivery occurs largely from below the plane of the double bond. The predominant axial stereochemistry at C-1 and C-I’ of the glycoside products 12 and 15 probably arises from the kinetic anomeric effect. These results are consistent with an A d ~ 2(addition-electrophilic-bimolecular) mechanism for a first step of protonation, followed by a glycosyl transfer usually directed by the kinetic anomeric effect. Theoretical calculations [ 131 have shown O h

Scheme 3. Protioglycosylation of a glycal

15 Special Problems in Glycosylation Reactions: 2-Deoxy Sugars

370

16

17

18

Scheme 4. P-Galactosidase-catalyzed addition of glycerol to D-galactal 16.

that the diastereofacial selectivity in protonation of D-glycals can be explained by a polarization of the R bond below the plane, with the exception of D-allal. 15.2.2 Enzyme-Catalyzed Additions to Glycals As early as 1972, Lehmann [14,151 reported that glycals could be stereospecifically converted to 2-deoxyglycosides using glycosidase enzymes. For example, D-galactal2-d 16 was transferred to glycerol using a p-galactosidase from Escherichia coli, 18 (Scheme 4). Delivery of the affording glyceryl 2-deoxy-~-~-galactopyranoside proton to C-2 of D-galactal took place from the bottom face to afford apparent trans addition. But it has been shown [ 161 that an enzymic nucleophile E-Nu is first bound on the a-face in a cis addition similar to the above acid-catalyzed process; the a-adduct 17 thus obtained is then displaced by glycerol from the B-face. Such a mechanism is the exact reversal of the enzyme's natural and stereospecific cleavage reaction. These results have been applied by Beau [17] for the synthesis of various p-2deoxyglycosides including disaccharides. With simple a-galactosides as acceptors, only p-(17-6) linkages are formed, the highest yield (50'!40) being obtained with the methyl glycoside. Limited yields are not the consequence of competing hydration or cleavage of the glycoside, but seem to be due to mutiple transfers. 15.2.3 Halogenation of Glycals Addition of halogen to double bonds occurs by a bimolecular process, with the approach of a halogen molecule perpendicularly to the n system of the enol ether from above or below the plane of the molecule. The resulting charge-transfer complex may rearrange, in the rate-limiting step, to intermediate ions (syn ion pairs, halonium ions, or open carbocations) which, according to their relative stabilities, control the stereochemical course of the reaction. Igarashi [18] reported that chlorine addition to 3,4,6-tri-O-acetyl-~-glucal10 gave the four 1,2-dichlorides 23-26 in proportions which depended on the solvent used. In nonpolar solvents, cis-addition products 24 (p-D-manno) and 25 (a-D-gluco) were predominantly obtained through ion pairs 19 and 20 in which the chloride ion is associated on the same side of the original plane from which chlorine attack first occurred. Moreover, because of the anomeric effect, ion pair 20 and therefore compound 25 were much more favored than ion pair 19 and compound 24 (Scheme 5).

311

15.2 Electrophilic Additions to Glyculs

+

ActaACO *

CI~..CI A

c AcO

Z

S

i

slow

Clp

AcO

charge-transfer complex

AcO

ACO

23

AcO

25 'I CI

24

CI

CI

AcO

CI

26

Scheme 5. Chlorine addition to 3,4,6-tri-O-acetyl-D-glucal.

In polar solvents, trans-addition products 23 (a-D-munno)and 26 (p-D-gluco)were predominantly obtained in nearly equal amounts through chloronium ions 21 and 22. The study of bromine addition to glycals is complicated by the easy anomerization of the 1,2-di-bromides. However, Boullanger and Descotes [ 191 observed that 29 in CC14 gave predominantly an a - ~ bromination of 3,4,6-tri-O-benzyl-~-glucal glum 1,2-dibromide 31, whereas bromination of 3,4,6-tri-O-acetyl-~-glucallO in the same solvent gave mainly the CL-D-WXUZY~O 1,2-dibromide 28 (Scheme 6 ) . Horton [20]

A

c

O

S

&

6r2

LOAC ei 27 cc14

10

29

AcO & *# Br

=OAC

28

-h E ! q

30

Scheme 6. Influence of protecting groups in bromine addition to D-glUCa1.

31

Br

372

15 Special Problems in Glycosylation Reactions: 2-Deoxy Sugars

demonstrated that electronic effects, rather than purely steric effects, can explain this spectacular influence of protecting groups upon the stereoselectivity of the reaction, at least in nonpolar solvents: an electron-withdrawing substituent at C-6 disfavors the formation of ion pairs or oxycarbenium ions and therefore stabilizes symmetrical bromonium ions such as 27, but has no influence upon the addition of chlorine. As seen later on for the addition of N-iodosuccinimide to glycals, preferential approach of bromine from above the plane of the acetylated glycal might result from electrostatic effects, namely orientation of the net dipole moment of the molecule. When substituents with low inductive effects (benzyl groups or 6-deoxyglycal) are present, attack from below the molecular plane leads to a syn ion pair 30 favored by the anomeric effect and readily collapsing to the a-cis isomer 31. In polar solvents, the formation of oxycarbenium ions instead of syn ion pairs will be favored and the attack by bromide ions from below the molecular plane (kinetic anomeric effect) should be easier when bromine at C-2 is above (intermediate 2 in scheme 1); higher proportions of an a-trans addition product are therefore expected whatever the protecting groups are. The results obtained in electrophilic additions of bromine (and other electro29 have been recently reexamined [21]. philes) to 3,4,6-tri-O-benzyl-~-glucal

15.2.4 Bromo- and Iodoalkoxylation of Glycals In 1977 Tatsuta et al. [22] reported that 3,4,6-tri-O-acetyl-~-glucal 10 reacted with various alcohols in the presence of N-bromosuccinimide in acetonitrile to give mainly the 2-bromo-2-deoxy-a-~-mannopyranosides in 70-90% yields. With the exception of methanol, the a-mannolP-gluco ratio was estimated to be better than 95 : 5 when the reaction was performed at 5 “C; higher temperatures increased the amount of P-glucosides. As for protonation, this procedure is limited to sugar primary alcohols such as 32 leading to an a-(1+6) disaccharide 33 in excellent yield (Scheme 7). Radical debromination was then performed with tributyltin hydride and catalytic amounts of azobis(isobutyronitri1e) (AIBN). To explain the good stereoselectivity of the glycosylation reaction, the authors postulated that the gluco bromonium ion (below-plane attack) is destabilized by the “reverse anomeric effect”. This effect, which was put forward [23] to explain the preference of anomeric cationic substituents for the equatorial position on a pyranose ring, was recently shown [241 to be not consistent with theories of molecular

NBS,CH&N % * : A AcO

+

10

% A/ AcO

OAC

32

OAc

-5”C, e 94%

AcFq A&

33 Ac*o-cAC: OAc

Scheme 7. N-bromosuccinimide-promoted glycosylation of 3,4,6-tri-O-acetyl-~-glucal.

15.2 Electrophilic Additions to Glycals

373

structures. It is more probable that in acetonitrile an intermediate oxycarbenium ion with an axial bromine atom at C-2 is preferentially attacked by an alcohol coming from the a-side (kinetic anomeric effect), as it has been postulated above for the addition of bromine. The iodoalkoxylation of acetylated glycals was initially reported by Lemieux [25] as early as 1964. As an application of well-known Prevost and Woodward methods, the reaction was promoted by a mixture of iodine, silver perchlorate and collidine (or iodonium di-collidine perchlorate [25],IDCP) in benzene. In 1978 Thiem [26] developed a very useful variation of this reaction where N-iodo-succinimide (NIS) is used as an iodonium donor. Later on, Horton [27] reported a thorough study of Thiem’s procedure where he examined the influence of the solvent, the nucleophile and the structure of the glycal in relation to the ratio of the final products. Iodonium ions being much more stable than corresponding bromonium or chloronium ions, it was not expected that the substituent at C-6 of the glycal would have a significant effect. Observing that the proportion of stereoisomers (only trans-) in acetonitrile remained the same regardless of the nucleophile used, it was concluded that in this solvent the electrophilic addition of iodine to the double bond was irreversible and the proportion of final products reflected the proportion of attack of iodine from below or above the plane of the double bond. In tetrahydrofuran (or in methanol) a complex is formed with iodine, so that the first step will be fast and reversible; then the complex collapses slowly to give the final product (Figure 1). In such a situation, the ratio of stereoisomers will be controlled by the tendency for trans-diaxial opening of the cyclic ion, and by steric factors in the glycal and the nucleophile. It appeared that in acetonitrile steric factors do not exert a major influence on formation of the iodonium ion; electrostatic efects might be responsible for the stereochemistry of the first step. 3,4-Di-O-acetyl-~-fucal34 in the favored ’H4 (L) conformation has an axial acetoxy group at C-4 and is still iodinated mainly from below the plane of the molecule. The dipole moment of the ring oxygen atom, the equatorial acetoxy group at C-3, and especially the axial acetoxy group at C-4 combine to give a net dipole moment with the negative pole below the plane, favoring electrophilic attack from this side and leading to an a-L-fucopyranoside 36 through the iodonium ion 35 (Scheme 8). 3,4-Di-O-acetyl-~-rhamnal 37 with a C-4 acetoxy group equatorial also in the preferred ’H4 (L) conformation has a smaller dipole moment and shows therefore a

3- -

R

R

\+ ,o-I

R

R

fast

0

OR’

+‘A R OH

I

Figure 1. NIS-promoted addition of an alcohol to a double bond in ether solvents.

374

15 Special Problems in Glycosylation Reactions: 2-Deoxy Sugars ROH

Me

AcO

AcO

OAc

34

I

36

35

Scheme 8. NIS-promoted glycosylation of 3,4-di-O-acetyl-~-fucal.

37

10

Figure 2. Iodonium intermediates with acetylated L-rhamnal37 and D-glucal 10.

slight decrease in reaction from below the plane of the molecule. 3,4,6-Tri-O-acetylD-glucal 10 has the same relative configuration as L-rhamnal and consequently is mainly attacked by iodine from above the plane of the molecule in its 4H5 (D) conformation (Figure 2). This interpretation could be complicated by the conformational equilibrium 4H5/ 5H4 which gives to 3,4,6-tri-O-acetyl-~-glucal10, for instance, a 40% contribution of the 5H4 (D) conformer with all its substituents axial. A vinylogous anomeric effect could explain this unusual distribution [28]. However, assuming a lower energy barrier for this conformer equilibrium in comparison to the subsequent reactions, the Curtin-Hammett principle can be applied and the product ratio will be governed by the activation energy of the electrophilic addition, rather than by the actual concentration of conformers. Thiem [29] found that methyl 3,4-di-O-acetyl-~-glucuronal 38 adopts almost exclusively an inverted 5H4 (D) half-chair conformation. A 60:40 mixture of a-D-manno-pyranoside 39 [in a flattened 1,4B(D) boat conformation] and p-Dgluco pyranoside 40 was obtained when 38 was treated with cyclohexanol and N-iodosuccinimide in acetone (Scheme 9). Other methyl glycuronals (L-ribo, D-Z~XO and L - x ~ ~ oadopt ) the usual conformation with an equatorial methoxycarbonyl

Scheme 9. NIS-Promoted glycosylation of methyl 3,4-di-O-acetyl-~-glucuronal.

15.2 Electrophilic Additions to Glycals

375

group and give almost exclusively the expected a-glycoside. A “conformational effect”, where an increased ratio of inverted half-chair conformation in the glycal shifts the a : p anomer ratio towards increased P-glycosides, was suggested. In the N-iodosuccinimide-promoted procedure the succinimide anion is frequently observed to compete with a less potent nucleophile. This leads to glycosyl succinimides which can adopt predominantly an inverted C d (D) chair conformation [30]. To this respect, nucleophilicity of the alcohol can be increased by a previous 0-tributylstannylation which allows better reactivity in NIS-promoted addition to glycals (excellent results are thus obtained with the axially configurated 4-hydroxy group of various L-fucose derivatives) [3 11. Since the stereochemistry of the iodoglycosylation reaction is most often governed by trans-diaxial addition, this methodology is quite successful for the synthesis of 2‘-deoxy-a-disaccharides. The trisaccharide sequences of dihydroaclacinomycin A, an anthracycline antibiotic, and kijanimycin, a macrolide antibiotic, containing three 2,6-dideoxysugar with a-(1-4) and a-(1-3) linkages respectively, were efficiently synthesized by Thiem using the NIS-procedure [3]. The 2-deoxy function is generated from the 2-iodo derivatives most often by radical reduction (Bu3SnH or Ph3SnH and catalytic AIBN), or sometimes by hydrogenation in the presence of palladium-charcoal and a base to neutralize HI, or with NiCl2NaBH4. Applying the “armed-disarmed’’ concept, Danishefsky [32] extended the NISprocedure to the coupling of an armed glycal donor protected by benzyl(29) or silyl ethers with one hydroxyl group of a disarmed glycal acceptor (41) otherwise protected by acyl groups. N-Iodosuccinimide or sym-collidine iodonium perchlorate are the “I+ equivalent” reagents. To reiterate the procedure, it is necessary to enhance the nucleophilicity of the obtained disaccharide glycal42 toward I+, so that it would function as a glycosyl donor to the next acceptor. For this purpose, benzoate groups are exchanged to silyl ethers. The armed disaccharide glycal43 can now be coupled with the disarmed glycal 41 to give a trisaccharide glycal 44, which can react further with a sugar alcohol to give a tetrasaccharide (Scheme 10).

’

Scheme 10. Trisaccharide assembly employing the iodoglycosylation reaction.

376

15 Special Problems in Glycosylation Reactions: 2-Deoxy Sugars

45

Scheme 11. An a-iodoacetate as a-glycosyl donor.

Using this strategy, Danishefsky [33] performed the synthesis of ciclamicin 0, an anthracycline antibiotic containing a trisaccharide with three a-(1-4) linked 2,6dideoxy sugars and an a-linkage to the daunomycin aglycone (See Figure 10 for structure of ciclamicin 0). This work and other uses of glycals for the assembly of oligosaccharides and glycoconjugates have been recently reviewed [ 341. Olivomycin A, a member of the aureolic acid family (See Figure 7 for structure), contains a 2,6-dideoxy branched sugar, olivomycose, which is a-(1-3) linked to another 2,6-dideoxy sugar. Instead of running a direct iodoglycosylation of olivomycal 45, Roush [35] preferred to convert this glycal into the iodoacetate 46 by action of acetic acid and N-iodosuccinimide at -78 "C; a 85 : 15 ratio of diaxial 46 and diequatorial isomers was thus obtained (Scheme 11). It appeared that iodoacetate 46 is an excellent glycosyl donor when activated by trimethylsilyl triflate at low temperatures; high a-stereoselectivity and excellent yields were obtained with warious acceptors including glycals such as 47 which gave a disaccharide glycal48 incorporated into the total synthesis of olivomycin A [361. At reflux the reaction of a glycal49 with acetic acid and NIS gives a 1 : 1 mixture of P-D-gluco 50 and a-D-manno 51 isomers. After separation, the manno isomer can be reduced back to the starting glycal with lithium iodide in tetrahydrofuran

PI.

Another route, completely stereospecific at C-2, to a p-D-glUC0 isomer 54 starts from a 1,6-anhydro compound 53 easily obtained [38] by iodocyclization of D-glucal 52 (Scheme 12). The 2-deoxy-2-iodo-~-glucopyranosyl acetates such as 55 appeared as excellent glycosyl donors with various sugar alcohols including glycals (56) using trimethylsilyl or tert.butyldimethylsilyl triflate as promoters. The 2'-deoxy-2'-iodo-Pdisaccharides (57) were obtained in high yield and with a stereoselectivity often higher than 9 : 1. The high levels of stereochemical control in glycosylations with either a-iodoacetates (46) or P-iodoacetates (55) are undoubtedly due to the fact that corresponding iodonium ions of type 4 or 5 respectively are configurationally stable under the glycosylation conditions (TMSOTf, low temperatures); no equilibration of these ions occurs.

15.2 Electrophilic Additions to Glycals

’

-Ro% -

49

Lil, THF, 8246%

317

50

1. (Bu,Sn),O

2.12, CHBCN

R0

OAc

80%

52

HO

H

T W O OAc

TWO

+

F

53

54

O h TMSOTf, -78°C

En0

80%

I

55

i

56

57

Scheme 12. a-Iodoacetates as p-glycosyl donors.

15.2.5 Epoxidation of Glycals

The efficient epoxidation of glycals was reported in 1989 by Danishefsky [34, 391. Since acyl protecting groups can participate in the ring opening reactions of the epoxide, they have to be replaced by ether or acetal functions. Moreover, glycal epoxides being highly sensitive to nucleophilic attack at the anomeric carbon, peracids commonly used for epoxidation of alkenes lead to products of heterolysis; the use of anhydrous potassium fluoride in the mchloroperbenzoic acid-mediated epoxidation of glycals was, however, reported [40] to be successful. 2,2-Dimethyldioxirane (DMDO) appeared as the proper reagent, since the by-product, acetone, is not expected to react with the 1,2-anhydrosugar. 3,4,6-Tri-O-benzyl-~-glucal 29 reacted smoothly with DMDO to give a 20 : 1 mixture of am-gluco 58 and p-D-manno epoxides in nearly quantitative yield (Scheme 13). Epoxidation of D-allal with an axial 3-O-substituent occurred selectively on the p-face, whereas D-gulal with hindering substituents on both faces of the double bond gave a 1 : 1 mixture of a- and 0-epoxides. Neat methanol reacted smoothly with the a-D-yluco-epoxide 58 to give quantitatively the methyl P-o-glucopyranoside. More complex acceptors, like sugar alcohols (59), required a promoter (anhydrous zinc chloride, or zinc triflate with stannylated alcohols) and gave predominantly p-Dglucopyranosides (60). Yields can be improved by constraining the C-3 and C-4 or C-4 and C-6 oxygen functions of the glycal epoxide into a cyclic motif (carbonate or acetal). The uniquely generated 2’-hydroxyl group arising from opening of the 1,2epoxide donor can therefore be exploited in the synthesis of 2-deoxy-~-glycosides [41].Deoxygenation into 62 was accomplished by free radical reduction of the 2’-

318

15 Special Problems in Glycosylation Reactions: 2-Deoxy Sugars

0

ZnCI, , THF 50% S

29

E

W

0

77%

1

R = OC(S)OPhF561

90%

K

R=H62

Scheme 13. Synthesis of a 2’-deoxy-P-disaccharide with a glycal epoxide as glycosyl donor.

pentafluorophenylthiocarbonate 61 (Scheme 13). Moreover, the glycosyl acceptor can be a glycal which will be compatible with the deoxygenation process. Glycal epoxides also react with phenols [41, 441 in the presence of potassium carbonate and a crown-ether to give (3-aryl glycosides; opening of the epoxide by phenate ions occurs according to a mechanism of S N type. ~ Deoxygenation at C-2 as above affords aryl 2-deoxy-(3-glycosides which are found in aureolic acids and other antibiotics. However, with 2,6-disubstituted aryl rings elimination of the anomeric function competes with the deoxygenation reaction and starting glycals are obtained. Under acidic conditions (ZnC12) 1,2-anhydro sugars react with phenols to give mixtures of a- and (3-glycosides; the mechanism of these glycosylations has been recently investigated [45]. In the presence of silver tetrafluoroborate stannylated primary alcohols react with the a-D-gluco-epoxide 58 to give a-glycosides [42]. When the acceptor is a glycal such as 63, an a-disaccharide glycal 64 is obtained, upon which the process can be reiterated to give a trisaccharide 65 (Scheme 14). 1. AgBF4

2. A 5

8

0

~ ~Pyr. o, 52%

*

1. 2,2-Dimethyldioxirane 2.63, AgBF4 3. Ago. Pyr.

P 51%

0-6

Scheme 14. Synthesis of an a-linked trisaccharide with glycal epoxides as glycosyl donors.

1.5.2 Electrophilic Additions to Glycals

319

4-Penten-1-01 h0 OH

ZnCI, ,63%

50% p:a6:1

66

58

67

0

-

Brio*

F

BnO

68

OH

Scheme 15. Conversion of a glycal epoxide to other p-glycosyl donors.

Reactions with secondary acceptors proceeded in lower yields and exhibited sharply diminished stereoselectivities. It should be mentioned that 1,2-epoxides can be converted [43] to other glycosylating agents, thiophenyl glycoside 66, pentenyl glycoside 67, fluoride 68, which can be helpful after acylation at C-2 for the synthesis of 2-deoxy-P-glycosides(Scheme 15). 15.2.6 Addition of Sulfur Based Electrophiles to Glycals Phenylsulfenyl chloride. Addition of phenylsulfenyl chloride, PhSCl, to 3,4,6-tetraO-benzyl-D-glucal29 was first performed by Schmidt [46]. The adduct arising from a below-plane addition was hydrolyzed with Na2C03 (80% overall yield), then converted into an w-trichloroacetimidate which in the presence of BF,.OEt2 and alcohols delivered mainly 2-deoxy-2-(phenylthio)-~-~-glucopyranosides. As for protonation, the preferred a-face selectivity in the addition of PhS+ was corroborated by theoretical studies [13]. The addition of PhSCl was thoroughly reinvestigated by Roush [47] who came to the following conclusions: the stereoselectivity of the reactions of D-glucal derivatives with PhSCl is highly dependent on the presence of an electronegative heteroatom substituent at C-6, as well as on the functionality at C-4. The C-6 substituent strongly influences the conformational preferences of the D-glucal derivatives, and greatest stereoselectivity is obtained with those glycals that exist preferentially in the inverted 5H4 (D) half-chair conformation [29]. Addition to the glycals is slow compared to 5H4/4H5 conformational interconversion, especially if the reaction is performed at low temperature. Therefore, the Curtin-Hammett principle must apply and the product distribution will be determined by the relative energies of the competing transition states and not the distribution of ground state conformers. Addition of PhSCl to the bottom face of the 4H5 (D) conformer of D-glucal 69 leads to the episulfonium ion 71 and should be favored relative to the top face addition leading to 70, owing to non-bonded interactions of the PhS- unit with the adjacent TBS ether in the latter case. This diastereofacial selectivity, however,

380

15 Special Problems in Glycosylation Reactions: 2-Deoxy Sugars

R&'

TBSO

f- TBSO

TBSO

72 a-D-Manno

S

70

O

7

GI

I

%

X

R O + ~ @

-k 'Ph 71

Ro SPh

+

*?&o PhS CI

73 a$-D-Gluco

A

Scheme 16. Stereochemistry of addition of PhSCl to D-glUCal derivatives.

should not be expected to be overwhelmingly large, according to what is observed [48] for glycals that are locked in this conformation. On the other hand, the two faces of the 5H4 (D) conformer are differentiated to a much greater extent: the bottom face is shielded only by the axial C-4 substituent. A charge-dipole interaction will stabilize the transition state when that substituent is an acyloxy group 72 > 25 : 1) was indeed obtained [27]. The best selectivity (a-D-glUCO 73/a-~-manno with bromo and chloroacetoxy substituents at C-6 and C-4 respectively (Scheme 16). This mechanistic analysis has to be, however, readjusted with glycals of other configurations, since a D-fucal derivative, which has no polar substituent at C-6, gives a bottom face addition with an excellent selectivity [49]. The 2-deoxy-2-(thiopheny1)-a,P-D-glucopyranosylchlorides such as compound 74 were hydrolyzed with Na2C03 or DBU in tetrahydrofuran-water to give lactols which can be converted [50] into aryl 2-deoxy-P-glycosides by a Mitsunobu reaction, followed by desulfurization or into a-trichloroacetimidates such as 75 which were found configurationally stable at C-2 under the TMS triflate-promoted glycosylation conditions [ 5 I]. The stereoselectivity in the glycosylation of various alcohols was found to be dependent on the substrate; the least sterically hindered alcohols gave the best selectivity for the desired P-glycosides 76 (a direct S N ~ displacement or the substitution of a tightly solvated trichloroacetamideoxycarbenium ion pair was probably occurring). The fact that the a-glycosides such as 78 comprise up to 20-50% of the product in glycosylation of hindered secondary alcohols supports the thesis that the reaction stereoselectivity is not governed by the intermediacy of episulfonium ions 77a or 77d, but rather that substitution reactions of oxycarbenium ions 77b or 77c play a dominant role (Scheme 17). The episulfonium ions have been shown [52] by theoretical methods to be less stable than the corresponding oxycarbenium ions. This subject is still a matter of high controversy [53].

15.2 Electvophilic Additions to Glycals

381

1. DBU, THF-HZO

TMSOTf, ROH CH& , -78OC

2. NaH, CC13CN

ROH Felkin-Anh control AcO

n b

SPh

PhS

nc control

76

AcO

s ' h 77d

70

OR

Scheme 17. Stereochemistry of glycosylation with a 2-thiophenyl-a-~-glucopyranosyl donor.

Desulfurization is achieved with Raney-nickel in refluxing ethanol or tetrahydrofuran at room temperature [46], or Bu3SnH-AIBN in toluene at 100°C [49]. Sulfonium salts and sulfenates. Arylbis(ary1thio)sulfonium salts were shown by Franck [48] to undergo electrophilic addition to glycals in the presence of alcohols to form principally P-glycosides. Ester-blocked glycals cannot be used since they are decomposed through a Ferrier rearrangement. The p-tolyl reagent gave the best P-selectivity, but although a major below-plane attack of sulfur onto the 3,4,6tri-0-benzyl-D-ghcal was always observed, the face selectivity was found to be highly dependent on the nature of the nucleophile. A case of double diastereodifferentiation was also uncovered with a racemic alcohol. Even 4,6-O-benzylidene or -isopropylidene D-glycals, which cannot flip to the inverted half-chair conformation, show a preference for below-plane attack. The conformational effect underlined above for iodine and PhSCl additions should therefore be discounted here. A powerful steric effect of a 3-axial alkoxy group will direct the electrophile to the opposite face. These results suggest that the reactive species associates the electrophile and the nucleophile in a dithiosulfenate structure (Figure 3). It can be seen that the aglycon and the glycal interact through space as the glycal begins to bond to the sulfur, and a face-differentiating interaction can be envisaged. The first-formed intermediate would then be a sulfurane. Franck [54] applied this methodology to the synthesis of an aureolic acid disaccharide made of two 2,6dideoxy sugars with a p-(1-3) linkage. This approach is strongly reminiscent of the reaction reported by Ito and Ogawa [55],where phenylsulfenate esters are added to

382

15 Special Problems in Glycosylation Reactions: 2-Deoxy Sugars \-

OR

SAr

Arylbis(arylthio) sulfonium

Dithiosulfenate

Figure 3. Addition of an arylbis(ary1thio)sulfonium salt and an alcohol to a glycal.

glycals in the presence of TMS triflate to give mainly P-glycosides (Figure 4). But although a similar attack of the TMS-activated sulfenate on the glycal is occurring, a close interaction between aglycon and glycal is not required here. In both cases the reaction can evolve to an episulfonium ion which is trans opened by the alcohol or a nucleophilic equivalent. 15.2.7 Addition of Selenium Based Electrophiles to Glycals

Beau and Sinay [ 561 reported the glycosyloxyselenation of glycals as an approach to 2'-deoxy-a-disaccharides. Addition of phenylselenenyl chloride, then an alcohol 79 and collidine to a protected D-glucal such as 29 in acetonitrile gave 9 : 1 mixtures of a-D-manno 80 and P-D-gluco glycosides in good yields (Scheme 18). The selective formation of a 1,2-episelenonium ion above the molecular plane was explained by complexation with benzyloxy groups. Unlike sulfur addition, the selenation is reversible and attempts to produce P-2-deoxy-glycosides in a one-step procedure were therefore investigated [57].It turned out that only trans diequatorial acetoxyselenides such as 81 can be obtained with a high selectivity (P: a 9 : 1) by adding PhSeC1, then silver acetate to the glycal in toluene. The stereochemistry of this addition was strongly influenced by the solvent polarity and the nature of the glycal protecting groups (a 4-0-benzoyl group reversed the selectivity). Acetoxy-selenides (81) were able to glycosylate even hindered secondary alcohols such as 1,2 : 5,6-di0-isopropylidene-a-D-glucofuranosein the presence of TMS triflate to give the P-disaccharide 82 in excellent yield (Scheme 18). The high P-selectivity was given

Figure 4. Addition of a phenylsulfenate ester to a glycal.

15.2 Electrophilic Additions to Glycals

383

PhSeCI, CHsCN, 0°C

%q/J then

r.t.

BnO

29

collidine -

1

79 AcHN o

h

*

80

61%

PhSeCI, Tol. then AgOAc

DlPGlcf TMSOTf

BnF* Brio

29

81%

81

SePh

a-Manno : g-Gluco 1 : 9

97%

82

a-Manno : g-G/uco I : 10

T\

Scheme 18. Stereoselective syntheses of 2/-deoxy 2'-phenylseleno-a- and 0-disaccharides.

by the efficient directing effect of the phenylselenyl group at C-2 provided that diethylether was used as a solvent. Other solvents like dichloromethane induced epimerization at C-2. Roush [47, 511 reinvestigated this two-step P-glycosylation method and came to the same conclusions as for the addition of PhSCl to glycals: the stereoselectivity of the reaction of ~-glucalderivatives with PhSeCl is highly dependent on the presence of an electronegative heteroatom substituent X at C-6 (X = Br or OTs better than X = H or OBn), as well as the functionality at C-4. A possible equilibration of episelenonium species, which does not seem to have been taken into account in other studies [ 581, makes the procedure, however, slightly less stereoselective than trichloroacetimiwith PhSCl. The 2-deoxy-2-(selenenophenyl)-a-~-glucopyranosyl dates are also highly reactive, but do not give better selectivity than the sulfur analogs. Deselenation is commonly performed with a tin hydride and AIBN in toluene at 100 "C. In a novel approach to 2'-deoxynucleosides, Castillon [59, 601 reported also a thorough study of factors controlling the stereoselectivity of the selenium-mediated N-glycosylation of pyranoid and furanoid glycals by silylated bases. The organoselenium chemistry was also applied [61] to the conversion of glycals into spiroortholactones (86), a type of 2-deoxyglycoside linkage found in orthosomycins (see Figure 8 for orthosomycin structure). Glycosyloxyselenation of 3,4,6tri-O-benzyl-D-glucal 29 by alcohol 83 was followed by unmasking of the second hydroxyl group to be glycosylated, then oxidation into a selenoxide 84. Because syn-elimination away from the two oxygen atoms is not possible for stereochemical reasons, forcing conditions provide a transient and reactive ketene acetal 85 which reacts intramolecularly and under stereoelectronic control with the hydroxyl function to give a new axial a ( l h 6 ) glycosidic linkage. The isomeric ortholactone resulting from an equatorial addition was not detected in the reaction mixture (Scheme 19).

384

15 Special Problems in Glycosylation Reactions: 2-Deoxy Sugars

29

Tol.-VinylOAc BnO

85

Scheme 19. A selenium-mediated synthesis of a 2-deoxy-glycosyl-spiroortholactone.

15.3 The Cycloaddition Way to Glycosyl Transfer Following the elegant synthesis of 2-aminosugar glycosides by [4+2] cycloaddition of azodicarboxylates to glycals [62], Franck [63, 641 developed new bicyclic donors for the synthesis of 2-deoxy-P-glycosides. A diacylthione 88 generated in situ from a phthalimidosulfenyl precursor 87 undergoes cycloaddition reactions with a variety of glycals (pyranoid 1 and furanoid); an excellent a-face selectivity leading to cycloadducts 89 was observed in all cases, except with the allal where the pseudoaxial substituent at C-3 directs the addition above the plane (Scheme 20). The resulting heterocycle 89 is a vinyl glycoside which should behave as a glycosyl donor by activation with an electrophile. The acyl side chain or the sulfur atom could provide neighboring group participation, making the glycosylation highly stereoselective. The cycloaddition appears to be a reaction with inverse electron demand, since the smallest differences in energy are between the HOMO of the glycal dienophile and the low-lying LUMO of the heterodiene. Acetates lower too much the HOMO of the glycal, so that no reaction was observed. When the reaction is performed in methanol, a 50% yield of methyl glycosides is obtained along with cycloadducts. This result was interpreted as a two-step process where episulfonium species or corresponding oxycarbenium ions are initially formed, then rearrange into cycloadducts or are intercepted by methanol.

lutidine CHzClz , r.t.

SNPhth

87

88

CHzClz, pyridine, r.t. 5 days, 80% (R = En)

Scheme 20. Cycloaddition of a diacylthione to a protected D-glucal.

15.4 Fluoroglycosylation of Glycals

385

1 . 2 R’OH, TfOH CH2CI2, - 20% RO 2. Raney nickel RO RO -OR r

liCI4, 76% 92

Scheme 21. Synthesis of 2-deoxy-P-glycosides with cycloadduct glycosyl donors.

The cycloadducts 89 cleanly opened in the presence of p-toluenesulfonic acid only with methanol. A further survey of the reaction [65] showed that the a-D-gluco cycloadduct exhibited in fact reversibility with its glycoside product, wheras the 0D-manno adduct did not. The differing anomeric effects exerted in the ground states were proposed as the force which explains the differences in behavior. As a consequence, alcohol acceptors should be added in large excess to the glum donor. The a-D-gluco 89 and D-galacto cycloadducts could be, however, activated by methylenation of the ketone function with the Nysted Reagent 90 and TiC14. affording dienes 91 which are now extended vinyl glycosides. Triflic acid in dichloromethane proved to be the promoter of choice for their conversion into P-glycosides in good yields and with excellent stereoselectivities, even with less nucleophilic secondary sugar alcohols (Scheme 2 1). The addition of tetrabutylammonium triflate suppressed the anomerization of the P-glycoside and also prevents the acidcatalyzed 0- to C-rearrangement of an aryl glycoside. Raney nickel desulfurization affords the 2-deoxy-~-glycosides92.

15.4 Fluoroglycosylation of Glycals Although 2-deoxy-2-fluoro-glycosides cannot be precursors of the corresponding 2deoxy-glycosides, they are highly appreciated for their biological activities and the stabilization of the glycosidic linkage against hydrolysis brought by the fluorine atom. In the field of electrophilic addition to glycals, it seems therefore justified to mention the recent work of Wong [66] where the electrophilic fluorinationnucleophilic addition reactions with Selectfluor-type reagents 93 upon glycals have been studied and optimized. Selectfluor 93 adds first to a glycal such as 10 predominantly from above the plane of the molecule to give a syn adduct, which is susceptible to a slow anomerization and is further displaced by the alcohol 59; heating improves yields and also affects anomeric configuration. A major 2’-deoxy-2’-fluoro-a-~-mannodisaccharide 94 is thus produced (Scheme 22). A judicious choice of protective groups was shown to improve the stereoselectivity of both fluorination and nucleophilic addition.

386

15 Special Problems in Glycosylation Reactions: 2-Deoxy Sugars

Scheme 22. Fluoroglycosylation of 3,4,6-tri-O-acetyl-~-glucal 10 with Selectfluor 93.

15.5 Glycosyl Donors with a C-2 Heteroatom Other methods than electrophilic additions to glycals can afford glycosyl donors with a C-2 heteroatom substituent that will guide the stereochemical course of the glycosylation by anchimeric participation and will be reductively removed to provide the 2-deoxyglycoside. Most often, the glycosylating intermediate (onium ion, ion pair or oxycarbenium ion) will be identical to the one obtained from a glycal. But the nature of the promoter (metal salt, Lewis acid) can strongly affect the stereoselectivity of the glycosylation. Glycosyl donors with an equatorial C-2 heteroatom should normally give access to 2-deoxy-P-glycosides.

15.5.1 2-Bromo-2-deoxyglycosyl bromides Thiem [31 introduced the use of 2-bromo-2-deoxy-a-~-glucopyranosyl bromides 96 for the p-selective glycosylation of complex aglycones involved in aureolic acid antibiotics. These bromides were prepared by the reaction of dibromomethyl methyl 95 and gave ether [67] with a methyl 2,3-O-isopropylidene-a-~-mannopyranoside predominantly (P :a 10 : 1) 2-deoxy-2-bromo-~-glucopyranosides in their silver triflate-promoted reactions with sugar alcohols. Radical debromination afforded the 2-deoxy-P-glycosides97 in excellent yields (Scheme 23). The presence of a labile fonnyl substituent at 0 - 3 of 97 is highly valuable for the

-

R O Y & ,

95

1. R"OH, AgOTf CH3N02-Tol. Ro= -78"c OCHO

BrgF

_____)

OM9

06

Br Br 2. Bu3SnH AIBN, Tol., A

OR"

07

Scheme 23. 2-Bromo-2-deoxyglucopyranosyl bromides in the synthesis of 2-deoxy-P-glycosides.

387

15.5 Glycosyl Donors with a C-2 Heteroatom

further elongation of an oligosaccharide chain at that position, a strategy that was successfully used by Thiem for the synthesis of a P-linked trisaccharide of mithramycin, an antibiotic of the aureolic acid family [68]. 15.5.2 2-Deoxy-2-(thiophenyl)-glycosylfluorides

Nicolaou [69, 701 reported that phenylthio glycosides with a trans OH group at C-2 undergo 1,2-migration of the thiophenyl group when they are treated with (diethylamino)sulfur trifluoride (DAST) at 0 "C. The glycosyl fluorides with an inverted configuration at C-2 are obtained in high yields. A starting phenylthio a-D-mannopyranoside 98 will afford the 2-deoxy-2thiophenyl-P-D-glucopyranosylfluoride 99. Activation of the fluoride with SnC12 in diethyl ether, a tin-complexing solvent, allows the glycosylation of alcohols to give P-glycosides 100 with an excellent P-stereoselectivity through a 1,2-episulfonium intermediate. In the absence of a complexing medium, a-glucosides 101 are predominantly obtained, most probably because the promoter is then coordinated to the sulfur atom and thus prevents it from participating in the coupling reaction (Scheme 24). Desulfurization with Raney nickel affords the desired 2deoxyglycosides. This strategy was applied [711 to the synthesis of the 2,2'-dideoxy-P-disaccharide B-C found in the antibiotic everninomicin, a member of the orthosomycin family (see Figure 8 for complete structure). Phenylthio a-D-mannopyranosides 102 and 104 were converted into fluoride 103 and P-methyl glycoside 105 respectively. The P-disaccharide 106 (BC), obtained by coupling of 103 and 105 in diethyl ether and subsequent protective group adjustment, was then acylated at 0-4' by fluoride 107 (A,) and a-glycosylated at 0-3' by the evernitrose fluoride 108 (A) in dichloromethane to give the AlB(A)C fragment of everninomicin (Scheme 25). Similarly, thioglycosides with a phenoxythiocarbonyl ester group at C-2 (109) undergo 1,2-migration of the anomeric phenyl- or ethylthio group when they are activated (110) by NIS/cat. TfOH. The intermediate episulfonium ion 111 is trapped by an alcohol to give P-Dglucopyranosides 112 when the starting compound is a phenyl- or ethylthio a-D-

- EtZNSF3 CHzCIz, 0°C

98

SPh

R'OH, SnCI, EtpO, - 15°C + : R Ro

RO

Ro *R

99

I

PhS

OR'

loo

PhS

101

PhS OR'

R'OH. SnCI, CHzCI2,-15"C RO !!

Scheme 24. Stereospecific syntheses of 2-deoxy-2-thiophenyl-P- and a-glycosides via 1,2-migration of the PhS group.

388

15 Special Problems in Glycosylation Reactions: 2-Deoxy Sugars

Twkq 1. DAST, CH2C12

Me

~o~~~

2. MeOH. SnCI, , EbO

104

F

Sph 3. BUdNF, THF

105

95%

SPh

1. SnCI, , Et,O. -15"C, 78% 2. Raney Ni, MeOH, 65OC 3. BnBr, NaH. Bu4NI, DMF, 0°C 4. BUdNF, THF, 25°C

Me

BnO

C Me0

109

2. DDQ 3.BF3 OEt, , CH,CI,,

NO2

95% (a-glycosylation)

Scheme 25. Synthesis of an everninomicin fragment.

mannopyranoside 109 (Scheme 26) [72]. 2'-Ethylthio disaccharides are, however, desulfurized by Raney nickel much more sluggishly than the corresponding 2'phenylthio compounds. 15.5.3 2,6-Anhydro-2-Thio-Glycosyl Donors Tatsuta and Toshima [73] designed conformationally rigid 2,6-anhydro-2-thio sugars for the stereocontrolled synthesis of 2,6-dideoxy-a- and -P-glycosides which are very commonly found in bioactive natural products. These new donors have a very rigid structure (boat conformation) and the stereoselectivity of the glycosylation should not be affected by the anomeric effect in the same manner as the more usual chair conformers.

S B

n

?

!

BnO

109

R = Ph or Et

- - -

OPhNIS

TtOH

SR

~

n

Bno

110

OAOPh ?

t

SR

q

B

n

?

!

Bno

R'OH

B n E o w Bno

111

:+

SR

112

R

Scheme 26. 1,2-Ethyl (or phenyl) thio group migration for the stereospecific synthesis of 2-deoxy-Pglycosides.

15.5 Glycosyl Donors with a C-2 Heteroatom

389

The required methyl 2,6-anhydro-2-thio-u-~-altropyranoside 113 was prepared according to known procedures [74, 751, then converted into a phenylthio glycoside 114,a fluoride 115 or a p-acetate 116.Activation of the thioglycoside 114 with Nbromosuccinimide or methyl triflate, and of the fluoride 115 with various Lewis acids (SnC12, SnC12-AgC104, SnCI2-ZnClz, Cp2 HfC12-AgC104, CpzZrC12-AgBFq, TMSOTf) allowed the glycosylation of various alcohols to give a-glycosides 117 in high yields and nearly complete stereoselectivity . Strikingly, that selectivity was independent of the solvent, the anomeric configuration of the donor and the nature of the promoter. In a drastic contrast, the P-glycosides 118 were almost exclusively obtained when the p-acetate donor 116 was activated by various Lewis acids (TMSOTf, Tf20, SnC14, TrC104); only in diethyl ether and tetrahydrofuran was the selectivity reversed, because of deactivation of the promoter. Anomerization of ainto p-glycosides by the Lewis acid alone prooved that a thermodynamic control is responsible for the high P-stereoselectivity of the glycosylation reaction (Scheme 27). Methanolysis, then hydrogenolysis with Raney nickel or radical reduction with Bu&H afforded the 2,6-dideoxy-a-(119) and ~-(120)-~-ribo-hexopyranosides. Moreover, 2,6-anhydro-2-thiomannopyranosyldonors 121 (Scheme 28) having an equatorial 3-0-substituent can be obtained by oxidation of the 3-hydroxyl group

OAc

4

OAc ‘ I ”

I

”*

OH

120

-& hF :$=

NBS.DAST

____)

AcO

AcO

OM

OAc

AcO

SPh

OAc

113

OAc

114 E2, ROH

1 . NaOMe 2.[HI E2 = SnClz or SnC12-AgCIC or SnCi2-ZnCI, OR

119

OAc

117

OR

or Cp2HfCI2-AgCIO, or Cp2ZrCI2-AgBF4 or TMSOTf E, = TMSOTf or Tf20 or SnCi, or TrCIO,

Scheme 27. 2,6-Anhydro-2-thio sugars as glycosyl donors for the stereospecific synthesis of 2deoxy-a- and P-glycosides.

390

A+4- %& 15 Special Problems in Glycosylution Reactions: 2-Deoxy Sugars

TPSO

AcO

NIS,TMSOTf

TPSO

121

SPh

Ho

122

TPSO

SPh

AcO

TPSO

123

0

SPh

1. LAH. THF 2. AQO, Pyr. 3. NBS, ROH

Scheme 28. Armed-disarmed glycosylation with 2,6-anhydro-2-thio sugars.

into a ketone, then stereoselective reduction with diisobutylaluminum hydride; they all give 2,6-dideoxy-a-~-arabino-hexopyranosides. Since the configuration of the glycosidic bond is completely independent of the configuration at the anomeric center of the glycosyl donor, these glycosylation reactions must involve an oxycarbenium ion as an intermediate. Under kinetic conditions, a repulsive electronic interaction between the sulfur atom of the 2,6bridge and the alcohol approaching from the p-face impedes the reaction and aglycosides are exclusively obtained. In contrast, under thermodynamic conditions, an a-glycoside with an axial 3-0-substituent will anomerize to the more stable pglycoside (Figure 5). When the 3-0-substituent is equatorial, no 1,3-diaxial interaction is present and the a-glycoside is thermodynamically stable. The high reactivity of 2,6-anhydro-2-thioglycosyldonors results from the electron-donating nature of the bridging sulfur atom. Indeed, the derived sulfoxides 122 and sulfones have no glycosylating power and can thus be implied in block syntheses exploiting the “armed-disarmed effect” (Scheme 28). Coupling of the armed phenylthio glycoside 121 with the disarmed acceptor 122 gives the a-disaccharide 123 in 89% yield. Activation of 123 by lithium aluminohydride (LAH) reduction of the sulfoxide group and subsequent reacetylation allows to glycosylate an alcohol in an a-selective procedure (124 is obtained in 98% yield when R = cyclohexyl). I

H

0-R

&-

AcO OAc

EZ‘

Figure 5. Repulsive interactions in glycosylation with a 2,6-anhydro-2-thio glycosyl donor.

15.5 Glycosyl Donors with a C-2 Heteroatom

HO

391

OH

OM W

O

H Cladinose

Me

SPh

Me

Erythromycin A

S

Figure 6. 2,6-Anhydro-2-thio-cladinose as a glycosyl donor in erythromycin A synthesis.

This highly stereocontrolled method of glycosylation was applied to the synthesis of erythromycin A and the C-D-E trisaccharide of olivomycin A [76]. The 3hydroxyl group of erythromycin A aglycone is sterically hindered and forms a hydrogen bond with the C-1 carbonyl group. Its glycosylation by a L-cladinose donor is a difficult task, as noticed by Woodward in his total synthesis of erythromycin A in 1981. The use of a phenyl2,6-anhydro-2-thio-~-cladinose thioglycoside activated by N-iodosuccinimide and triflic acid allowed the stereocontrolled synthesis of the desired a-glycoside in 90% yield (Figure 6). The olivomycin A trisaccharide (under the form of a P-cyclohexyl glycoside) was assembled through three 2,6-anhydro-2-thio sugars. The P-glycosidic bonds of units C and D were made (90% yields, no a-anomer) with a p-acetate donor which has an axial 3-O-substituent, the conversion to the reauired equatorial configuration being made after each coupling. The a-glycosidic bond of unit E was built in 89% yield and a fully stereoselective glycosylation reaction from a phenylthio glycoside derived [77] from olivomycose (Figure 7).

HO &eM

OAc

?Me OH

DEIPSO

D ‘PrOCO

HO E

Olivomycin A

Figure 7. Synthesis of the CDE trisaccharide of olivomycin A with 2,6-anhydro-2-thio sugars as glycosyl donors.

392

15 Special Problems in Glycosylation Reactions: 2-Deoxy Sugars

15.5.4 1,2-Di-O-Acetyl-fi-Hexopyranoses and N-Formylglucosamine Derivatives An efficient, general and stereospecific preparation of 2'-deoxy-P-disaccharides reported by Sinay [78, 791 makes use of a 2-0-acetyl substituent as a stereodirecting 126 (and -D-galacto-)pyranoses group. 1,2-Di-0-acetyl-3,4,6-tri-O-benzyl-P-~-glucowere obtained from the known corresponding orthoesters 125 (they could have been also prepared from glycal epoxides). They reacted with various sugar alcohols in the presence of TMS triflate at -20 "C to give P-disaccharides in high yields (75-90%) and a total stereoselectivity attributed to the efficient anchimeric assistance of the 2-0-acetyl group. After removal of this group, the Barton-McCombie procedure allowed deoxygenation at C-2' to give the 2-deoxy-P-glycosides 127 in 70-95% yields (Scheme 29). This method could be extended to 6-deoxysugars giving thus a straightforward access to 2,6-dideoxy-P-glycosides. If necessary, a 2-0-acylpyranosyl bromide can be used in place of the 1,Zdiacetate; Konigs-Knorr conditions are then applied to obtain the P-disaccharide [80]. The only limitation lies in the impossibility of using acylated glycosyl acceptors, inasmuch as regiospecific deoxygenation at C-2' would be problematic. This can be circumvented, however, by the following procedure [78, 791. 1,3,4,6-Tetra-0-acetyl-2-deoxy-2-fo~amido-~-~-glucopyranose 128, which is readily obtained from glucosamine, can be converted to an a-chloride 129 or an atrichloroacetimidate 130. These three glycosyl donors, when properly activated at room temperature (TMS triflate for the acetate 128 and the imidate 130, silver tri131 (Scheme flate for the chloride 129),give 2'-deoxy-2'-formamido-~-disaccharides 29). The excellent stereocontrol of the glycosylation reaction results from the participating property of the N-formyl group (oxazolinium ion).

1. ROH, TMSOTf, - 20°C 2. NaOMe, MeOH 3. NaH, irn., CSz, Me1 4. Bu,SnH. AlBN OR

126

127

Me ROH 1. POC13, NEt3 2. Bu,SnH, AlBN

TMSoTf A AcO c ;

A-li

a

O

A

r.t.

c

~

AcO

128 NHCHo 13' HCONH

OR

AcO

NHCHO

t

132

1ax=cI ROH, AgOTf, r.t. 130 X = OC: NHCCI, ROH, TMSOTf, r.t.

Scheme 29. fi-stereocontrol with 2-0-acetyl or 2-N-formyl groups in the synthesis of 2-deoxy-Pglycosides.

15.6 2-Deoxyglycosyl Donors

393

Q Figure 8. Synthesis of the BCDE tetrasaccharide fragment found in orthosomycins.

When the donor is correctly adjusted to the acceptor, yields are high (64-97%). However, acidic conditions of the glycosylation reaction may induce partial anomerization at the reducing end of the obtained disaccharide. In that case, more reactive N-phthaloylglucosamine donors can be used; a sequential treatment converts then the N-phthaloyl group into the required N-formyl substituent. Radical deamination at C-2' through an isonitrile delivered 2/-deoxy-P-disaccharides 132 in 65-89% overall yields (Scheme 29). This methodology was applied to the synthesis of the B-C-D-E tetrasaccharide fragment found in orthosomycins (Figure 8) [82].* The interglycoside spiroortholactone junction between disaccharide B-C and D-E fragments was constructed by glycosyloxyselenation of the glycal, oxidation at selenium, and syn-elimination of the resulting selenoxide, according to the procedure already described [611.

15.6 2-Deoxyglycosyl Donors In the absence of an heteroatom at C-2 a pyranosyl donor is expected, when activated by a promoter, to give an oxycarbenium intermediate which will be attacked

*After completion of this chapter, this methodology has been used as a key step for the total synthesis of everninomicin 13,384-1 by Nicolaou's group (Angew. Chem. Int. Ed. 1999,38, 3334-3350).

394

15 Special Problems in Glycosylation Reactions: 2-Deoxy Sugars

-

Montrnorillonite K-10 ~d-0

CH&,

25’C

Me

TBSO

67%

133

134

om

135 @a 69:31

OMe

Montrnorillonite K-10

CHPCI~, 25°C 136

137a:P 89:19

OMe

Scheme 30. Montmorillonite-promoted glycosylations of olivose.

by an alcohol preferentially on its a-face (kinetic anomeric effect) if no impeding steric interaction is present. However, in a few cases (insoluble silver silicate catalyst, anomeric phosphites), a reasonable 0-stereoselectivity can be obtained. 15.6.1 2-Deoxy-Hexopyranoses

It has been recently reported [ 831 that the stereocontrolled glycosylation of olivose (2,6-dideoxy-~-arabino-hexose) by a primary alcohol such as 134, using montmorillonite K-10 as a solid acid promoter as well as a dehydrating agent, leads predominantly to p-glycosides (135) when a 4-0-acetyl group is present (133); a participating effect of that group was postulated [ 851. With a 4-0-tert-butyldimethylsilyl substituent like in 136, a major a-glycosylation leading to 137 is observed as expected from the kinetic anomeric effect (Scheme 30). In his total synthesis of the macrolide avermectin Bla, Ley 1841 reported the in situ activation of an oleandrose unit (2,6-dideoxy-3-0-methyl-~-arabino-hexose) by carbonyldiimidazole or thiocarbonyl-diimidazole to give imidazolylcarbonyl and imidazolyl-thiocarbonyl glycosides respectively which undergo predominant aglycosylation in the presence of silver perchlorate in tetrahydrofuran. 15.6.2 Tert-Butyldimethylsilyl 2-Deoxyglycosides

Tert-Butyldimethylsilyl 2-deoxy-p-glycosides have been used for the glycosylation of primary [86]and secondary [ 111 sugar alcohols in the presence of TMS triflate at low temperatures. 2’-Deoxy-a-disaccharides are obtained in excellent yields. 15.6.3 1-0-Acyl- and Acetimidyl-2-Deoxy-Hexopyranoses

Remarkable selectivity for an attack from the a-side was observed [87a] when the anomeric p-nitro-benzoate of evernitrose 138 was reacted with the sugar alcohol

15.6 2-Deoxyglycosyl Donors

395

Scheme 31. Stereoselective a-glycosylation of the branched-chain nitro sugar evernitrose.

139 in the presence of TMS triflate at low temperature. In spite of a repulsive 1,3syn-axial interaction between the 3’-methyl branch and the anomeric substituent, a 2’-deoxy-a-disaccharide 140 (AB fragment of everninomicin) was exclusively obtained in a 64% yield (Scheme 31). As early as 1979 and as an application of the “imidate procedure”, Sinay [87b] reported the use of anomeric (N-methy1)acetimidates of 2-deoxysugars for the highly stereoselective synthesis of 2’-deoxy-a-disaccharides in excellent yields. 15.6.4 2-Deoxyglycosyl Bromides and Fluorides ~ of an Insoluble silver salts such as silver silicate favor direct S N displacement anomeric bromide by alcohols [ 881. When applied to 2-deoxy-pyranosyl bromides, this procedure can lead to 2‘-deoxy-P-disaccharideswith a good stereoselectivity. For instance (Scheme 32), bromide 141, where an ester group at 0-4 can also favor [85] P-glycoside formation, reacted with the sugar alcohol 142 in the presence of silver silicate to give the disaccharide 143 in 85% yield (P:a 5.4: 1) [89]. 2-Deoxy-a- and P-pyranosyl fluorides react with sugar alcohols in diethyl ether in the presence of TiF4 to give predominantly glycosides of the same anomeric configuration; a fluoride/catalyst/solvent complex is postulated in a “double s~~displacement”. A non-nucleophilic solvent such as hexane can neither form a complex nor stabilize an oxycarbenium ion and, therefore, in such solvents major S N ~ reactions are observed [90]. 2-Deoxyglycosyl fluorides have also been activated in neutral conditions by LiC104-diethyl ether mixtures 1911. Wiesner [92] introduced the concept of 1,3-participation of an axial 3-0substituent (p-methoxy-benzoate or N-methylcarbamate) for the stereoselective A 1,3-acyloxonium or a P-glycosylation of digitoxose, a 2,6-dideoxy-~-riho-hexose. 1,3-bridging iminium ion was proposed as an intermediate species where the a-face of the donor would be blocked during the glycosylation step (Figure 9).

+

++ozB BnO

;horn + / ; p & Silver silicate

~

BZO

BnO

05%

141

Br

142

Scheme 32. Silver silicate-promoted P-glycosylation.

143 @:a5.4:l

O M

396

R

15 Special Problems in Glycosylation Reactions: 2-Deoxy Sugars

O

or

W

R

O

R"OH w

W

0.

R

O

w

o

R

.

R'COO

R' = pmethoxyphenyl

+o

or NHMe

+ OMe Figure 9. Hypothetical 1,3-~articipationof a p-methoxybenzoate or a carbamate in glycosylation.

However, several attempts [93, 941 to apply this concept with digitoxosyl halides under silver triflate activation (or thioalkyl digitoxosides under mercury salt activation) showed no reliability in enhancing the P :a ratio of 2-deoxyglycosides. 15.6.5 S-(2-Deoxyglycosyl)phosphorodithioates

Michalska [95, 961 introduced S-(2-deoxyglycosyl)phosphorodithioatesfor the synthesis of 2-deoxy-glycosides. 0,O-Dialkylphosphorodithioic acids add to acylated (144) or benzylated glycals in benzene (Scheme 33) to give predominantly crystalline a-adducts (145). Protonation at C-2 occurs from the a-face as it has been seen with Ph3P.HBr, and overall the addition is cis [ 121. Simple alkoxides react upon these a-adducts to give 2-deoxy-P-glycosides ac~ [97]. Activation by thiophilic promoters such as silver cording to a S N mechanism perchlorate in acetonitrile [96], or NIS in dichloromethane, or IDCP in acetonitrile [ 981 leads with alcohols mainly to 2-deoxy-a-glycosides. 2'-Deoxy-a-disaccharides (147) can thus be prepared in good yields, but with a : P ratios better than 9 : 1 only in a few cases (sugar alcohol 146). Anomeric a$-S-phosphorodithioates can also be obtained [99] from glycal epoxide 58. After pivaloylation at 0-2, they can be activated by methyl triflate to give with sugar alcohols exclusively 0-disaccharides in good yields. Selective deoxyge-

-

vc S

S

II

HS P(OEt);!

PhH, r.t.

BZO W

144

Bzd

OEt OEt

$ BzO

s

'

IDCP, CHSCN, 0°C

80%, a:t3 9:l

145

BZO BZO

Scheme 33. 0,O-Diethylphosphorodithioic acid-mediated a-glycosylation of a glycal.

147

15.6 2-Deoxyglycosyl Donors

391

Scheme 34. Generation of anomeric 2-deoxy-a-glycosyl phosphate donors by a radical rearrangement.

nation at 0-2’ as already described [41, 78, 791 would then afford 2’-deoxy-pdisaccharides. 15.6.6 2-Deoxyglycosyl Phosphates, Phosphoramidites and Phosphites An anomeric 2-deoxy-a-glycosyl phosphate 150 is generated by a radical 2-1 migration of the phosphate ester group, which occurs when bromide 148 is irradiated in the presence of tributyltin hydride [loo]. The intermediate anomeric radical 149 (in a boat conformation) rearranges into a C-2 radical in an usual 4C1 (D) chair conformation; an hydrogen atom is then trapped mostly from the less shielded above face (Scheme 34). The unstable phosphate 150 reacts with alcohols in the presence of magnesium perchlorate to give mainly 2-deoxy-a-glycosides 132 in good yields (a: p ratio 2.34.5 : 1). 3,4,6-Tri-O-benzyl-2-deoxy-~-arabino-hexopyranose 151, easily obtained [ 101 from 3,4,6-tri-O-benzyl-~-glucal 29, was converted into an anomeric N,Ndiisopropyl phosphoramidite 152 which, under activation by TMS triflate at low temperatures, reacted with sugar alcohols to give 2’-deoxy disaccharides 153 in high yields and excellent a-stereoselectivity (Scheme 35) [ 1011. 1,2-Cis-2-(p-methoxypheny1thio)-a-D-glycopyranosylphosphoroamidate donors, with an equatorial 2-SAr participating group and the anomeric OP:O(NMe*)Z

EtOP(NiPr,),, CH2C12 HNiPr, - Tetrazole

YEt

BnO

&lo

OH

151

o4

p .

152 a:p1.6:1

OH

BnO NiPr,

- fi 153

CIP(0Et)z

151

ROH, CHzCIZ TMSOTf, -78°C

OR

ROH. Tol. TMSOTf, -94OC

90%

OR

154 a : p 9 4 : 6

OP(OEt)p

127

Scheme 35. 2-Deoxyglycosyl phosphoramidites and diethyl phosphites for respective a- and glycosylations.

p-

398

15 Special Problems in Glycosylation Reactions: 2-Deoxy Sugars

Scheme 36. 2-Deoxy (2-pyridy1)thioglycosides as a-glycosyl donors

leaving group, have been previously [ 1021 used for the P-glycosylation of alcohols. This variety of anomeric phosphates was claimed [ 1031 to be more stable than other phosphates such as 152. Diethyl phosphite 154, when activated in the same conditions (except toluene as solvent) as 152, gave predominantly 2-deoxy-~-glycosides127 (Scheme 35) [ 1041. The highest P-stereoselectivity was observed with primary alcohols (P :a 9 : 1). The glycosylation reaction was assumed to proceed through the intermediacy of a thermodynamically more stable 2-deoxy-a-~-glucopyranosyl triflate (or its tight a-ion pair). Anomeric a,P-bis(trichloroethy1) phosphites undergo mainly a-glycosylation when activated by boron trifluoride etherate or tin (11) triflate at room temperature [105]. 15.6.7 2-Deoxy Thioglycosides

2-Deoxy-(2-pyridyl)thioglycoside 155, when activated by methyl iodide, reacts smoothly with sugar alcohol 156 to give 2'-deoxy-a-disaccharide 157 in good yield and high stereoselectivity (Scheme 36) [ 1061. Tatsuta [ 1071 reported that highly a-stereoselective glycosylation of phenyl 2deoxy-3,4-0-isopropylidene- 1-thio-D-fucopyranoside 158 could be accomplished because of its rigid and unusual boat conformation. 2-Deoxy-a-glycosides (161 versus 159) were obtained with a much lower selectivity from 2-deoxyfucosyl donors 160 in an usual 4C1 (D) chair conformation (Scheme 37). It was also shown that

0

%

" QNBS

0

T SPh

99% 0 -

158

159 a:P>99:1

A d

A m ! & ,

160

97 idem %

SPh

AcO

AcO&

161 a$ 11:l

\

I

-o

0

Scheme 37. Highly stereoselective glycosylation by conformational assistance.

15.6 d-Deo,~yglycosylDonors

399

selectivity came from conjormational ussistunce, not from a 1,3-diaxial interaction due to the configuration of the 3,4-O-isopropylidene group, and was highly independent of alcohol acceptors. This effect was largely exploited in the 2,6-anhydro-2thio-glycosyl donors as already seen [73, 761. 15.6.8 2-Deoxyglycosyl Sulfoxides The “armed/disarmed concept” takes into account that the electron-donating/ withdrawing power of substituents in carbohydrates has a profound impact on the reactivity of the anomeric center. It can therefore be exploited to construct tri- and larger oligosaccharides in a one-step reaction. In 1993 Kahne [lo81 reported the synthesis of the trisaccharide of ciclamycin 0, an anthracycline antibiotic, in 25% yield by a “polymerization reaction” using two 2-deoxyglycosyl sulfoxides B and C and a phenyl2-deoxy-thioglycoside A as monomers (Figure 10). The reactivities of the sulfoxide donors and nucleophiles were tuned such that the linkage between A and B would occur first (a p-methoxybenzyl sulfoxide B is more reactive than an unsubstituted phenyl sulfoxide C), followed by the linkage to C (a silyl ether AB is a good acceptor when triflic acid is used as a promoter, but reacts more slowly than an unprotected alcohol). Methyl propiolate is used as a sulfenic acid scavenger. The trisaccharide has an anomeric phenyl sulfide on the A ring and can be readily oxidized into a sulfoxide, allowing an iterative strategy for oligosaccharide synthesis. The use of p-methoxybenzyl (PMB) groups in place of benzyl ethers was required for final deprotection of ciclamycin 0, but was found to decrease the yield of the one-step coupling reaction. Therefore, a stepwise approach to the trisaccharide was finally performed with noticeable improvements in glycosylation steps [ 1091. Activation of an anomeric sulfoxide with triflic anhydride releases phenylsulfenyl tri-

TfOH, Et20-CHZCIZ HC&CO,Me -70to-70°C 25%

c

1I

Ciclarnycin 0 b

p

C

0 ’

Figure 10. One-step synthesis of the trisaccharide of ciclamycin 0.

SPh

Me& A

15 Special Problems in Glycosylation Reactions: 2-Deoxy Sugars

400

Me$sph

TO’

kSPh

-&+Me?g-7 2 ‘:SPh

HO OPMB

AcO

AcO

PhSOTf

- q m l V e n g m J

AcO

OPMB

PhSSPh

Me0 -

m

;

P

Me$sph

h

Me0

b@o OPMB

b&o AcO

-

Me@

OPMB

Me#o AcO

OPMB

OPMB

Figure 11. Formation of a trisaccharide by-product by competitive PhSOTf activation of a phenylthio glycoside.

flate, PhSOTf, which causes problems by activating thiophenyl glycosides even at low temperatures; a trisaccharide resulting from such an activation was identified among by-products (Figure 1 1). It was found that 4-allyl-1,2-dimethoxybenzene could trap efficiently the phenylsulfenyl triflate, although some trisaccharide was still present. A second strategy was then examined: a hindered sulfide (2,6-dichlorophenyl) with decreased nucleophilicity would react more slowly with PhSOTf, providing a better opportunity for the scavenger to operate more efficiently. Finally, an “inverse addition” procedure (the sulfoxide is slowly added to the mixture of acceptor and promoter) minimizes the formation of anomeric sulfenates. With these modifications disaccharide A-B, then trisaccharide A-B-C could be obtained in 82 and 68% yields respectively. Oxidation of trisaccharide A-B-C with 2,2-dimethyldioxirane gave a sulfoxide (900/,)which was coupled to the aglycone of ciclamycin 0 under the modified conditions in 75% yield (Figure 10). After removal of the p-methoxybenzyl ethers with DDQ the ciclamycin 0 was obtained in a 17% overall yield and with only six steps from the starting monomers.

15.7 Other Approaches to 2-Deoxyglycosides All the approaches which have been described so far involved glycosylation reactions, that is to say the creation of an exocyclic C-1-0 bond on the pyranose ring.

15.7 Other Approaches to 2-Deoxyglycosides

162

I

1. PhSeCl 2. Bu3SnH

401

%& OR

127

Scheme 38. Electrophile-mediated intramolecular cyclization of an acyclic enol ether.

However, it should be also possible to construct a 2-deoxy-glycoside by formation of the endocyclic C-1-0 bond from an acyclic sugar, or the C-1-H-1 bond from an anomeric radical. 15.7.1 Cyclization of Acyclic Sugars Acyclic sugars obtained by Wittig reactions. Acyclic enol ethers 163 obtained by Wittig reaction upon 2,3,5-tri-O-benzyl- D-arabinofuranose 162 undergo intramolecular cyclizations when they are treated by an electrophile. N-Iodosuccinimide gives predominantly 2-deoxy-2-iodo-a-~-mannopyranosides which can be deiodinated to give 2-deoxy-a-glycosides 153. The reaction of phenylselenenyl chloride with 163, followed by radical deselenation, gives mainly the 2-deoxy-P-glycosides 127 (Scheme 38) [110, 1111. Redox glycosylation via thionoester intermediates. Barrett [ 112, 1131 introduced the concept of “redox glycosylation”, where the usual glycosylation reaction is replaced by the esterification of an aldonic acid 164, followed by reduction to a reactive intermediate (enol ether from an ester or 0,s-acetal 166 from the thionoester 165 obtained by treatment of the ester by the Lawesson’s reagent). Cyclization of 166 by silver tetrafluoroborate gives the 2-deoxyglycoside 167 in good yield (Scheme 39).

1. ROH, DCC 2. Lawesson’s reagent

165

OL.OR S

OMe 167

Scheme 39. Redox glycosylation via thionoester intermediates.

166

II

A m 4

collidine

402

15 Special Problems in Glycosylation Reactions: 2-Deoxy Sugars 1. Lawesson's reagent 2.286-Di-t-butyl-4-methy~En~+ Mel. MeOH, 40°C

Eno+

pyridine

sMe-

BnO 168

169

' 0

81 %

En0

om

OMf

170 p:a6:1

Scheme 40. Alkoxy-substituted anomeric radicals for the construction of 2-deoxy-P-glycosides.

15.7.2 Use of Alkoxy-Substituted Anomeric Radicals The hemithio orthoester 169 was prepared from lactone 168 by treatment with Lawesson's reagent, then reaction with methyl iodide and methanol. Radical desulfurization of 169 gives predominantly a 2-deoxy-P-glycoside 170 by transfer of hydrogen to an intermediate methoxy-substituted anomeric radical preferentially from below (Scheme 40). P-Selectivity is, however, enhanced when an alkoxy substituent is present at C-2 [ 1141. Crich [ 115-1 171 generated anomeric alkoxy-substituted radicals by irradiation of "Barton esters" (0-acyl thiohydroxamate 176) in the presence of a thiol. Conversion of a protected D-galactall71 into the anomeric sulfone 172, then into the thiophenyl glycoside of a methyl 2-ulosonate 173, was followed by a classical 0-glycosylation (alcohols or phenols). The carboxylic acid 174 was coupled with the salt 175 to give the 0-acyl thiohydroxamate 176. Its photolysis in the presence of a thiol generated an anomeric radical by decarboxylation; hydrogen transfer took place stereoselec177 were tively from below as for compound 169 and 2-deoxy-~-galactopyranosides predominantly produced (Scheme 41).

RO

- -

1. HCI, tol. 2. PhSH, Et(iPr),N OR 3. MCPBA. NaHC03

1 . Buti, O:C(OMe),

2.U Naphth., (PhS)2

RO

171

172

RO

OR'

+-

177

R = tBuMe,Si

Scheme 41. Reductive decarboxylation of 3-deoxy-ulosonic acid glycosides

I I SPh

179

1 . R'OH, HgCI, or NBS

R'SH CHZC12 hv, 5°C

RO

CO2Me

RO

References

403

References 1. A.C. Weymouth-Wilson, Nut. Prod. Rep., 1997, 14, 99-110. 2. R.W. Franck, S.M. Weinreb in Studies in Natural Products Chemistry (Ed.: Atta-UrRahman), Elsevier, Amsterdam, 1989, p. 173. 3. J. Thiem, W. Klaffke, Top. Curr. Chem., 1990, 154, 285-332. 4. K. Toshima, K. Tatsuta, Chem. Reo., 1993, 93, 1503-1531. 5. A. Kirschning, A.F.-W. Bechtold, J. Rohr, Top. Curr. Chem., 1997, 188, 1-84. 6. A.J. Kirby, The Anomeric Effect and Related Stereoelectronic Effects at Oxygen, Springer Verlag, New York, 1983. 7. P. Deslongchamps, Stereoelectronic Effects in Organic Chemistry, Pergamon Press, Oxford, 1983. 8. R.W. Franck in Conformational Analysis and Stereochemistry of Six-membered Rings (Ed.: E. Juaristi), VCH Publishers, Inc., New York, 1995, Chapter 5 , p. 157-198. 9. V. Bolitt, C. Mioskowski, S.-G. Lee, J.R. Falck, J. Org. Chem., 1990, 55, 5812-5813. 10. S. Sabesan, S. Neira, J. Org. Chem., 1991, 56, 5468-5472. 11. C. Kolar, G. Kneissl, Angew. Chem. Znt. Ed. Engl., 1990, 29, 809-81 1. 12. N. Kaila, M. Blumenstein, H. Bielawska, R.W. Franck, J. Org. Chem., 1992, 57, 4576-4578. 13. R.W. Franck, N. Kaila, M. Blumenstein, A. Geer, X.L. Huang, J.J. Dannenberg, J. Org. Chem., 1993,58, 5335-5337. 14. J. Lehmann, E. Schroter, Carbohydr. Res., 1972, 23, 359-368. 15. J. Lehmann, B. Zieger, Carbohydr. Rex, 1977, 58, 73-78. 16. O.M. Viratelle, J. Yon, Biochemistry, 1980, 19, 4143-4149. 17. J.-M. Petit, F. Paquet, J.-M. Beau, Tetrahedron Lett., 1991, 32, 6125-6128. 18. K. Igarashi, T. Honma, T. Imagawa, J. Org. Chem., 1970,35, 610-616. 19. P. Boullanger, G. Descotes, Carbohydr. Res., 1976, 51, 55-63. 20. D. Horton, W. Priebe, 0. Varela, J. Org. Chem., 1986, 51, 3479-3485. 21. G. Bellucci, C. Chiappe, F. D’Andrea, G. Lo Moro, Tetrahedron, 1997, 53, 3417-3424. 22. K. Tatsuta, K. Fujimoto, M. Kinoshita, Carbohydr. Res., 1977, 54, 85-104. 23. R.U. Lemieux, A.R. Morgan, Can. J. Chem., 1965,43,2205-2213. 24. C.L. Perrin, M.A. Fabian, J. Brunckova, B.K. Ohta, J. Am. Chem. Soc., 1999, 121, 69116918 and references therein. 25. a) R.U. Lemieux, S. Levine, Can. J. Chem., 1964, 42, 1473-1480; b) R.U. Lemieux, A.R. Morgan, Can. J. Chem., 1965,43, 2190-2198. 26. J. Thiem, H. Karl, J. Schwentner, Synthesis, 1978, 696-698. 27. D. Horton, W. Priebe, M. Sznaidman, Carbohydr. Rex, 1990, 205, 71-86. 28. D.P. Curran, Y.G. Suh, Carbohydr. Rex, 1987, 171, 161-191. 29. J. Thiem, P. Ossowski, J. Carbohydr. Chem., 1984, 3, 287-313. 30. J. Thiem, S. Kopper, J. Schwentner, Liebigs Ann. Chem., 1985, 2135-2150. 31. J. Thiem, W. Klaffke, J. Org. Chem., 1989, 54, 2006-2009. 32. R.W. Friesen, S.J. Danishefsky, J. Am. Chem. Soc., 1989, 111, 6656-6660. 33. K. Suzuki, G.A. Sulikowski, R.W. Friesen, S.J. Danishefsky, J. Am. Chem. Soc., 1990, 112, 8895-8902. 34. S.J. Danishefsky, M.T. Bilodeau, Angew. Chem. Int. Ed. Engl., 1996,35, 1380-1419. 35. W.R. Roush, K. Briner, D.P. Sebesta, Synlett, 1993, 264-266. 36. W.R. Roush, R.A. Hartz, D.J. Gustin, J. Am. Chem. Soc., 1999, 121, 1990-1991. 31. W.R. Roush, C.E. Bennett, J. Am. Chem. Soc., 1999,121, 3541-3542. 38. D. Tailler, J.-C. Jacquinet, A.-M. Noirot, J.-M. Beau, J. Chem. Soc., Perkin Trans. I , 1992, 3 163-3 164. 39. R.L. Halcomb, S.J. Danishefsky, J. Am. Chem. Soc., 1989, 111, 6661-6666. 40. G. Bellucci, G. Catelani, C. Chiappe, F. D’Andrea, Tetrahedron Lett., 1994, 35, 8433-8436. 41. J. Gervay, S.J. Danishefsky, J. Org. Chem., 1991, 56, 5448-5451. 42. K.K.-C. Liu, S.J. Danishefsky, J. Org. Chem., 1994, 59, 1895-1897. 43. D.M. Gordon, S.J. Danishefsky, Carbohydr. Rex, 1990,206, 361-366. 44. R.G. Dushin, S.J. Danishefsky, J. Am. Chem. Soc., 1992, 114, 3472-3475.

404

15 Special Problems in Glycosylution Reactions: 2-Deoxy Sugars

C. Chiappe, G. Lo Moro, P. Munforte, Tetrahedron, 1997, 53, 10471-10478. R. Preuss, R.R. Schmidt, Synthesis, 1988, 694-697. W.R. Roush, D.P. Sebesta, C.E. Bennett, Tetrahedron, 1997,53, 882558836, G. Grewal, N. Kaila, R.W. Franck, J. Org. Chem., 1992, 57, 2084-2092. W.R. Roush, X.-F. Lin, J. Am. Chem. Soc., 1995, 117, 2236-2250. W.R. Roush, X.-F. Lin, J. Org. Chem., 1991, 56, 5740-5742. W.R. Roush, D.P. Sebesta, R.A. James, Tetrahedron, 1997, 53, 8837-8852. D.K. Jones, D.C. Liotta, Tetrahedron Lett., 1993, 34, 7209-7212. T.J. Dudley, I.P. Smoliakova, M.R. Hoffmann, J. Org. Chem., 1999, 64, 1247-1253. R.W. Franck, N. Kaila, Carbohydr. Rex, 1993,239, 71-83. Y. Ito, T. Ogawa, Tetrahedron Lett., 1987, 28, 2723-2726. G. Jaurand, J.-M. Beau, P. Sinay, J. Chem. SOC.,Chem. Commun., 1981, 572-573. M. Perez, J.-M. Beau, Tetrahedron Lett., 1989, 30, 75-78. A. Kaye, S. Neidle, C.B. Reese, Tetrahedron Lett., 1988,29, 2711-2714. A. El-Laghdach, M.I. Matheu, S. Castillon, Tetrahedron, 1994, 50, 12219-12234. Y. Diaz, A. El-Laghdach, S. Castillon, Tetrahedron, 1997,53, 10921-10938. G. Jaurand, J.-M. Beau, P. Sinay, J. Chem. Soc., Chem. Commun., 1982, 701-703. Y. Leblanc, B.J. Fitzsimmons, J.P. Springer, J. Rokach, J. Am. Chem. Soc., 1989, I l l , 29953000. 63. R. W. Franck, C.H. Marzabadi, J. Org. Chem., 1998, 63, 2197-2208 and references therein. 64. A. Dios, A. Geer, C.H. Marzabadi, R.W. Franck, J. Org. Chem., 1998,63, 6673-6679. 65. A. Dios, C. Nativi, G. Capozzi, R.W. Franck, Eur. J. Org. Chem., 1999, 1869-1874. 66. S.P. Vincent, M.D. Burkart, C.-Y. Tsai, Z. Zhang, C.-H. Wong, J. Org. Chem., 1999, 64: 5264-5279, and references therein. 67. K. Bock, C. Pedersen, J. Thiem, Carbohydr. Rex, 1979, 73, 85-91. 68. J. Thiem, M. Gerken, B. Schottmer, J. Weigand, Carbohydr. Res., 1987, 164, 327-341. 69. K.C. Nicolaou, T. Ladduwahetty, J.L. Randall, A. Chucholowski, J. Am. Chem. Soc., 1986, 108, 2466-2467. 70. K.C. Nicolaou, C.W. Hummel, N.J. Bockovich, C.-H. Wong, J. Chem. Soc., Chem. Commun., 1991, 870- 872. 71. K.C. Nicolaou, R.M. Rodriguez, H.J. Mitchell, F.L. van Delft, Angew. Chem. Znt. Ed., 1998, 37, 1874-1876. 72. H.M. Zuurmond, P.A.M. van der Klein, G.A. van der Marel, J.H. van Boom, Tetrahedron, 1993,49, 6501-6514. 73. K. Toshima, S. Mukaiyama, Y. Nozaki, H. Inokuchi, M. Nakata, K. Tatsuta, J. Am. Chem. SOC.,1994, 116, 9042-9051 and references therein. 74. J. Kocourek, Carbohydr. Rex, 1967, 3, 502-505. 75. A.B. Foster, J.M. Duxbury, T.D. Inch, J.M. Webber, Chem. Commun., 1967, 881-883. 76. K. Toshima, Y. Nozaki, S. Mukaiyama, T. Tamai, M. Nakata, K. Tatsuta, M. Kinoshita, J. Am. Chem. Soc., 1995,117, 3717-3727. 77. K. Toshima, T. Yoshida, S. Mukaiyama, K. Tatsuta, Carbohydr. Res., 1991, 222, 173188. 78. M. Trumtel, A. Veyrieres, P. Sinay, Tetrahedron Lett., 1989, 30, 2529-2532. 79. M. Trumtel, P. Tavecchia, A. Veyrieres, P. Sinay, Carbohydr. Rex, 1989, 191, 29-52. 80. C. Zagar, H.-D. Scharf, Carbohydr. Rex, 1993, 248, 107-118. 81. P. Tavecchia, M. Trumtel, A. Veyrieres, P. Sinay, Tetrahedron Lett., 1989, 30, 2533-2536. 82. M. Trumtel, P. Tavecchia, A. Veyrieres, P. Sinay, Carbohydr. Rex, 1990,202, 257-275. 83. T. Jyojima, N. Miyamoto, Y. Ogawa, S. Matsumura, K. Toshima, Tetrahedron Lett., 1999, 40, 5023-5026. 84. S.V. Ley, A. Armstrong, D. Diez-Martin, M. J. Ford, P. Grice, J.G. Knight, H.C. Kolb, A. Madin, C.A. Marby, S. Mukherjee, A.N. Shaw, A.M.Z. Slawin, S. Vile, A.D. White, D.J. Williams, M. Woods, J. Chem. Soc., Perkin Trans. I , 1991, 667-692. 85. C.A.A. van Boeckel, T. Beetz, S.F. van Aelst, Tetrahedron, 1984, 40, 4097-4107. 86. E. Petrakova, C.P.J. Glaudemans, Carbohydr. Res., 1995, 268, 35-46. 87a. P. Jutten, H.-D. Scharf, G. Raabe, J. Org. Chem., 1991, 56, 7144-7149. 87b. J. Szymoniak, P. Sinay, Tetrahedron Lett., 1979, 6, 545-548. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62.

References

405

88. a) H. Paulsen, W. Kutschker, 0. Lockhoff, Chem. Ber., 1981, 114, 3233-3241; b) H. Paulsen, 0. Lockhoff, Chem. Ber., 1981, 114, 3102-3114. 89. a) R.W. Binkley, D. J. Koholic, J. Org. Chem. 1989, 54, 3577-3581; b) R.W. Binkley, M.R. Sivik, J. Carbohydr. Chem., 1991, 10, 399 -416. 90. J. Jiinnemann, I. Lundt, J. Thiem, Liebiys Ann. Chem., 1991, 759-764. 91. H. Schene, H. Waldmann, Synthesis, 1999, 1411-1422. 92. K. Wiesner, T.Y.R. Tsai, H. Jin, Helu. Chzm. Acta, 1985, 68, 300-314. 93. R.W. Binkley, D.J. Koholic, J. Carhohydr. Chem., 1988, 7, 487-499. 94. J. Thiem, S. Kopper, Tetrahedron, 1990, 46, 113-138. 95. J. Borowiecka, P. Lipka, M. Michalska, Tetrahedron, 1988, 44, 2067-2076. 96. H. Bielawska, M. Michalska, Tetrahedron Lett., 1998, 39, 9761-9764. 97. M. Michalska, J. Borowiecka, J. Carhohydr. Chem., 1983, 2, 99-103. 98. L. Laupichler, H. Sajus, J. Thiem, Synthesis, 1992, 1133-1136. 99. O.J. Plante, P.H. Seeberger, J. Org. Chem., 1998, 63, 9150-9151. 100. A. Koch, C. Lamberth, F. Wetterich, B. Giese, J. Org. Chem. 1993, 58, 1083-1089. 101. H. Li, M. Chen, K. Zhao, Tetrahedron Lett., 1997,38, 6143-6144. 102. S. Hashimoto, Y. Yanagiya, T. Honda, S. Ikegami, Chem. Lett., 1992, 1511-1514. 103. S. Hashimoto, Y. Yanagiya, T. Honda, H. Harada, S. Ikegami, Tetrahedron Lett., 1992, 33, 3523-3526. 104. S. Hashimoto, A. Sano, H. Sakamoto, M. Nakdjima, Y. Yanagiya, S. Ikegami, Synlett, 1995, I27 1-1273. 105. T. MuIler, R. Schneider, R.R. Schmidt, Tetrahedron Lett., 1994, 35, 4763-4766. 106. D. Ravi, V.R. Kulkarni, H.B. Mereyala, Tetrahedron Lett., 1989, 30, 4287-4290. 107. K. Toshima, Y. Nozaki, K. Tatsuta, Tetruhedron Lett., 1991, 32, 6887-6890. 108. S. Raghavan, D. Kahne, J. Am. Chem. Soc., 1993, 115, 1580-1581. 109. J. Gildersleeve, A. Smith, K. Sakurai, S. Raghavan, D. Kahne, J. Am. Chem. Soc., 1999,121, 6176-61 82. 110. K. Suzuki, T. Mukaiyama, Chem. Lett., 1982,683-686. 111. K. Suzuki, T. Mukaiyama, Chem. Lett., 1982, 1525-1528. 112. A.G.M. Barrett, B.C.B. Bezuidenhoudt, A.R. Howell, A.C. Lee, M.A. Russell, J. Org. Chem., 1989,54, 2275- 2277. 113. A.G.M. Barrett, B.C.B. Bezuidenhoudt, L.M. Melcher, J. Org. Chem., 1990,55, 5196-5197. 114. D. Kahne, D. Yang, J.J. Lim, R. Miller, E. Paguaga, J. Am. Chem. Soc., 1988, 110, 87168717. 115. D. Crich, T.J. Ritchie, J. Chem. Soc., Chem. Commun., 1988, 1461-1463. 116. D. Crich, T.J. Ritchie, Carbohydr. Rex, 1989, 190, C3-C6. 117. D. Crich, T.J. Ritchie, J. Chem. Soc., Perkin Trans. I , 1990, 945-954.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

16 Orthogonal Strategy in Oligosaccharide Synthesis Osamu Kan ie

In the pursuit of efficient chemical synthesis of oligosaccharides, important progress has been made. Landmarks include chemoselective glycosylations and polymersupported oligosaccharide synthesis. Although polymer-supported synthesis is advantageous in respect of the isolation process, it needs to be tuned to be a more realistic prospect, especially in oligosaccharide synthesis. Also, because the final product is released along with deletion byproducts, delicate purification manipulations, e.g. by HPLC must usually be performed at the final stage. The advantage of the chemoselective activation strategy is recognized as its potential that no manipulations are required before the coupling reactions because the glycosyl acceptor has a pre-installed leaving group which also protects the anomeric center. The number of available protecting groups and chemoselective glycosylation conditions, however, limit the chemoselective sequential glycosylation strategy. This chapter addresses the problem of the chemoselective activation strategy and describes the successful introduction of a novel coupling strategy-an orthogonal glycosylation strategy in which two chemically distinct glycosylation reactions are utilized-and its application to polymer-supported oligosaccharide synthesis.

16.1 Introduction Much effort has been devoted to identification of the functions of the oligosaccharides found on the cell surface, where they covalently attached to a variety of organic molecules such as proteins and lipids. Because of the diversity of oligosaccharide structures, they are often described as the ‘finger print’ of a certain cell. The possibility that the complex structures are directly related to their functions has prompted many synthetic chemists to synthesize glycoconjugates. During the last two decades the technology for the synthesis of these structures has, therefore, improved dramatically. The key reaction in the construction of oligo-

408

16 Orthogonal Strategy in Oligosaccharide Synthesis

saccharides is, of course, the glycosylation reaction, in which stereo- and regiospecificities must be carefully controlled. There are several ways of controlling the stereochemistry of glycosidic bond formation: e.g.: 1) 1,Ztrans glycoside formation by means of a C-2 participating group [ l , 21; 2) axial glycoside formation resulting from the anomeric effect [3]; 3) equatorial glycoside formation using solvation of the carbonium cation by acetonitrile [4]; 4) axial glycoside formation by SN2 displacement of P-halides formed by halide equilibration in the presence of quaternary ammonium halide [ 51; 5) P-mannosylation by means of insoluble silver salts [6]; and 6) P-mannosylation by intramolecular aglycone delivery [71. The generalized scheme for the synthesis of oligosaccharides is shown in Figure 1 as an example of complex oligosaccharide synthesis. One must decide which method (1-6) should be applied for each coupling step before the synthesis. Also, a protecting group must be removed after each glycosylation step to release the hydroxyl function required for the next glycosylation reaction.

16.2 Analysis of the Strategic Aspects of Oligosaccharide Synthesis 16.2.1 General Aspects

To achieve efficient synthesis, both stepwise and convergent methods must be employed. A Stepwise method in which synthesis starts from the reducing end has classically been used for oligosaccharide synthesis because of the difficulty of transforming an anomeric protecting group into a leaving group. The recent development of anomeric protecting groups has enabled alternative synthesis which starts from the non-reducing end and a convergent synthesis (Figure 2). These anomeric protecting groups include the allyl, methoxyphenyl (MP), npentenyl, 2-trimethylsilylethyl (SE), t-butyldimethylsilyl, and t-butyldiphenylsilyl groups. Removal of this class of protecting groups releases the anomeric hydroxyl function to be converted into leaving groups [8].

16.2.2 The Pursuit of Efficiency in Oligosaccharide Synthesis

The other significant contribution in oligosaccharide synthesis (see Figure 3) was the introduction of ‘stable’ leaving groups which can be used as protecting groups until exposed to certain activation conditions. The thioglycosides are typical examples of this class of leaving group. Thioglycosides can be activated under alkylating

16.2 Analysis of the Strategic Aspects of Oligosaccharide Synthesis

409

Suitably protected sugar [A]

1

Anomeric deprotection

1

GI~COSYI

Activation

acceptor

Glvcosvl donor

p2d (NP) Sugar [A-B]

J Figure 1. Generalized oligosaccharide synthesis. Iterative reactions consisting of protection, deprotection, and glycosylation reaction are performed to obtain oligosaccharides. L = leaving group; P = protecting group.

or oxidative conditions but are stable under the traditional glycosylation conditions such as the Koenigs-Knorr method, so that halides can be chemoselectively activated without affecting them (Figure 3C). In addition, they can be transformed into halides or other leaving groups (Figure 3B). For this reason, thioglycosides are ideal intermediates in a flexible synthetic strategy [9]. The possibility of chemoselective activation of anomeric leaving groups has thus emerged [9-111. The armed and disarmed concept (Figure 3C), which employs a single potential leaving group, n-pentenyl glycosides, for both the donor and acceptor has been developed on the basis of the observation that the reactivity of glycosyl donors is

410

16 Orthogonal Strategy in Oligosaccharide Synthesis B

A

reducing end

m]

C

non-reducingend

product

: glyoosyl donor

Figure 2. Two stepwise syntheses (A and C ) and block condensation (B). One of the stepwise methods in which synthesis starts from the reducing end (A) has classically been used for oligosaccharide synthesis. The recent development of anomeric protecting groups has enabled alternative synthesis starting from the non-reducing end (C) and a convergent one (B).

affected by the protecting groups (i.e. ether or ester) especially at 0-2 [12a]. The utility of n-pentenyl glycosides is obvious, because small fragments of oligosaccharides can be synthesized in a very efficient manner and the n-pentenyl group can be temporarily protected as the dibromide to be used as the acceptor for block condensations. The armed and disarmed concept has proven to be applicable also to glycals [ 12b], thioglycosides [ 12~1,selenyl glycosides [ 12d], and glycosyl phosphoroamidates [ 11, 12el.

16.3 The Introduction of the Orthogonal Glycosylation Strategy 16.3.1 Limitation of Current Concepts

The advantage of the chemoselective activation strategy lies in its efficiency, because the glycosyl acceptor has a potential leaving group which also temporarily protects the anomeric center. Therefore, in principle, there is no need for deprotection or protecting group manipulations before the coupling reactions. Because the methodology relies on the relative reactivities of donor molecules, and these are controlled by the protecting groups and/or different leaving groups available, one must very carefully select the combination of leaving groups to be used, and thus this naturally imposes a limitation on this chemoselective sequential glycosylation strategy. Therefore, a different methodology which does not rely on the different reactivities of molecules is needed to reduce the complexity of the coupling process. An idea of such methodology can be derived by comparing the reactivities of glycosy1 halides and thioglycosides as an example (Figure 4).Although their reactivities are different in each group (A),there is a possibility of selective activation of halides and thioglycosides (B) in both directions, which might open a new coupling strategy.

16.3 The Introduction of the Orthogonal Glycosylution Strategy

A '$-L

41 1

B

, H O e P

O

Reactivity of leaving groups L' > L2> L3> L"

L

H O O P

Reactivity of leaving groups L1> L2(conditions A) L2> L1(conditions B)

P O G L H O G P

F

t P P O G L HO-P

O

*

G

P

t e

L

2T M S O G P

pod.,, Figure 3. Schematic representation of stepwise synthesis. (For A-D the direction of elongation is non-reducing end + reducing end and for E and F it is reducing end 3 non-reducing end. A: Most traditional method in this series (A-D) involving glycosylation reaction, deprotection of anomeric protecting group, and activation of the anomeric hydroxyl group into a donor. B: This method eliminates a deprotection step but involves direct transformation of the anomeric protecting group into a leaving group. C: This method utilizes a leaving group as the protecting group and so the number of reaction steps is minimized. This group includes the armed-disarmed and chemoselective glycosylations. D: This method also features the minimum number of steps in the synthetic scheme. It does not, however, rely on the hierarchy of the reactivities but on orthogonality and, therefore, hypothetically, eliminates the limitation of the number of coupling reactions. E: Most traditional method involving glycosylation reaction and deprotection to release the hydroxyl group for the next coupling reaction. F: This method utilizes a carefully selected combination of leaving groups and hydroxyl protecting groups, which enables the shortest, one-pot, synthetic scheme.

412

16 Orthogonal Strategy in Oligosaccharide Synthesis

labile

4

stable

silver salts

Br

l

CI

x n

5 oddative m conditions

'-Me

'\Ph

Figure 4. Comparison of the reactivities of glycosyl halides and thioglycosides as examples. Although their reactivities are different in each group (A) there is a possibility of selective activation of halides and thioglycosides (B) in both directions, which might lead to a new coupling strategy.

16.3.2 The Orthogonal Coupling Concept

The minimum requirement for this scheme would be to combine two chemically distinct glycosylation reactions in which one of the leaving groups is activated while the other behaves as a protecting group, and vice uersa (Figures 3D, 4 and 5). To fulfil the requirement for this orthogonal system, each selected leaving group should be unaffected under the conditions used to activate the other. Also, both leaving groups should tolerate routine manipulations of temporary protecting groups. For this orthogonal strategy, a set of leaving groups and activation conditions for each group-L', phenylthio group and L2, fluoride, and (a), NIS-TfOH (or AgOTf) [ 131 and (b),Cp2HfClz-AgC104 [ 141-are selected as candidates.

Figure 5. A scheme for orthogonal glycosylation. The combined use of two chemically distinct glycosylation reactions, where one of the leaving groups is activated while the other behaves as a protecting group and vice versa, enables continuous coupling without any transformations. L = leaving group; P = protecting group; a = activation condition for L1; b = activation condition for L2. N.B. L2 is stable under condition a and L' is stable under condition b.

16.3 The Introduction o j the Orthogonal Glycosylation Strategy

41 3

Figure 6. Manipulations of orthogonal protecting groups. Orthogonal protecting groups are “a set of completely independent classes of protection groups, such that each class can be removed in any order and in the presence of all other classes” as defined by Baranay and Merrifield. X, Y, Z = functional groups; A, B, C = orthogonal protecting groups; a, b, c = orthogonal deprotection conditions.

16.3.3 What is Orthogonality Anyway? The term ‘orthogonality’ is frequently used in a generalization of its mathematical meaning to describe sets of primitives or capabilities that span the entire ‘capability space’ of the system and are non-overlapping or mutually independent. The term ‘orthogonal system’ in chemistry was first defined by Baranay and Merrifield as “a set of completely independent classes of protection groups, such that each class can be removed in any order and in the presence of all other classes” (Figure 6) [ 151. Manipulations of orthogonal protecting groups are widely accepted in peptide chemistry and can also be seen in other fields [16]. Despite the importance of the idea in the synthesis of polymeric materials such as peptides, oligonucleotides, and oligosaccharides, the term is used to describe the independence of a set of protecting groups during deprotection, and is not used to describe condensation reactions. The synthesis does, therefore, require multiple steps before the coupling reactions. To reduce the number of synthetic steps, the challenge is to apply the idea of orthogonality to the reactive site of coupling. In this system, each synthon which is involved in the coupling reactions must have a functional group with orthogonal behavior in the coupling reactions, e.g. inertness and reactiveness under the given conditions. The basis of the orthogonal condensation system is the combined use of individual reactions which enable the synthesis of polymeric compounds without deprotection or activation steps, as explained above. It should, furthermore, be

414

16 Orthogonal Strategy in Oligosaccharide Synthesis

noted that the number of conditions to be employed is not limited in a system as long as the conditions are orthogonal to each other. 16.3.4 Orthogonal Glycosylation and Solid-Phase Oligosaccharide Synthesis Such systems are effectively suited to solid-phase oligosaccharide synthesis [ 171. Although the main advantage of solid-phase synthesis is the elimination of tedious purification processes, one drawback of this advantage is that each reaction conducted on the support must give nearly quantitative yield or the unreacted reaction center must be capped (again, the capping must be quantitative). These are required to ensure elimination of the possibility that there are deletion compounds in the mixture of compounds obtained. When a mixture that consists only of product and precursors for each step is given after several reaction cycles on a solid support, they can be separated by conventional purification protocols such as HPLC. If, on the other hand, the final mixture contains compounds in which some of the synthons are randomly deleted, isolation will not be a practical proposition. To avoid this problem, we usually give the above two answers; when, however, the orthogonal glycosylation strategy is applied to the solid-phase synthesis, these problems can be solved simultaneously (Figure 7), because (i) the hydrolyzed byproduct cannot be a candidate for the glycosyl donor in the next coupling cycle [17d], and (ii) the activation step is no longer necessary because the leaving group is incorporated as the protecting group for the previous reaction. Thus, in this system, one of the leaving groups is activated in the presence of hemiacetals and the other leaving group. There is, therefore, no need for capping. High yielding coupling reactions are, of course, needed but it is not necessary if one can efficiently extract the desired product from a product mixture. A tag might be incorporated at the end of the coupling reactions, because the only compound that has leaving group is the desired compound and others are presumably hydrolysis or elimination by-products formed at some stage in the orthogonal glycosylation strategy. Such tags include hydrophobic aglycones; these have proved to be extremely useful for the rapid solid-phase extraction of oligosaccharides in the assay system known as the SepPak assay [18].

16.4 The Orthogonal Glycosylation Strategy 16.4.1 Orthogonal Chain Elongation of Homo-Oligosaccharides: Synthesis of Chito-Oligosaccharides 1191 To assess the orthogonality of the above-mentioned combination of reactions, the interconversion of the two anomeric potential leaving groups phenylthio and fluoro group was performed. Thiophenyl glycoside 1 was treated under Nicolaou’s conditions [20] (u) to yield the glycosyl fluoride 2 quantitatively. The fluoride 2 could,

16.4 The Orthogonal Glycosylution Strutegy

415

Orthogonal condensation

PO Sugar [2] Polymer

-

PO Sugar [ l ] PO Sugar [3]

&O

Polymer-

Activate Y

PO

Sugar [l+]-Tag deprotection

L L O HOW*

Ho

H0-1-

L o

Ho

OX/* HO

HO

O

L o O HO w* HO

Sugar [l-n]-Tag HO etc

Figure 7. The overall idea of solid-phase oligosaccharide synthesis on the basis of the orthogonal glycosylation strategy. The orthogonal glycosylation strategy can be used in solid-phase oligosaccharide synthesis and can accelerate the process by dramatically reducing the number of reaction steps. A hydrophobic tag can be incorporated at the end of the synthesis and this is used to extract the desired compound from the accumulated by-products.

16 Orthogonal Strategy in Oligosaccharide Synthesis

416

-

OAc

.OAc

a quant

AAGO c O s S pSPh h NPhth 1

F NPhth

A&& AcO* AcO

b

2

93%

Scheme 1. An interconversion of potential donors. Reagents and conditions: (a) SnC12-AgC104PhSH in CH2C12, room temp.; (b) NBS-DAST in CHZC12, room temp.

on the other hand, be activated under Mukaiyama's conditions [21] (b)in the presence of thiophenol to give the thioglycoside 1 in 93% yield (Scheme 1). This successful interconversion reaction prompted us to investigate further. Using this orthogonal set of reactions, a set of glycosylation reactions was investigated using 1 and 2 as donors and 3 and 4 as acceptors. It should be noted here that because the selected anomeric functional groups are orthogonal, all the compounds used were synthesized from a single compound. The thioglycoside 1 was reacted with the acceptor 3 under conditions (a)to afford disaccharide 5 (90%). The fluoride 2 was also successfully activated under conditions (b) and reacted with 4, without affecting the thioglycosidic linkage, to give disaccharide 6 (78%) (Scheme 2). Self-condensed product was not detected in either sequence. These results indicate that less reactive acyl protected donors were activated in preference to the potentially more reactive ether-protected acceptors; the reactions are therefore clearly distinguishable from armed-disarmed reactions. The chosen set of reactions is therefore shown to be orthogonal; chemoselective in both directions. To show the applicability of this strategy to the synthesis of longer-chain oligosaccharides, iterative couplings were further examined by constructing the heptasaccharide 12 in which ail synthons were derived from 1. Firstly, thioglycoside donor 7 was coupled with acceptor fluoride 3 under conditions (a) to give the disaccharide fluoride 8 (85%) which was then reacted with the acceptor 4 to produce 9

Ester protected donors Ether protected acceptors

AcO AcO&Sph

NPhth

+

1

i$&&F

NPhth

8_ 90%

Ap*CO-og BnO NPhth

3

NPhth F

5

NPhth

AcO 2

+

! f o a SNPhth p h 4

&

A

NPhth BnO

NPhth ~ SPh

6

Scheme 2. A set of orthogonal reactions showing complete distinctness of the reactions used. Reagents and conditions: (a) NIS (1.3 equiv.)-AgOTf (0.1 equiv.) in CH2C12, -50 "C + room temp.; (b) Cp2HfClz (1.3 equiv.)-AgC104 (2.6 equiv.) in CH2C12, -78 "C .+ room temp.; (c) NaOMe.

16.4 The Orthogonal Glycosylution Strategy

417

[conditions (b), 72%]. Subsequent reaction of 9 with 3 [conditions (a), 65%] gave tetrasaccharide 10. Having accomplished the synthesis of a tetrasaccharide by the orthogonal strategy, we examined a block condensation approach. Tetrasaccharide acceptor 11, prepared by Zemplen deacetylation of 10, was coupled with its precursor 9 to give compound 12 [conditions (a), 670/0],which is again ready for further use as a oligosaccharide donor. The orthogonal glycosylation reactions were put together with the block condensation to show its efficiency and flexibility (Scheme 3 ) .

BnO

-+&+ -

4%

“

4

w

SNPhtP

h

-

a

65%

9

10 R = A c ‘(11 R = H

NPht

-$&++-F

12

Scheme 3. An example of the orthogonal glycosylation strategy showing only the continuous coupling reactions required to obtain the oligomer. Reagents and conditions: (a) NIS (1.3 equiv.)AgOTf (0.1 equiv.) in CH2C12, -50°C 4room temp.; (b) CpzHfClz (1.3 equiv.)-AgClOa (2.6 equiv.) in CHIC12, -78 “C + room temp.; (c) NaOMe, 91%.

4 18

16 Orthogonal Strategy in Oligosaccharide Synthesis

fj-D-GalNAc-(1 -3)-a-D-Gal-( 1-3)-p-D-Gal-( 1-3)-p-D-GalNAc-( 1-4)-fj-D-Gal-( 1-4)-fj-D-Gle( 1-)-Cer a-L-FuC(1-2)

J

13

14

Figure 8. The structure of a novel extended blood-type B glycosphingolipid (13)and the target structure (14).

16.4.2 Orthogonal Coupling for Hetero-Oligomer Synthesis [22] To verify generality of the aforementioned orthogonal system, a branched tetrasaccharide 14, a partial structure of a novel blood-type related glycosphingolipid 13 (Figure 8), was chosen as a target molecule. Four monosaccharide units 15, 16, 18, and 22 were synthesized as either glycosyl fluorides or thioglycosides to fulfil the criteria of the orthogonal coupling concept. All necessary selective protections and anomeric transformations were completed at the monosaccharide stage, thereby minimizing manipulations during elongation of the sugar chain. The protecting groups especially at C-2 position of each synthon were carefully chosen for the stereoselective glycosylation reactions (Scheme 4). The synthesis was started from the suitably protected galactosyl fluoride 15 (a : P = 1.7 : 1) which was glycosylated with GalNAc precursor 16 under conditions (a) to afford 17. The stereochemical assignment was confirmed by 'H NMR; J1,,2, % 8.5 Hz for H-1' indicated the P-D-configuration for the GalNAc residue. No anomerization of the reducing-end fluoride was observed during the glycosylation reaction, which eliminated the potential ambiguity of this strategy. The disaccharide 17 was used directly as the donor for the next glycosylation reaction with the phenylthioglycoside 18. Glycosylation under conditions (b) afforded the desired trisaccharide 19. The a configuration of the newly formed glycosidic linkage was evident from the coupling constant of the anomeric proton at 64.77 ( J = 3.6 Hz). Glycosylation of 19 with octanol was then conducted under conditions (u) to yield the octyl glycoside 20. The applicability of the current strategy was proven by showing direct orthogonal coupling reactions three times for introduction of various forms of interglycosidic linkage. Deprotection of the levulinoyl group on trisaccharide 20 by use of hydrazine acetate [23] afforded 21 (85.4%) which was then used as an acceptor for the a-

16.4 The Orthogonal Glycosylution Strutegy

419

OBn I .-Oh

AcO

SPh 18

AcO

73%

OBn 17

Octyl alcohol

19

OLev

SPh % +

Scheme 4. Synthesis of blood-type-related oligosaccharides by the orthogonal strategy. Reagents and conditions: (a) *NIS-AgOTf in CH2C12, -50 + -20°C; **NIS-AgOTf in CH2C12, 10°C; (b) Cp2HfC12-AgC104 in -20°C + room temp.; *** DMTST in benzene-CH2Cl2, 0 CH2C12-Et20, -78 "C ---t room temp.; (c) NHzNH2.AcOH; (d) (i) H2NCH2CH2NH2-EtOH, 90 "C; (ii) Ac20-MeOH-Et3N (cat.), room temp.; (e) Hz-Pd/C-MeOH, H+. ---f

420

16 Orthogonal Strategy in Oligosaccharide Synthesis

fucosylation reaction. Thus the levulinoyl group was used not only as a participating group to yield the P-glycoside but also as a selectively removable protecting group. The coupling reaction of 21 using methylthio glycoside 22 [24] gave the desired tetrasaccharide (88%). Ethylenediamine [25]treatment of the tetrasaccharide then selective acetylation of the resulting amine yielded 14 (85%) after hydrogenolysis under acidic conditions. In the same manner, compound 21 was deprotected to afford 25. We have demonstrated the applicability of the orthogonal glycosylation strategy to the synthesis of a branched oligosaccharide with four different types of glycosidic linkage. The synthesis of the target tetrasaccharide 14 was thus achieved in seven steps (38%) overall yield) from protected monosaccharide units. With the trisaccharide 25 and the known blood-type B trisaccharide available it might be possible to utilize these oligosaccharides as probes to investigate an unknown biosynthetic pathway. 16.4.3 Application to Polymer-Supported Synthesis 1261 The elongation of the sugar chain from the non-reducing end in polymer-supported oligosaccharide synthesis is less straightforward, mainly for two reasons: 1) accumulation of by-products on the support together with the desired compound; and 2) a transformation reaction must be performed to obtain the glycosyl donor. When, however, the orthogonal glycosylation strategy is applied to polymersupported synthesis, an important improvement, namely pre-installation of a potential leaving group, is achieved. This has potential to make the methodology more advantageous-because the leaving group is attached, there will be no need for the activation reaction and no need to protect the anomeric hydroxyl group produced by hydrolysis. As a result, again only one reaction step is required per cycle; this is the same as was shown for solution synthesis. In addition, because only the last introduced sugar unit has a leaving group, one can introduce a tag to ease isolation of the final product after deprotection. The recoverability of excess glycosyl acceptor used in the coupling reaction should be noted in addition to the above advantages as a common feature of polymer-supported synthesis starting from non-reducing end terminus. To test this hypothesis, the first attempt was made using ‘soluble’ poly(ethy1ene glycol) (PEG) as the polymer [17b]; the trimannoside portion (27) of the highmannose type glycoprotein (26) was chosen as the target. In addition, the 2-(trimethylsily1)ethyl (SE) group was used as a hydrophobic tag [27] in this case in addition to the temporary protecting group [8d,e] (Figure 9). Initial coupling of the polymer, PEG monomethyl ether (average MW = 5000)supported methylthiomannoside 28 with the fluoride 29 was performed in the presence of DMTST, prepared in situ from MeOSOzCF3 and MeSSMe [conditions (a)], to afford the disaccharide-PEG conjugate 30 in 89% yield. Subsequent coupling with the reducing-end mannoside was performed by using SE-glycoside 31 as aglycone under conditions (b)to afford the trisaccharide-PEG conjugates 32.

16.5 Conclusions and Prospects a-0-Man-(1-2))-a-D-Man-(l-6)

42 1

1

a-D-Man-(1-6) a-D-Man-(l-2)-a-D-Man-(l-6) p-D-Man-(l-.+p-D-GicNAc-(

1

1-.4)-B-D-GlcNAc-Asn-Protein

a-D-Man-(l-2)-a-D-Man-(l-+2)-a-D-Man-( 1 ’ 3 ) -1 26

I

27

OSE

Figure 9. The structures of the high mannose-type N-linked oligosaccharide 26 and the target tri. mannoside 27.

Having shown the construction of a trisaccharide sequence, PEG was cleaved off under basic conditions and the crude trisaccharide was hydrogenated over Pd(OH)*. The resulting mixture was then applied to a short column of CISreversed-phase silica gel, which was first eluted with water to elute all by-products. Subsequent elution with 50% MeOH gave a mixture of trisaccharides from which 27 was isolated by eluting with a linear gradient from water to 50% MeOH (Scheme 5). This strategy, featuring a novel combination of polymer-support chemistry, the concept of orthogonal glycosylation, and simple isolation by incorporating a hydrophobic tag has proven successful. It should also be noted that the aglycone in 27, which is used as a hydrophobic tag, can be selectively cleaved if one wants to conjugate the formed oligosaccharide with other structures. One expansion of this strategy can be seen in connection with the intramolecular aglycone delivery approach, where a polymer-substituted benzyl group was used as both the linker and the tether [28] (Scheme 6 ) . Thus, methylthiomannoside 33 was coupled with a fluorinated GlcNAc derivative 3 to yield only P-linked mannoside which can be used as an acceptor itself or as a donor after protection of released 2OH. The advantage of this particular use is that the only compound to be released from the polymer is the desired product.

16.5 Conclusions and Prospects An orthogonal glycosylation strategy has been developed by combined use of phenylthioglycosides and glycosyl fluorides both as donors and acceptors. By use of this

422

16 Orthogonal Strategy in Oligosaccharide Synthesis

BnO BnO BnO BnO

30

SEO :

BnO BnO

'siWo 'I

32

OSE

H O T OH

HO HO

HO

OH

HO

HO *H r HO

Wafer eluent

27

OSE

Organic (alcohollc) eluent Scheme 5. Polymer-supported synthesis of trimannoside on the basis of the orthogonal strategy. Reagents and conditions: (a) DMTST in CHzC12, room temp., (b) CpzHfClz-AgOTf in CH2CI2, 0 "C + room temp., (c) (i) NaOMe; (ii) Hz-Pd(OH)z in MeOH-EtOAc. *Yields were calculated on a weight basis.

16.5 Conclusions and Prospects

423

SMle

34

,OBn

lb TBDPSO

BnO

NPhth

36

NPhth

d 85%

35 pTBDPSO hTo0?~*+;& NPhth

N3 NPMh

37

Scheme 6. Combination of polymer-supported chemistry, P-mannosylation, and orthogonal strategy. Reagents and conditions: (a) DDQ, 4 A mol. sieves in CH2C12, room temp., (b) AczO-Pyr., 48% from 3, (c) Cp*HfClz-AgOTf in ClCH2CH2C1, -15°C + room temp.

strategy, syntheses of several oligosaccharides have been achieved. As was clearly demonstrated in the synthesis of the chito-oligosaccharide, the novel strategy is free from the effects of the protecting groups, unlike the armed-disarmed concept. The usefulness of the strategy is obvious, because by use of this method an oligosaccharide can be synthesized highly efficiently with the lowest number of reactions. The syntheses of homo- and hetero-oligosaccharides has been demonstrated successfully. Further, the strategy was also examined for polymer-supported synthesis; this showed the potential of the methodology in solid-phase oligosaccharide synthesis. A hydrophobic tag was incorporated at the end of the synthesis to demonstrate the idea of simple isolation of the final product and/or the possibility of direct use of the isolated material for biological evaluations. The orthogonal coupling strategy can be applied to the synthesis of other polymeric compounds. Several approaches have been reported for construction of polymeric and dendritic structures on the basis of the orthogonal coupling strategy [29] (Figure 10).

424

16 Orthogonal Strategy in Oligosaccharide Synthesis

conditions a

PPhB - DEAD

b

Pd(PPh,)$l2

-

- CUI

Pd(0Ac)z - P(o-tOlyl), NBu3

Pd2(dba)3- DPPF - NaOtBu

NaH

Et3N DMAP

-

Figure 10. Some examples of the orthogonal coupling strategy in the synthesis of polymers and dendnmers.

Acknowledgments

The author thanks Professor Tomoya Ogawa and Dr. Yukishige Ito for their continuous encouragement and suggestions. Professors Hans Paulsen, Ole Hindsgaul, Robert Field, Chi-Huey Wong, Makoto Kiso, Tadahiro Takeda, and Dr. Frank Barresi are also acknowledged for their encouragement. Most of all, the author would like to express his sincere respect and gratitude to the late Professor Akira Hasegawa. References 1. (a) Igarashi, K. The Koenigs-Knorr reactions. In: Adu. Curbohydr. Chem. Biochem., 1977, 34, pp. 243-283. (b) Paulsen, H. Angew. Chem. Znt. Ed. Enyl., 1982, 21, 155-175. 2. (a) Lemieux, R. U., Takeda, T., Chung, B. Y. A. C.S. Symp. Ser., 1976, 4, 90-1 15. (b) Ito, Y., Ogawa, T. Tetrahedron. 1990, 46, 89-102. (c) Okamoto, K., Goto, T. Tetrahedron. 1990, 46,

5835-5857. 3. Lemieux, R. U. In: Molecular Rearrangements, de Mayo, P. Ed., Interscience, New York, 1964, p. 709. 4. Brccini, I., Derouet, C., Esnault, J., Heme du Penhoat, C., Mallet, J.-M., Michon, V., Sinay, P. Carbohydr. Res., 1993, 246, 23-41. and references cited therein. 5. Lemieux, R. U., Hendriks, K. B., Stick, R. V., James, K. J. Am. Chem. Soc., 1975, 97,40564062. 6. (a) Paulsen, H., Lockhoff, 0. Chem. Ber., 1981,114, 3102-3114. (b) Garegg, P. J., Ossowski, P. Acta Chem. Scand., 1983, B37, 249-250. (c) Van Boeckel, C. A. A., Beetz, T., van Aelst, S. F. Tetrahedron, 1984, 40, 4097-4107. 7. (a) Barresi, F., Hindsgaul, 0. J. Am. Chem. Soc., 1991, 113, 9376-9377. (b) Stork, G., Kim, G. J. Am. Chem. Soc., 1992, 114, 1087-1088. (c) Barresi, F., Hindsgaul, 0. Synlett, 1992, 759761. (d) Bols, M. Tetrahedron, 1993, 49, 10049-10060. (e) Barresi, F., Hindsgaul, 0. Can. J. Chem., 1994, 72, 1447-1465. (f) Ito, Y., Ogawa, T. Angew. Chem. Int. Ed. Engl., 1994, 33, 116551767.

References

425

8. (a) Sugimoto, M., Ogawa, T. Glycoconjugate J., 1985, 2, 5-9. (b) Murakata, C., Ogawa, T. Carbohydr. Rex; 1992, 235, 95-1 14. (c) Fraser-Reid, B., Udodong, U. E., Wu, Z.: Ottosson, H., Merritt, J. R., Rao, C. S., Roberts, C., Madsen, R. SYNLETT., 1992, 927-942. (d) Jansson, K., Frejd, T., Kihlberg, J., Magnusson, G. Tetrahedron Lett., 1988, 29, 361-362. (e) Jansson, K., Ahlfors, S., Frejd, T., Kihlberg, J., Magnusson, G. J. Org. Chem., 1988, 53, 56295647. (f) Murase, T., Kameyama, A,, Kartha, K. P. R., Ishida, H., Kiso, M., Hasegawa, A. J. Carbohydr. Chem., 1989, 8, 265-283. (g) Kanie, O., Takeda, T., Ogihara, Y. Carbohydr. Res., 1989,190, 53-64. (h) Kinzy, W., Schmidt, R. R. Liebigs Ann. Chem., 1985, 1537-1545. (i) Nakahara, Y., Ogawa, T. Tetrahedron Lett., 1987,28, 2731-2734. 9. (a) Koto, S., Uchida, T., Zen, S. Bull. Chem. Soc. Jpn., 1973, 46, 2520-2523. (b) Nicolaou, K. C., Dolle, R. E., Papahatjis, D. P., Randall, J. L. J. Am. Chem. Soc., 1984, 106, 4189-4192. (c) Lonn, H. Carbohydr. Rex, 1985, 139, 105-113. (d) Fiigedi, P., Garegg, P. J., Lonn, H., Norberg, T. GlycoconjugateJ., 1987, 4, 97-108. ( e ) Nilsson, M., Norberg, E. Carhohydr. Rex, 1988, 183, 71-82. 10. Chemoselective glycosylation reaction in general: (a) Paulsen, H.; Tietz, H. Carbohydr. Rex, 1985,144, 205-229. (b) Friesen, R. W., Danishefsky, S. J. J. Am. Chem. Soc., 1989,111, 66566660. (c) Trumtel, M., Veyrieres, A., Sinay, P. Tetrahedron Lett., 1989, 30, 2529-2532. (d) Veeneman, G. H., van Boom, J. H. Tetrahedron Lett., 1990,31,275-278. (e) Veeneman, G. H., van Leeuwen, S. H., Zuurmond, H., van Boom, J. H. J. Carbohydr. Chem., 1990, 9, 783-796. (f) Mori, M., Ito, Y., Uzawa, J., Ogawa, T. Tetrahedron Lett., 1990, 31, 3191-3194. (g) Mehta, S., Pinto, B. M. Tetrahedron Lett., 1991, 32, 4435 4438. (h) Marra, A,, Gauffeny, F., Sinay, P. Tetrahedron, 1991,47, 5149-5160. (i) Zegelaar-Jaarsveld, K., van der Marel, G. A,, van Boom, J. H. Tetrahedron, 1992, 48, 10133-10148. (j) Jain, R. K., Matta, K. L. Carbohydr. Rex, 1992, 226, 91-100. (k) Marra, A., Esnault, J., Veyrieres, A., Sinay, P. J. Am. Chem. Soc., 1992, 114, 6354-6360. (1) Hashimoto, S., Yanagiya, Y., Honda, T., Kobayashi, H., Ikegami, S. Presented at 18th Symposium on Progress in Organic Reactions and Syntheses-Applications in the Life Sciences, Sapporo, Japan, October 1992, Abstr. pp 276-280. (m) Raghavan, S., Kahne, D. J. Am. Chem. Soc., 1993, 115, 1580-1581. (n) Mehta, S., Pinto, B. M. J. Org. Chem., 1993, 58, 3269-3276. ( 0 ) Sliedregt, L. A. J. M., Zegelaar-Jaarsveld, K., van der Marel, G. A., van Boom, J. H. SYNLETT, 1993, 335-337. (p) Yamada, H., Harada, T., Miyazaki, H., Takahashi, T. Tetrahedron Lett., 1994, 35, 3979-3982. (4) Yamada, H., Harada, T., Takahashi, T. J. Am. Chem. Soc., 1994, 116, 7919-7920. (r) Chenault, H. K., Castro, A. Tetrahedron Lett., 1994, 35, 9 145-9148. 11. Latent-active strategy: (a) Roy, R., Andersson, F. O., Letellier, M. Tetrahedron Lett., 1992, 33, 6053-6056. (b) Boons, G.-J., Isles, S. Tetrahedron Lett., 1994, 35, 3593-3596. 12. Armed-disarmed strategy: (a) Mootoo, D. R., Konradsson, P., Udodong, U., Fraser-Reid, B. J. Am. Chem. Soc., 1988, 110, 5583-5584. (b) Friesen, R. W., Danishefsky, S. J. J. Am. Chem. Soc., 1989: I l l , 6656-6666. (c) Veeneman, G. H., van Leeuwen, S. H., van Boom, J. H. Tetrahedron Lett., 1990, 31, 1331-1334. (d) Grice, P., Ley, S. V., Pietruszka, J., Priepke, H. W. M., Walther, E. P. E. Synlett, 1995, 781-784. (e) Hashimoto, S., Sakamoto, H., Honda, T., Abe, H., Nakamura, S., Ikegami, S. Tetrahedron Lett., 1997, 38, 8969-8972. 13. (a) Konradsson, P., Mootoo, D. R., McDevitt, R. E., Fraser-Reid, B. J. Chem. Soc., Chem. Commun., 1990, 270-272. (b) Veeneman, G. H., van Leeuwen, S. H., van Boom, J. H. Tetrahedron Lett., 1990,31, 1331-1334. (c) Konradsson, P., Udodong, U. E., Fraser-Reid, B. Tetrahedron Lett., 1990,31,4313-43 16. 14. (a) Suzuki, K., Maeta,; H. Matsumoto, T., late Tsuchihashi, G. Tetrahedron Lett., 1988, 29, 3571-3574. (b) Suzuki, K., Maeta, H., Matsumoto, T. Tetrahedron Lett., 1989,30, 4853-4856. 15. (a) Barany, G., Merrifield, R. B. J. Am. Chern. Soc., 1977, 99, 7363-7365. (b) Barany, G., Merrifield, R. B. J. Am. Chem. Soc., 1980, 102, 3084-3095. 16. Schelhaas, M.: Waldmann, H. Angeiv. Chem. Znt. Ed. Engl., 1996, 35, 2056-2083. 17. For the recent advances of the solid-phase ohgosaccharide synthesis, see: (a) Yan, L., Taylor, C. M., Goodnow Jr., R., Kahne, D. J. Am. Chem. Soc. 1994,116, 6953-6954. (b) Douglas, S. P., Whitfield, D., Krepinsky, J. J. J. Am. Cliem. Soc. 1994, 117, 2116-2117. (c) Schuster, M., Wang, P., Paulson, J. C., Wong, C.-H. J. Am. Chem. Soc. 1994, 116, 1135-1136. (d) Randolph, J. T., McClure, K. F., Danishefsky, S. J. J. Am. Chem. Soc. 1995, 117, 5712-5719. (e) Rademann, J., Schumidt, R. R. Tetrahedron Lett. 1996, 37, 3989-3990. (f ) Shimizu, H., Ito, Y . ,

426

18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29.

16 Orthogonal Strategy in Oligosaccharide Synthesis

Kanie, O., Ogawa, T. Bioorg. Med. Chem. Lett. 1996, 6, 2841-2846. (8) Nicolaou, K. C., Winssinger, N., Pastor, J., DeRoose, F. J. Am. Chem. Soc. 1997, 119, 449-450. Palcic, M. M., Heerze, L. D., Pierce, M., Hindsgaul 0. GlycoconjugateJ. 1988, 5, 49-63. Kanie, O., Ito, Y., Ogawa, T. J. Am. Chem. Soc., 1994, 116, 12073-12074. Nicolaou, K. C., Caulfield, T. J., Kataoka, H., Stylianides, N. A. J. Am. Chem. Soc., 1990, 112, 3693-3695. Mukaiyama, T., Murai, Y., Shoda, S. Chem. Lett., 1981, 431-432. Kanie, O., Ito, Y., Ogawa, T. Tetrahedron Letters, 1996, 37, 4551-4554. Koeners, H. J., Verhoeven, J., van Boom, J. H. Rec. Trav. Chim. Pays-Bas., 1981, 100, 65-72. Sato, S., Ito, Y., Nukada, T., Nakahara, Y., Ogawa, T. Carbohydr. Rex, 1987, 167, 197-210. Kanie, O., Crawley, S. C., Palcic, M. M., Hindsgaul, 0. Carbohydr. Res., 1993, 243, 139-164. Ito, Y., Kanie, O., Ogawa, T. Angew. Chem. Int. Ed. Engl., 1996, 35, 2510-2512. Ito, Y., Paulson, J. C. J. Am. Chem. Soc. 1993, 115, 1603-1604. Ito, Y., Ogawa, T. J. Am. Chem. Soc. 1997, 119, 5562-5566. (a) Zeng, F., Zimmermann, S. C. J. Am. Chem. Soc. 1996,118,5326-5327. (b) Maddux, T., Li, W., Yu, L. J. Am. Chem. Soc. 1997, 119, 844-845. (c) Deb, S. K., Maddux, T. M., Yu, L. J. Am. Chem. Soc. 1997, 119, 9079-9080. (d) Frost, C. G., Mendonp, P. Chem. Lett. 1997, 1 159-1 160.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

17 Protecting Groups: Effects on Reactivity, Glycosylation Stereoselectivity, and Coupling Efficiency Luke G. Green and Steven V. Ley

17.1 Introduction In general, protecting groups are seen as an expedient, performing the uninteresting task of masking functionality that would otherwise interfere with a carefully planned strategy. A protecting group or protecting group strategy is usually chosen on the basis of its compatibility (in protection, deprotection and lability to other transformations), selectivity (in protection) and sequence (in deprotection when multiple protecting groups are used) [I]. Ultimately the same criteria apply in oligosaccharide assembly where the protecting group pattern is principally dictated by the target structure. Even within these constraints there are, however, options which can mean the difference between success and failure, efficiency and tedium. Protecting groups are not innocent bystanders in carbohydrate chemistry and to relegate them in the synthesis design is foolhardy. It is not the purpose of this chapter to review methods for installing and removing carbohydrate protecting groups, because this has already been covered by other texts [2]. Rather, it is intended to outline the mechanisms by which they effect glycosylations thereby enabling readers to decide which protecting group strategy best serves their purpose. Before investigating the roles various protecting groups can play it is pertinent to analyze briefly the glycosidic mechanism, because this is central to oligosaccharide assembly processes. There have been many detailed studies of the processes that occur during a glycosylation [ 3 ] but much of the evidence used to substantiate proposed inter-glycosyl coupling mechanisms is anecdotal or circumstantial. Monitoring actual glycosylation reactions is complex and often not very illuminating, given the number of factors involved and the instability of the intermediates. The mechanism presented here is not intended to be definitive but serves well as a working model.

428

I7 Protecting Groups

17.2 Glycosidic Mechanism The first rule is that glycosylations do not, under most reaction conditions, proceed through a free oxonium ion in an s N 1 manner [4]. Jencks has estimated the lifetime s, a time too short to of a glycosyl oxonium ion in water to be approximately enable solvent equilibration. It therefore seems unlikely that in typical, low-polarity glycosylation solvents the ion-pairs can be separated sufficiently to generate the free glycosyl oxonium ion as an intermediate. Hence, glycosylations can be considered as proceeding through an 'exploded' transition state in which the incipient oxacarbenium moiety is stabilized electrostatically by both entering and leaving groups [ 5 ] , as in Capand CBataken from Lemieux's mechanism (Figure 1) [6].

R.

*.

!?

X

Aa

Ba

R.

Figure 1. Lemieux's glycosidic mechanism.

17.2 Glycosidic Mechanism

429

Lemieux's mechanism (based on earlier work by Rhind-Tutt and Vernon [7], and Ishikawa and Fletcher [S]) was put forward to explain the high u-selectivity observed on alcoholysis of anomeric bromides with an ether-type protecting group on C-2. The first step involves dissociation of the glycosyl bromides to tight ion-pairs (B) [9]. This process is promoted by donation from the oxygen lone pair (n) into the carbon oxygen antibonding orbital (o*), requiring the two orbitals to be appropriately aligned [ 101. For the u-halide, with the required orbitals naturally in an antiperiplanar arrangement, this interaction only results in flattening of the pyran ring as the oxygen takes on more sp2 character. For the (3-halide, however, the sugar must reorganize to a boat conformation (AD) to place the orbitals anti-periplanar (there is mounting evidence that the reaction probably proceeds through a synperiplanar arrangement via the lower energy twist-boat formation) [ 111. The resulting ion pairs are then intercepted by the acceptor alcohol (SOH) with inversion, in the rate limiting step, leading to the protonated glycosides (E). The product from the a-halide pathway proceeds through a boat-like intermediate (Ep, whereas the (3-halide leads to a lower energy chair-like intermediate (Ea). This smaller activation barrier, coupled with the higher ground-state energy of the (3-halide (relative to the anomerically stabilized u anomer), more than compensate the energy penalty incurred in aligning the oxygen lone pair with the antibonding orbital (CurtinHammet principle) [ 121. Hence the p-bromide is more reactive than the u-bromide. Even though the u-bromide is predominant (as it is the more thermodynamically stabilized species) interconversion between u- and (3-bromides (C ) is considerably faster than glycosylation (especially when lowly nucleophilic alcohols such as glycosy1 acceptors are used and a source of bromide ion is present-% situ anomerization protocol') [6] and glycosylation occurs preferentially on the more reactive (3bromide, resulting in the u-glycoside as the major product. This is one of the earliest examples of a dynamic kinetic resolution. This mechanism is readily translated to other glycosylation systems if an activation step is added (Figure 2). The leaving group (X) is initially coordinated by the activating group (M+Y,-) to form a catalyst-coordinated ion-pair. Although this pair can be attacked by the alcohol directly, affording products with inverted stereochemistry of the initial ROU

Figure 2. Generalized glycosylation process.

430

17 Protecting Groups

leaving group, in general [ 131, this does not occur and anion exchange occurs before alcohol attack. Schuerch first demonstrated that silver triflate activation of glycosyl halides generates glycosyl triflates [ 141 and it has been shown recently that anion exchange occurs in a variety of other systems [15]. As with the bromide, any functionality sufficiently reactive to act as a leaving group will not be configurationally stable and the same mechanism for glycoside formation applies. The a-selectivity of these reactions is not, however, usually as consistent as for the 'in situ anomerization protocol' of anomeric bromides.

17.3 Electronic and Torsional Effects The first step of any glycosylation involves dissociation into ion-pairs, a process which results in reorganization of the glycoside and the development of a positive charge on the ring oxygen. Clearly, any factors that influence this process will have profound consequences on the outcome of a glycosylation. Riiber et al. noticed that the removal of the oxygen at C2 of glycosides greatly increased the rate at which the methyl glycosides were hydrolysed (Scheme 1) [ 161. Overend et al. [17], amongst others [18], further investigated the effects of deoxygenation at other positions on the rates of hydrolysis of methyl glycosides (Table 1). Richards proposed the explanation that the oxygen atoms, owing to their electronegative nature, inductively destabilize the developing positive charge required for the formation of the ion-pairs (B) [ 181. Hence, their removal results in an increase in the rate of hydrolysis, with the C2 oxygen, because it is nearest the reacting center, having the greatest effect. Even the C6 can, however, affect the reaction, as is evi-

HO

H+

I

*

&HH

OMe

OH

Scheme 1. Reagents and conditions: HCl(aql.

Table 1. Rates of hydrolysis of methyl glucopyranosides at 58 "C in 0.01 M HCI. Glucopyranoside

Relative rate

Methyl-a-D-glucopyranoside Methyl-2-deoxy-a-~-glucopyranoside Methyl-3-deoxy-a-~-glucopyranoside Methyl-4-deoxy-a-~-glucopyranoside

1 2090 20 40

17.4 Influence o j Protecting Group on Donor Reactivity

43 1

Table 2. Comparison of the rates of hydrolysis of methyl glycosides and their 6-deoxy analogs. GIycoside

Rate relative to Me-a-D-glucopyranoside

Methyl-a-D-mannopyranoside Methyl-a-L-rhamnopyranoside Methyl-a-D-galactopyranoside Methyl-6-deoxy-a-~-galactopyranoside

2.4 8.3 5.2 28.2

Figure 3. Torsional strain in the incipient oxonium ion

dent from comparing the rates for methyl-a-D-mannose and methyl-a-L-rhamnose [19], or for methyl-a-D-galactose and its 6-deoxy analog [17] (Table 2). The enhanced lability of the methyl-6-deoxygalactopyranosideis mirrored by the acid-sensitivity of a-fucosyl linkages in general [20]. Another pattern that emerges from these data is the relative reactivities of the differing hexoses-galactosides > mannosides > glucosides. Although there are many factors responsible for this sequence (ground state energies in particular) one contributing factor, as Edwards suggested [2I], arises from the eclipsing interactions that develop between the C2-C3 and C4-C5 substituents, disfavoring formation of the ion-pairs (Figure 3 ) . An axial substituent will interact less with its neighbor and therefore provide less resistance to the conformational reorganization. Overend et al. later argued that the larger substituent on C5 results in the C4-C5 interaction being more influential than the C2ZC3 interaction [ 171, purportedly explaining why galactosides were more reactive than mannosides. Although this explanation is incomplete in rationalizing the complex processes involved in glycoside activation, it was the first recognition that torsional effects could be a controlling influence in a glycosylation; an idea that did not resurface for many years [22]. Inductive effects, however, were all too evident to early researchers; Fletcher et al. demonstrated that increasing numbers of ester protecting groups reduced the reactivity of glycosyl halides under methanolysis conditions (Table 3 ) [23].

17.4 Influence of Protecting Group on Donor Reactivity The origin of t h s effect lies in the destabilizing effect of the exocyclic ring oxygens on ion-pair formation. An ester protecting group, owing to its electron-withdrawing

432

17 Protecting Groups

BnO

1

3

2

Table 3. Influence of protecting group on the rates of methanolysis of arabinosyl chlorides. Substrate

Relative rate of methanolysis

1

106

2 3

13 1

nature, increases the electronegativity of the oxygen it is protecting thereby increasing the deactivating effect of that oxygen. This deactivation effect translates to increased resistance to incipient oxonium ion formation i. e. the leaving group is stabilized. In contrast, a benzyl group does not disturb the oxygen in the same manner (not being electron withdrawing) and thus the fugacity of the leaving group is untempered. Ley et al. have investigated this effect in detail, quantifying the individual contributions of a variety of protecting groups to the different hydroxyl positions of rhamnosides and mannosides [24]. Two differently protected glycosyl donors were made to compete for a standard glycosyl acceptor (Scheme 2). A carbohydrate

Donor A 2 equiv.

Donor 0 2 equiv.

Disaccharide AC

Acceptor C (R'OH) 1 equiv.

Disaccharide BC

Scheme 2. Reagents and conditions: NIS (N-Iodosuccinimide), TfOH (cat.), 4 (CH2C1)2-Et20.

A molecular sieves,

17.4 Influence of Protecting Group on Donor Reactivity

433

Table 4. The possible protecting group patterns of ethyl 1-thio-a-L-rhamnoside. R'

R2

R3

Bn BZ Bn Bn Bz BZ Bn BZ Bn

Bn Bn BZ Bn BZ Bn BZ BZ CDA

Bn Bn Bn BZ Bn BZ BZ BZ

acceptor was specifically chosen to simulate actual glycosylation conditions because the steric profile of the alcohol can influence reaction kinetics through variations in its nucleophilicity [26]. All possible combinations of benzyl and benzoyl protecting group patterns were prepared (Table 4). The torsionally deactivating protecting group cyclohexane-1,2diacetal (CDA), introduced for the selective protection of trans diequatorial 1,2diols (Scheme 3) [25], was also included in the survey of donor activity. The rigidity imposed by the fused cyclic acetals resists the conformational changes required for glycosylation, hence the reactivity of the sugar is diminished. The ratio of the product disaccharides AC and BC was determined by 'H NMR (each disaccharide had already been prepared discretely) thus providing a quantitative analysis of reactivity of the various protecting group patterns (Table 5). Not surprisingly the amount of electronic deactivation arising from a benzoyl protecting group varies with proximity to the ring oxygen/reacting center in a manner similar to that in the deoxygenation experiments of Overend et al. (cf. Table 1); i.e. the order of importance is C2 > C4 > C3 [27]. The torsional deactivation effect from the CDA group is marked, being slightly greater than a benzoate at C2 and slightly less than the 3,4-dibenzoyl compound. As a simplifying approximation, the individual contributions from each protecting group can be treated as being independent of any other protecting groups in the system. Although the validity of this approximation is questionable [28] it does

Me0 i

55% UH

--

D

OH -. .

M e0-

Scheme 3. Reagents and conditions: 1,1,2,2-tetramethoxycyclohexane,CSA (1 0-Camphorsulfonic acid), CH(OMe)3, MeOH.

434

I7 Protecting Groups

Table 5. Selected results from competition experiments. Donor A

Donor B

Quotient AC/BC

TriBn TriBn TriBn 2,4-DiBz 2,3-DiBz 3,4-DiBz 3,4-DiBz 3,4-CDA

3-BZ 4-BZ 2-BZ TriBz TriBz TriBz 3,4-CDA 2-BZ

3.1 8.9 26.6 2.5 13.0 24.2 1.7 1.7

enable semi-quantitative prediction of donor reactivities by multiplication of the quotients or deactivation factors (DFs) for each position. For example: reactivity of 2,4-DiBz (Di-benzoyl) = 8.9 x 26.6 = 236.7 reactivity of 2,3,4-TriBz (Tri-benzoyl) = 3.1 x 8.9 x 26.6 = 733.9 therefore ratio of 2,4-DiBz/TriBz = 733.9/236.7 = 3.1 observed ratio = 2.5 That the predicted result is of similar magnitude to the observed value is indicative of the usefulness of this approximation; the difference cannot, however, be explained away merely as statistical error and should in part be attributed to the failure of the approximation. A similar set of competition experiments was performed for thiomannosides which included the second generation of diacetal protecting groups; butane diacetal (BDA) [29] (Table 6). The relative importance of the various positions is unchanged but C6 emerges as the position of most significance after C2, which could be further tuned by varying the electronics of the benzoate. Making the benzoate more electron rich with a donating p-methoxy group reduced the deactivation; conversely the

Table 6. Standardized DFs for non-Bn protecting groups on thiomannosides. Position of non-Bn group

DF

3-BZ 4-BZ 6-(p-MeOBz) 6-BZ 6-(p-N02Bz) 3,4-CDA 3,4-BDA 2-BZ

1.1 5.0 6.6 8.2 18.9 13.9 16.5 33.6

17.4 Influence o j Protecting Group on Donor Reactivity

BzO BzQ

BzO Bz&

BZO

1.o

To’ BZO

BzO Bz&

2.3

435

To1 HO

3.1

BzO

5.1

11.8

Figure 4. Relative reactivities of selected thiogalactosides.

more electron deficient p-nitro group served to increase the deactivation. Again the torsional effects of the diacetal protecting groups are amply demonstrated, reducing the reactivity of the donor by a factor of around 15. Wong et al. have investigated protecting group effects on p-methylphenyl thiogalactosides and p-methylphenyl thioglucosamines employing a similar competition strategy except with the significant difference of using methanol as the acceptor alcohol [28]. This facilitated examination of effects of hydroxyl groups on donor reactivity by suppressing homocoupling that might otherwise occur if a less nucleophilic acceptor (e.g. carbohydrate alcohol) was used. Curiously, the order of importance of the positions seems to be dramatically altered with C2 and C6 being demoted in favor of C4 and C3 (Figure 4).This result seems to be at odds with the pattern so far established in this chapter, even allowing for the variations that might be expected from changes in stereochemistry of the hydroxyl and leaving group (axial uersus equatorial). Whether this is particular to galactose [30] or is the result of comparing a benzoyl group with a hydroxyl function whose ability to hydrogen bond (providing some torsional resistance) will vary with position and adjacent protecting groups, has yet to be determined. It is, however, clear that a hydroxyl group can be considered as more activating than a benzyl protecting group (Figure 5).

pP

pP Bnb

1.o

2.8

Figure 5. Relative reactivities of a benzylated and an unprotected thiogalactoside.

436

17 Protecting Groups

Wong et al. also demonstrated that the deactivating power of the different electron withdrawing groups investigated was -=Nj> -0AcCl > -NPhth > -0Bz > NHTroc (Troc = 2,2,2-trichloroethurycurbonyl> -0Bn. It should now be clear that protecting groups profoundly affect the reactivities of carbohydrates through torsional and electronic effects. The size of this effect varies with the position and nature of the protecting group and for multiple protecting groups the effects are additive. The reactivity difference can be significant enough to enable successful competitive coupling of two sugars containing identical leaving groups (see Chapters 6(?) and 19(?)).This reactivity tuning can also influence the stereoselectivity and yield of a glycosylation.

17.5 Stereoselectivity 17.5.1 Neighboring-Group Participation The most obvious way in which a protecting group can influence the stereoselectivity of a glycosylation is through anchimeric assistance (neighboring-group participation) [31]. When there is a carbonyl functionality present at C2 [32] (e.g. ester, carbonate, phthalimide, etc.) it can attack the incipient oxonium ion to form the thermodynamically more stable dioxonium species 5 (Figure 6). The position of this equilibrium will depend upon the nucleophilicity of the carbony1 oxygen which varies with the electronics of substituent R. A trichloroacetate, for example, cannot participate whereas a rnonochloroacetate does, but with lower efficiency than a simple acetate [33]. Attack by an alcohol on 5 occurs anti to the carbonyl bond resulting in exclusive generation of the trans glycoside [34]. Hence, donors bearing a 2-acyl protecting group react to give predominantly trans glycosides as they react through the lower energy dioxonium species. If, however, there are severe steric clashes (vide infra Section 17.7) between donor and acceptor that disfavor reaction on the dioxonium species, reaction will instead occur on the incipient oxonium ion with concomitant loss of stereoselectivity [35]. This problem can sometimes be resolved by deactivating the donor; stabilizing the sugar makes intermolecular attack on the incipient

*x

?R

-

-% Lo

4

Figure 6. Anchimeric stabilization.

x-

R

5

437

17.5 Stereoselectivity

Ac

AcA&B & :sph AcO

Ac

P hthN

6

Bn

+

BnO BnO

i

*

BnO

86%

Bn BnO

Me

p:a 15:l 7

OMe

* 9

74% +

I

10

7

OMe

P:a30:l

Scheme 4. Reagents and conditions: (i) NIS, AgOTf, 4 A molecular sieve, CH2C12-toluene, -50 "C to -30 "C; (ii) AgOTf, 2,6-lutidine, 4 molecular sieve, CHzCl2-toluene, -50 " C .

A

oxonium ion a less favorable process, forcing the reaction to proceed through the dioxonium species. Such a situation arose during the synthesis of the glycans of glycodelin (Scheme 4) [36].Coupling of lactosamine donor 6 to trimannoside acceptor 7 provided pentasaccharide 8 as a 15: 1 mixture of anomers (p : a). The stereoselectivity of the coupling was improved to 30 : 1 (p : a) if the glucosamine residue was protected with ester groups instead of benzyl groups. Although it is possible that the observed improvement in stereoselectivity originates from potential alteration of the steric profile of the lactosamine donor as a result of changing the protecting groups, a completely acetylated lactosamine donor [37] gave identical selectivity to 9. 17.5.2 Reactivity Control

If, however, the C2 is functionalized by a group that cannot participate (e.g. azide, benzyl ether, etc. ) the stereoselectivity of a glycosylation is more unpredictable. The two competing pathways, reaction on the a-leaving group and reaction on the pleaving group, are of more similar activation energy than the incipient oxonium ion and the dioxonium species. The choice of protecting group can, however, still effect

438

17 Protecting Groups

Bn BnO Br

12

11

Nc%

BnoBr

14

13

02%

15

BnO

,

16

Table 7. Stereoselectivity of methanolysis of glucosyl halides (molar ratio of methanol to monomer -390: 1). Substrate

a : p Ratio

11 12

7:93 16:84 16:84 37 : 63 92:s 16:84

13 14 15 16

the stereochemical outcome electronically, as was noted by Frechet and Schuerch (Table 7) [38]. They investigated the effect of different protecting groups at C6 on the stereoselectivity of methanolysis of glucosyl halides (building on earlier work by Ishikawa and Fletcher) [ 81. A clear pattern emerged-electron-rich benzoates promoted pselectivity and the more electron-withdrawing the benzoate became (p-nitro being the most deactivating) the greater the proportion of a-glycoside formed. This can be rationalized in terms of the reactivities of the donors; from the data in Table 6 it is clear that the more electron-deficient the benzoyl, the more the reactivity of the donor is diminished (the greater the DF) owing to the inductive effects highlighted above (Section 17.3). This results in an increase in the activation energies for glycosylation of both the a- and P-bromide. The a-anomer, with its lower ground-state energy (anomerically stabilized), becomes too unreactive to glycosylate methanol at a rate that is competitive with the P-bromide pathway (4). Interconversion of the ato P-bromide is fast and the reaction therefore (for 15) generates the a-glycoside as the major product (Figure 7). If, however, the reactivity of the donor is reduced too much, as in 16, in which the leaving group is changed to the chloride (some 20-30 times less reactive), the rate of

17.5 Stereoselectivity U-D-GIY-1-X

a-D-Gly+, x-

p -~-Gly-l-X

p - ~ - G l y +X,

439

(3) MeOH

(4) M eOH

Methyl 0-D-glycopyranoside

Methyl a-D-glycopyranoside

Figure 7. Methanolysis of glycopyranosides.

interconversion of a- to P-chloride (2) is reduced (relative to glycosylation). Because the a-chloride is present at a higher concentration (because it is the thermodynamically favored arrangement) than the p-chloride, reaction proceeds through the achloride pathway (3) and the p-glycoside is the major product. Alternatively, if the donor is made sufficiently reactive (as in ll), the a-bromide is reactive enough to glycosylate methanol at a rate competitive with the lower energy p-pathway, aided by its increased concentration, such that the P-glycoside predominates. Unfortunately these results were not reproduced when carbohydrate acceptors were used; only a-glucosides were obtained, irrespective of the protecting group on C6 [39]. This is a reflection of the sensitivity of the kinetics of glycosylation-a carbohydrate alcohol is considerably less nucleophilic than methanol for reasons both of size and electronics. This reduced nucleophilicity ensures that the a- to phalide equilibration process is faster than the glycosylation and hence reaction proceeds through path (4) [40]. The reduced concentration of alcohol relative to donor (ratios of 390: 1 are not practical for synthesis of oligosaccharides) would also ensure that equilibration was faster than glycosylation, thus favoring formation of the a-glycoside [41]. The underlying principles, however, are still valid and can be successfully applied in many systems; in general, deactivation of the donor improves the a-selectivity of a glycosylation if the donor species can still anomerize. Such an occasion arose during various groups’ syntheses of glycosyl phosphatidyl inositol membrane anchors [42], in which an oligosaccharide is linked to an inositol moiety via an a-glucosamine residue. a-Linkages in gluco-configured sugars are particularly troublesome because, unlike manno-configured sugars whose axial C2 hydroxyl serves to stereoelectronically promote a-selectivity [43], the equatorial C2 imparts no particular bias. Ley’s initial approach employed glycosyl donor 17 but even under optimized conditions [44] the best selectivities obtained consisted of a 3 : 1 mixture (a : p) (Scheme 5). The in situ-generated glycosyl triflate/perchlorate was too reactive for the two glycosylation pathways ((3) and (4) in Figure 7) to be cleanly differentiated. Ley’s solution was to change the donor species to the less reactive iodide (as opposed to the triflate/perchlorate), this deactivation afforded the desired disaccharide 19 with near complete a-selectivity [42a]. Schmidt’s solution was also to deactivate the donor but instead of deactivating the active leaving group (triflate), he deactivated the donor by employing acetyl protecting groups

440

T Bn B

17 Protecting Groups

S

O

~

F

N3 I

17

+

b

60% a:p3:1

19

18

T Bn B s

o

20

G

ii

Br

+ 18

b

65%

TBs4& Bn

Al I

Bn

a$ >30:1 19

Scheme 5. Reagents and conditions: (i) Yb(OTf)Z, CaC03, Ba(C104)2,Et20, room temp; (ii) TBAI (tetra-N-butyl-ammonium iodide), CH2C12, room temp; (iii) TMSOTf, Et20, room temp. Mnt = menthyl.

(21), affording the same result [42c].The different protecting groups on the acceptor alcohol 22 may also have contributed to this improvement in selectivity (uideinfru). A similar problem was encountered during the synthesis of the glycans of glycodellin-S [36]. The synthesis of LewisY epitope 26 required bis-a-fucosylation of lactosamine moiety 24 (Scheme 6). The tri-0-benzylated fucosyl donor 25 provided the tetrasaccharide 26 with acceptable selectivity but unfortunately as an inseparable mixture. Recourse to a bromide donor under halide ion catalysis (as in the previous example) was not successful because the deactivation was so great it made glycosylation of the hindered hydroxyls prohibitively slow. Deactivating the donor species by acetylating the 3- and 4-positions (27) [45], however, increased the a-selectivity to the point that the other anomer was unobservable. There are other ways to influence the stereochemical outcome by affecting the

441

17.6 Injluence of the Protecting Group on the Acceptor

i OH

PhthN

24

*

57%

++% PhthN

+ P Bn

OBn Bn

BnO

25

i OH

PhthN

24 +

26

a$ 15:l

*

SEt

47%

F' AcO

OBn AcO

24

28

Scheme 6. Reagents and conditions: (i) AgC104, 2,6-lutidine, Et,O, THF, -35 "C.

anomerization equilibrium; lower temperatures 'freeze out' the a-donor species promoting p-selectivity, especially when the donor is deactivated [38, 461 (see Chapter 16(?) on p-mannose synthesis) and the polarity and donating ability of the solvent can alter the equilibrium [47] (see chapter on solvent effects). As has already been mentioned, the nucleophilicity of the acceptor alcohol can also effect the stereoselectivity of glycosylation. The more nucleophilic alcohols can react on the a-ion pair whereas less nucleophilic alcohols glycosylate more slowly, allowing the donor species to equilibrate to the more reactive p-ion pair, generating the a-glycoside. Hence glycosylation of the primary C6 is usually less a-selective than any of the secondary positions. The choice of protecting group can also electronically influence the nucleophilicity of the acceptor alcohol.

17.6 Influence of the Protecting Group on the Acceptor In the same way that protecting groups influence the electron density on the anomeric carbon, tuning the reactivity, they also effect the electron density of the hydroxyl groups within the sugar, adjusting their nucleophilicity. Sinay noted in the synthesis of lactosamine derivative 32 that the protecting group at C3 of the glucosamine had a dramatic effect on the glycosylation yield with galactosyl bromide 29 (Scheme 7) [48]. Although the acetylated acceptor 30 was such a poor nucleophile that the coupling with 29 only produced a 5% yield of the desired

442

I 7 Protecting Groups

Scheme 7. Reagents and conditions: (i) HgBr2, (CH2C1)2, room temp.

i A

* Ac

Ac H OBn

Br

AcO

OBn

R= Ac, 30 R= Bn,31

29 Acceptor

Yield (%)

30 31

5 87

32

product, benzyl protection increased the nucleophilicity of acceptor 31 affording 32 in 87% yield [49]. The protecting group at the anomeric position also influences hydroxyl reactivity; thioglycosides tend to be less nucleophilic than the equivalent 0-glycoside [ 501. This can result in reduced glycosylation yields for difficult couplings as was the case in the synthesis of disaccharide 36, taken from van Boom's synthesis of a pentasaccharide from Crytococcus neoformans (Scheme 8) [ ~ O C ] .Coupling of xyloside 33 with thiomannoside 34 provided disaccharide 36 in a disappointing 38% yield, methyl mannopyranoside 35, however, coupled with 33 in 68% yield. The final mechanism by which protecting groups can influence the outcome of a glycosylation is through the way they alter the steric profile of the donor and acceptor.

Scheme 8. Reagents and conditions: (i) TMSOTf, (CH2C1)2, room temp

A%*

i

AcOO

-

;&-% A

AcO

Bn

33

BnO Bn

R=SEt, 34 R=OMe, 35 Acceptor

Yield ('%)

34

38 68

35

R

36

17.7 Steric EfSects on Glycosylution

443

17.7 Steric Effects on Glycosylation Of all the ways in which protecting groups can influence a glycosylation, steric effects are perhaps the least quantifiable and yet ultimately, the most influential. Although the rate-limiting step for most glycosylations is the activation of the leaving group [51] the product determining step is the approach of the acceptor to the donor (step C in Figure 1) [52]. The different steric interactions in the two transition states CaPand CPaaffect the activation energies of each pathway, which determine the selectivity (and yield) of a glycosylation [53]. This is obvious when the protecting group is close to the reacting center, as in Scheme 9 [54].Attack by the rhamnoside 40 on the torsionally deactivated mannosyl a-triflate, preformed from 37 at low temperature by activation with triflic anhydride, proceeded only with modest P-selectivity when the C2 of mannosyl donor was protected with a bulky tert-butyldimethylsilyl (TBDMS) group. Replacement of this group with the smaller trimethylsilyl (TMS) or benzyl group greatly improved the P-selectivity (for optimized conditions see Chapter 16 (?) on p-mannose). Even remote positions, the C6 for example, have, however, been reported to have an effect on the stereoselectivity of glycosylation [55]. The issue is unfortunately clouded by the change in electronic properties which frequently accompanies a change in protecting groups. Perhaps the most dramatic example of a protecting group affecting the outcome of a glycosylation comes from a synthesis of a glycan from laminin by Zhu et al. (Scheme 10) [56].Attempted coupling of lactose donor 42 with P-methylglycoside 43p failed to produce any of desired hexasaccharide 44p. When coupling was attempted on the a-methyl glycoside 43a under identical conditions, however, this

Scheme 9. Reagents and conditions: (i) Tf20, DTBMP, Et,O, -78 "C. Me P

BnQ

i

q

+

Bn

VEt R=TBDMS,37 0 R=TMS, 38 R=Bn, 39

41

40

Donor

Selectivity (j3 : a )

37 38 39

1.6: 1

19

5.1 : 1 6.1: 1

91

Yield (%)

85

\

444

17 Protecting Groups

Scheme 10. Reagents and conditions: (i) BF3 . OEtz, CH2C12,-10°C to room temp.

Ac

AcO

Ac i

I

Ac BnO

*

Ac

44

AcO

OH

T&

BnO BnO

B -nBn

R

43 BnO

R’

A 43a 43P

R

R’

Yield (%)

Me0 H

H Me0

65 0

single stereochemical alteration was sufficient to rectify the situation, furnishing hexasaccharide 44u in 65% yield. Rationalization of this reversal of fortune lies in the different conformations of 43u and 43p. Molecular modeling demonstrated that the steric hindrance about the hydroxyl group of 43p was greater than that for 43u, making the hydroxyl less amenable to glycosylation. Add to the equation the steric hindrance about the donor site and the net ‘steric mismatch’ is enough to prevent coupling between donor 42 and 43p. It is unusual for such a controlling element to be so readily identified or modified and steric mismatch, whose likelihood increases with increasing size of coupling partners, can be the Nemesis of block coupling-based strategies [ 571.

17.8 Conclusions In the last 25 years much progress has been made in the development of glycosyl donors and activating systems [58],with few corners of the periodic table remaining untouched in the search for an ever-better system. With few exceptions [59], however, they all depend on the displacement of an anomeric leaving group (frequently a triflate) in a bimolecular fashion and are thus all subject to the influences outlined hitherto. That different donors provide varying results for the same glycosylation

445

17.8 Conclusions

i

BnO

30%

45

-

All

-

‘&OAIl

47

PhthN

46

a:P 23:l

Scheme 11. Reagents and conditions: (i) 0.07 M LiC104, CHZC12, room temp.

can be ascribed to their differing rates of activation under the conditions used. This was encountered by Waldmann and Schmid in their investigations into the synthesis of blood group determinants (Scheme 1 1) [60]. Coupling of the a-fucosyl fluoride 45 with the acceptor 46 proceeded with good stereocontrol to afford the a-( 1,4)-linked disaccharide 47 albeit in poor yield. When, however, the more reactive P-fluoride was employed under identical conditions, the desired disaccharide was obtained in better yield (68%) but with worse stereoselectivity ( a : P 4: 1). In this example even though the donors are of the same type, the difference in stereochemistry was enough to affect the course of the reaction. The reaction mechanism has not changed, merely the positions of the various equilibria. This result is not unique to a certain leaving group but to the entire system-the activator, the solvent, the temperature, the donor, the acceptor, the protecting groups, and the leaving group-potentially requiring that each individual reaction be optimized. This predicament prompted carbohydrate chemists, as early as 1973, to bemoan “unfortunately the reactivity of C1 and the stereoselectivity of its reaction depend in a complex way on the leaving group, the blocking groups, the configuration of the sugar and the structure of the molecule bearing the entering hydroxyl function” [611. Although the glycosylation reaction is by no means tamed, owing to the unpredictable interplay of these factors (steric effects in particular), its behavior has mostly been characterized, even if not entirely understood. Protecting groups, as first summarized by Paulsen in his seminal review [62], can alter the course of a reaction in a predictable manner; providing the shrewd chemist with some control over the reaction. To gain complete control of a glycosylation will require the development of a coupling strategy which evades these influences, one that does not depend on a bimolecular collision as its key step [63]. Acknowledgments

We thank Professor A. J. Kirby and Dr S. L. Warriner for helpful discussion in the preparation of this manuscript.

446

I 7 Protecting Groups

References 1. P. J. Kocienski Protecting Groups;Georg Thieme Verlag: Stuttgart: 1994. 2. (a) T. W. Greene, P. G. M. Wuts Protective Groups in Organic Synthesis, Wiley: New York: 1991 (b) ref [ I ] (c) M. Bols Carbohydrate Building Blocks, Wiley: New York: 1996 (d) S. Hanessian Preparative Carbohydrate Chemistry, Marcel-Dekker: New York: 1997 (e) G.-J. Boons Carbohydrate Chemistry, Blackie Academic & Professional: London: 1998. 3. B. Capon Chem. Rev. 1969, 69, 407. 4. M. L. Sinott, W. P. J. Jencks J. Am. Chem. Soc. 1980, 102, 2056. 5. (a) T. L. Amyes, W. P. J. Jencks J. Am. Chem. SOC.1989, 111, 7888 (b) M. L. Sinott Chem. Reu. 1990, 90, 1171. 6. R. Lemieux, K. B. Hendriks, R. V. Stick, K. James J. Am. Chem. SOC.1975,97, 4056. 7. A. J. Rhind-Tutt, C. A. Vernon J. Chem. Soc. 1960, 4637. 8. T. Ishikawa, H. G. Fletcher J. Org. Chem. 1969, 34, 563. 9. R. U. Lemieux, G. Huber Can. J. Chem. 1955,33, 128. 10. (a) P. Delongchamps, C. Moreau, D. FrCchet, P. Atlani Can. J. Chem. 1972, 50, 3402 (b) P. Deslongchamps ACS Symposium series: The Anomeric Effect and Associated Stereoelectronic effects; G. R. J. Thatcher Ed.; American Chemical Society: Washington, D. C., 1993; Vol. 549. 11. (a) A. J. Ratcliffe, D. R. Mootoo, C. Webster Andrews, B. Fraser-Reid J. Am. Chem. SOC. 1989, 111, 7661 (b) P. Deslongchamps, P. G. Jones, S. Li, A. J. Kirby, S. Kusela, Y. Ma J. Chem. SOC.,Perkin Trans. I 1997, 12, 2621. 12. D. Y. Curtin Rec. Chem. Proc., 1954, 15, 111. 13. There are of course exceptions: (a) where there is no available counter-ion e.g. BF30Et2 activation of anomeric fluorides and trichloroacetimidates but N.B. the catalyst coordinated ionpair is also not configurationally stable either. Some silver salts e.g. AgZC03, AgzO, and Ag silicate also operate via a ‘push-pull’ mechanism--(;. Wulff, G. Rohle, U. Schmidt Chem. Ber. 1972, 105, 111-however, in this case the initial bromides are not usually configurationally stable. ZnC12 mediated concerted nucleophilic opening of deactivated glycal epoxides relies on direct attack for selectivity. (b) In cases of solvolysis, as the nucleophile is present in the solvent cage, attack immediately follows activation of the leaving group, leading to product with inversion of the leaving group stereochemistry. A similar special case includes the reaction of trichloroacetimidates with carboxylic acids-R. R. Schmidt, J. Michel Angew. Chem. Znt. Ed. Engl. 1980,19,731-and dibenzyl phosphate-R. R. Schmidt, M. Strumpp Liebig Ann. Chem. 1984, 680-where the nucleophile is again present in the solvent cage at the point of activation of the leaving group. 14. (a) V. Marousek, T. H. Lucas, P. E. Wheat, C. Schuerch Carbohydr. Res. 1978, 60, 85. Anomeric perchlorates had been reported previously-K. Igarashi, T. Honma, J. Irisawa Curbohydr. Res. 1970, 15, 329. 15. D. Crich, S. X. Sun Tetrahedron 1998, 54, 8321. 16. C. N. Riiber, N. A. Ssrensen Kgl. Norske. Videnskab. Selskabs Skrifter 1938, I , 1. 17. W. G. Overend, C. W. Rees, J. S. Sequeria J. Chem. SOC.1962, 3429. 18. G. N. Richards Chem. Znd. 1955, 228. 19. M. S. Feather, J. F. Harris J. Org. Chem., 1965, 30, 153. 20. E. A. Kabat, H. Baer, A. E. Bezer, V. Knaub J. Exp. Med., 1948, 88, 43. 17. J. T. Edward Chem. Znd. 1955, 1102. 22. B. Fraser-Reid, A. Wu, C. Webster Andrew, E. Skowronski J. Am. Chem. SOC.1991, 113, 1434. 23. C. P. J. Glaudemans, H. G. Fletcher J. Am. Chem. SOC.1965, 87,4636. 24. N. L. Douglas, S. V. Ley, U. Lucking, S. L. Warriner J. Chem. SOC.,Perkin Trans. I , 1998, 51. 25. L. R. Schroeder, J. W. Green, D. C. Johnson J. Chem. SOC.B, 1966, 447. 26. (a) S. V. Ley, H. W. M. Priepke, S. L. Warriner Angew. Chem. Znt. Ed. Engl. 1994, 33, 2290 (b) P. Grice, S. V. Ley, J. Pietruszka, H. W. M. Priepke, S. L. Warriner, J. Chem. SOC. Perkin Trans. I , 1997, 351. 27. A plausible explanation for the increased importance of the C4 relative to the C3 based upon hyperconjugation with the endocyclic oxygen lone-pairs has been put forward-C. A. A. van Boeckel, T. Beeta, S. F. van Aelst Tetrahedron, 1984, 40, 4097.

References

447

28. Z. Zhang, I. R. Ollmann, X.-S. Ye, R. Wischnat, T. Baasov, C.-H. Wong J. Am. Chem. Soc. 1999,121, 734. 29. (a) N. L. Douglas, S. V. Ley, H. M. I. Osborn, D. R. Owen, H. W. M. Priepke, S. L. Warriner Synlett, 1996, 793 (b) U. Berens, D. Leckel, S. C. Oepen J. Org. Chem. 1995, 60, 8204 (c) J.-L. Montchamp, F. Tian, M. E. Hart, J. W. Frost J Org. Chem. 1996, 61, 3897. 30. That the C4 of galactose has a particular bearing has been noted by others (a) M. Miljkovic, D. Yeagley, P. Deslongchamps, Y. L. Dory J. Org. Chem. 1997, 62, 7597 (b) 0. Kwon, S. J. Danishefsky J. Am. Chem. Soc. 1998, 120, 1588. 31. R. U. Lemieux Adu. Carbohydr. Chem. 1954, 9, I . 32. Although participation from other positions has been postulated by many early investigators, there is little evidence to substantiate that this is possible. 33. R. U. Lemieux, C. Brice, G. Huber Can. J. Chem. 1955, 33, 134. 34. In truth this is a simplification; although attack at C1 of 6 leads to the thermodynamic product, kinetic attack occurs first at the carbonyl carbon leading to an orthoester product. If the reaction is performed under acidic conditions, the acid sensitive orthoester rearranges to form the trans glycoside. However if the donor is very deactivated the orthoester may require a higher acid concentration to induce rearrangement which can result in transesterification (to the acceptor alcohol) predominating, especially if an acetate is used. Destabilizing the orthoester by activating the donor or using benzoyl or pivaloyl protecting groups can resolve this problem. 35. This problem has been reported by many other authors (a) S. Oscarsson Topics in Current Chemistry: Glycoscience; H. Driguuez, J. Thiem Ed.; Springer: Berlin Heidelberg, 1997; Vol. 186 (b) A. V. Nikolaev, T. J. Rutherford, M. A. Ferguson, J. S. Brimacornbe Bioorg. Med. Chem. Lett. 1994, 4, 785 (c) T. Ziegler, B. Adams, P. Kovac, C. P. J. Glaudemans Carbohydr. Chem. 1990, 9, 135 (d) H. Jiao, 0. Hindsgaul Angew. Chem. Int. Ed. Engl. 1999, 38, 346 (e) N. Kochetkov The Synthesis of Polysaccharides in The Total Synthesis of Natural Products; J. ApSimon Ed.; Wiley-Interscience: New York, 1992; Vol. 8, p 245. 36. D. Depre, A. Duffels, L. G. Green, R. Lenz, S. V. Ley, C.-H. Wong Chem. Eur. J. 1999, 5, 3326. 37. J. Arnop, M. Haraldsson, J. Lonngren J. Chem. Soc., Perkin Trans. 1 1982, 1841. 38. J. M. Frechkt, C. Schuerch J. Am. Chem. Soc. 1972, 94, 604. 39. J. M. Frechkt, C. Schuerch Polymer Supported Reactions in Organic Synthesis; P. Hodge Ed.; Wiley: Chichester, 1980. 40. Schroeder et al. demonstrated that a-selectivity increased with increasing size of the alcoholL. R. Schroeder, J. W. Green, D. C. Johnson J. Chem. SOC.B 1966, 447. 41. J. E. Wallace, L. R. Schroeder J. Chem. Soc., Perkin Trans. 1 1976, 1938. 42. (a) D. K. Baeschlin, A. R. Chaperon, V. Charbonneau, L. G. Green, S. V. Ley, U. Lucking, E. Walter Angew. Chem. Int. Ed. Engl. 1998,37, 3423 (b) A. S. Campbell, B. Fraser-Reid J. Am. Chem. Soc. 1995,117, 10387 (c) T. G. Mayer, B. Krazer, R. R. Schmidt Angew. Chem. Znt. Ed. Engl. 1994, 33, 2177 (d) C. Murakata, T. Ogawa Curbohydr. Res. 1992,235, 95. 43. The A2 effect-R. E. Reeves Adu. Carbohydr. Chem. 1960, 15, 11-the axial oxygen at C2 serves to both to reinforce the anomeric effect, further lowering the ground state energy of the a-leaving group relative to the P-leaving group and sterically shields the p-face thus promoting a-glycosidation. 44. W. S. Kim, S. Hosono, H. Sasai, M. Shibisaki Heterocycles 1996, 42, 795-perchlorate counter-ions have been proposed to provide higher a-selectivities than triflates owing to their slightly reduced fugacity and ether type solvents at room temperature also serve to promote aselectivity (see chapter on solvent effects). 45. Others have used deactivated fucosyl donors-(a) M. Djeter, H. M. Flowers Carbohydr. Res. 1972, 23, 41 (b) V. Behar, S. J. Danishefsky, Angew. Chem. Int. Ed. Engl. 1994, 33, 1468 (c) R. Windmuller, R. R . Schmidt, Tetrahedron Lett. 1994, 65, 7927 (d) T. Kiyoi, Y. Nakai, H. Kondo, H. Ishida, M. Kiso, A. Hasegawa, Biocirg. Med. Chem. 1996, 4, 1167. 46, J. E. Wallace, L. R. Schroeder J. Chem. Soc., Perkin Trans. I1976, 1633. 47. R. Eby, C. Schuerch Carbohydr. Res. 1974, 34, 79. 48. P. Sinay Pure Appl. Chem. 1978, 50, 1437. 49. Boons and Zhu have exploited this effect for the synthesis of small oligosaccharides-G.-J. Boons, T. Zhu Synlett 1997, 809.

448

17 Protecting Groups

50. (a) V. Pozsgay Carbohydrates: Synthetic Methods and Applications in Medicinal Chemistry: Synthesis of Oligosaccharides Related to Plant, Vertebrate, and Bacterial Cell-wall Glycans: H. Ogura, A. Hasegawa, T. Suami Ed.; Kodansha, VCH: Tokyo, New York, 1992, p 188 (b) H. Paulsen, W. Rauwald, U. Weichert Liebig. Ann. Chem. 1988, 75 (c) K. Jaarsveld Zegelaar, S. A. W. Smits, G. A. van der Marcel, J. H. van Boom Bioorg. Med. Chem. 1996,4, 1819. 51. e.g. for sulfoxides D. Kahne, S. Raghavan J. Am. Chem. SOC.1993, 114, 1580. 52. For glycosyl bromides under halide catalysis this is the rate determining step [6]. 53. N. M. Spijker, C. A. Boeckel Angew. Chem. Int. Ed. Engl. 1991,30, 180. 54. D. Crich, S. X. SunJ. Org. Chem. 1997,62, 1198. 55. inter alia (a) M. Wakao, Y. Nakai, K. Fukase, S. Kusumoto Chem. Lett. 1999, 27 (b) R. Rodebaugh, S. Joshi, B. Fraser-Reid, H. M. Geysen J. Org. Chem. 1997, 62, 5660. 56. X.-X. Zhu, M . 3 . Cai, R.-L. Zhou Carbohydr. Res. 1997, 303, 261. 51, (a) Y.-M. Zhang, A. Brodzky, P. Sinay, G. Saint-Marcroux, B. Perly Tetrahedron: Asymmetry 1995, 6, 1195 (b) V. Pozsgay, E. P. Dubois, L. J. Pannell J. Org. Chem. 1997, 62, 2832. 58. K. Toshima, K. Tatsuta Chem. Rev. 1993, 93, 1503. 59. These include: (a) direct alkylation of the anomeric position-R. R. Schmidt, Angew. Chem. Int. Ed. Engl. 1986, 25, 212 (b) attempts at SNi based strategies-@) M. E. Behendt, R. R. Schmidt Tetrahedron Lett. 1993, 34, 10733; (ii) T. Iimori, T. Shibizaki, S. Ikegami Tetrahedron Lett. 1996, 37, 2267 iii) G. Scheffler, R. R. Schmidt Tetrahedron Lett. 1997, 38, 2943 iv) G. Scheffler, R. R. Schmidt J. Org. Chem. 1999, 64, 1319 (c) tethered strategies (see Chapter 16 (?) on 0-mannose) (d) redox glycosylation-A. G. M. Barrett, B. C. B. Bezuidenhoudt, A. F. Gasiecki, A. R. Howell, M. A. Russell J. Am. Chem. SOC.1989, 111, 1392 (e) acetal reduction-H. Ohtake, T. Iimori, S. Ikegami Tetrahedron Lett. 1997,38, 3413. 60. H. Waldmann, U. Schmid Chem. Eur. J. 1998,3, 494. 61. C. Schuerch Acc. Chem. Res. 1973, 6, 184. 62. H. Paulsen Angew. Chem. Int. Ed. Engl. 1982, 21, 155. 63. F. Baressi, 0. Hindsgaul Modern Synthetic Methods: Glycosylation Methods in Oligosaccharide Synthesis, B. Ernst, C. Leuman, Ed.; VCH, Basel, 1995.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

18 Intramolecular Glycosidation Reactions Jacob Mudsen and Mikael Bols

18.1 Introduction Carbohydrate derivatives found in nature are almost always linked to each other by glycosidic bonds. There has been increasing interest in the synthesis of biologically active oligo- and polysaccharides and reliable methods for the formation of glycosidic linkages have been the subject of intensive study. Two problems are usually encountered in the synthesis of a glycosidic bondunreactive hydroxyl groups and lack of diastereoselectivity (i.e. formation of a and epimers). To overcome these problems in the last decade chemists have explored a new approach-intramolecular glycosidation reactions. In this review we will define the term intramolecular glycosidation reaction as a type of reaction in which the glycosyl donor (the glycone) is linked to the glycosyl acceptor (the aglycone) by means of a tether producing a 0-glycosidic bond (Scheme 1). The tether can be either a temporary tether which is removed during the glycosidation reaction or it can be a stable tether which is cleaved after the glycosidation procedure. This chapter is divided into two main parts: 1) reactions in which the tether participate in the reaction; and 2) reactions in which the tether does not participate in the reaction.

Non-tethered intramolecular glycosidations which produce spiroketals [ 11 or natural products [2, 31 have been reported; these will not be discussed here.

18 Intramolecular Glycosidation Reactions

450

GIycosyl

X

donor

7d@& Tetherin procedure

L

Glycosyl acceptor

Ho przdure GI sidationb L Tethered disacchari d

RoL*

X

Scheme 1. Outline of tethered intramolecular glycosidation reactions. (T1 = temporary tether, T2 = stable tether)

18.2 Reactions in which the Tether Participates in the Reaction 18.2.1 Tethering to the Glycosyl Donor Carbon Tethers

In 1991 Barresi and Hindsgaul reported the first example of an intramolecular glycosidation procedure and introduced the expression ‘intramolecular aglycone delivery’. The idea was to use this approach in syntheses of 1,2-cis glycosidic linkages in the difficult mannose case [4]. The set-up for this reaction was thioglycoside 1, a glycosyl donor with a 2-0-acetate group (Scheme 2). The acetate was converted to methylene derivative 2 with the Tebbe reagent. Treatment of 2 with the alcohols a, b or c in the presence of a catalytic amount of acid formed the key acetals 3a-3c in moderate yield. Activation of 3a-3c with 5 equiv. N-iodosuccinimide (NIS) resulted in selective intramolecular glycosidation at the p face via intermediates 4a-4c, which on aqueous work-up gave the desired pmannopyranosides 5a-5c. In a later paper an improved method, in which the base 2,6-di-t-butyl-4methylpyridine (DBMP) was added in the activation step, was reported [5].This increased the yields of 5b and 5c significantly. Further investigation, by means of competition experiments, showed that the reaction was intramolecular, but also revealed some limitations of the method. The mixed acetals 3a-3d were relatively unstable and had to be prepared carefully to avoid hydrolysis to the corresponding monosaccharides or dimerization. Furthermore the yield dropped drastically when trisaccharides, e. g. 5d, or tetrasaccharides were synthesized [6]. A similar approach was taken by Ito and Ogawa who introducedp-methoxybenzyl (PMB) as carbon tether [7]. Alcohols protected with PMB, e.g. 6, can be oxidatively deprotected with DDQ [8] or CAN [9] presumably via the oxocarbenium intermediate 7 (Scheme 3 ) .

451

18.2 Reactions in which the Tether Participates in the Reaction

~~7% A

- Bny%FiR

reagent, Tebbe

BnO

ROH

BnO

cat. TsOH

2

SEt

SEt

BnO

4"C,20min

1:z

SEt

NIS (and base)

R- OH

BnO BnO O *"

Mixed

P-glycoside 5 without basel with base

51%

61%

-

45%

42%

17%

r

-I

Brio OMe

a

(OBn HO%

BnO Brio OMe b

OBn BnO BnO

Hosoe C

38%

BnO

NPhthBno d

-

21%

O(CHz)&H3 NF'hth

Scheme 2. Isopropylidene-tethered glycosidation.

r

6

L

+ l

7

J

CHO

Scheme 3. Protection and deprotection of PMB ethers.

OR

5

452

18 Intramolecular Glycosidation Reactions

OMe

I

Q Brio\

OPMB R-OH

BnO BnO

8

F

'

/

BnO

-

L

R- OH

~

P-glycosides

9

F

yield[%]*'

OH

Brio* BnO

OBn OBn

10a

61-14'

BnO BnO

a

OR

10

OBn

OBn

Ho?&& BnO

OBn NPhth

1oc

40

P

* Variations in different solvents(CH,CI,, ** Total yields for two reactions.

Et,O, MeCN).

Scheme 4. PMB-tethered glycosidation in the synthesis of P-mannopyranosides.

The basic idea was to trap intermediate 7 under anhydrous conditions and react it with another alcohol producing the corresponding mixed acetal. The authors showed that treatment of mannosyl donor 8 with DDQ and alcohols a-c gave the mixed acetals 9a-9c (Scheme 4). Without purification, 9a-9c were activated with AgOTf and SnC12 in the presence of DBMP. Aqueous work-up gave the expected P-mannopyranosides lOa-lOc in good overall yields. By use of this strategy the core penta- and hexasaccharides found in asparaginelinked oligosaccharides were synthesized in good yield [ 10- 121. It has recently been reported that the choice of protection groups also plays an mannose derivative important role. Thus using a 4,6-O-cyclohexylidene-protected as glycosyl donor gave the desired P-mannopyranosides in high yields [13]. Ito and Ogawa have also reported the synthesis of P-mannopyranoside linkages by use of the 'intramolecular aglycone delivery' strategy with PMB as tethering group linked to a solid phase (Scheme 5) [ 141. It is notable that the products 13a-13d contain free hydroxyl groups. By this

18.2 Reactions in which the Tether Participates in the Reaction

453

R-OH

DDQ,CH2C12 room temp.

TBSO $3hp

ph3%+oR TBSO

1

12 SMe

11 sMe

MeOTf

%&H2Cl

P-rnannopyranoside

R- OH

Yield[%]

-.aoBn 139

BnO

50

%"P

OBn

P

TBSO

6

BnO BnO

OBn

b

13b

43

13c

37

13d

54

13

OR

OBn

Ho&F BnO NPhth C

QBn HO-,H,

N3 d

Scheme 5. Solid-phase version of PMB-tethered glycosidation.

means the P-mannopyranoside derivatives will be the only compounds not connected to the polymer phase after the glycosidation reaction. Unreacted starting material, hydrolysed mixed acetals, or 1,2 elimination products will stay bound to the polymer phase. This means that the disaccharides should be easily isolated. A variation of the PMB strategy was used by Krog-Jensen and Oscarson to produce (3-D-fructofuranoside linkages (Scheme 6) [ 15, 161. Reaction of the a-Dfructofuranoside derivative 14 with mannose derivatives 15 or 16 and DDQ gave the corresponding mixed acetals 17 which were not characterized. Immediately after work-up the acetals were treated with dimethyl(methylthi0)sulfonium trifluoromethanesulfonate (DMTST) in CHZC12 to give P-D-fructofuranoside derivatives 19, both in 77% yield.

454

18 Intramolecular Glycosidation Reactions

DMTST

Bnow

____)

CHZCIZ, 71%

OBn

OBn

-

1

17

19

Scheme 6. PMB-tethered glycosidation in the synthesis of P-fructofuranosides

Other promoters (MeOTf, NIS, iodonium dicollidine perchlorate (IDCP), and iodonium dicollidine triflate (IDCT)) have also been investigated, but DMTST usually seems to be the right choice [ 161. Silicon Tethers The use of silicon as a temporary tether in intramolecular reactions has found widespread application in organic chemistry [ 171. This concept was first described by Stork et al. in 1992 for intramolecular stereocontrolled glycosidation reactions. With a silyl group connected to the 0-2 position of a P-mannosyl donor it was demonstrated that the glycosyl acceptors could contain primary and secondary hydroxyl groups [18, 191. Treatment of sulfoxide 20 with MezSiClz and alcohols a-f gave the mixed silyl acetals 21a-21f in good yields (Scheme 7). Activation of the sulfoxide with trifluoromethane sulfonic acid anhydride (Tf2O) and base, presumably resulting in the five-membered intermediates 22a-22f responsible for exclusive glycosidation at the [3 face, gave 0-mannopyranosides 23a-23f. Special interest was now devoted to glycoside 23d because it was produced in 12% yield only. It was found that compound 24, containing a silicon bridge, could be isolated as the major product in 82% yield (Scheme 8). A similar observation was made when silyl acetal 21f was activated, producing the expected glycoside 23f in 54% yield and 41% of a debenzylated compound [ 191. A similar approach was simultaneously investigated by Bols et al. Initially thioglycoside 25 was reacted with Me2SiC12 and primary, secondary, or tertiary aliphatic alcohols or phenol in the presence of pyridine to produce the corresponding silyl acetals 26a-26d in good yields (Scheme 9). Activation of 26a-26d with NIS and a catalytic amount of TfOH resulted in exclusive formation of a-glucopyranosides 28a-28d [20].

455

18.2 Reactions in which the Tether Participates in the Reaction

+B ;n

? '

. l

Me$iC12/imidazole

* : nB

BnO

ROWTHF

Hh

BnO

CH2CI2, Et2O

. l

BnO BnO

21 OGs'Ph

20 OGS'Ph

R- OH

DTBP/Tf20

BnO

'x,,

silyl a d s 21

P-mannopyranosides23

89%

92%

84%

65%

88%

82%

Brio OMe a

BnO, Bno*OMe BnO OH b

&:nB HO Brio OMe

BnO BnO

OR

23

98%

12%

60%

48%

d

HBnO O% Brio OMe e

BnO.

54%

+ @-O-debenzylated product.(as 24) 41%

Scheme 7. Dimethylsilyl-tethered glycosidation in the synthesis of fi-mannopyranosides.

It is worthy of note that only small differences were found between the reactivity of primary and tertiary alcohols a-c in the above glycosidation reactions. The study was therefore extended to include carbohydrate derivatives e, f, and g, with one unprotected hydroxyl group as glycosyl acceptor. It was found that no acid catalyst was needed when the solvent was changed from CH2C12 to MeN02. Thus activation of silyl acetals 26e-26g with NIS in MeN02 gave the desired a-glucopyranosides 28e-28g in good yield. No P-glucopyranosides were detected [21, 221. In addition

456

18 Intramolecular Glycosidation Reactions

"*,.sil BnO BnO B+ O \*OBn

OBn

BnO

'... 1 Tf20, DTBP

OMe

BnO

' g o - O B n

Et20, CH2C12,829:'

OMe

Brio BnO

o/'i'ph 24

2ld

Scheme 8. Side-reaction in dimethylsilyl-tethered glycosidation.

+

AcO AcO

2 ROSiMe&I

-

OAc

Aco%sph

Pyridine

THF, 2h, 25'C

AcO

25

..

Activation of 2694: 2.5 eq.NIS cat. TmH, CH2C12, 10 min.

I

Si

silyl acetals 26

a-glumsides28

15%

59%

Cyclohexyl b

14%

62%

terr-Butyl

16%

61%

54%

12%

R

n-Octyl

'

'OR 26

Activation of 26e-g: 2.5 eq. NIS, MeNO,

a

C

Phenyl d

I

(OBn

66%

BnO* BiO

74%

I

OMe AcO% AcO

O @ OfF

12%

85%

Ho OR

28

39%

Scheme 9. Dimethylsilyl-tethered glycosidation in the synthesis of a-glucopyranosides,

18.2 Reactions in which the Tether Participates in the Reaction

451

AcO MezSiCl2,pyridine

AcO “&SPh

alcohole,73%

*

OH

29

5

OMe

30

I) NIS, MeN02 NIS, MeNOz

OAc

31(31%)

+

AcO

OMe

0 OMe

31 R = Bn (32%) 32 R = H (49%)

Scheme 10. Debenzylation observed in dimethylsilyl-tethered glycosidation.

the intramolecular nature of the glycosidation reaction was proven by means of a competition experiment [ 231. Treatment of thiogalactopyranoside 29 with MezSiCl2 and glucose derivative e gave silyl acetal30 in 73% yield (Scheme 10). Activation of 30 with NIS in MeN02 produced the desired a-galactopyranoside 31 in 320/0yield and, highly unexpectedly, another a-galactopyranoside 32 (identified by careful spectroscopic and chemical studies) which had lost a benzyl group, in 49% yield. Compound 33 could, furthermore, be isolated when the reaction mixture was not quenched with acid, so it was proposed that debenzylation occurred via an intramolecular pathway involving silicon [22]. It has recently been demonstrated that treatment of disaccharide 31 with NIS produces the debenzylated compound 32, probably via a benzylidene intermediate and without participation of silicon [24]. In the glycosidation reactions described above the silyl groups are attached to the 0-2 positions of the glycosyl donors. Bols and Hansen investigated the effect of having the silyl group attached to other positions of the donor [25].This could be a useful strategy because a free hydroxyl group, which could be a target for further glycosidation, is obtained in this type of glycosidation reaction. Thioglucopyranoside 34 was silylated to give 35 in 77% yield. Activation of 35 with NIS in MeN02 gave a 1 : 4 mixture of glucopyranosides 36 in 22% yield (Scheme 11). Higher yields and better selectivity were obtained when the silyl group was anchored to the 0-4 position of thioglucoside derivative 37, giving exclusively the a-

458

I8 Intramolecular Glycosidation Reactions OBn


CISiMe20C8HI,, pyridine SMe HO

* C*H,,OMqSiO BnO-&&

SMe

77%

OBn

OBn

35

34

BnO& HO

0 OBn

36 a$

1:4

Scheme 11. Dimethylsilyl-tethered glycosidation with tethering to the 3 position of the donor.

glucopyranoside 38 in 45% yield (Scheme 12). Silylation of 39 at the 0 - 6 position gave P-glucopyranoside 40 and tribenzyl levoglucosan 41 in 2 :7 ratio. Finally, silylated ribose derivative 42 was treated with NIS in MeN02 to give P-riboside 43 in 63% yield. That complete stereocontrol is observed in the glycosidation of 37, 39, and 42 indicates that this approach could lead to new strategies for oligosaccharide synthesis. Kojitriose has been prepared with complete stereocontrol by use of silicon tethers at two points in the synthesis (Figure 1) [26].

6

C,H,,OMqSiO BnO

NIS, MeNO, >

SMe

--.S;i BnO

45%

BnO

OBn

37

38

OBn

arn

39

40

yJal\sph 63% YYo-

OSiMqOC,H,,

.

.

O 42 2

OBn

41 211

OH

NIS, MeNO,

.

.

O 43 X

Scheme 12. Dimethylsilyl-tethered glycosidation with tethering to the 4, 5 and 6 positions of the donor.

18.3 Reuctions in which the Tether does not Participate in the Reaction

459

HO HO

HO

HO

Figure 1. Kojitriose.

HO

Kojitriose

18.2.2 Tethering to the Leaving Group

In contrast with the above strategies, Schmidt and Behrendt investigated another approach in which the glycosyl acceptor was attached directly to the leaving group [27]. Treatment of 44 with LDA and benzaldehyde resulted in 45 as a 1 : 1 mixture of diastereomers in 90% yield (Scheme 13). Without separation of the diastereomers 45 was reacted with triflates a-c, in the presence of NaH and 15-crown-5 as base, to give predominant P-glucopyranosides 48a-48c and p-lactone 49 in good yields. It was not possible to obtain glycopyranosides 48a-48c directly from 44 and no alkylation on the hydroxyl group in 45 was observed when 45 was treated with triflates a-c, apparently because of steric hindrance. In support of this idea some alkylation was found when the a isomer of 45, which has a less hindered hydroxyl group, was treated with triflate c. It was, therefore, proposed that alkylation occurred at the oxide of the carbonyl group via intermediate 46 and orthoester 47 to produce 48a-48c and 49. A similar approach was taken by Hanaoka et al., who used CO~(CO)S as complexing agent in the glycosidation step [28]. A variation of the above strategy has been applied by Ikegami et al. Decarboxylation of a carbonate tether under acid conditions was claimed to produce intramolecular glycosidic bonds, even though anomeric selectivity was not very high [29]. Schmidt et al. have re-investigated this decarboxylative glycosidation reaction and by means of a competition experiment have found evidence that the mechanism is intermolecular rather than intramolecular [30].

18.3 Reactions in which the Tether does not Participate in the Reaction Ziegler et al. have explored another approach in an attempt to produce p-mannoside [31] or a-glucoside [32] linkages. A succinyl linker between the glycosyl donor and the glycosyl acceptor was used to tether the glycosyl donor and acceptor together. It is noteworthy that the glycosyl acceptor bears a free hydroxyl group.

460

18 Intramolecular Glycosidation Reactions

-

,OBn

-

OBn B O nFo* BnO BnO

44

LDA,THF, PhCHO, -80°C &O -nB BnO

R-OTf

Wh(1:I)

NaH, lkmwn-5 45

O

.B BnO n o

OBn

& c q

Ph

R

-

49

R-OTf

BnO

BnO

BnO

,OBn

-0H

H. 'Ph

46

47

Bno*ORBnO Bnb

48a-c

R-OTf

/OTf

B-rrlucoovranosides

Yield

48a

90%

P

48b

7Ph

P:a8:1

a

C,,H,OTf

b (OTf

Scheme 13. Leaving group-tethered glycosidation.

In contrast to the intramolecular aglycone-delivery strategy the glycosidic bond is here formed directly between the free hydroxyl group and the glycosyl donor producing a bridged disaccharide. Finally the tether can be removed. Treatment of thioglycoside 50 with 10 equivalents of succinic anhydride 51 gave mannopyranoside derivative 52 in almost quantitative yield (Scheme 14). Reaction of 52 with glucose derivative 53 then regioselective opening of the benzylidene acetal with NaCNBH3 gave the tethered saccharide 54 in good yield (Scheme 15). Activation of 54 with NIS and a catalytic amount of TMSOTf gave the bridged a-mannopyranoside 55 in 540/0yield (Scheme 16).

18.3 Reactions in which the Tether does not Participate in the Reaction

461

Scheme 14. Succinoylation of a glycosyl donor.

BnO,

I ) DCC, CH2C12,60%

52

+

HO

2) NaCNBH3 THF HCI in ether, 87%

53

SEt

54

Scheme 15. Tethering with a succinic acid tether.

0

BnO

54

0

55

Scheme 16. Succinic acid-tethered glycosidation.

The effect of connecting the succinyl linker to positions other than 0 - 2 and 0-3’ was also investigated and prearranged saccharides 56 and 57 were prepared by use of the same strategy as described above. Activation with NIS and TMSOTf gave mixtures of bridged a- and b-mannopyranosides 57 and 59 (Scheme 17). There has also been much interest in the formation of P-L-rhamnosides (6-deoxy13-L-mannopyranosides) in which the glycosyl donors and acceptors are tethered with succinyl, malonyl, or phthaloyl bridges [33-351. It has been pointed out that not only the configuration of the glycosyl donors but also the configuration of the glycosyl acceptors plays an important role in the anomeric selectivity obtained. Comparison of reaction 57 + 59 with reaction 57a + 59a, in which the glycosyl acceptor has been changed from a D-glucose derivative to an L-glucose derivative, reveals that the anomeric selectivity has changed significantly (Scheme 18) [36].

462

18 Intramolecular Glycosidation Reactions

Bnr+hr&$, 0

~~0OBn NIS, MeCN,-30°C cat. TMSOTft

BnO

BnO SEt

56 0 NIS, cat.TMSOTf MeCN, -30°C BnO BnO

BnO BnO SEt

45%~ 25%p

0 59

51

Scheme 17. Succinic acid-tethered glycosidation

22Ya

NIS,cat. TMSOTf McCN, -30°C

OBz

BnO SPh 57a

0 59a

Scheme 18. Stereodifferentiation in succinic acid-tethered glycosidation.

It has recently been found that anomeric selectivity is sometimes strongly dependent on the nature of the activating agent. Thus treatment of 60 with MeOTf or NIS resulted in the corresponding a- or p-glycopyranosides, respectively (Scheme 19). It was, furthermore, found that use of galactopyranoside derivatives with a free hydroxyl group in the 4 position as glycosyl acceptors gave P-mannopyranosides exclusively [ 371. One of the main drawbacks of all these strategies is the time-consuming task of synthesizing monosaccharides with the desired protection group patterns. To avoid too much protection-group chemistry Valverde et al. investigated the possibility of having more than one unprotected hydroxyl group on the glycosyl acceptor [38]. Tethered saccharides 61 and 62 were prepared by using phthalic anhydride to tether the glycosyl donor and glycosyl acceptor [39]. Activation with NIS-TfOH resulted in the bridged a-mannopyranoside derivatives 63 and 64,with stereo- and regiocontrol, in good yields. The tether was finally removed and the free hydroxyl groups acetylated to give the corresponding a-mannopyranosides 65 and 66 (Scheme 20).

463

18.3 Reactions in which the Tether does not Purticipate in the Reaction

Brio>

-I 00 -

60

I

OBn

Brio\ 5 eqv. NIS, TMSOTflcat.) CH&N, -3OoC, 10 min.,

OBn

/OBn

BnO

z%

Scheme 19. Dependence of anomeric selectivity on activating agent. (OTBDPS

/ 1)NaOMelMeOH

2) A%OPyridine

OAc OMe

Me0 AcO-

OMe

65

MeO. M

nu

62

64.

n,n/

YLV

Ad)

1OMe

66

,

Scheme 20. Phthalic acid-tethered glycosidation.

Interestingly, it was found that different anchoring positions on the glycosyl acceptor resulted in differences between the selectivity of the free hydroxyl groups. Thus anchoring to the 0-2' position gave the 3' a-glycoside 65 whereas anchoring to the 0-6' position gave the 4' a-glycoside 66. Treatment of 67 with CpzHfClz and AgOTf has been found to give the regioselective intramolecular glycosidation product 68 in 37% yield (Scheme 21) [40]. This regioselectivity was in agreement with Monte Carlo calculations.

I OMe

464

18 Intramolecular Glycosidution Reactions

C&HfClz, AgOTf

CH,CI,, 4A-MS, 31%

68

67

Scheme 21. Phthalic acid-tethered synthesis of a trisaccharide.

a, a’-dibromo-rn-xylene has recently been used to produce the tethered compounds 69 and 70 which, upon activation with NIS/TMSOTf, resulted in the macrocyclic derivatives 71 and 72 containing 14- or 15-membered rings respectively (Scheme 22). In both instances p( 1-4)-linkages were formed exclusively [41].

18.4 Conclusion Since the first report of an intramolecular glycosidation reaction in 1991 chemists have shown much creativity within this area. The new and interesting chemistry

1.3 eq. NIS

BnO BnO OBn

x

x = Pentenyioxy

BnO BnO

BnO BnO

OBn OMe

71

69

X = SEt

1.3 eq. NIS

BnO BnO

1 eq. TMSOTf 84%

SEt

OBn

OBn

70

Scheme 22. rn-Xylene-tetheredglycosidation.

BnO BnO

OBn

OBn

72

References

465

which has been developed reveals much about glycoside reactions and also demonstrates the difficulty in stereoselective formation of glycosidic bonds. For everyday use the PMB method seems very attractive; it is the only method extended to solid-phase chemistry. So far, however, interest has been shown only in the formation of P-mannopyranosides and fLfructofuranosides, whereas the use of silyl tethers been very well documented and successfully used in stereoselective formation of a variety of glycosidic linkages. Methods in which the glycosyl acceptor contains more then one unprotected hydroxyl group have occasionally been shown to be stereoselective but the selectivity is relatively unpredictable. Occasionally, however, theoretical calculations have been consistent with the results obtained.

References 1. P. Preuss, K-H. Jung, R. R. Schmidt, Liebigs Ann., 1992, 377-382. 2. M. Nakata, T. Tamai, T. Kamio, M. Kinoshita, K. Tatsuta, Tetrahedron, 1994, 35, 19, 30093102. 3. M. Nakata, T. Tamai, T. Kamio, M. Kinoshita, K. Tatsuta, Bull. Chem. Soc. Jpn., 1994, 67, 3057- 3066. 4. F. Barresi, 0. Hindsgaul, J. Am. Chem. Soc., 1991, 113, 9376-9378. 5. F. Barresi, 0. Hindsgaul, Synlett, 1992, 759-761. 6. F. Barresi, 0. Hindsgaul, Can. J. Chem., 1994, 72, 1447-1465. 7. Y. lto, T. Ogawa, Angew. Chem. Int. Ed. Engl., 1994,33, 17, 1765-1767. 8. Y. Oikawa, T. Yonemitsu, Tetrahedron Lett., 1982, 23, 885-888. 9. R. Johansson, B. Samuelsson, J. Chem. Soc. Chem. Commun., 1984,201-202. 10. A. Dan, Y. Ito, T. Ogawa, J. Org. Chem., 1995,60,4680-4681. 11. A. Dan, Y. Ito, T. Ogawa, Tetrahedron Lett., 1995, 36, 41, 7487-7490. 12. A. Dan, M. Lergenmiiller, M. Amano, Y. Nakahara, T. Ogawa, Y. Ito, Chem. Eur. J., 1998,4, 11, 2182-2190. 13. Y. Ito, Y. Ohnishi, T. Ogawa, Y. Nakahara, Synlett, 1998, 1102-1 104. 14. Y. Ito, T. Ogawa, J. Am. Chem. Soc., 1997, 119, 5562-5566. 15. C. Krog-Jensen, S. Oscarson, J. Org. Chrm., 1996, 61, 4512-4513. 16. C. Krog-Jensen, S. Oscarson, J. Org. Chem., 1998, 63, 1780-1784. 17. M. Bols, T. Skrydstrup, Chem. Rev., 1995, 95, 1253-1277. 18. G. Stork, G. Kim, J. Am. Chem. SOC.,1992,114, 1087-1088. 19. G. Stork, J. J. la Clair, J. Am. Chem. SOL..,1996, 118, 247-248. 20. M. Bols, J. Chem. SOC. Chem. Commun., 1992,913-914. 21. M. Bols, J. Chem. Soc. Chem. Commun., 1993, 791-792. 22. M. Bols, Tetrahedron, 1993, 49. 44, 10049-10060. 23. M. Bols, Acta Chern. Scand., 1993, 47, 829-834. 24. J. Madsen, M. Bols, Angew. Chem., 1998, 110, 22, 3348-3350. 25. M. Bols, H. C. Hansen, Chem. Lett., 1994, 1049-1052. 26. M. Bols, Acta Chem. Scand., 1996, SO, 931-937. 27. M. E. Behrendt, R. R. Schmidt, Tetrahedron Lett., 1993, 34, 42, 6733-6736. 28. C. Mukai, T. Itoh, M. Hanaoka, Tetrahedron Lett., 1997, 38, 26, 4595-4598. 29. T. Iimori, T. Shibazaki, S. Ikegami, Tetrahedron Lett., 1996, 37, 13, 2267-2270. 30. G. Scheffler, R. R. Schmidt, Tetrahedron Lett., 38, 17, 2943-2946. 31. T. Ziegler, G. Lemanski, A. Rakoczy, Tetrahedron Lett., 1995, 36, 49, 8973-8976. 32. T. Ziegler, A. Ritter, J. Huttlen, Tetrahedron Lett., 1997, 38, 21, 3715-3718. 33. T. Ziegler, R. Lau, Tetrahedron Lett., 1995, 36, 9, 1417-1420. 34. R. Lau, G. Schiile, U. Schwaneberg, T. Ziegler, Liebigs Ann., 1995, 1745-1754. 35. G. Schule, T. Ziegler, Liebigs Ann., 1996, 1599-1607.

466

18 Intramolecular Glycosidution Reactions

36. T. Ziegler, G. Lemanski, Eur. J. Chem., 1998, 163-170. 37. T. Ziegler, G. Lemanski, Angew. Chem., 1998,110, 22, 3367-3369. 38. S. Valverde, A. M. Gomez, A. Hernandez, B. Herradon, J. C. Lopez, J. Chem. SOC. Chem. Commun., 1995, 2005-2006. 39. S. Valverde, A. M. Gomez, J. C. Lopez, B. Herradon, Tetrahedron Lett., 1996, 37, 7, 11051108. 40. H. Yamada, K. Imamura, T. Takahashi, Tetrahedron Lett., 1997,38, 3, 391-394. 41. U. Huchel, R. R. Schmidt, Tetrahedron Lett., 1998,39, 7693-7694.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

19 Classics In Total Synthesis of Oligosaccharides and G1ycoconjugates Jean-Maurice Mullet and Pierre Sinuy

19.1 Introduction Several chapters in this volume illustrate various selected methods which have been developed over the years to achieve efficient glycosylation reactions. It is quite amazing to witness the progress which has been made in the art of oligosaccharide construction, probably the most central challenge in carbohydrate chemistry. Our aim in this brief chapter is to highlight these achievements and illustrate various strategical aspects such as the judicious choice of protecting groups and of the activation of the anomeric centre, while describing a few ‘classic’ syntheses of complex oligosaccharides of potential biological importance. One inherent difficulty associated with such a chapter is the selection of what should be considered as ‘classic’ in the art of synthetic oligosaccharide construction. We are sure that the carbohydrate community will understand that many alternative brilliant accomplishments may have equally been selected.

19.2 Syntheses of Nod factors 19.2.1 Introduction

The two Nod factors 1 and 2 (Scheme 1) are secreted by bacteria fixing atmospheric nitrogen [ l , 21. The synthetic challenge associated with 1 or 2 can be summarized as follows: how to selectively differentiate, as shown in Scheme 2, hydroxyl and amino groups present in a chitin type fragment. More precisely, the strategical problems inherent to the total synthesis of Nod factors are as follows:

468

19 Classics In Total Synthesis of Oligosaccharides and Glycoconjugates OS03Na

H HO O&o&,O& HO O~oH HN

AcHN

HO

HO AcHN

-

O

1 2

AcHN

A

R=H R=Ac

Scheme 1. The structure of two classical Nod factors.

0-acetylation (Nod 2 only)

sulfation

HO HO NH2

NH2

NH2

Scheme 2. What has to be achieved to build up the Nod factors.

1. The four primary hydroxyl groups have to be carefully differentiated; the presence of a sulfate group and, in compound 2, of an additional acetate, forces the introduction of suitable temporary protecting groups on the primary positions. 2. Similarly, one amino group has also to be differentiated from the others. 3. Last, but not least, the presence of double bonds in the final product precludes the classical use of hydrogenation conditions, subsequent to the introduction of the unsaturated chain. This means for example that the very classical benzyl ether cannot be used as a protecting group in the last steps. This kind of problem is similar to the one met in the synthesis of sphingoglycolipids where a double bound is also present in the sphingosine moiety. 19.2.2 The K. C. Nicolaou Synthesis (1992) [3] The key protected tetrasaccharide to be synthesized in this approach is compound 3, the retrosynthetic analysis being presented in Scheme 3. This strategy illustrates the classical elongation process from three monosaccharidic blocks 4, 5 and 6, starting from the protected reducing end, that is compound 4. The three required blocks 4, 5 and 6 are easily prepared from a common glucosamine precursor 7 (Scheme 4). There are two noticeable features in the assembly of the monosaccharidic blocks which have to be commented for the benefit of the practitioner. First the use of the

469

19.2 Syntheses of Nod factors ,OTBDMS

OTBDPS ' MB &$LOMP NPhth

M B & M & M & M & o M p

MB

NPhth

AcHN

AcHN

AcHN

3

,OMP F

,

$ &

- O c A MB NPhth 5

Scheme 3. Retrosynthetic analysis of the Nicolaou's strategy. Abbreviations: MP (p-methoxyphenyl), MB (p-methoxybenzyl), NPhth (N-phtalimido), TBDMS (tertbutyldimethylsilyl), TBDPS (tertbutyldiphenylsilyl).

phtalimido group (see specific comments in Box 1) and second the use of glycosyl fluorides for the construction of the glycosidic linkage. The combination Cp2MC12AgC104 (M = Zr, Hf) is highly efficient for the activation of glycosyl fluorides [4], Zr-system exhibiting the most powerful reactivity. In this piece of work, the silver perchlorate has been replaced by silver triflate (Scheme 5). The blockwise synthesis selected here allows the transformation in due time of three phtalimido groups into three N-acetyl groups (see compound 3) and of the last one into the required N-acyl group: the problem of differentiation of one amino group among four is thus easily solved, as illustated in scheme 6. The rational for the choice of the various protecting groups present in blocks 4, 5 and 6 - that is

A Ac c O A S P h

Aco&oAc Ac

NPhth

a

1

NPhth 7

steps

5 and 6

+1:n

LAcoAc

9

OMe

steps

4

Scheme 4. Reagents and conditions: a) PhSH, SnC14, CH2C12, 25"C, 74%; b) MeOC6€&0H, SnC14, CHzC12, 25 "C, 72%.

470

19 Classics In Total Synthesis of Oligosaccharides and Glycoconjugates

"'AF

OTBDPS

MB

+

OTBDPS

5

H*MB oMp

NPhth

4

5

AcO&O&& MB

NPhth

MB NPhth

NPhthOMP

OTBDPS

MB

NPhth 5

MB

+ MB

MB NPhth

6

A NPhth M

d

O NPhth M

P

11

OMP NPhth

MB NPhth

Id

12

H&ooMP&oMp

MB NPhth

M

Ic

Aco&&e

FS. . .

H

+

MB

MB

AcHN

AcHN

AcHN

13

Ie

M M & BM &M &G MB

OMP NPhth

AcHN

AcHN

AcHN

3

Scheme 5. Reagents and conditions: a) 5, AgOTf, Cp2ZrCl2, 2,6-ditertbutyl-4-methyl pyridine, MS 4A, CHzC12, 0-25"C, 56%; b) K2C03, THF, 25"C, MeOH; c) 5, AgOTf, Cp2HfCl2, 2,6-ditertbutyl-4-methyl pyridine, MS 4A, CH2C12, 0-25"C, 60%; d) N2H4, EtOH, PhH, 100°C then AczO, MeOH, CH2C12, 72%; e) 6, AgOTf, CpzHfCl2, 2,6-ditertbutyl-4-methyl pyridine, MS 4A, CH2C12,0-25 "C, 50%.

in the resulting protected tetrasaccharide 3 - is fully apparent when contemplating the chemical conversion of 3 into the two final targets 1 and 2, as illustrated in Scheme 6 . Final comments: An interesting feature of the Nicolaou's strategy is the judicious choice of three different protecting groups for the primary alcoholic positions, which on one hand withstand the glycosylation conditions, and on the other hand can be sequentially removed. Such a nicely tuned strategy indeed allows the only synthetic access described so far to the acetylated Nod factor 2. Weaker points are the moderate yields resulting from the use of the fluoride glycosylation method and

19.2 Syntheses of Nod factors M B o g M & M & M & o M p

a

3

47 1

MB NHZ

AcHN ,4

AcHN

AcHN

M B o g M & MB&M&M&Mm?DFg -

RCOOH

MB

b

HN 0

-

AcHN

AcHN

-

AcHN

15

C

OMP

MB

H 0

-

AcHN

AcHN

-

AcHN

16

d

M B o & M & M & M & OMP

MB HN - - - - - -O

e --t

2 (Nod-factor 2)

AcHN

f

AcHN

-

AcHN

17 1 (Nod-factor 1)

Scheme 6. Reagents and conditions: a) N2H4, EtOH, lOO"C, 87%; b) RCOOH (3eq), 2-chloro-lmethyl pyridinium iodide, Et3N, CH3CN, 25 "C, 73%; c ) PPTS, EtOH, 25 "C then Ac20, Et3N, DMAP (cat), CH2C12, 2 5 T , 72%; d) TBAF, THF then Me3NSO3, Pyr. 25"C, 85%; e) CAN, CH3CN, H20, 25"C, 30%, f ) MeONa, MeOH, 2 5 T , 75%.

the rather low yield of the simultaneous removal of MB and MP protecting groups on a complex molecule. 19.2.3 The J.-M. Beau Synthesis (1994) 112)

This is another example of synthesis of an oligosaccharide through stepwise addition of monosaccharide blocks. But in sharp contrast to the previously described achievement, it is representative of a nonconventional construction of a linear oligosaccharide starting from the nonreducing end of the oligomer. The protected tetrasaccharide is built up from two precursors as shown in Scheme 7. The detailed synthetic route is shown in Scheme 8.

472

19 Classics In Total Synthesis of Oligosaccharides and Glycoconjugates

19.2 Syntheses of Nod factors ,OPiv

l

,OPiv

,OPiv

473

,OAc

18

,OAc

Scheme 7. Retrosynthetic analysis of the Beau’s strategy. Abbreviations: Z (benzyloxycarbonyl), Bn (benzyl), Piv (Pivaloyl).

The 1,2-trans selectivity of the first glycosylation reaction is insured by a carbamate participating group, which is not so frequently used (see Box 2) compared to the now very classical phthalimido group. It is worth noting that the acetolysis of l16-anhydrosugars is a rather smooth process, compatible with glycosidic linkages present in the molecule. Such a behaviour qualifies the 1,6-anhydro ring for this rather atypical type of left to right elongation. A feature of this strategy is the daring choice of a nonparticipating azido group at C-2 to successfully achieve a 1,2-trans glycosylation through a clean SN2 process. There are now more and more examples of this, so that such an option should not be disregarded in a synthetic strategy (see Box 3). Looking at the final steps of this strategy (Scheme 9), it is apparent that the fatty acid chain is introduced at the very last step, permitting the use of benzyl ethers as temporary blocking groups. Final comments: It has been stated in the previous section (see Box 1) that N-phthalimido group insures a perfect (3-selectivity, but is sometimes difficult to remove. Especially, such a removal is not compatible with acetates. The J.-M. Beau synthesis highlights an attractive option: the use of the nonparticipating azido group, the selectivity being intrinsically insured by a clean S N process ~ under appropriate experimental conditions. 19.2.4 The T. Ogawa Synthesis (1994) 1161

The strategy designed by T. Ogawa et al. uses three blocks, one being a disaccharide donor (Scheme 10). This strategy (Scheme 10) is rather close to that described by K. C. Nicolaou. The central block 37 has been prepared from a chitobiose derivative (171. The option of the N-phthalimido participating group has also been selected (see Box 1). The in-

19 Classics In Total Synthesis of Oligosaccharides and Glycoconjugates

414

B KBn r & & , ZHN

&g a

+

OCNHCC13

OH

19

C

N3

N3

21R=Ac 22R=Piv

NHZ

-

B"&m+

Bn&o+

d

Bn

NHZ

NHZ

N3 OAc

23

N3 OCNHCCI,

24

R

20

0

__t

NHZ

e

bl

Bn

20

Bn

*n&O

25

B Bn n & B * BNHZ 4

N3

f

N,

20 __t

'

26

"&&Bn ~&o

'"OCNHCCI,

9

N3 NHZ

N3

h

1

27R=Ac 28R=Piv

N3

Scheme 8. Reagents and conditions: a) BF3 EtzO, Toluene, -78°C > -2O"C, 91%; b) MeONa, MeOH then PivC1, DMAP, Pyr. 2 0 T , 100%; c) Ac20, TFA, rt; d) BnNH2, Et20, rt, then CCl3CN, DBU, CH2C12, rt, 73% 3 steps; e) 20, BF3 EtzO, CHzC12 -78°C > -2O"C, 85%; f ) Ac20, TFA, rt; 73%; BnNH2, Et20 rt, then CC13CN, DBU, CH2C12, rt, 73% 3 steps, g) 20, BF3 Et20, CH2C12 -78°C > -2O"C, 91%; h) MeONa, MeOH then PivC1, DMAP, Pyr. 20"C, 100%.

troduction of the unsaturated fatty acid is performed at the very last step, a feature which allows the initial protection with benzyl ethers (see also J.-M. Beau). The protected tetrasaccharide is obtained as described in Scheme 11. The deprotected tetrasaccharide 34 is easily obtained as depicted in Scheme 12 and is identical to the one previously obtained by J.-M. Beau. Final comments: As persistently stated in this chapter, the stability of protecting groups during the removal of N-phthalimido has to be carefully considered while selecting a strategy. This work points out the stability of a primary sulfate function during hydrazinolysis of the N-phthalimido group (Scheme 12).

19.2 Syntheses of Nod factors

A

A

C

O

S

S

~+

HJ

COW

NH-CO-Ct+CCI,

EE rt

-r

415

AK Al-Jg&Jcorm

NH-CC-CH&CI~ 85%

19.2.5 The Y. Z. Hui Synthesis (1992) [lS] The strategy used by Hui et al. illustrates the so-called block synthesis, more precisely the synthesis of a protected tetrasaccharide from two appropriate disaccharidic moieties, as shown in Scheme 13. The disaccharidic glycosyl donor 49 was first prepared as described in Scheme 14. As previously pointed out in Box 3 and exemplified in J.-M. Beau et al. ~ approach, it is possible to achieve a 1,2-trans glycosylation through an S N process,

19 Classics In Total Synthesis of Oligosaccharides and Glycoconjugates

476

a

28

OR NHZ

C

N3

bJ

N3

N3

N3

N3

NHAc

OBn NHAc

NHAc

NHAc

NHAc

NHAc

29 R= AC 30 R = OCNHCC13

B n & B & & B &

OBn NHZ

N3 18

d

Bn&B&&Boi& NHZ

-

NHAc

Bn&B&&Boscg

31

.

8

OBn NHZ

NHAc

32

-----t f

Bn&Bn &B&&OBn NHZ

NHAc

33

-

H & & & &

9

OH NH2

AcHN

AcHN

h

AcHN

34

Scheme 9. Reagents and conditions: a) TFA, Ac20, 81%; b) BnNH2, Et20 then CC13CN, DBU, CH2C12;c) BnOH, BF3.Et20,toluene, 79%; d) NaBH4, NiC12, DME-EtOH, then Ac20, Fyr.; e) MeONa, MeOH, 52%; Me3NS03, DMF, 50"C, 98%; f ) MeONa, MeOH, 88% g) H2, Pd/C AcOEt, MeOH, H20; h) N-acylation 60%.

using a nonparticipating azido group at C-2. Another feature of this synthesis is the very classical choice of the ally1 group as a temporary protecting group of an anomeric centre. The disaccharidic acceptor 50 is obtained as described in Scheme 15. A noticeable aspect of this synthetic sequence is the use of a monochloroacetate protecting group, which is easily removed under mild conditions fully compatible with N-phtalimido and pivaloyl groups.

19.2 Syntheses of Nod factors

471

AcO AcO NPht

NHAc

NHAc

fl AcO AcO &F

NHAc

35

A BnO c o & O ~ CBnO N H C C I 3 NPht

NPht 37

36

HBnO o g

NPht

o

B

n

38 NPht

Scheme 10. Retrosynthetic analysis of the Ogawa's strategy.

AcO BnO

CNHCC13

+

-

HO BnO

MP

NPht

BnO NPht

NPht

39

40

41

NPht

PMP

b

I t "

AcO AcO*O*CNHCC13 BnO Rnn BnO &O&CNHCCl3 BnO NPht 37

AcO BnO NPht

NPht

d

OBnAcOAcO +F

BnO NHAc

*

-

A0&.& BnO NHAc

n

C

OBn NPht

NPht

HO

BnO

-

Ho*OBn4 O B Brio Brio 38 NPht

U

NPht

*

NHAc

e

43

35

Scheme 11. Reagents and conditions: a) BF3.Et20, CH2C12, -78 "C?82%; b) CAN, CH3CN, H20, then CC13CN, DBU; c) BF3.Et20, CH2C12, -78"C, 69%; d) N2H4, EtOH, then Ac20, MeOH, 99%; e) Cp2HfCl2, AgOTf, (CH2Cl)2,79%.

I9 Classics In Total Synthesis of Oligosaccharides and Glycoconjugates

478

35

a

AcO AcO NPht

NHAc

NHAc

OBn NHAc

NH2

NHAc 46

NHAc

OBn NHAc

1

45 R= S03Na

C

HO HO

d

HOHO &o

- & o &HOo g o H NH2

HO

AcHN

HO AcHN

AcHN

34

R4-h 47 0

Scheme 12. Reagents and conditions: a) CAN, CH3CN, H20; b) Et3NSO3, DMF, 50"C, 70% two steps, c) N2H4, EtOH; d) Pd-black, Hz, MeOH, H20, 88% two steps; e) MeOH, Hz0, 97%.

BnO BnO

N3

NPhth

BnO

Bn NPhth

NPhth

48

BnO BnO

-C(NH)CC13 N3

HBnO O & A OBnO B n

NPhth 49

Scheme 13. Retrosynthetic analysis of the Hui's strategy.

NPhth

50

NPhth

19.2 Syntheses of Nod factors

BnO% BnO

+

51

b

bC(NH)CC13

BnO BnO - &

a

HBnO O G , , ,

Bno&o&Al~ BnO

BnO

NPhth

N3

NPhth

52

& N3 O BnO 54 a

479

53

H NPhth L

BnO BnO &O&

N3 BnO

NPhth I

55

BnO-*BnO

BnO

NPhthO-C(NH)CCI,

N3 49

Scheme 14. Reagents and conditions: a) BF,.Et20, MS 4p\, CH2C12, -20"C, 61%; b) PdCI2, NaOAc, aq AcOH, 5O"C, 84%; c) (COC1)2,DMF (cat), 88%; d) CC13CN, DBU, CH2C12, 84%.

57

58

BnO

BnO

NPhth

59

O-C(NH)CC13 NPhth

60

DPiv Ho BnO *OBn

'\

61

HO BnO

BnO NPhth 50

NPhth

OBn NPhth

NPhth

62

Scheme 15. Reagents and conditions: a) ClCH2COC1, Pyr. O T , 88%; b) PdC12, NaOAc, aq AcOH, 50"C, 89%; c) (COC1)2,DMF (cat) ; 89%; d) CC13CN, DBU, CHzC12,85%;e) AgOTf, MS 4A, CHzC12, -1O"C, 58%; f ) BF3.Etz0, CHZC12, -15"C, 76%; g) Thiourea, Pyr, EtOH, 7 0 T , 86Yo.

480

19 Classics In Total Synthesis of Oligosaccharides and Glycoconjugates

The final steps of the synthesis are shown in Scheme 16. The imidate 49 appeared as a better glycosylationg agent than the chloride 55. It has to be pointed out that the final amine 34 is the one previously obtained by J.-M. Beau and T. Ogawa. Final comments: A recurrent theme in the synthesis of Nod factors is the differentiation of the terminal amino group. It is achieved here in a very simple way, using a combination of azido and N-phthalimido groups.

19.2.6 Conclusion There are always two questions when launching an oligosaccharide synthesis: 1. which protecting groups have to be used. 2. which glycosylation methods have to be selected. There are many glycosylating agents developed so far. Nevertheless, only two methods have been selected by the four groups: the fluoride method and the trichloroacetimidate method. As far as the protecting groups are concerned, a larger diversity has been exploited. A careful comparison of these four achievements illustrates many facets of the art of oligosaccharide synthesis.

19.3 Synthesis of the Antithrombin-Binding Pentasaccharide Sequence in Heparin (1984) [19, 201 19.3.1 Introduction Heparin is an extraordinary complex sulfated glycosaminoglycan (GAG) with well known anticoagulant and antithrombotic activities. The coagulation of blood is the result of a cascade of events implying a series of proteases called activated bloodcoagulation factors. Antithrombin (AT) is the main physiological inhibitor of these factors and heparin is known to bind to AT and to amplify its action. A breakthrough in heparinology came in the early 1980’s with the recognition [21, 221 of a specific pentasaccharide sequence within the polysaccharide, capable of binding to AT. The total synthesis of this pentasaccharide 67 (Scheme 17) has first been reported in 1984 [ 191, confirming the specific nature of the interaction heparin/AT. We can consider this selected piece of work as a typical classic because:

1. This is the first synthesis of an oligosaccharidic fragment of heparin with a prominent biological activity [23]. 2. It opened the door to a series of syntheses of other biologically active compounds containing variants of this pentasaccharide.

19.3 Synthesis of the Antithrombin-Binding Pentusacchuride Sequence 55

+

48 1

50

49 +50

b

BnO BnO N3

NPhth

NPhth

OBn NPhth

63

C

BnO BnO

N3

BnO

NHAc

NHAc

NHAc

NHAc

OBn NHAc

64 d __t

BnO BnO

N3

e

NHAc

65 HOHO +o

& o &HO o~:H NH2

HO

AcHN

L1

HO AcHN

AcHN

34

Scheme 16. Reagents and conditions: a) AgOTf, MS 4A, CH2C12, -2O"C, 50%; b) BF,.Et20, CH2C12, -15"C, 72%; c) N2H4, H20, EtOH, 9 5 T , then AczO, Pyr., 2 0 T , then KOH, MeOH, THF 20"C, 66% in 3 steps; d) Pyr.SO3, DMF, 50°C; e) Hz, 10% Pd/C, THF, EtOH, H20, 83%; f ) 66, Et3N (cat), EtOH, 45 "C, 45%.

3. This synthetic pentasaccharide and analogues are potential antithrombotic drugs devoid of certain detrimental side effects when compared with natural standard heparin [24].

19.3.2 An Overview of the Synthesis of the Protected Pentasaccharide 73 The synthesis, which is summarized in Scheme 18, is based on the use of three blocks: two disaccharides and one monosaccharide. The hydroxyl groups which are going to be sulfated are protected by acetyl groups, the hydroxyl groups being free in the final pentasaccharide are protected as benzyl ethers.

482

19 Classics In Total Synthesis of Oligosucchurides and Glycoconjugutes

OH

67

oso3-

Scheme 17. The structure of the specific heparin pentasaccharide sequence binding to AT.

19.3.3 Synthesis of the Disaccharidic Bromide Donor 68 The preparation of this bromide is described in Scheme 19. The bromide 75 is easily prepared from the corresponding hemiacetal using the Vilsmeier reagent. This represents a mild and convenient way to achieve this kind of transformation (see Box 4). The selective formation of a P-glucoside from 75 goes through a so called push-pull mechanism, that is an SN?process at the surface of an insoluble silver salt such as silver carbonate. This method complements the combination a-trichloroacetimidate/ BF3.Et20, which has been put into practice in the previous classic of this chapter.

19.3.4 Synthesis of the Disaccharidic Acceptor 69 The glycosylating species which has been used is the orthoester 79 (Scheme 20). This derivative of methyl L-iduronate is an interesting compound. The tertbutyl orthoester fulfills three functions: it insures the protection of positions 1 and 2, a necessary step for the selective synthesis of 81; it reacts with an alcohol under acidic conditions to give a 1,2-trans glycoside, with the concomitant introduction of an acetate function at position 2 (a progenitor of the sulfate group). This is an example of the orthoester procedure nicely developed by the Kotchetkov group in the 1960s [28]. Such a method is rarely used nowadays. The synthesis of 69 is presented in Scheme 20: the orthoester glycosylation is followed by the selective removal of the monochloracetate on position 4.

19.3.5 Synthesis of the Protected Pentasaccharide 73 The two glycosylation reactions 68 & 69 then 71 & 72 (see Scheme 18) illustrate the use of a bromide, in the presence of silver triflate, to synthesize an a-glycoside. It should be noted that the presence of a participating acetyl group at C-6 probably insures an excellent a-selectivity (Scheme 2 1). Without this participation, the re-

19.3 Synthesis of the Antithrombin-Binding Pentasaccharide Sequence

M Bn c A & A 4 BnO

q i

+ 68

N3Br

483

HO

1

69 ZHN OBn

OAc

a

O , Ac

70

I OAc

I b

t

.OAc

,OAc

72

lc 73

I

OAc

I OAc

Scheme 18. Reagents and conditions: a) AgOTf, collidine, CHZC12, -2O"C, 55%; b) thiourea, pyr. EtOH, 100 "C, 82%; c) AgOTf, collidine, CH2C12, -20 "C, 70%.

sulting selectivity is more difficult to predict and relies on subtle factors, such the nature of the solvent and the temperature. 19.3.6 Synthesis of the Active Site of Heparin

The final steps are shown in Scheme 22. This is now a classical sequence of reactions in the field of glycosaminoglycans. Final comments: With eight sulfonato groups on specific oxygen and nitrogen atoms of the molecule, with the presence of the very unusual L-iduronic acid residue, this molecule represents a formidable synthetic challenge. It is amazing to ob-

484

19 Classics In Total Synthesis of Oligosaccharides and Glycoconjugates

Bn

OH

BnO

-

Bn

Bn

---+

M~A&~& Bn

C

BnO

75

74

-

b 77

68

d

BnO 78

N3 O

A ~

Scheme 19. Reagents and conditions: BrCH=N+Me*, Br-, collidine, CH2C12, 0 "C, 90%; b) Ag2C03, MS 4 A , CH2C12, rt,70%; c) TFA, Ac20, rt, 86%; d) TiBr4, CH,Cl,/AcOEt, rt.

Box 4 Preparation of anomeric bromides

,OAc

79

\

80

Scheme 20. Reagents and conditions: a) 2,6-dimethylpyridinium perchlorate, PhCl, 130 "C, 40%; b) thiourea, Pyr. EtOH, lOO"C, 86%.

19.4 Totul Synthesis of VIM-2 Ganglioside

485

Scheme 21. A possible remote acetyl group participation in a glycosylation reaction.

,OAc

Scheme 22. Reagents and conditions: a) NaOH, CHCI3, MeOH, H20, rt then CH2N2 73%; b) Me3N:S03, DMF, 5 0 T , 90%; c) NaOH, MeOH, H20, 90%; d) H2, 10% Pd/C, MeOH,

H20, 93%; e) Me3N:SO3, H20 (pH = 9.5), 36%

serve that closely related analogues of this compound have now been prepared in bulk quantities as drug substances. Following excellent pharmacokinetics and tolerance studies in man [29], clinical investigations are currently underway to confirm the therapeutic potential of such complex carbohydrate derivatives. This is an excellent example of the power of organic synthesis in the field of glycoscience [30].

486

19 Classics In Total Synthesis of Oligosaccharides and Glycoconjugates

H

84

Scheme 23. Structure of a natural human sphingoglycolipid called VIM-2.

OTBDPS

f'

p- glycosylation

HO-p-13H27 N3

c

85

Reduction and fatty acid amidification

Scheme 24. Overall strategy used by Hasegawa et al. to synthesize VIM-2.

19.4 Total Synthesis of VIM-2 Ganglioside [31] 19.4.1 Introduction The so called VIM-2 ganglioside 84 (Scheme 23) is a natural sphingoglycolipid which is regarded as a tumor associated carbohydrate antigen. It has been isolated from human chronic myelogenous leukemia cells. Many compounds of this type have been synthesized. We would like to present in this section the achievement of the late Hasegawa. It is a typical example of the block synthesis approach of a sphingoglycolipid. 19.4.2 The Total Synthesis of VIM-2-a

General Strategy

A key step of the strategy is the glycosylation of compound 85 [32] (Scheme 24) with a complex glycosyl donor which is fully acylated. The presence of any benzyl ether protecting group is precluded at this point of the synthesis, due to the presence of an easily reducible double bond in compound 85; catalytic hydrogenolysis should be avoided. In the case of VIM-2, the fully acylated glycosyl donor is the trichloroacetimidate 87, which has been obtained from the 2- (trimethylsilyl)ethyl glycoside 86 (Scheme25). The 2-(trimethylsilyl) ethyl glycoside, popularized by G. Magnusson 1331 is often used as a temporary protecting group of the anomeric position, and is easily trans-

19.4 Total Synthesis of VIM-2 Ganglioside B

z

W

~

Bzo' OBz

OAcAc

NHAc 86

a, b

~

Ac OAc

487

&

~

dOAc 1 NHAc

OAcAc

0 OAc

OAc OAc

Scheme 25. Reagents and conditions: a) CF3COOH/CH2C12; b) CC13CN, DBU.

formed in a two step sequence here exemplified in an efficient glycosyl donor. It nicely complements in this respect the ally1 group (see the Hui's synthesis of the Nod factor).The main part of the synthesis is thus the preparation of the 2-(trimethylsilyl) ethyl glycoside 86. Let's now analyse the main characteristics of this achievement. 19.4.3 Preparation of the Key Protected Octasaccharide 86

A protected trisaccharidic Lewis X thiophenyl glycoside 92 is prepared as shown in Scheme 26. Several features of this synthesis have to be pointed out. In the diol89, the position 4 is more reactive than position 3, so that there is no need for a temporary protection at position 3, an observation which simplifies the synthetic scheme. This is due to the presence of the bulky neighbouring phthalimido group. This diol is thus an interesting pivotal acceptor for a double glycosylation sequence. The thiophenyl glycoside 89 can be glycosylated with the bromide 88 (so called orthogonality, which has not always been perfectly observed in this case). The bromide 88 is benzylated at the primary position, a feature which will be commented later on. The thiophenyl Lewis X trisaccharide 92 is now condensed without any problem with the easily available alcohol 93 to give the pentasaccharide 94, as shown in Scheme 27. De-0-benzoylation of 94 gave the triol 95. This triol is next glycosylated with the previously reported thiophenyl glycoside 90, as shown in Scheme 28. This critical step of the synthesis demonstrates that the

./

32

488

19 Classics In Total Synthesis of Oligosaccharides and Glycoconjugates

+

H&SPh

Bz

NPhth

Bzo Br

aa

a9

a

BZ 90

Bzs G

b

Bz

h

OBz

?

O

B

M

T

h

n

OBn OBn 92

Scheme 26. Reagents and conditions: a) AgOTf, toluene, -4O"C, 75%; b) NIS, TfOH, toluene, - 15 "C, 93%.

93

W

O

B

n

94

OBn OBn

Scheme 27. Reagents and conditions: a) NIS, TfOH, CH&, 98%.

-6O"C, 86%; b) MeONa, MeOH,

19.4 Total Synthesis of VIM-2 Ganglioside

90+95

-

489

a

BZ OBn

96 b

NHAc OBn

97

Scheme 28. Reagents and conditions: a) NIS, TfOH, CH2C12, -6O"C, 83% b) N2H4, EtOH, H 2 0 , reflux, then AczO, MeOH, 40 "C, 93%.

trio1 95 can be selectively glycosylated at position 3 , a feature which again considerably simplifies the strategy. We now understand why the primary position of the terminal galactosyl residue should be benzylated, and not benzoylated (see compound 88). The more reactive primary hydroxyl group has to be protected by a persistent blocking group. At this stage, the N-phthalimido groups are converted into N-acetyl groups, an operation which is more problematic when the sialyl residue is present. The sialylation of the polyol 97 is finally most easily achieved using the sialyl donor 98, as shown in Scheme 29. It is remarkable that the regioselective sialylation of the required hydroxyl group is achieved in 72% yield, in the presence of several non protected hydroxyl groups. Final steps are now uneventful and are shown in the last Scheme 30. Final comments: Such an overall strategy can be utilized to synthesize a variety of oligosaccharides of the polylactosamine type, with additional possible L-fucose residues. It is thus of general interest and can be considered as a classic. Many considerations in this synthesis have also been delineated by other groups, and this achievement reflects a nice strategical manipulation of many interesting observations done over the years.

19.5 Epilogue The aim of this chapter was to convince the reader that the chemical organic synthesis in solution of oligosaccharides of any complexity is now feasible. More than

490

19 Classics In Total Synthesis of Oligosaccharides and Glycoconjugates

97 +

BzM C02Me

a_

OBz

98

AcH

99

b

86

Scheme 29. Reagents and conditions: NIS, TfOH, CH3CN, -43"C, 72%).b) H2, Pd/C, MeOH/ EtOH/AcOH, then Ac20, Pyr., 63%.

85

+

87

AcH

OAc OAc

100

Scheme 30. Reagents and conditions: a) TMSOTf, CH2C12, 0"C, 47%; b) Ph3P, PhH, H20; c) octadecanoic acid, 2-chloro-1,3-dimethylimidazolium chloride, PhH. 80%; d) NBQF, CH3CN; e) MeONa, MeOH then water, 87%.

enough glycosylation methods are now available, and a great deal is known about the control of the selectivity. Nowadays, organic chemists are more than ever in a position to collaborate with glycobiologists, to contribute to the development of gly coscience. References 1. P. Lerouge, P. Roche, C. Faucher, F. Maillet, G. Truchet, J.-C. Promt, J. Denarie, Nature, 1990,344, 781.

References

49 1

2. P. Roche, P. Lerouge, C. Ponthus, J.-C. Prome, J. Biol. Chem., 1991,266, 10933; G. Truche, P. Roche, P. Lerouge, J. Vasse, S. Camut, F. De Billy, J-C. Prome, J. Denarie, Nature, 1991, 351, 670; M. Schultze, B. Quiclet-Sire, E. Kondorosi, H. Virelizier, J. N. Glushka, G. Endre, S. Gero, A. Kondorosi, Proc. Nat. Acud. Sci. USA, 1992, 89, 192. 3. K. C. Nicolaou, N. J. Bockovich, D. R. Carcanague, C. W. Hummel, L. F. Even. J. Am. Chem. Soc. 1992,114, 8701. 4. T. Matsumoto, H. Maeta, K. Suzuki, G.4. Tsuchihashi, Tetrahedron Lett., 1988, 29, 3567. 5. S. Akira, T. Osawa, Chem. Pharm. Bull., 1960, 8, 583. 6. R. U. Lemieux, T. Takeda, B. Y. Chung, In Synthetic Methods For Carbohydrates, Ed. Hs. El. Khadem, ACS Symposium Series. 39, 1976, 99. 7. P. L. Durette, E. P. Meitzner, T. Y. Shen, Tetruhedron Lett., 1979, 4013. 8. 0. Kanie, S. C. Crawley, M. M. Palcic, 0. Hindsgaul, Curbohydr. Res. 1993, 243, 139. 9. D. R. Bundle, S. Josephson. Can. J. Chem., 1979, 57, 662. 10. J. S. Debeham, R. Madsen, C. Roberts, B. Fraser-Reid, J. Am. Chem. Soc., 1995, 117, 3302. 11. a) J. C. Castro-Palomio, R. R. Schmidt, Tetrahedron Lett., 1995, 36, 5343. b) Zbid, Liebigs Ann., 1996, 1623. 12. a) D. Tailler, J.-C. Jacquinet, J.-M. Beau, XVltlz Intern. Carbohydr. Symp., Paris 1992, A298; b) D. Tailler, J.-C. Jacquinet, J.-M. Beau. J. Chem. Soc, Chem. Commun. 1994, 1827. 13. a) H. Paulsen, B. Helpap, Carbohydr. Res., 1991,216, 289. b) D. Tailler. J.-C. Jacquinet. A-M. Noirot, J. M. Beau, J. Chem. Soc. Perkin. I , 1992, 3163. c) M. Schultz, H. Kunz, Tetrahedron Lett. 1992,33, 5319. d) E. Kaji, F. Lichtenthaler, Y. Osa, S. Zen, Bull. Soc. Jp., 1995,68, 1172. e) A. Vargas-Berengel, M. Meldal, H. Paulsen, K. Bock. J. Chem. Soc. Perkin 1, 1994, 2615. 14. J. Banoub, P. Boullanger, D. Lafont, Chem. Rev., 1992, 92, 1167. 15. W. Kinzy, R. R. Schmidt, Carbohydr. Rex, 1989, 193, 33. 16. S. Ikeshita, A. Sakamoto, Y. Nakahara, Y. Nakahara, T. Ogawa. Tetrahedron Lett., 1994, 3123. 17. H. Kuyama, Y. Nakahara, T. Nukada, Y. Ito, Y. Nakahara, T. Ogawa, Carbohydr. Res. 1993, 243, C1. 18. a) L. X. Wang, Y. Z. Hui, XVIth Intern. Carbohydr. Symp., Paris A203. 1992; b) L. X. Wang, C. Li, Q. W. Wang, Y. Z. Hui, Tetrahedron Lett., 1993, 34, 7763; c) Ihid., J. Chem. Soc. Perkin I, 1994, 621. 19. P. Sinay, J.-C. Jacquinet, M. Petitou, P. Duchaussoy, I. Lederman, J. Choay, Carbohydr. Rex, 1984, 132, C5. 20. M . Petitou, P. Duchaussoy, I. Lederman, J. Choay, P. Sinay, J.-C. Jacquinet, G. Torii, Curbohydr. Res. 1986, 147, 221. 21. J. Choay, J.-C. Lormeau, M. Petitou, J. Fareed, Ann. N. Y. Acad. Sci. 1981, 370, 644. 22. L. Thurnberg, G. Backstrom, U Lindahl, Carbohydr. Res. 1982, 100, 393. 23. see also: C. A. A van Boeckel, T. Beetz, J. N. Vos, A. J. M. de Jong, S. F. van Aelst, R. H. van den Bosch, J. M. R. Mertens, F. A. van der Vlugt, J. Carbohydr. Chem. 1985, 4, 293. 24. M. Petitou, J.-P. Herault, A. Bernat, P.-A. Driguez, P. Duchaussoy, J.-C. Lormeau, J.-M. Herbert Nature, 1999, 398, 417. 25. D. R. Hepburn, H. R. Hudson, J. Chem. Soc. Perkin 1, 1976, 754. 26. H. Paulsen, C. Kolar, W. Stenzel, Chem. Ber. 1978, 111, 2358. 27. F. W. Lichtenhaler, E. Kaji, S. Weprek, J. Org. Chem. 1985, 50, 3505. 28. N. K. Kochetkov, A. J. Khorlin, A. F. Bochkov, Tetrahedron, 1967, 26, 693. 29. B. Boneu, J. Necciari, R. Cariou, A.M. Gabaig, G. Kieffer, J. Dickinson, G. Lamond, H. Moelker, T. Mant, H. Magnani, Thromb. Huemost., 1995, 74, 1468. 30. P. Sinay, Nature, 1999, 398, 377. 31. T. Ehara, A. Kameyama, Y. Yamada, H. Ishida, M. Kiso, A. Hasegawa, Curbohydr. Res., 1996,281; 237. 32. A. Hasegawa, T. Nagahama, H. Ohki, M. Kiso, J. Carbohydr. Chem., 1992, 11, 699. 33. K. Jansson, S. Ahlfors, T. Frejd, J. Kihlberg, G. Magnusson, J. Org. Chem. 1988, 53, 5629.

Part I Volume 1

I1 Synthesis of Oligosaccharide Mimics

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

20 Synthesis of C-Oligosaccharides Troels Skrydstrup, Boris Vnuzeilles, and Jean-Marie Beau

20.1 Introduction Cell-surface carbohydrates are well known for participating in key molecular recognition events with protein receptors [l]. Such interactions can be beneficial, as in cell-to-cell recognition and binding, or detrimental, as in bacterial [2] and viral infections 131, and inflammation [4]and metastasis 151, the initial steps of which can involve carbohydrate-protein interactions. To combat such diseases it would be ideal to prepare small, soluble carbohydrate-based pharmaceuticals which inhibit these recognition events by binding preferentially to the protein receptors. The potential instability of such small, soluble oligosaccharides to extra- and intracellular glycosidases has, however, prompted many chemists to develop hydrolytically stable carbohydrate mimics [ 6 ] . C-Glycosides are an important class of such glycomimetics, and are defined as carbohydrates in which the interglycosidic linkage has been substituted by a methylene group. This minor but important modification renders these analogs completely resistant to hydrolysis [7]. Because of their close analogy with the parent 0glycoside, C-disaccharides and C-oligosaccharides have become important compounds for study of carbohydrate-protein interactions in lectins and glycosidases. The incorporation of a methylene group in the glycosidic linkage not only enables binding of the C-oligomer to the receptor without sensitivity to hydrolysis but might also give information about the conformation of the bound sugar around the glycosidic linkage by 'H NMR analysis [8, 91. Realizing that C-glycosides seem to be promising drug candidates against the above detrimental cellular interactions, or are important biological tools for the study of carbohydrate-protein interactions in lectins and glycosidases 19, lo], several groups have developed interesting methodologies for the construction of such compounds. The synthesis of these compounds is not always easy, considering that a C-C bond generation is required not only at the terminal sugar but also at the reducing end.

496

20 Synthesis of C-Oligosuccharides

In this review the different methods for the synthesis of C-disaccharides and higher C-oligomers will be highlighted, covering the literature to the end of 1998. The synthetic approaches to these glycomimetics will be divided into four chapters depending on whether the methodology applied by individual groups was: 1) by direct coupling of intact carbohydrate units via (i) anionic or (ii) radical pathways; 2) by the de novo synthesis of one .or all of the sugar units in the C-disaccharide or C-oligomer; or 3) by applying a cycloaddition reaction to connect the two sugar units. We include only work which represents a true mimic of the sugar, i.e., mimics with the correct number of linking carbons and endocyclic oxygen atoms. A great many series of disaccharide analogs have been synthesized either having no carbon linkage or one carbon more than the correct mimic, or where the ring oxygen has been replaced by nitrogen [ 1 I , 121. For these studies the reader is referred to several other reviews [7, 131.

20.2 The Anionic Approach 20.2.1 C5-Alkynyl Anions An approach to the first synthesis of a true C-disaccharide was reported by Rouzaud and Sinay in 1983 [14]. Having previously worked on the synthesis of C-(alkyny1)-Pglucosides [ 151, the Sinay group took advantage of this chemistry for the synthesis of a Glc(P1-6)Glc C-disaccharide employing a C5-alkynyl anion as shown in Scheme 1. This method is highly effective for the construction of 1,6-linked C-disaccharides owing to the ease of preparation of the starting sugars and the high-yielding coupling steps exemplified below. The glucose derivative 1 was converted in two steps

4

BnO BnO

1. DMSO, (COC1)2,

Br

2. CBq, PPh3 0

BnO OMe 2

1

2+3

-

BuLi BnO 92% B

n BnoOH

BnO O 4

Scheme 1.

OMe

OBn 3

2. 1. EgSiH, H,, PdKBF3.Et20 HHO O L h BnO 52% HO OMe

(

o

i 0

5

l

20.2 The Anionic Approach

BuLi 0

BnO BnO 6

75%

BnO

497

BnO

7

8 1. Et,SiH, BF,.OEt, 2. Hz, Pd/C

74%

OH

I 6, BuLi 67% 1 Me0

9

APBn

n..

Bnow BnO[uB:

OBn

BnO BnOW

\ \O

O

OBn B

*

n

HO

74%

OMe HO

4 steps 29%

BnO B n

0

&

~

&

OBn

BnO BnO BnO BnO

'?

OBn

BnoOH

H

OBn 71% 12

Scheme 2.

to the vinyl dibromide 2 which was transformed to the acetylenic anion with 2 equiv. BuLi at -50 "C. After treatment with the glucopyranolactone 3, the hemiacetal4 was obtained in high (92%) yield. Stereospecific reduction of the hemiacetal with Et3SiH and BF3.Et20 involving hydride attack on the u face of a cyclic oxonium ion intermediate then afforded the b-C-disaccharide 5 after catalytic hydrogenolysis. Applying this methodology, the same group was able to prepare the Ctetrasaccharide 12 in an iterative sense, in addition to the dimer 9 and trimer 11 obtained from intermediates along the synthetic route (Scheme 2) [ 161. Starting this time with the vinyl dibromide 6 and galactopyranolactone 7, a similar reaction sequence as described above was adopted for the preparation of the protected Cdisaccharide 8. After removal of the anomeric hydroxyl group and selective reduc-

OMe HO

498

20 Synthesis of C-Oligosaccharides 1. AllylMgBr

3. oso4, 2. MeZC(OMe)Z, NMO H+

mu-n OBnOG0 B BnO BnO 0 3

BnO% BnO

~

-0BnO

BnO OH 13

84% 84%

6 8 I

0

'

A

+ diastereomer (2:3)

PivCI, Pyr BnO,

B n O S

2, BuLi NaOMe, MeOH HCI C1C02CCI,, Py

86%

OPiv

B n O p un

15

36%

-)-0

14

0

1. C02(CO)* 2. BF,:OEt, 3. CAN

55%

I. TEA, Py, H20 2. PdC, H?

RO- ' "

Ro

17 R=H

OMe

(a$=1:l)

8R

Aczo'Py

R=Ac(9l%)

16

Scheme 3.

tion of the alkyne functionality, a two-step hydrolysis and oxidation sequence gave the corresponding lactone which could be coupled again with 6 to give a Ctrisaccharide intermediate 10. In the same manner, these steps were repeated and followed by a deprotection step to produce the C-tetrasaccharide 12, the parent 0glycoside of which is known as a plant elicitor. Schmidt and coworkers have recently extended this approach for the synthesis of C-disaccharides of ketoses as depicted in Scheme 3 [ 171. In their attempt to prepare a glycosyl transferase inhibitor the Schmidt group reacted the Sinay intermediate 2 with the hemiacetall3, readily prepared in three steps from the glucopyranolactone 3. Although, no coupling was observed in this case, the reaction proceeded when 13 was first transformed to the open-chain compound 14 with pivaloyl chloride and then treated with the alkynyl anion. This led to the formation of two propargylic alcohols, as a 3 :2 diastereomeric mixture in a 61% yield, which were subjected to protecting group manipulations to give 15. Ring closure was best achieved by per-

20.2 The Anionic Approach

499

forming a Nicholas reaction whereby the alkynes were first transformed to their dicobalthexacarbonyl complexes. Thus treatment of either diastereomer with BF3.Et20 readily led to the formation of a stabilized carbocation with immediate cyclization to give a 1 : 1 mixture of a : p anomers 16 (only a anomer shown) after decomplexation with cerium ammonium nitrate. A two-step deprotection protocol of either anomer then gave the corresponding unprotected C-disaccharides 17 (only a anomer shown). Interestingly, with the diastereomer of ketone 14 only one stereoisomer was obtained upon reaction with the vinyl dibromide 2. In addition, ring closure of the coupling product is highly selective, affording an a : p mixture in a 10: 1 ratio. Employing the monosubstituted alkyne 19, easily prepared from vinyl dibromide 2, van Boom recently developed a stereoselective ZnC12-assisted ring-opening of epoxide 18, enabling rapid access to the C-isomaltoside 21 via the alkyne 20 (Scheme 4) [18]. This surprising cis selectivity observed in the epoxide opening is rationalized by the initial formation of a zinc acetylide species (see Scheme 5), followed by an intramolecular delivery of the alkyne moiety in the ion-pair 23 or the p chloride 24. A Nicholas reaction [ 191 involving the triflic acid-induced isomerization [20] of the dicobalthexacarbonyl complex of alkyne 20 can be performed to access the p anomer 22.

4

BnO BnO

+

BuLi, ZnC1, BnO

0

BnoOMe

18

59%

19 O *:B i BnoOMe

H,, Pd/C

80%*

R

E

O

20

X

RoOMe

21 R=H Aczo’Py

R=Ac(92%)

BdOLMe

Scheme 4.

22

500

20 Synthesis of C-Oligosaccharides Sugar-Li

i

B

n

O

G

Sugar

OBn

*ZnCl

BnO 0

18

23

(OBn

20

24

Scheme 5.

20.2.2 C1-Glycal Carbanions An attractive method employing stabilized C1-glycal anions for the synthesis of pC-disaccharides was developed by Schmidt and coworkers (Scheme 6). The Konstanz group had observed earlier that the deprotonation of protected glycals is greatly facilitated by the presence of a phenylsulfinyl group at the C2 position [21]. To obtain a glycal derivative, such as 25, an easy two-step synthesis from tribenzylD-glucal was employed with a 63% overall yield. C1-substituted glycals are then formed by the coupling of the corresponding anion 26 with aldehydes. Reductive removal of the phenyl sulfinyl group and a stereospecific hydroboration/oxidation step affords P-C-glycosides 27 with an equatorial C2-hydroxyl group. This strategy was first reported in 1989 for the preparation of a C-cellobiose derivative as illustrated in Scheme 7 [22]. The electrophilic species was prepared

OBn 1. PhSCI; DBU BnO BnO&

0 2.mCPBA

63%

B g ( 3 0 q

LDA-100°C

Bpo)

Li

PhSO 25

PhSO 26

RCHO R PhSO 21

Scheme 6.

501

20.2 The Anionic Approach

HBnO O

1. DMSO, DCC 2. Ph3P=CH2 3. H3B.Me2S;H202

q

* 26%

BnOOBn 28

30

29

OBn

1. Raney Ni

25,LDA

BnO

___)

3. PdC, H2

74%

OH

OBn 31

32

Scheme 7.

starting from the appropriately protected glucoside 28. DMSO-assisted oxidation of the C4 hydroxyl group then Wittig reaction led to formation of an exocyclic alkene, which upon borane addition, employing the borane dimethylsulfide complex, and subsequent oxidation, afforded in modest yield the C-branched sugar 29 with the gulucto configuration in addition to the reversed borane addition product. Steric control from the C3 and C5 positions is undoubtedly the cause of the high stereoselectivity observed in the hydroboration step. Oxidation of 29 then gave the corresponding aldehyde which could be isomerized by mild basic treatment to the isomer 30 with the gluco configuration. Addition of the C1-anion to aldehyde 30 gave compound 31 in good yield with high diastereoselective control at the new stereogenic center (10 : 1). Transformation of the Cl-substituted glycal to the C-cellobiose derivative 32 then required a threestep sequence involving reductive removal of the phenylsulfoxide with Raney nickel, a stereoselective hydroboration-oxidation step, and finally catalytic hydrogenation. Since this preliminary example, the Schmidt group has adapted this approach for the construction of other p-C-glycosides employing either the 2-(phenylsulfinyl)-~glucal 25 or the galactal counterpart 33 (Scheme 8) [23-261. Many improvements BnO B

n

OBn O

e HO

PkO

HO

HO

35

OH 36

Scheme 8.

OMe

34

OH

HO

Ho I

HO

33

502

20 Synthesis of C-Oligosaccharides 1. BnBr, NaH 2. CS,, NaH; Me1

33 + 30

steps

4.H2. Pd/C

BnO

OH

OBn

Scheme 9.

were made to increase the yields of the synthesis of the electrophilic coupling partner, notably in the alkenylation and hydroboration/oxidation steps. It was found that Tebbe’s reagent gave good yields of the desired alkene where the Wittig reagent failed, whereas 9-BBN proved much more effective than borane, securing consistently higher regioselectivities in the hydroboration step. In one example a cyano group was used for the introduction of a C4-branched formyl glycoside [25], an approach which was also reported by Vasella [27] and Armstrong [28] for onecarbon homologations of uloses. Several hydroxy C-disaccharides, 34-36, have thus been prepared by this approach, as illustrated in Scheme 8. The presence of the hydroxyl group in the carbon linkage might even be an advantage in that it enables the introduction of other functional groups at this position, and provides a conformational preference around the two exocyclic C-C bonds. In one case, the hydroxy C-disaccharide was deoxygenated using the Barton protocol (NaH, CS2, MeI, then Bu3SnH, AIBN) providing an efficient overall entry to C-lactose 37 (Scheme 9) [24]. 20.2.3 Anomeric Samarium Species Application of a Cl-lithiated sugar with a protected C2-hydroxyl group, and its direct coupling with a carbonyl substrate at low temperature, does not lead to the formation of the C-glycoside owing to a rapid, competing p-elimination which leads instead to the corresponding glycal [29]. In contrast, anomeric samarium species have proven to be much more resistant to this process, even at room temperature, and hence have been used for the rapid assembly of C-disaccharides. In 1995, Skrydstrup and Beau reported the synthesis of the C-disaccharide derivative 40 of a-C-mannopyranosyl( 1-2)-~-glucopyranose, via the reductive samariation of the mannosyl pyridyl sulfone 38 in the presence of aldehyde 39 (Scheme 10) [30]. Because this aryl sulfone group has a sufficiently low-lying LUMO energy level, electron transfer from the mild reducing agent, samarium diiodide, is facilitated, resulting in a fragmentation step with concomitant formation of the thermodynamically more stable a-anomeric radical (the anomeric effect), as illustrated in Scheme 11 [31]. A second electron transfer to 49 from an additional equivalent of samarium diiodide then leads, under kinetic control, to the organosamarium species 50 which is configurationally stable because of a fast conformational change, placing the C1 and C2 substituents in a more stable equatorial position, as in 51. Its

20.2 The Anionic Approach

oms

BF%

+

~,g-&+

2.Bu4NF 1. Sm12

503

~

BnO

38

OB n

75%

o'

S02PY

OBn

HO

OBn 40

39

BnO

~$3 ~{3 BnO

S02PY 41

OMe

73%

BzO BnO

42

BnO

1. Im2CS 2. FjPhOH, Ph3SnH

70%

2. I.B MeONa u~NF

OTBS

.,,H BzO BzO BnO

3. H2, Pd/C

+H

*

79% HO HO

.O

-0 I

OMe

OMe

44

Bz% + BnO

B BnO

45

q P

*

oJ S02PY 46

OMe 47

48

OMe

Scheme 10.

B;;oEh BnO

Scheme 11.

syn-elimination to glycal

I

504

20 Synthesis of C-Oligosuccharides

53

Ph

Ph

1. O3 2. Bu$nH, AIBN

,

39

BzO

63% P h d P h

OMe

OMe 42

Scheme 12.

coupling with aldehydes under Barbier conditions therefore produces the a-Cdisaccharides, 52, only [30]. The synthesis of the electrophilic partner 39, reported in 1993 by Jung and Choe, is interesting because of its originality in respect of the introduction of a formyl group into the sugar ring (Scheme 12) [32]. In this work, the authors demonstrated that substrate 53, obtained by iodoglycosylation of tribenzylglucal with 1-phenylprop2-en01 and then ozonolysis, readily underwent a 5-ex0 radical cyclizationfragmentation process resulting in the intramolecular transfer of a formyl group to C2 and generation of a more stable benzyl radical. Not only was the formyl group transferred stereospecifically to give the 1,2-cis relationship in 39, owing to the 5-membered cyclic intermediate in the transfer mechanism, but the original functionality at the anomeric center was then transformed to the well-known benzyl protecting group. An adaptation of this approach has recently been exploited by Skrydstrup and Beau for the preparation of aldehyde 42 required in the synthesis of methyl a-Cmannobioside 45, a C-glycoside analog of the Mycohacterium tuberculosis capping disaccharide (Scheme 10) [33].Rather than a formyl group transfer from C1, the C3 position was selected affording the derivatized sugar 42 with a CZaxial substituent (Scheme 12). Subsequent coupling with the mannosyl pyridyl sulfone 41 under the influence of SmI2 afforded only one C-glycoside, 43, which was transformed to the unprotected C-disaccharide 45 via a deoxygenation step to 44 and subsequent deprotection. Recently, the C-dimer of u-D-Man(1-3)a-~-Man48 has been prepared by a similar approach employing the mannosyl pyridyl sulfone 46 and aldehyde 47 P41. Other extensions of this C-glycosylation methodology have been reported for the preparation of two C-linked 1,6-disaccharides. Reaction of the pyridyl sulfone of N acetylglucosamine 54 with aldehyde 55 produced mainly the a-C-disaccharide 56 (Scheme 13) [35].The stereoselectivity at the anomeric center is explained in this case by internal coordination of the N-acetyl group with the metal ion of the intermediate glycosyl samarium species 57, retarding anomerization to the ther-

20.2 The Anionic Approach

505

Bnr% AcHN SO2Py

BnO

bMe Bnb

54

55

bMe

56 (a$= 4: I )

57

Scheme 13.

modynamically more stable p isomer as observed for the corresponding glucose derivative. Reductive samariation of tetrabenzylmannosyl chloride 58 (X = C1) or phosphate 58 (X = OPO(OPh)2) in the presence of 55 also served as a useful reaction for the synthesis of C-disaccharides as exemplified in the synthesis of C-dimer 59 (Scheme 14) [36, 371. Recently the glycosyl pyridyl sulfones have been applied by the Linhardt group for the synthesis of a C-disaccharide containing N-acetylneuraminic acid (Scheme 15) [38]. The coupling of glycosyl pyridyl sulfone 60 with the previously reported aldehyde 63 gave a high yield of the (1,3)-C-disaccharide64 with complete stereoselectivity both at the anomeric center and at the new exocyclic stereocenter. It is important to point out that these reactions proceed through a samarium enolate intermediate 65 in contrast to 2-deoxy sugars lacking a C1 carboxylate group which involve true anomeric samarium species and which are non-selective at C1. The coupling step can even occur with the glycosyl phenyl sulfone 61 or chloride 62 because the initial homolytic cleavage step after the first electron transfer is favored by the stability of the radical formed (captodative effect). An extension to a KDNcontaining disaccharide derivative 66 has recently been published by the same team with the same aldehyde 63 having identical stereoselectivity [39].

55

+

goq

F o x

Sm12

b

x 58

Scheme 14.

x=c1

X= OPO(OPh),

54% 73%

BnO

BnO 59

BnO

OMe

506

20 Synthesis of C-Oligosaccharides

OAc

OMe

60 X= ~lS02pV 61 X= aS02Ph 62 X= PCl

63

64

63, SmI2 Bn

OAc

X= aS02Ph

66

OAc

a-face

65

x=SCl Scheme 15.

20.2.4 C-Branched Carbanions p-Glycosyl nitromethanes are effectively prepared from the parent sugar in two steps [40]. Because the corresponding nitronate anions do not undergo ring opening, chain elongation at the a carbon is possible and has been exploited by Martin and Lai for the synthesis of both p(1-6)- and p,p(l-l’)-linked disaccharides [41, 421. Treatment of the C-glucoside 67 with KF in the presence of the galactose-derived aldehyde 68 led to the rapid formation of the derivatized C-dimer which was transformed to the unsaturated nitro compound 69 (Scheme 16). Further reduction and denitration involved two steps with eventual deprotection providing the meth1-6)-~-Gal. ylene bridged analog 70 of p-~-Glc(

A AcO c

O

G NO2

~

To%

OAc

67

68 ,OAc

1. N a B Q 2. BqSnH, AlBN 40%

Scheme 16.

1. KF, 18-crown-6 2. Ac20, Py * 47%

20.2 The Anionic Approach

507

1.80% AcOH 2. AclO, BF,.Et,O

1.67, KF, 18-crown-6 2. A c ~ OPy ,

c

36%

44% OPiv 71

OPiv

u I 70

OH 74

73

Scheme 17.

Elaboration of this approach for the synthesis of C-P,P-trehalose was successfully achieved as illustrated in Scheme 17. The nitroaldol condensation of 67 with the aldehyde-glucose derivative 71 provided an epimeric mixture of unstable nitroalditols which were immediately dehydrated to give mainly the E-nitroalkene 72. The sensitive deacetalization and cyclization step is best brought about by heating 72 under reflux in 80% AcOH. In this way, both deprotection and ring formation take place to produce the second p-C-glycoside of the required C-disaccharide via attack of the internal nucleophile on the si face of the double bond. A substantial amount of the C-pyranosyl-C-furanosyl isomer was likewise isolated. Radical denitration and deprotection then afforded the C-p, b-trehalose 74. Reduction of the nitro group in 73 also gave rise to a C-disaccharide containing a bridging methylenamine group; similar compounds were demonstrated by Schmidt to be good glycosidase inhibitors. A similar approach was also published by Witczak and collaborators for the preparation of 3-deoxy-C-cellobiose (Scheme 18) [43]. The starting glycosyl nitro-

-? B

n

O

w +

BnO

Qo

BnO NO,

0

BnoOH 75

76

1. BqSnH, ABCN 2. BF3:OEt2, Et&H 3. L-Selectride 5 . Pd(OH)& 4. EgSiOTf, AczO C,H,,, EtOH

-

-& AAcO

28% OAc 78

Scheme 18.

508

BBnO n

20 Synthesis of C-Oligosaccharides

O

G

+

k0

BnO BnO

BnoOCH3

OBn

KF, 18-crown-6 52%

BnO-- ..BnO

BnO

BnO

80

79 1. BuLi, PhOC(S)Cl

2. Bu$nH, AIBN 3. TsNHNH2, AcONa 4.Na, NH3 10%

81

HO

B"bbCH3

OH

HO

Scheme 19.

methane 75, easily prepared in two steps, underwent stereospecific conjugate addition to the sterically less hindered exo face of the bicyclic enone 76 under mild basic conditions (KF or triethylamine). Tributyltin hydride reduction of the nitro group in 77 followed by removal of the anomeric hydroxyl group with Et$iH/BF3.Etz0 and stereoselective reduction of the ketone subsequently afforded the C-disaccharide 78 after a deprotection-acetolysis sequence. Reversing the roles of the two coupling partners in the nitroaldol condensation is also an option for construction of C-linked (1-6)-disaccharides [44]. In this instance aldehyde 80 was coupled to the C6-nitro derivative 79 to afford adduct 81 (Scheme 19). Tin hydride-induced reduction of the corresponding thionocarbonate then led to the C-dimer 82 after two reduction steps. Schmidt and coworkers, on the other hand, have developed an approach to Clinked (3-disaccharides employing CCbranched carbanions which can be prepared by lithium-halogen exchange as exemplified with the 1,6-anhydro sugar 85 in Scheme 20 [45]. This precursor was prepared via a stereoselective hydroborationoxidation of the exocyclic alkene in 84 obtained in two steps from alcohol 83. Addition of the anion generated from 85 to gluconolactone 3 gave a hemiketal which was stereoselectively reduced to the C-disaccharide 86. Further standard manipulations provided the octaacetate derivative of (3-~-Glc( 1-4)-~-Gal87. The peracetylated derivative of p-D-Gal(1-4)-~-Gal88was prepared in a similar manner. 20.2.5 CdPhosphoranes Wittig olefination is another attractive and rapid route to (1,6)-C-disaccharides. The Dondoni team demonstrated that the readily accessible phosphonium ylides at C6 of sugars coupled well with C-formyl glycosides available through thiazole-based formylation of glyconolactones [46]. In Scheme 21, a representative of this approach is shown for the synthesis of the C-dimer of p-~-Gal(l-6)-~-Gal [47]. Hence, treat-

20.2 The Anionic Approach

1. DMSO,DCC 2. Ph3P=CH2

HO

83

pq

39%

OBn

;:;wy;

H202

+ 67%

OBn

I

1. Ac20, TFA 2. HZ, WIC

3.Ac20,Py

B 0

%no

40%

OBn

OBn

86

+ . &+& o ? i A

OBn 85

84

1. BuLi 2. Et3SiH, BF3.Et20 85 + 3 * 34%

509

OAc

87

OAc

OAc

88

AcO

OAc

OAc

Scheme 20.

BnO

6-step formylation

53%

BnO BnO 80

80 + 89

Scheme 21.

ment of the phosphonium salt 89 with base and then aldehyde 80 led to the formation of the cis olefin 90, which upon Pd-catalyzed hydrogenation subsequently afforded the C-disaccharide 91. A set of ten (1-6)-linked p-C-disaccharides could therefore be prepared by means of this approach starting from C-glycosyl formaldehydes 80 and 93-98 with the phosphonium salts 89 and 92 (Scheme 22) [48]. The basic conditions of the Wittig reaction nevertheless prohibit the formation of the corresponding u-C dimer; e. y. starting with the u-C-glucosyl precursor, the corresponding P-C-disaccharide is obtained by initial isomerization at the anomeric center.

510

20 Synthesis of C-Oligosaccharides 'PPh3 I-

BnO% BnO

BBnO n BnoOMe 92

y

%:nBBnO

' a0

93

96

BnOR R

94 R = H, R = CHO 95R= CHO,R'=H

98

91

Scheme 22.

steps

BnO

c

P

BnO

70%

68

99

BE+ 101

-P

102

Ho

Scheme 23.

The reverse strategy has recently been reported by the same group for the preparation of a similar C-mimic of the p(1-6) tetragalactoside as shown in Scheme 2 [49]. Condensation of the D-galactopyranosyl-derived phosphonium salt 99 and aldehyde 68 led to the unsaturated dimer 100 (Scheme 23). Repetition of the sequence twice with phosphonium salt 99 furnished tetramer 101. These steps are, however, characterized by low coupling efficiencies (36% for the trimer and 11% for the tetramer) mostly because of the base-induced a,p-unsaturation of the aldehyde substrates. The desired tetrasaccharide 102 was then obtained after standard deprotection and double-bond hydrogenation.

70

20.3 The Radical Approach

51 1

20.3 The Radical Approach 20.3.1 Intermolecular Anomeric Radical Addition In principle, intermolecular radical-mediated addition of anomeric radicals to an exocyclic double bond of the C-glycosyl acceptor should provide an efficient route to C-oligosaccharides leading directly to the desired mimic. In practice, however, these reactions with unactivated alkenes are not favorable as they lead solely to products of direct H-abstraction. Reducing the LUMO of the alkene with electronwithdrawing groups, typically a,P-unsaturated carbonyl compounds or similar derivatives, greatly improves the overall efficiency of these coupling reactions. For example, Giese demonstrated that the anomeric radical obtained by reaction of acetobromoglucose with Bu3SnH adds to the a-methylene y-lactone 103 in good yields and with high a-selectivity to give the C-dimer 104 (Scheme 24) [50].In addition, hydrogen abstraction at the new radical center at C2 of the reducing sugar is stereoselective leading to the predominant formation of the product with an equatorial oriented CZsubstituent. Further modification involving reduction of the lactone to lactol and acetylation afforded the protected C-mimic of kojibiose 105. Two other examples 106 and 107 have also been successfully prepared, as shown in Scheme 24. In a similar approach, Vogel described the use of ‘naked sugars’ such as 108 as radical acceptors, as shown in Scheme 25 [51, 521. The coupling product 109 was further elaborated before a true C-disaccharide was reached, with the key step being Baeyer-Villiger oxidation to introduce the fully oxygenated ring system. Several

Ace% OAc

AcO

nBo- \+

Bu$nH, AIBN 70%

AcoB,

OBn 104

103 1. Na[A1(OC2H,0Me)2(OEt)H1 2. AQO

64%

OBn

105

BnO BnO 106

Scheme 24.

AcO

Me0B-n

I OAc AcO

OAc

107

5 12

20 Synthesis of C-Oligosucchurides

1. Bu3SnH, AIBN 2. mCPBA, NaHC03 OAc AcO OAc 1 0 9 ( a : p =5.5:l) OAc

66%

108

C02Me

1. NaHC03, MeOH 2. A c ~ O DMAP, , Py

1.NaBH4 2. HCI OAC 3. A c ~ ODMAP, , Py

*

* 36%

0 0

n

Acd

OAc

Scheme 25.

variations of this approach have now been reported for the synthesis of several Cdimers with different substitution patterns of the radical acceptor in addition to the use of acetobromogalactose (Scheme 26) [ 531. In a different approach, Martin reported the use of an exocyclic radical derivative The as a means of preparing the a(1-6) C-linked disaccharide 113 (Scheme 27) [54]. corresponding radical from the cobaltoxime derivative 111, prepared from iodide

1. Bu3SnH, AIBN + pl,sed\

Atgo% AcO AcoBr

c;

0

2.NaBH4

PhSe, 48%

1. mCPBA 2. Ac7O 3. osb, 4. Ac,O

I\

r5

c

OH

86%

R mCPBA

steps

R

AcO 0

OAc

Scheme 26.

20.3 The Radical Approach

2. 1. A hv,c NaOH, ~O 95% E

t *O

111 + 112 30%

"

"

"

~

Bk'

Bu3SnH

AcO AcO

*

AcO

AcO

AcO

OMe

513

113

AcO

OMe

Scheme 27.

Bu3SnH

Bno%I BnO

+

BnoOH

114

Q 0 76

ABCN 26% 0

BnO BnO

Scheme 28.

110, was generated photochemically and coupled to the nitronate derivative of the C6 functionalized glucoside 112. Tin hydride-promoted denitration then afforded the desired C-dimer. An exocyclic radical approach to C-disaccharides was also employed by Witczak starting with the readily available iodide 114 (Scheme 28) [43]. An exo-facial selective radical addition to levoglucosenone 76 afforded, in low yield, the C-dimer 115, with direct reduction being the major competing pathway. This compound was then converted to the 3-deoxy derivative of C-cellobiose as depicted in Scheme 18. 20.3.3 Intramolecular Anomeric Radical Addition

If the radical addition reaction is performed in an intramolecular fashion, olefin activation with an electron-withdrawing group will not be a prerequisite for Cglycosylation, because the cyclization event competes well with hydrogen abstraction. In such examples, ring size and the stereochemistry of the ring substituents greatly influence the stereoselectivity and efficiency of the cyclization reaction. With the above points in mind, Sinay's group developed a direct approach to Clinked disaccharides employing a temporary silicon tether approach to graft the

5 14

20 Synthesis of C-Oligosaccharides

o& &no

SePh

BuLi, Me$iC12; Imidazole

BnO

OH

116

117

2. 1. HF, BqSnH, THF AIBN *

BnbAMe

.

95%

BnO& BnO

SePh

9,Sic

B&(30% B

40%

n

BnO

118

4 BnoOMe

BnoOMe

Scheme 29.

radical donor and acceptor together [55]. These reactions are surprisingly efficient considering the medium-sized ring intermediates formed, which upon liberation from the silicon linker afford the C-dimer. Illustrative of this approach is the synthesis of a C-linked analog of methyl amaltoside (Scheme 29) [56]. Linkage of the two glucose derivatives 116 and 117 with dimethyldichlorosilane followed by tin hydride-promoted 9-endo-cyclization afforded the C-disaccharide 118 in 40% yield after liberation from the linker. The anomeric effect seems to dominate in this reaction with exclusive formation of the a anomer, whereas H-abstraction by the resulting intermediate radical after cyclization affords the equatorially substituted product. The nature of the attacking sugar radical has a substantial impact on the cyclization yields and stereoselectivites of these reactions-when the method was extended to the galacto series 60% yield of three isomeric Clinked disaccharides was obtained in the ring-closure step [57]. The ring size and tethering positions are equally important, as is observed in the 8-endo cyclizations of the gluco and galacto derivatives 119 and 121, respectively (Scheme 30). In the first case an inseparable mixture of all four isomers 120 was obtained [56],whereas the C4-epimer 121 led to the C-linked disaccharide 122 in 45% yield [ 5 8 ] . Quite remarkably, the anomer was furnished in the latter case,

6 1

OBn Bno& -BnO $S -,ePh

OBn

/

BnO

\

BnO !&&SePh

121

Scheme 30.

O

w

o

B OBn

120

OBn

\

B BnOn

50%

119

\&

OBn

O\Si//

OBn

2. 1. Bu+nH, HF, THF AIBN

OMe

1. BusSnH, AIBN 2. BQNF

45% (tethering incl.)

OBn 122

n

515

20.3 The Radical Approach

~ ~ 0 % SO2R B n g o M C

~1. .SmI2 HF

BnO

\si’o

/

50%, R = Ph 70%, R = Py

\

BnO

118

OMe

Scheme 31.

Br:o&

0

CSA, 4..4 MS

Tebbe’s reagen:

81%

87%

BnO

SePh

iePh

-

117

123 1. BqSnH, AIBN

BnO

Brio\

OH

BnO

BnO

35% 124

125

+

pproduct (3%)

BnOOMe

Scheme 32.

where there was undoubtedly a strong conformational bias induced by the tether because attack of the anomeric radical species occurred on the p face. The standard tin hydride conditions for generating an anomeric radical can be replaced by the divalent samarium-induced reductions of glycosyl aryl sulfones [ 59, 601. In one instance the Sinay group employed the 2-pyridyl sulfone group, introduced by Skrydstrup and Beau in 1994 [30, 311, which was rapidly reduced by SmI2 at room temperature (Scheme 31). In this way, a significant increase in the cyclization yield of 118 was achieved [60]. The ketal connectors which were exploited earlier in 0-glycosylations proved equally efficient in the synthesis of C-linked disaccharides as shown in Scheme 32 [61]. With acetate 123 a two-step sequence via vinyl ether formation with Tebbe’s reagent and acetalization with alcohol 117 produced the linked sugars 124. Upon tin hydride-promoted cyclization and removal of the acetal tether the a-D-Man(1 4 ) - ~ - G l cC-disaccharide 125 was obtained in 35% yield. Similarly, a benzylidene acetal linkage as in 128 could be produced via a DDQmediated oxidation of the p-methoxybenzyl group in 126 in the presence of the fucose derivative 127 (Scheme 33) [62], a strategy introduced in 0-glycosylations by Ito and Ogawa [63]. Subsequent radical cyclization and acetal hydrolysis afforded the a-~-Fuc(lL2)-~-Glc analog 129 in good (60%) yield. A phosporamidic linker also proved viable, as reported in the synthesis of the Cdisaccharide analog of a-~-GalNAc( 1-4)-~-Glc(Scheme 34) [64]. Intermediate 131 prepared in two steps from the galactosamine derivative 130 and the unsaturated -

5 16

20 Synthesis of C-Oligosucchurides

1. Bu3SnH, AIBN 2. DDQ, HZO

60%

129

Scheme 33.

1. PhPCl,, Et3N, THF 2. tBuOOH

BnO

BnO 80%

OMe

SePh

130

117

1. Bu+nH, AIBN 2. HCI, MeOH 3. MeONa, MeOH 4.AQO, Py

BnO OBn

+ p product (4%)

BnOQ+++

16% 132

OMe

Scheme 34.

alcohol 117 underwent 9-endo cyclization to give the C-dimer 132 after a deprotection and acetylation sequence. An approach to P-C-disaccharides was reported by the same group whereby the roles of the two tethered sugars are interconverted as exemplified with silaketal 133 in Scheme 35, in which the radical acceptor has an anomeric exo methylene [65]. Radical addition to the double bond led to formation of an a-oriented anomeric radical which subsequently abstracts a hydrogen leaving the C1 carbon branch in the j3 position as in the C-disaccharide 134. Because of the facile access to these C-dimers by use of this intramolecular approach, multigram quantities of the analogs are available and have been exploited for the synthesis of mixed C/O-glycosides of biologically relevant oligosaccharides. In this way, the analogs of sialyl LewisXand the pentasaccharide re-

517

20.3 The Radical Approach (OBn 1. Bu$nH, AIBN 2. B u ~ N FTHF ,

OMe

37% (tethering incl.) '

OH

-

\

OH 134

133

Scheme 35.

135

136

Scheme 36.

sponsible for the antithrombin I11 binding region of heparin, 135 [66] and 136 [67], respectively, have been prepared (Scheme 36). Predictable stereochemistry at C 1 would be expected when applying h x o cyclizations in silicon-tethered radical reactions for the construction of Cdisaccharides. A sole example has been provided by Skrydstrup and Beau using glycosyl pyridyl sulfones for the generation of the anomeric radical [31]. This approach has been successfully exploited in the synthesis of peracetylated methyl Cisomaltoside 137 (Scheme 37).

1. SmL 2. B u 4 h 3. H2, PdC 4. A c ~ OPy ,

,OAc

a

*

48%

AcO AcO 137

Scheme 37.

Ac'O 1 OMe

5 18

20 Synthesis of C-Oligosaccharides

20.4 The Partial de Novo Approach In the partial de nuuu synthesis of C-oligosaccharides one of the carbohydrate units is constructed from smaller building blocks. The advantage of this approach is that in the designed synthesis of a C-mimic little variation can easily lead to other stereoisomers, which has resulted in the preparation of a large collection of C-di- and trisaccharides. Much of the work in this area has been dominated by Kishi and coworkers, primarily because of their interest in studying the solution conformation of the parent oligosaccharides and their bound conformation on carbohydrate binding proteins [9], an interest which was sparked from their phenomenal synthetic work on the marine toxin, palytoxin [68]. Kishi and coworkers found that the solution conformations of C-disaccharides strongly resemble those of the corresponding O-glycosides and can therefore be exploited to predict the conformational preferences of the natural sugars. Because the C-analogs have two protons on the connecting carbon atom their solution conformations are readily extracted from simple 'H NMR experiments, which is not the case for the parent sugars. As a result of other NMR experiments Jimenez-Barber0 suggested that C-disaccharides are more flexible around the methylene bridge, and it was observed that in certain instances such compounds are bound to lectins in conformations different from that of the normal disaccharide [ 10, 691. In general, the de nuuu approach applied by the Kishi group for the synthesis of C-oligosaccharides relies on acyclic stereocontrol for the construction of the contiguous chiral centers of at least one of the carbohydrate moieties. The first examples of this approach were reported by this group in 1987 and involved the synthesis of four C-disaccharides, including the C-analogs of methyl maltoside and cellobioside, from a common synthetic intermediate [70]. Initial Wittig reaction of the readily available arabinose derivative 138 and the phosphonium salt 139 led to the cis alkene 140 (Scheme 38). Stereoselective dihydroxylation, exploiting the 1,3-allylic strain effect, then monoprotection of the diol, oxidation and cleavage led to the cyclic hemiacetal 142 after protection of the primary alcohol. Reduction of the hemiacetal was best performed with n-Pr$iH, affording exclusively the p anomer 143. Deprotection and acidic methanolysis then led to the C-disaccharide analog 144 of P-~-Man(l-4)-~-Glc. On the other hand, simple oxidation of the C2-hydroxyl group in 143 followed by stereoselective reduction converted the mannu series to the gluco series with the eventual formation of C-cellobioside 145. Preparation of the corresponding a isomers of these C-dimers requires only slight modification of the syntheses, as illustrated in Scheme 39 by simple reversion of the order of the synthetic steps in the preparation of the acyclic precursor. Hence, the isopropylidene in intermediate 141 was hydrolyzed, followed by diol cleavage and preparation of the anomeric p-nitrobenzoate 146. Introduction of the C6' carbon could be achieved by stereoselective allenylation resulting in formation of the manno derivative 147 after chair inversion. Oxidative cleavage then reduction of the allene moiety and subsequent debenzylation and methyl glycoside formation led to the a - ~ - M a n1-4)-~-Glc ( carbon analog 148. Access to methyl C-maltoside 149 was

519

20.4 The Partial de Novo Approach

138

139

140

1. DMSO, (COC1)2, Et3N 2. HC1 3. BzCl, Py

1. Os04 2. PMBBr, NaH

B z * BnO 75% BnO

O

G

Y

n

OH 142

2. 1. DMSO, BH3.Et3N (COCI),, Et3N

nPr$iH, BF3.Et20

B

z

O

G

BnO BnO OH 143

90%

1

1. H2, Pd(OH)2/C 2. MeOH, HC1 90%

1


145

Scheme 38.

1. AcOH, HzO 2. Pb(OAc)4 3. pN02PhCOCl

//SiMe,,

*

141

PNBO

*

I OBn

1. 03; NaBH4 2. Hz, Pd(OH)2/C 3. MeOH, HC1

-

HO

146

H 148

Scheme 39.

BnO BnO

BF,.Et,O

,OH

W

HoOMe

:

,

,

520

20 Synthesis of C-Oligosaccharides 1. 0 3 ; MezS 2. CBr4, PPh? 3. BuLi 4. 12, morpholine 5 . KOzCN=NC02K. AcOH

I

BnO

150 1. Hz, WA1203

DHP, PPTS 3. B u ~ N F

OBn

151

OBn

/OBn

L.

- '\?oI?;1,

1. NaH, Me1 2. Hz, Pd(OH t>/I

5 . DMSO, 4. TsOH (COC1)2, Et3N

BnO% BnO BnO

'

'"'OH

OBn BnO

OTBS

152

- OH

nn

153

" HO :

o

UDll

x

HO 21

154

HoOMe

Scheme 40.

achieved by a sequence of reactions similar to that used for the synthesis of the Ccellobioside 145. The Kishi-Nozaki reaction [71] is the key step in the preparation of methyl Cisomaltoside (Scheme 40) [72]. Coupling of vinyl iodide 150 with the xylose derivative 151, readily available in six steps from L-xylose, afforded the corresponding allylic alcohol with a diastereomeric ratio of 15 : 1 in favor of 152. Finally, selective hydrogenation of the double bond and oxidation of the primary alcohol led to the C-disaccharide 153, which could be converted to its methyl glycoside 21. A useful dideuterated derivative of C-isomaltoside 154 for solution conformation studies was also accessible from this synthetic approach via the cis-deuteration of the alkene intermediate 152 with Pt on A1203 and D2. A similar strategy was applied for efficient synthesis of the C-glycoside analog of sucrose, starting from the vinyl iodide 155 and the aldehyde 156 (Scheme 41) [73]. The wrong diastereomer at the C3' position is obtained in the anti-selective Cr(I1)mediated reaction; this can nevertheless be corrected via a two-step Mitsunobu reaction and hydrolysis sequence. An anti-favored epoxidation and a regioselective acid-catalyzed ring formation by attack of the C5' hydroxyl group at the quaternary center of the epoxide finally completes the C-disaccharide 157 after a debenzylation step. Nicotra and coworkers have reported another synthesis of a C-glycoside analog of sucrose, as shown in Scheme 42, although missing the C2/-hydroxymethyl group [74].Wittig reaction of the stabilized phosphorane 159, prepared in four steps from C-ally1 glucose 158, with the isopropylene derivative of D-glyceraldehyde 160 rep-

521

20.4 The Partial de Novo Approach

1. DEAD, PhCOOH

2. KzCO~,MeOH 3. mCPBA

-OH 1. CSA

+

30% overall from 155

OH OH 157

Scheme 41.

Bi:oT

kc

Hg(OAc)2, NaCl97%t B g 3 0 Y o H i : 3. PPh,, Et,N

Brio

t

36%

HgCl

158

BnO0v

?%

-

160

88%

BnO

\

1. oso4, NMO 2. AczO, Py

-

159

PPh,

,OBn

52%

161 0 162

FeCI,.SiOz

-Bi:o%

1. Florisil 2. K2C03, EtOH 3. H2, Pd/C

HoOAc

78%

OAc

89%

OH

163

OH

164 OH

Scheme 42.

resents the key coupling step in this synthesis. Subsequent stereoselective dihydroxylation of alkene 161 afforded adduct 162 as the major product (4: 1 ratio) after acetylation; this was subjected to acid hydrolysis furnishing the pyranose derivative 163. Florisil treatment and two-step deprotection then gave the C-sucrose analog 164 in which the furanose and pyranose forms are in equilibrium. The C-dimer of trehalose (cL-D-G~c(~-~’)-cY,-D-G~c) can also be prepared by use of

522

20 Synthesis of C-Oligosaccharides 1. 03;Me$ 2. (Me0)2P(O)CHN2, tBuOK , i B 3. 12,morpholine t

QBn Y

BnO

88%

158

O

G%

O

165

166

1. H2, Lindlar cat. 2. BnBr, NaH

165 + 166

Bn

PBn

57%

(OBn 1. oso4 2. BuzSnO;

CsF, PMBCI, TBAI 3. A c ~ ODMAP, , Py

* 85%

*

63%

PMB~ 169

168 1. DDQ 2. TsOH

1. DMSO, (COC1)z. Et3N 2. BH,.THF 3. AcOH, H20 4. Hz. Pd(OH)I/C *

-

4. 3. MMTr, KzCO3, Py MeOHBi:o\

76%

OMMTr

74%

Hgo%

'OH

HO HOoH

BnOoBn

170

171

Scheme 43.

a Cr(II)/Ni(II)-mediated coupling reaction as reported by Kishi and Wei in 1994 (Scheme 43) [75].Condensation of the alkynyl iodide 165, prepared in three steps from C-ally1 glucose 158, with the four-carbon fragment 166 favored the formation of the syn-propargylic alcohol 167 with a diastereomeric excess of 60%. Further cis reduction of the trip bond and a hydroxyl group-protecting step, set the stage for a stereoselective dihydroxylation exploiting the 1,3-allylic strain effect. Regioselective protection of the diol then gave the protected polyol 168, which could be transformed in three steps to the primary epoxide 169. Exclusive formation of the tetrahydropyran 170 was then achieved by removal of the p-methoxybenzyl group and subsequent acid-catalyzed epoxide opening. Finally, by inverting the C2'stereocenter via an oxidation-reduction sequence and deprotection, the C-analog of a,a-trehalose 171 was made available.

20.4 The Partial de Novo Approach

523

woMe

HO

OH

X M e P O H HOoH

172 X = OH, Y = CH20H 173 X = H, Y = CH20H 174 X = OH, Y = H 175 X = Y = H

1

y

-

Me@OH

HO 176 X = OH, Y = CH2OH 177 X = H, Y = CH2OH 178 X = OH, Y = H 179 X = Y = H

Scheme 44.

A most impressive illustration in the application of the de nouo approach to the synthesis of C-oligosaccharides was provided by the same group in their synthesis of the C-trimer of the type II(H) cellular antigen trisaccharide (a-~-Fuc( 1-2)p-DGal(1-4)(3-~-GlcNAc(Scheme 44). In contrast to other stratagems, a sufficiently flexible aldol route was adopted to introduce specific structural modifications, so that a series of trimer analogs 172-179 could be prepared [8b, 761. As shown in Scheme 44, the bond disconnections require a glucosamine (or glucose) unit with a three-carbon appendage on the C4 carbon, a Cl-formyl fucose unit, and a fourcarbon acyclic fragment representing the future central galactose sugar. To prepare the glucosamine unit, the epoxide 180 was transformed in six steps to the C4branched sugar 182, including opening of the epoxide via a Cu(1)-catalyzed Grignard reaction and subsequent treatment with base to give the C2,C3 epoxide 181 (Scheme 45), which was then subjected to azide opening, benzylation, azide reduction, and phthalimide formation. Opening of the I ,6-anhydro derivative 182 and allylation at the C1 position led stereoselectively to the (3-C-glycoside 183; the stereochemical outcome can be explained by the anchimeric assistance of the C2phthalimido group. A five-step procedure then converted 183 to the ketone 184 which coupled smoothly with aldehyde 185 affording the aldol product 186. A sequence involving desilylation, thioglycoside formation, reduction and oxidation then set the stage for the next aldol coupling between the C-dimer 187 and aldehyde 188. In this reaction, optimum equilibrating conditions are required to produce the equatorial alcohols at the C2-position of the galactose unit in 189. Two stereoselective reductions on the enone product 190, obtained by base-promoted elimination of the corresponding mesylate of 189, led to the C-trimer 191 which was fully deprotected and converted to the polyol 176 in the form of its lithium carboxylate. Armstrong and Sutherland have reported a combinatorial approach to Cdisaccharides in which one sugar unit was kept constant while the other was varied to provide all possible stereoisomers [77]. To illustrate this concept, the C-allylated glucose derivative 158, was chain-extended to give the diene ester 192 in four conventional steps (Scheme 46). Dihydroxylation and DIBALH reduction afforded four stereoisomeric lactols with the DIL-galacto and the D/L-ido configurations

524

20 Synthesis of C-Oligosaccharides 1. NaN3, TBAI

I

1. A M g C I , CuI

3. Hz. Lindlar cat. 4.phthalic anhydride; Ac20 48%

12% 180 OTS

1. DBU, MeOH 2. BnBr, NaHMDS 3. 03;PPh3 4.PMBO(CH2)gPPh3Br,BuLi 5. Hz, (Ph,P),RhCl

1. Ac20, BF3.EtzO

~

-

sTMSOTf ,

w

14%

NPhth

183

NPhth

YMB

(CH2)s

OPMB

184

186

1. B u ~ N F 2. MeSH, BF3.Et20 3. Ph3SnH, AIBN 4.DMSO, (COC1)2, Et3N

t

187

OBn

Me OBn BnO

B

p

(

61%

p

H

1. MsCl, 2,6-lutidine 2. NH,

OBn

BnOOBn

189

HO OBn OBn

1 OBn BnO

191

)

8

,

OPM B

80% Me

OBn

2

OBn

t

1. Ph3SnH. AIBN 2. NaBH4

Scheme 45.

OPMB

OBn

-0PMB

OH

Po

OBn BnO LHMDS, TMEDA. MgBr2

85%

BnO

-

26%

OTBS

& Bn

Bn

OPMB

OBn 190

1. NH2NHz; A c ~ O 2. DDQ 3. (Ph3P)3RuCIZ 4.NaC1OZ,NaH2P04 5. Hz. Pd(OH),; LiOH

-

34%

176

20.4 The Partial de Novo Approach I. 9-BBN;HzOz 2. DMSO, (COC1)2, Et3N 3. (Et0)20P*COOEt, 4.12, PhH

tBuOK

525

I . 0~01. NMO

OEt 2. DIBALH

*

20%

69% 158

192

OH OH 0

QH OH 0 P H + \ * H

+

\

OH OH

OH OH

OH OH 5: 1

D/L-gaiacto

OH OH

D/L-ido

HO

HO HO

HO L-Gal

D-Gal 193

OH

HO

194

D-Id0 195

L-Id0 196

Scheme 46.

(represented in their acyclic form), which were subjected to hydrogenolysis to give the corresponding C-disaccharides 193-196. The same group has provided another example in their synthesis of C-glycoside analogs of the Lewis type I blood group determinant, of general formula a - ~ Fuc( 1-2)hexose( 1-3)p-~-GlcNAc,whereby combinations of the central hexose unit are constructed by non-selective transformations to give a mixture of diastereomers [ 781. On the other hand, predictable stereochemical transformations enable access to a sole C-trisaccharide. This strategy begins by the Nozaki-Kishi reaction between vinyl bromide 197, prepared in three steps from L-fucose, and the aldehyde 198, obtained in four steps from D-glucosamine, to give, after silylation, a 2: 1 mixture of allylic alcohols 199 and 200, respectively (Scheme 47). Diastereoselective hydroboration was performed on the major isomer 199, which after oxidation, then homologation with ally1 Grignard reagent and subsequent desilylation led to the diols 201 and 202 (2 : 1 ratio). Cyclization of 201 by initial epoxidation of the alkene and then deprotection affords a 1 : 1 mixture of C-trisaccharides 203 and 204. In this way, 11 C-trimers were synthesized. To access single components of reactions affording inseparable diastereomeric mixtures, these non-selective steps (marked by

OH

526

20 Synthesis of C-Oligosucchurides

Me OBn 197

Me

198

200(1) 1.9-BBN, H202 2. periodinane *3. A M g B r 4. BQNF 199

BnOoBn *l. mCPBA 2. cat. CSA 3. Hi,Pd(OH)$ 201

Me

OBn 201 (2)

-

OBn

I OBn BnO

202 ( I )

+

55% h Z U d O H

Me

I OH HO

203 (1)

HAOH

OH 204 (1)

Scheme 47.

an asterisk in Scheme 47) are substituted with selective steps, an approach described by the authors as a recursive stereochemical deconvolution strategy. Applying a conceptually different approach to the de novo design of Coligosaccharides, Vogel and coworkers described the use of stereoselective transformations on the rigid oxa-norbornane bicyclic system; some examples, shown in Schemes 25 and 26 in Section 20.3.1, rely on radical reactions for assembly of the two sugar units [51-531. Recently, Vogel and Gerber have also applied an ionic transformation, depicted in Scheme 48, starting from the optically pure ketone 205 [79]. A six-step procedure converted 205 to enone 206, a precursor of both ring units of the target C-dimer. Stereoselective conversion of this enone to the poly-oxygenated system 207 was followed by regioselective Bayer-Villiger oxidation, opening of the uronolactone, and subsequent dithioacetal hydrolysis to give the methyl uronic ester derivative 208. Aldol condensation of this aldehyde with the aluminum enolate 209, prepared from enone 206, gave rise to the C-disaccharide precursor 210 after carbonyl reduction. Application of a similar sequence of steps as for the preparation of 208 then afforded the protected form of the C-linked ~-~-mannuronic(l-3) dimer 211, which was ready for the introduction of a third sugar unit by the above-described aldol reaction.

20.5 The Cyclouddition and Rearrangement Approach

527

1. t-BuOOH, DBU 2. NaBH4 3. BzCI, Py 4. B u ~ N F

30%

60%

BnO

205 1. periodinane 2. CF$OOH, H,O 3. AcCI, Py

79%

1. mCPBA, NaHC03 2. EtSH, TfOH; MeOH 3. Hg(C104). H20; Ag2CO3

* 59%

O&iZ OAc BnO

OBz

MeOOC OAc Bz 0 208

207 1.208

Me2AISPh 206

2. NaBH4

6 steps

BnO

TBs%sph

OAIMe2

210

209

AcO BzO 211

I SEt COOMe

Scheme 48.

20.5 The Cycloaddition and Rearrangement Approach In a rather different strategy to access C-disaccharides, Paton and co-workers applied the 1,3-dipolar cycloaddition of nitrile oxides derived from pyranoses with unsaturated sugars as the key step for linking the two sugar units [SO, 811. As shown with nitrile oxide 213 (Scheme 49), generated by the dehydration of the nitromethyl sugar derivative 212, its cycloaddition to alkene 214 led to the regioselective formation of the isoxazoline cycloadducts 215 in a 4: 1 ratio. Deacetylation of the major isomer and two sequential reduction steps yielded a mixture of diastereomers, with the major one, 216, being hydrolyzed to the hydroxymethylene-bridged analog O f p-D-Xyl(1-6)p-D-Glc 217. In another example, nitrile oxide 213 was cyclized with an excess of the 2,3unsaturated sugar 218 generating, after a similar sequence of steps, a 1 : 1 mixture of p(1-2)- and P(1-3)-C-disaccharides 219 and 220 (Scheme 50). In this example, the facial selectivity is controlled by the substituents of the em-pyranoside.

528

20 Synthesis of C-Oligosucchurides

'

bo OBn

"20-X OAc

tolylene di-isocyanate Et,N Reflux *

93%

Ob

212 X = CH2NO2 213 X = CNO 1 . KCN, MeOH 2. H2, Raney Ni, B(OH), 1

.,.n,l

I I

0-

215 (dr = 4: 1)

(4 equiv)

OH

H

.I\

ot\

214

g

o

a

O

OH B

l rn \

92%

216

HO

Of\

217

Scheme 49.

OAc 0

steps

213 +

-1 218

OEt

OEt

HO.

OH

HO

OH 219

1

Hd

OH

OEt

220

Scheme 50.

References 1. (a) T. Feizi, Curr. Opin. Struct. Bio. 1993, 3, 107; (b) A. Varki, Proc. Nail. Acad. Sci. USA 1994. 91, 7390.

2. J. Beuth, H. L. KO, G. Pulverer, G. Uhlenbruck, H. Pichlmaier, GlycoconjugateJ. 1995, 12, 1, and references therein. 3. S. A. Wharton, W. Weis, J. J. Skehel, D. C. Wiley, The Influenza Virus; Plenum, New York, 1989. 4. (a) P. Sears, C.-H. Wong, Proc. Nutl. Acad. Sci. USA 1996, 93, 12086; (b) L. L. Kiessling, N. L. Pohl, Chem. Biol. 1996,3, 71; (c) L. Lasky, Annu. Rev. Biochem. 1995, 64, 113. 5. (a) S. Hakamori, Curr. Opin. Zmmun. 1991,3, 646; (b) T. Toyokuni, A. K. Singhal, Chem. SOC. Rev. 1995,24, 231. 6. For some examples, see: C.-H. Wong, F. Moris-Varas, S.-H. Hung, T. G. Marron, C.-C. Lin, K. W. Gong, G. Weitz-Schmidt, J. Am. Chem. Soc. 1998, 119, 8152, and references therein. 7. (a) M. H. D. Postema, C-Glycoside Synthesis; CRC Press, Boca Raton, 1995; (b) D. E. Levy, C. Tang, The Chemistry of C-Glycosides; Pergamon Press, Exeter, 1995; (c) G. Casiraghi, F. Zanardi, G. Rassu, P. Spanu, Chem. Rev. 1995, 95, 1677. 8. (a) T.-C. Wu, P. G. Goekjian, Y . Kishi, J. Org. Chem. 1987, 52,4819; (b) P. G. Goekjian, T.-C. Wu, H.-Y. Kang, Y. Kishi, J. Org. Chem. 1987, 52, 4823; (c) S. A. Babirad, Y. Wang, P. G. Goekjian, Y. Kishi, J. Org. Chem. 1987, 52, 4825; (d) A. Wei, A. Haudrechy, C. Audin, H.-S. Jun, N. Haudrechy-Bretel, Y. Kishi, J. Org. Chem. 1995, 60, 2160 and previous work. 9. (a) A. Wei, K. M. Boy, Y. Kishi, J. Am. Chem. Soc. 1995, 117, 9432; (b) R. Ravishankar, A. Surolia, M. Vijayan, S. Lim, Y. Kishi, J. Am. Chem. SOC.1998, 120, 11297. 10. (a) J.-F. Espinosa, F. J. Caiiada, J. L. Asensio, H. Dietrich, M. Martin-Lomas, R. R. Schmidt, J. Jimenez-Barbero, Angew. Chem. 1996, 108, 323; Angew. Chem., Int. Ed. Engl. 1996, 35, 303;

Rejerences

529

(b) J.-F. Espinosa, F. J. Caiiada, J. L. Asensio, M. Martin-Pastor, H. Dietrich, M. MartinLomas, R. R. Schmidt, J. Jimenez-Barbero, J. Am. Chem. Soc. 1996, 118, 10862; (c) J.-F. Espinosa, E. Montero, A. Vian, J. L. Garcia, H. Dietrich, M. Martin-Lomas, R. R. Schmidt, A. Imberty, J. Caiiada, J. Jimenez-Barbero, J. Am. Chem. Soc. 1998, 120, 1309. 11. 0. Martin, L. Liu, F. Yang, Tetrahedron Lett. 1996, 37, 1991. 12. (a) B. A. Jones, Y . T. Pan, A. D. Elbein, C. R. Johnson, J. Am. Chem. Soc. 1997, 119, 4856; (b) A. Baudat, P. Vogel, J. Org. Chem. 1997, 62, 6252. 13. (a) J.-M. Beau, T. Gallagher, Topics Curr. Chem. 1997, 187, 1; (b) P. Vogel, R. Ferritto, K. Kraehenbuehl, A. Baudat, in Curbohjdrate Mimics (Ed. Y . Chapleur), Wiley-VCH, Weinheim, 1995, p. 19. 14. D. Rouzaud, P. Sinay, J. Chem. Soc. Chem. Commun. 1983, 1353. 15. J.-M. Lancelin, P. H. Amvam Zollo, P. Sinay, Tetrahedron Lett. 1983, 24, 4833. 16. Y.-C. Xin, Y.-M. Zhang, J.-M. Mallet, C. P. J. Glaudemans, P. Sinay, Eur. J. Org. Chem. 1999, 471. 17. H. Streicher, A. Geyer, R. R. Schmidt, Chem. Eur. J. 1996,2, 502. 18. M. A. Leeuwenburg, C. M. Timmers, G. A. van der Marel, J. H. van Boom, J.-M. Mallet, P. Sinay, Tetrahedron Lett. 1998, 38, 6251. 19. K. M. Nicholas, Acc. Chem. Res. 1987, 20, 207. 20. S. Tanaka, M. Isobe, Tetrahedron 1994, 50, 5633. 21. R. Preuss, R. R. Schmidt, Liebigs Ann. Chem. 1989, 983. 22. R. R. Schmidt, R. Preuss, Tetrahedron Lett. 1989, 30, 3409. 23. R. R. Schmidt, A. Beyerbach, Liebigs Ann. Chem. 1992, 983. 24. H. Dietrich, R. R. Schmidt, Liebigs Ann. Chem. 1994, 975. 25. T. Eisele, H. Ishida, G. Hummel, R. R. Schmidt, Liebigs Ann. Chem. 1995, 2113. 26. B. Patro, R. R. Schmidt, Synthesis 1998, 1731. 27. J. Alzeer, C. Cai, A. Vasella, Helv. Chim. Acta 1995, 78, 242. 28. S. M. Daley, R. W. Armstrong, Tetrahedron Lett. 1989, 30, 5713. 29. (a) J.-M. Lancelin, L. Morin-Allory, P. Sinay, J. Chem. Soc. Chem. Commun. 1984, 355; (b) V. Pedretti, A. Veyrieres, P. Sinay, Tetrahedron 1990, 46, 77. 30. D. Mazeas, T. Skrydstrup, J.-M. Beau, Angew. Chem. 1995, 107, 990; Angew. Cheni. Int. Ed. Engl. 1995,34, 909; T. Skrydstrup, 0. Jarreton, D. Mazeas, D. Urban, J.-M. Beau, Chem. Eur. J., 1998, 4 , 655. 31. D. MazCas, T. Skrydstrup, 0. Doumeix, J.-M. Beau, Angew. Chem. 1994, 106, 1457; Angew. Chem. In?. Ed. Engl. 1994, 33, 1383; T. Skrydstrup, D. Mazeas, M. Elmouchir, G. Doisneau, J.-M. Beau? Chem. Eur. J. 1997,8, 1342. 32. M. E. Jung, S. W. T. Choe, Tetrahedron Lett. 1993, 34, 6247. 33. 0. Jarreton, T. Skrydstrup, J.-M. Beau, J. Chem. Soc., Chem. Commun. 1996, 1661; 0. Jarreton, T. Skrydstrup, J.-F. Espinosa, J. Jimenez-Barbero, J.-M. Beau, Chem. Eur. J., 1999, 5, 430. 34. S. L. Krintel, J. Jimenez-Barbero, T. Skrydstrup, Tetrahedron Lett., 1999, 40, 7565. 35. L. Andersen, L. M. Mikkelsen, J.-M. Beau, T. Skrydstrup, Synlett 1998, 1393. 36. S.-C. Hung, C.-H. Wong, Tetrahedron Lett. 1996, 37, 4903. 37. S.-C. Hung, C.-H. Wong, Angew. Chem. 1996, ION, 2700; Angew. Clzem. Int. Ed. Engl. 1996, 35, 2611. 38. I. R. Vlahov, P. I. Vlahova, R. J. Linhardt, J. Am. Chem. Soc. 1997, 119, 1480; Y. Du, R. J. Linhardt, Carbohydr. Res. 1998, 308, 161; T. Polat, Y . Du, R. J. Linhardt, Synlett 1998, 1195. 39. Y. Du, T. Polat, R. J. Linhardt, Tetrahedron Lett. 1998, 39, 5007. 40. (a) L. Petrus, S. Bystricky, V. Bilik, Chem. Zvesti 1982, 36, 103; (b) A. Fortsch, H. Kogelberg, P. Koll, Carbohydr. Res. 1987, 164, 391. 41. 0. R. Martin, W. Lai, J. Ory. Chem. 1990, 55, 5188. 42. 0. R. Martin, W. Lai, J. Org. Chem. 1993, 58, 176. 43. Z . J. Witczak, R. Chhabra, J. Chojnacki, Tetruhedron Lett. 1997, 38, 2215. 44. W. R. Kobertz, C. R. Bertozzi, M. D. Bednarski, J. Org. Chem. 1996, 61, 1894. 45. R. Preuss, R. R. Schmidt, J. Carbohydr. Chem. 1991, 10, 887; R. Preuss, K.-H. Jung, R. R. Schmidt, Liebigs Ann. Cliem. 1992, 377. 46. A. Dondoni, M.-C. Scherrmann, J. Org. Chem. 1994,59, 6404.

530 47. 48. 49. 50.

20 Synthesis of C-Oligosucchurides

A. Dondoni, A. Boscarato, H. Zuurmond, Tetrahedron Lett. 1996,37, 7587. A. Dondoni, H. Zuurmond, A. Boscarato, J. Org. Chem. 1997, 62, 8114. A. Dondoni, M. Kleban, H. Zuurmond, A. Marra, Tetrahedron Lett. 1998, 39, 7991. B. Giese, T. Witzel, Angew. Chem. 1986, 98, 459; Angew. Chem. Int. Ed. Engl. 1986, 25, 450; B. Giese, M. Hosh, C. Lamberth, R. R. Schmidt, Tetrahedron Lett. 1988,29, 1375. 51. R. M. Bimwala, P. Vogel, Tetrahedron Lett. 1991, 32, 1429. 52. R. M. Bimwala, P. Vogel, J. Org. Chem. 1992, 57, 2076. 53. R. Ferrito, P. Vogel, Tetrahedron Asym. 1994, 5, 2077. 54. 0. R. Martin, F. Xie, R. Kakarla, R. Benhamza, Synlett 1993, 165. 55. P. Sinay, Pure & Applied Chem. 1997, 69, 459; P. Sinay, Pure & Applied Chem. 1998, 70, 407. 56. Y.-C. Xin, J.-M. Mallet, P. Sinay, J. Chem. Soc., Chem. Commun. 1993, 864. 57. A. Mallet, J.-M. Mallet, P. Sinay, Tetrahedron Asym. 1994, 5, 2593. 58. G. Rubinstein, J.-M. Mallet, P. Sinay, Tetrahedron Lett. 1998, 39, 3697. 59. A. Chenede, E. Rekai, E. Perrin, P. Sinay, Synlett 1994, 420. 60. A. Chenede, Ph. D. Thesis, Universite Pierre et Marie Curie, 1996. 61. B. Vauzeilles, D. Cravo, J.-M. Mallet, P. Sinay, Synlett 1993, 522. 62. E. D. Rekai, G. Rubinstein, J.-M. Mallet, P. Sinay, Synlett 1998, 831. 63. Y. Ito, T. Ogawa, Angew. Chem. 1994, 106, 1843; Angew. Chem. Int. Ed. Engl. 1994,33, 1765; A. Dan, Y. Ito, T. Ogawa, J. Org. Chem. 1995, 60, 4680. 64. G. Rubinstein, J. Esnault, J.-M. Mallet, P. Sinay, Tetrahedron Asym. 1997, 8, 1327. 65. B. Vauzeilles, Ph. D. Thesis, Universite Pierre et Marie Curie, 1996. 66. G. Rubinstein, Ph. D. Thesis, Universite Pierre et Marie Curie, 1996. 67. A. Helmboldt, J.-M. Mallet, M. Petitou, P. Sinay, Bull. Soc. Chim. Fr. 1997, 134, 1057; A. Helmboldt, M. Petitou, J.-M. Mallet, J.-P. Htrault, J.-C. Lormeau, P. A. Driguez, J.-M. Herbert, P. Sinay, Bioorg. Med. Chern. Lett. 1997, 7, 1507; M. Petitou, J.-P. Herault, J.-C. Lormeau, A. Helmboldt, J.-M. Mallet, P. Sinay, J.-M. Herbert, Bioorg. Med. Chem. Lett. 1998, 8, 1509. 68. R. W. Armstrong, J.-M. Beau, S. H. Cheon, W. S. Christ, H. Fujioka, W.-H. Ham, L. D. Hawkins, H. Jin, S. H. Kang, Y. Kishi, M. J. Martinelli, W. W. McWhorter Jr., M. Mizumo, M. Nakata, A. E. Stutz, F. X. Talamas, M. Taniguchi, J. A. Tino, K. Ueda, J.4. Uenishi, J. B. White, M. Yonaga, J. Am. Chem. Soc. 1989,111, 7525 and 7530. 69. (a) J.-F. Espinosa, H. Dietrich, M. Martin-Lomas, R. R. Schmidt, J. Jimenez-Barbero, Tetruhedron. Lett. 1996, 37, 1467; (b) J. F. Espinosa, M. Bruix, 0. Jarreton, T. Skrydstrup, J.-M. Beau, J. Jimenez-Barbero, Chem., Eur. J. 1999, 5, 442. 70. S. A. Babirad, Y. Wang, Y. Kishi, J. Org. Chem. 1987, 52, 1372; Y. Wang, S. A. Babirad, Y. Kishi, J. Org. Chem. 1992, 57, 468. 71. (a) H. Jin, J. Uenishi, W. J. Christ, Y. Kishi, J. Am. Chem. Soc. 1986, 108, 5644; (b) K. Takai, M. Tagashira, T. Kuroda, K. Oshima, K. Utimoto, H. Nozaki, J. Am. Chem. Soc. 1986, 108, 6048. 72. P. G. Goekjian, T.-C. Wu, H.-Y. Kang, Y. Kishi, J. Org. Chem. 1987, 52,4823; P. G. Goekjian, T.-C. Wu, H.-Y. Kang, Y. Kishi, J. Org. Chem. 1991, 56, 6422. 73. U. C . Dyer, Y. Kishi, J. Org. Chem. 1988, 53, 3384; D. J. OLeary, Y. Kishi, J. Org. Chem. 1993, 58, 304. 74. M. Carcano, F. Nicotra, L. Panza, G. Russo, J. Chem. Soc. Chem. Commun. 1989,642. 75. A. Wei, Y. Kishi, J. Org. Chem. 1994, 59, 88. 76. T. Haneda, P. G. Goekjian, S. H. Kim, Y. Kishi, J. Org. Chem. 1992, 57, 490. 77. R. W. Armstrong, D. P. Sunderlin, Tetrahedron. Lett. 1994, 35, 7743. 78. D. P. Sunderlin, R. W. Armstrong, J. Am. Chem. Soc. 1996, 118, 9802; D. P. Sunderlin, R. W. Armstrong, J. Org. Chem. 1997, 62, 5267. 79. P. Gerber, P. Vogel, Tetrahedron. Lett. 1999, 40, 3165. 80. R. M. Paton, K. J. Penman, Tetrahedron. Lett. 1994, 35, 3163. 81. I. M. Dawson, T. Johnson, R. M. Paton, R. A. C. Rennie, J. Chem. Soc. Chem. Commun. 1988. 1339.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

21 Synthesis of Oligosaccharide Mimics: S-Analogs Jon K. Fairweather and Hugues Driguez

21.1 Introduction It has been known for some time that carbohydrate-protein recognition is crucial in the biotransformation of natural oligo- and polysaccharides and, in recent years, it has also been considered that such interactions are the essential process for biological transfer of information into living organisms. Investigations of such recognition processes promote deeper understanding which might ultimately be applied to new concepts of enzyme engineering and glycotherapy. There are, essentially, two approaches to such investigations; site-directed mutagenesis of proteins and use of natural substrates or the use of native proteins and modified substrates. Whereas the former of these approaches, using the techniques of protein engineering, is becoming more popular [ l , 21, modification of substrates is somewhat more intriguing to the synthetic chemist. Some would argue that the introduction of specific alterations in the natural substrate and their interactions with native proteins are more challenging to analyze, and that only such an approach addresses the fundamental aspects of enzyme-substrate interactions-the nature of recognition, binding, and the catalytic process. Surely, one of the most versatile approaches for the study of glycosylhydrolases is the use of non-hydrolyzable oligosaccharide mimics, as competitive inhibitors, to map the active site. This can be achieved, in most respects, by substituting the natural glycosidic linkage (0)with a recognizable mimic-the analogs of choice thus far have been methylene groups (CH2) [3], free or functionalized amines (NR) [4], or sulfur atoms (S) [ 5 ] . This review will focus on thiooligosaccharides which have been synthesized with a biological application in mind.

532

21 Synthesis of Oligosaccharide Mimics: S-Analogs

21.2 General Synthesis The synthesis of thiooligosaccharides has traditionally been conducted via one of two strategies-(a) Lewis acid catalysed condensation between a glycosyl acceptor containing an SH group and a suitable glycosyl donor, or (b) S~2-likedisplacement of a leaving group of a glycosyl acceptor with a 1-thiolate or of a glycosyl halide by a sugar thiolate. Michael addition of thiolates to a,p-unsaturated systems [6], epoxide and aziridine ring openings [7], and chemoenzymatic approaches [8] have also been applied as alternatives to the traditional approaches. 21.2.1 Preparation of Thioglycoses 1-Thioglycoses

The most widely used synthons for the synthesis of thiooligosaccharides are the 1thioglycoses and their synthesis has been widely studied. Approaches for the synthesis of 1,2-trans-thioglycoses include nucleophilic sulfur substitutions at the anomeric center of the glycosyl halides with either pseudo-thourea derivatives or other thionucleophiles [9, 101. More recently, phase-transfer catalysis [ 111 and mild Lewis acid catalysed glycosidation of peracetylated sugars with thioacetic acid [ 121 have been applied. The first reported synthesis of peracetylated 1-thio-a-D-glucopyranose started from the Brigl’s anhydride [9] but traditionally, 1,2-cis-thioglycoses have been generated from the 1,2-trans-glycosylhalides by reaction of alkyl or benzyl xanthates in acetone [ 131, potassium thioacetate in hexamethylphosphoramide (HMPA) [ 141, or the tetrabutylammonium salt of thioacetic acid in toluene [15] (Scheme 1). A recent alternative is the peroxide-induced addition of thioacetic acid to hydroxy-glucal derivatives [ 161. 2-, 3-, 4-, 5, or 6-Thioglycoses

The action of thioacetate, thiobenzoate or thiocyanate anions on sugars bearing leaving groups (halides or sulfonyl), epoxides, and cyclic sulfates has afforded the thiosugars in good yield (Scheme 2).

+

-sx

____c

Br

Scheme 1. S~2-typesubstitution of glycosyl halides by thiolates.

21.2 General Synthesis

533

1,6-[17]

1,3 -[201

F-

TfO

sx

Scheme 2. General procedure for the synthesis of thioglycoses.

21.2.2 Selective S-Deprotection of Thioglycoses

The most widely used synthons for the preparation of thiooligosaccharides have been peracetylated 1-thioglycoses, which can be selectively deprotected either in a two step procedure or in situ and used as an activated species (Scheme 3). There are several methods for the selective S-deacetylation of thioglycoses; one of the most successful of these is demercuration of phenylmercury thio-a- or -P-Dglycoses [22, 231. Deprotection has also been achieved by use of sodium methoxide

R=Ac

Scheme 3. Activated thioglycosyl donors.

R=AcorH

R=AcorH

534

21 Synthesis of Oliyosaccharide Mimics: S-Analogs

at low temperature (less than -40°C) [24, 251 although 0-deacetylation is sometimes observed [24]. The thiolate anion is, however, more nucleophilic than its oxygen counterpart, and it should be noted that the fully deprotected 1-thiolates (usually generated in situ) can also be used in such reactions. A more preferential chemoselective deprotection has been achieved by the action of diethylamine in DMF [26], cysteamine in either acetonitrile [27] or HMPA containing dithioerythritol (DTE) [28], or hydrazinium acetate in DMF [ 111. The same procedures can also be used to obtain compounds with the free thiol at the 2-, 3-, 4-, 5-, or 6- positions. Although these derivatives are very prone to forming disulfide linkages, this can be overcome by treatment of such compounds with DTE in aqueous solution thereby generating the free thiol [29]. 21.2.3 Glycosylation Methods Both traditional and new approaches (often used in the preparation of 0-glycosides) have been applied to the establishment of thio linkages. Here, the synthesis of thiodisaccharides and thiooligosaccharides is described with reference to linkage specificity and method of generation.

21.3 Establishment of 1,6-Thio Linkages 21.3.1 6-Thiodisaccharides The first reducing thiodisaccharide synthesized was methyl a-thiogentiobioside (l), albeit in a very poor yield (19%), from the condensation of the tosylate (2) and the 1-thiolate (3) [30]. The low yield is probably attributable to the poor leaving ability of the tosylate group in acetone-water. Although the displacement of a glycosyl halide with a sugar thiolate overlaps the preparation of the 1-thioglycose itself, the synthetic utility of this reaction is noteworthy and, as an alternative, this work also compares the displacement of the glycosyl halide (4) with a sugar thiolate generated from 5 (Scheme 4). Methyl a-thiogentiobioside (1) was synthesized in this manner, in much higher yield (55%) [30].

OMe

2R=OTs 5R=SAc

3 R’ = SNa5R2 = H 4 R ’ = H. R = B r

OMe 1

Scheme 4. (i) K2CO3 (acetone/HzO); (ii) NaOMe (MeOH) then (acetone); (iii) AczO (pyridine).

535

21.3 Establishment of l,6-Thio Linkages

SH

R1 R2 R3 H OAc 7 OAc 9 OAc OAc H 10 NHAc OAc H

R1 R2 R3 8 OAc H OAc 11 OAc OAc H 12 NHAc OAc H

6

Scheme 5. (i) CsCO, (DMF).

Santoyo-Gonzales and co-workers [ 3 11 have recently developed a highly stereoselective and simple method for introducing thio linkages. Using the 4,6-cyclic sulfate acceptor 6, the galactosyl thiol 7 was condensed, in a cesium carbonatepromoted reaction, to yield the S-linked disaccharide 8 in excellent yield (89%) (Scheme 5). The versatile nature of this reaction has been applied to other donors; the glucosyl and N-acetylglucosaminyl thiols 9 and 10 have yielded the thiodisaccharides 11 and 12 in 52% and 68% yield, respectively. This reaction has also been generalized and a convenient route to P-1,4-gluco and P-1,3-allo thiodisaccharides has been established (see Section 21.4.1). An improved preparation of tri-O-acetyl-l,6-epithioP-D-glucose (13) [32] and the selective opening at C6 with 9 to yield the disaccharide 14 [33] (Scheme 6) could be an interesting alternative for the synthesis of complex thiooligosaccharides; the resulting thiodisaccharide or an analog refunctionalized from 15 can itself act as a donor for the synthesis of 0-glycosides. An approach to the synthesis of 1,6-thio linkages to mannose residues has recently been reported whereby the manno- 1,2-epoxide (16) was opened with the thiolate of 17 to generate the thiodisaccharide (18) in moderate yield (63%) [34] (Scheme 7). Although there are many other synthetic possibilities for the preparation of 1,6thio linkages, arguably none is easier or more effective than the displacement of a 6-deoxy-6-iodo derivative with a 1-thiolate. Recalling the poor yield of the methyl

i

AcO OAc

OAC

RO

SH OAc

SEt

RO ‘OR

13

9

Scheme 6. (i) Et3OBF4 (CH2C12); (ii) NaOMe (MeOH).

536

21 Synthesis of Oligosuccharide Mimics: S-Analogs

BnO ,OBn

\

OBn

16

17

18

Scheme 7. (i) NaH (DMF).

AcO AcO

OAc

ACO

AcO

SH AcO

-

19 R’ = OAc R2 = H 21 R’ = H, R’ = OAC

9 R’ = OAc, R2 = H 7 R’ = H, R2 = OAC

20

OAc

AcO

Scheme 8. (i) NaH (THF), then DMF; (ii) K2CO3 (acetone/H20).

a-thiogentiobioside (1) from 6-0-tosyl substitution (19%), the 6-thiogentiobiose derivative (19) has subsequently been obtained in good yield (73%) through halogen substitution of the 6-deoxy-6-iodoglucose (20) with the sodium salt of the thiol (9) in DMF [35]; 6-thioallolactose (21) was similarly obtained in moderate yield (59%) from the potassium salt of the donor 7 in acetone [36] (Scheme 8). Condensation of the iodide 20 with the thiolate 22 generated from 23, by selective S-deacetylationwith cysteamine in HMPA with DTE, afforded the 6-thioisomaltose (24) in good yield (69%) [37] (Scheme 9).

A AcOc

O OAc a OAc

+

AGO% ACO

i

AcO

R

AcO

Acfc+

AcO*AcO OAc OAc

20

Scheme 9. (i) cysteamine/DTE (HMPA).

22 R = S-

23 R = SAC

24

21.3 Establishment ojl,6-Thio Linkages ACO,

OAc 1

537 yOpMe

-

Acd

R' I

R*

R' R2 R3 25 OAc OMe H 26 NHAc OMe H 32 NHAc H O(CH2)2N3

R' R2 R3 R4 27 Br OAc OMe H 28 Br NHAc OMe H 31 OTS NHAc H O(CH2)2N3

29R=Ac 30R=Na

Scheme 10. (i) EtzNH (DMF); (ii) (DMF)

Other easily generated leaving groups have been investigated and two recent papers suggest that D M F is a good solvent for the effective substitution of tosylate and bromine at C-6 of sugars when used in conjunction with the appropriately activated donor. The N-acetylneuraminic acid derivatives 25 and 26 have been synthesized quite efficiently (82 and 75%, respectively) when the methyl 6-bromo-6-deoxya-D-glycosides (27 and 28, respectively) were coupled with the 2-S-acetyl-Neu5Ac (29) [26] (Scheme 10). The S-deacetylated and activated donor was prepared in situ by treatment of thioacetate 29 with diethylamine in DMF (room temperature). The sodium salt (30) could be isolated and was condensed with the tosylate (31) by heating in DMF, thereby generating the a(2-6) thiodisaccharide 32 in appreciable yield (76%) [38]. The sodium salt of 1-thio-a-L-fucopyranose (33) was generated by treatment of thiol 34 with NaH (DMF); subsequent condensation with the ally1 glycoside 35 afforded the a( 1-6) thiofucosyl disaccharide 36 in remarkable yield (99%) through a sequence of o-acetylation in the presence of dimethylancinopyridine (DMAP) and de-o-acetylation [7] (Scheme 11).

HO

-

Me%OH

i-iii

A AcOc Me&slA~ AcO

O

+

S0 NHAc

HO*HO

0 NHAc

33R=Na 34R=H

35

36

Scheme 11. (i) NaH (THF), then (DMF); (ii) Ac20-DMAP (pyridine); (iii) NaOMe (MeOH).

21 Synthesis of Oligosaccharide Mimics: S-Analogs

538

A AcO c

O

S

i OAc

AcO

38 R' = H,R2 = SC(=S)OEt 19 R' = H,R2 = OAc 39 R' = Br, R2 = H

37

Scheme 12. (i) NaOMe (CHC13), then (DMPU).

21.3.2 6-Thiooligosaccharides The dithiogentiotriose 37 was prepared by methoxide treatment of the thiodisaccharide 38, generated from 19 via the bromide 39, and the iodide 20 in 1,3dimethyl-2-oxohexahydropyrimidine (DMPU) (50%) [35] (Scheme 12). A similar approach has resulted in the synthesis of thiomaltose derivatives; the peracetylated 1-thio-a-D-glucopyranose 23 was coupled with 6-deoxy-6-iodomaltose (40) affording the thiotrisaccharide 41 (87%) [ 391; 6"-deoxy-6"-iodomaltose (42) afforded the thio-a-glycosyl derivative 43 in excellent yield (80%) [40]. In addition, the 6W-deoxy-6W-iodomaltooligosaccharides44 and 45 afforded the expected 6O-Sa-D-glucopyranosyl 6w-thiomaltooligosaccharides 46 (86%) and 47 (80%), respectively [41] (Scheme 13). The thiodisaccharide 18 (obtained as described earlier) was condensed in the presence of triethylsilyltriflate (TESOTF) with the trichloroacetimidate 48 to generate the thiotrisaccharide 49 in very good yield (88%) [34] (Scheme 14). This compound has been used as a probe of N-acetylglucosyltransferase [34].

21.3.3 Branched Thiocyclodextrins The selective deacetylation and in situ activation of thioacetates with cysteamine and DTE has been widely used for the synthesis of numerous cyclodextrin derivatives. The branched CDs 50 (82%), 51 (63%) and 52 (85%) were synthesized from 6-deoxy-6-iodocyclomaltoheptaose (53) and the S-activated glycoses of 23, 54, and 55, respectively [39, 421. Interestingly, only activated. 1-thio-a-D-mannopyranose (56) gave the expected CD (57) in high yield (83%) [42] (Scheme 15). Alternative routes to branched CDs have also been investigated. The branched thiocyclomaltoheptaoses 50 and 51 have also been obtained in 64 and 60% yield, respectively, from 6-0-tosylcyclomaltoheptaose (58) by s N 2 displacement with the I-thiolates 59 and 60 [43]. The utility of such a reaction for tethering cyclodextrins

21.3 Establishment of 1,6-Thio Linkages

AcO

+

539

Acfc*

k2c%&

23

OAc OAc

AcO

OAc OAc

40

41

AcO

+

23

Aco+oe) AcO

AcO

0 Ac

AcO

OAc

n

42n=1 44n=2 45n=3

43n=l 46n=2 47n=3

Scheme 13. (i) cysteamine/DTE (HMPA).

BnO

+

cc13

18

i

BnO BnO

O(CH2)7CH3

OBn 48

49

Scheme 14. (i) TMSOTf (CH2C12).

is exemplified by the more recent synthesis of a bis-thiocyclomaltooligosaccharide with an a,a-trehalosyl linker (61) upon condensation of 62 with 58 in DMF in the presence of sodium iodide [44]. Tosylate and bromide groups have a similar efficiency as leaving groups and the heptakis(6-bromo-6-deoxy)cyclomaltoheptaose (63) yielded the heptakis-6-thiogalactosylcyclomaltoheptaoses64 (80%) and 65

540

21 Synthesis of Oligosuccharide Mimics: S-Analogs , R’

R’ R2 53 I OAc 58 OTs OH 63 Br Br

R3 OAc OH OH

R2

23 54 55 56 59 60 66 67 62

R’ R2 R3 R4 R5 H SAC OAc H OAc SAC H OAc H OAc SAC H OAc H OAc H SH H OAcOAc SNa H OH H OH H SNa OH H OH SNa H OH H OH H SNa OH H OH

R7 OAc OAc H OAc OH OH H H

50 R‘ 51 R’ 52 R’ 57 R’ 64 R’ 65 R’

= a-S-Glc, R2 = OH = P-S-Glc, R2 = OH = p-S-Gal, R2 = OH = a-S-Man, R2 = OH

= R2 = a-S-Gal = R2 = p-S-Gal

61 R’ = o @ *:H i

:;*O~~OH HO 0

R6 H H OAc H H H OH OH

SNa

HO 0

OH ,R S-6p-CD

Scheme 15. (i) Cysteamine/DTE (HMPA), then NaOMe (MeOH); (ii) (DMPU), then NaOMe (MeOH); (iii) NaI (DMF), then NaOMe (MeOH).

(50%) although only when treated with the sodium salts of the 1-thiogalactoses 66 and 67, respectively, at elevated temperature [45] (Scheme 15). A new and exciting application to the synthesis of branched cyclodextrins has recently been reported, whereby acarbose (68), the known potent inhibitor of glucoamylase, has been tethered to a cyclodextrin via flexible spacers of different length [46]. The functionalized spacers (69 and 70) were S-deacetylated and the corresponding thiolates coupled easily to 6-deoxy-6-iodocyclodextrin(53),affording the branched fully acetylated cyclodextrins 71 (76%) and 72 (64%), respectively.

~=OH

21.4 Establishment ofl,4-Thio Linkages

541

These were refunctionalized as their iodides (73 and 74) and elongation of 73 with the spacer 69 under similar conditions generated the trityl ether 75 and thence the iodide 76. The thioacetate derivative of acarbose (77), was coupled to the cyclodextrin 53 and branched cyclodextrins 73, 74, and 76, and then subsequently deprotected, readily affording the acarbose tethered branched cyclodextrins 78 (63%), 79 (88%), 80 (83%) and 81 (84O), respectively, in a good yield (Scheme 16).

21.4 Establishment of 1,4-Thio Linkages 21.4.1 1,CThiodisaccharides General Approaches The first report of SN2-like displacement of glycosyl halides -y sugar thiolates for generating 1,4-thio linkages emerged in 1978 [ 141. Treatment of the 1,2-trans glycosyl halide 82 with the sodium salt generated from methoxide treatment of the methyl 4-thio-a-~-glucoside83, afforded methyl 4thiomaltoside (84), albeit in poor yield (34%) [ 141. Similarly, 4-thiocellobioside (85) and 4-thiodigalactobioside (86) were also synthesized but in only slightly better yield (52% and 56%, respectively) from the halides 4 and 87 and the corresponding thiolates from 83 and 88 [14] (Scheme 17). Methyl 4-thiolactoside (89) was also synthesized from the bromide 87 by a slight modification of this procedure. Somewhat surprisingly, however, the reverse coupling failed to give the expected thiodisaccharide [47]. A synthesis of the 4-thiogalabioside (90) has been reported whereby the thiolate of 91, generated by treatment with cesium carbonate, has been condensed with the 1,2-cis chloride 92 in good yield (85%) [48] (Scheme 18). The benzylated 1,6-anhydro P-D-glucose (93) has been coupled to trimethylsilyl4thio a-D-glucoside (94) under acidic conditions to afford the thiodisaccharide 95 in good yield (66%; a/P, 15: 1) [49] (Scheme 19). A noteworthy comment from the authors reveals, however, the poor reactivity of the 1,6-anhydro disaccharide (not shown). As eluded to earlier for the preparation of 1,6-thiodisaccharides (Section 21.3.1) the cyclic sulfate group has been targeted as a means of establishing thio linkages [31]. An extension of this has seen the thiodisaccharides 96 (73%), 97 (61%), and 98 (64%) produced from the regioselective opening of the 3,4-cyclic sulfate 99 by the thiols 7, 9, and 10, respectively (Scheme 20). On the other hand, nucleophilic attack of the same thiols (7, 9, and 10) on the corresponding methyl p-glycoside gave p(1,3)-thiodisaccharides in the gulo series. S~ZDisplacementon Triflates Whereas nucleophilic attack of an anomeric alkoxide on primary triflates is extremely efficient (50-80%), the attack on secondary triflates is somewhat lower

542

21 Synthesis of Oligosaccharide Mimics: S-Analogs

69 rn = 1, n = 2 70 m = 2. n = 3

68

I 77

I

vii-k

SAC AcO

Y

i-iii

71 rn = 1, n = 2, R = OTr 72 rn = 2, n = 3, R = OTr 73 rn = 1, n = 2, R = I 74 rn = 2, n = 3, R = I 75 rn = 4, n = 3, R = OTr 76 rn = 4, n = 3, R = I

J iv-vi

x-xi

CH3

HO

78rn=O 79 rn = 1, n = 2 80rn=2,n=3 81 m = 4 , n = 3

Scheme 16. (i) NaOMe (MeOH); (ii) (DMF); (iii) Ac20-DMAP (pyridine); (iv) HBF4/H20 (CH3CN); (v) MsCl (pyridine); (vi) NaI (DMF); (vii) A q O (pyridine); (viii) HBr/HOAc (CH2C12); (ix) KSAc (DMF); (x) Et2NH (DMF); (xi) NaOMe (MeOH), then NH40H.

21.4 Establishment of 1,l-Thio Linkages

a

civ

AcS% AcO

CI

AcO AcO

___t

OAC

OMe

83

82

84

-

83 AcO

543

i-iV

H HO

O

a HO

Br

4

OH

85 HO

Ace%

HO

i-iv

AcO

Br

OMe

87

88

83

OMe

86

i,ii,v,iv

HO "o

~

)geo"e

+

87

HO

89

OH

Scheme 17. (i) NaOMe (MeOH); (ii) (HMPA); (iii) Ac2O (pyridine); (iv) NaOMe (MeOH); (v) BzCl (pyridine).

Acop Ac;s .& [ii

AcO

OAc

~

OAc OEtTMS

+

HO*s 92

91

Scheme 18. (i) Cs2CO3 (DMF); (ii) NaOMe (MeOH).

90

OEtTMS

OH

544

21 Synthesis of OligosaccharideMimics: S-Analogs

i,ii OMe

OBn

BnO

OMe

95 (+ p-linked isomer)

94

93

Scheme 19. (i) ZnIz (CH2Cl2); (ii) KzC03 (MeOH).

i

SH

+

+& fs>

R

R3 2 e

;

V

o

M

e

AcO AcO OMe

R' R2 R3 7 OAc H OAc 9 OAc OAc H 10 NHAc OAc H

~

99

R'

OAc

R' R2 R3 96 OAc H OAc 97 OAc OAc H 98 NHAc OAc H

Scheme 20. (i) CsCO3 (DMF).

yielding. The main hindrance is the basicity of the alkoxide anion; this restricts the acceptor molecules used in the coupling reactions to those bearing only ethers and ketal protecting groups. The lower basicity and higher nucleophilicity of the sulfur atom, however, enables a I-thiolate to be generated in the presence of base-sensitive protecting groups. The majority of 1,4-thio linkages have, hence, been achieved by this method. 4-Thiomaltose (100) was obtained in good yield from either 4-O-triflate displacement of the 1,6-anhydro galactose (not shown) or of the methyl a-D-galactoside (101) with donor 60 [15]. These reactions were effected by condensation in HMPA and generated the thiodisaccharide 102 which was isolated and then subjected to acetolysis and deprotection (Scheme 21). 4-Thiocellobiose (103) was similarly obtained in an overall yield of 70% from the same acceptor (101) and the donor 59 via the thiodisaccharide 104 [50, 511 (Scheme 22). Tetra-O-acetyl-1-thio-P-D-galactoside (7), generated in situ from the corresponding pseudothiourea precursor, was coupled to the 4-O-triflyl-galactosyl sphingosine (105) yielding the thiodisaccharide (106) in 68% yield [52] (Scheme 22). The efficiency of this procedure extends to other series and under similar conditions, 4-thioxylobiose (107) was easily obtained (60% yield) by condensation of the thiolate (108) generated from the peracetylated l-thioxylose (109) and the triflate (110) [53]. A very recent investigation by Stick and co-workers [54] has afforded the pseudo-

21.4 Establishment of I,l-Thio Linkages

i,ii

545

-

BzO

OMe

60

101

102 R' = H,R2 = OMe, R3 = Bz, R4 = Ac

iii,iv

K 100 R',R2 = H,OH, R3 = R4 =OH

Scheme 21. (i) (HMPA); (ii) Ac20-DMAP (pyridine); (iii) H?S04/HOAc (AczO); (iv) NaOMe (MeOH).

R' R2 R3 R4 59Na OH OH H 7 H OAc H OAc

101 R' = H, R2 = OMe 105 R' = cerarnide, R2 = H

R3 R4 R5 R6 OBz OAc OAc H OH OH OH H OBz OAc H OAc

R' iili~104 H Io3 H,OH 106 cerarnide

TfO

BZO

108 R' = Na R2= H 109 R' = R'= Ac

110

A,+

"

\

OH

107

Scheme 22. (i) (HMPA); (ii) Ac20--DMAP (pyridine); (iii) H ~ S O ~ / H O A(Ac20); C (iv) NaOMe (MeOH); (v) K2C03 (acetone-MeOH).

thiodisaccharide 111; this was achieved in moderate yield (40%) by coupling of the condurithiol derivative 112 and the triflate 113 (Scheme 23). This molecule might prove interesting in xylosidase-inhibition studies. Thiodisaccharides in the N-acetylglucosamine (GlcNHAc) series have also been reported. N,N"-diacetylthiochitobiose (114) was achieved (50% and 63% yield respectively) via triflate substitution of the 2-acetamido-2-deoxy-galactosides 115 or 116 with thiolates generated from 117 or 10 [ 5 5 ] (Scheme 24). The ally1 glycoside 118 has also been coupled to the thiol 34 thereby generating the thiodisaccharide 119 in good overall yield (40%) [7] (Scheme 25).

546

21 Synthesis of Oligosaccharide Mimics: S-Analogs TfO

BnO-Q/ BnO

OBn SH

MOM.

+ BzO

BzO

112

i-v

H

~

'

~

113

s

~ OH O

111

Scheme 23. (i) DBU (toluene); (ii) NaOMe (MeOH-THF); (iii) Na (NH3(hq));(iv) Ac20-DMAP (pyridine); (v) NaOMe (MeOH).

OH

i-v -

AcO

HO H

a

t

v

AcHN o

AcNH

NHAc

117R=Ac 10R=H

O

R' R2 R3 115 OMe H NHAc 116 H OMeNHAc

H

OH

R4

114

OBz OBz

Scheme 24. (i) NaH (DMF) or cysteamine (DMF); (ii) NaOMe (MeOH); (iii) Ac20-DMAP (pyridine); (iv) HzSOd/HOAc (AczO); (v) NaOMe (MeOH).

HO

i-iii,ii,iv,ii,v

Me

OAC TBDMSO LI

AcO

NHAc

34

118

119

-

Scheme 25. (i) NaH (DMF); (ii) Ac20-DMAP (pyridine); (iii) B q N F (THF); (iv) H2S (pyridine/ H20); (v) NaOMe (MeOH).

21.4.2 1,4-Thiooligosaccharides Conventional Approaches

The peracetylated 1,4-thiodisaccharides of cellobiose [5 11 (120) and N,N"diacetylchitobiose [ 5 5 ] (121), obtained relatively easily by conventional chemical manipulation of their thiodisaccharide precursors, are key compounds in the synthesis of the 4,4"-dithiocellotrioside 122 and N,N'l,N1l'-triacetyldithiochitotrioside (123) by coupling with their respective triflates (101 and 116). The corresponding deprotected analogs (124 and 125) were obtained in good to fair yields (Scheme 26). An extension of this sequential modification and addition has afforded the tri-

M

547

21.4 Establishment ofl,4-Thio Linkages

120 R' = OAc 121 R1 = NHAc

101 or 116

R1 OMe OMe

122 123 124 125 126

OMe

127

H

R2 H H H H SCtjH4NH2

OMe H

(3)

R3 OAc NHAc OH NHAc OH

R4 Ac Ac H H H

OH

H

OH

H

OH

H

ggOMe 128

H AcHNOMe

129

H

:

e OH

H

f

j

HoOMe

Scheme 26. (i) Cysteamine/DTE (HMPA or DMF).

saccharide 126, the tetrasaccharides 127 and 128, and even a pentasaccharide 129 in reasonable yield [ 51, 55, 561. The tetrasaccharide 130 and the hemithiocellotetraoside 131 have also been obtained from the laminaribiosyl (132) or cellobiosyl (133) donor by use of the common lactosyl triflate acceptor (134) [57, 581 (Scheme 27). In the thiomaltose series, the first synthesis of thiomaltotriosides [59] (not shown) was considerably improved by the use of a versatile acceptor molecule. The trityl 1S-a-D-glycosides have masked a-thiol functionality at C-1 which could be easily and selectively activated for future condensation reactions [60, 611. The trityl 1-S-C~-Dglucoside 135 and galactoside 136 were easily obtained by S N displacement ~ of the p-chlorides 82 and 92 with the tetrabutylammonium salt of triphenylmethanethiol. The use of this aglycone proved exceedingly advantageous because it was easily transformed into the corresponding thioacetates 23 and 134 (Scheme 28). A synthesis of methyl 4,4"-dithiomaltotrioside (138) has subsequently been achieved in good overall yield (41%) by selective in situ deacetylation and activation of the thioacetate 23 and condensation with the trityl 1-S-galactoside 139 to yield the dithiodisaccharide 140. This in turn could be transformed into the donor 141 and condensed with the triflate 101 to yield the trisaccharide 142 [62] (Scheme 29). An extension of this work, using the donor 143 has resulted in the production of a suite of thio- and hemithiooligosaccharides up to degree of polymerization of 8 [63].

548

21 Synthesis of Oligosaccharide Mimics: S-Analogs

130

i,ii

132

TfO BzOs

B

z

.

z

.

O

M

e

134

ACO OAc

OAc

133

Scheme 27. (i) Et2NH (DMF); (ii) NaOMe (MeOH).

CI

iijii

A

OAc

82 R‘ = O A C ~ R =H ~ 92 R’ = H, R = OAc

~

STr

135 R’ = OAc R2 = H 136 R’ = H, Rp = OAc

SAC

23 R1= OAc, R2 = H 137 R’= H, R2 = OAc

Scheme 28. (i) BU4NSTr (toluene); (ii) PhHgOAc (MeOH/CH2C12); (iii) H2S (CH2Cl~/pyridine/ Ac20).

Chemoenzymatic Approaches Hemithiocellodextrins can essentially be viewed as a repeating chain comprising 4-thiocellobiosyl units linked by a P-1,4-O-glycoside or, conversely, as cellobiosyl units linked by P-1,4-S-glycosides. The enzymic oligomerization of 4-thio-Pcellobiosyl fluoride (144) by use of a cellulase in a mixed solvent system has resulted in the isolation of hemithiocellodextrins of different length (145 n = 1 (4.5%), n = 2 (7.5%), n = 3 (5.7%), n = 4 (5.0%), n = 5,6 (20%)) [64] (Scheme 30). A similar chemoenzymatic approach has been used for the synthesis of a series of

549

21.4 Establishment of 1,4-Thio Linkages

23

+

i

BzO

w

AcO

STr

140 R’ = Tr, R2 = Bz = Ac, R2 = BZ 143R’=R2=Ac

139

ii I 1 4 1 R’

141

+

Tfo%

i *

BzO

R20

OMe

101

R10

142 R’ = Bz, R2 = Ac 138 R’ = R2 = H

~.

17

R1oAM,

iii

Scheme 29. (i) Cysteamine/DTE (HMPA); (ii) PhHgOAc (EtOH), then H2S (Ac*O/CH*C12/ pyridine); (iii) NaOMe (MeOH).

hemithiocyclodextrins, viewed as a cyclic chain of repeating thiomaltose units linked through a-1,4-O-glycosides. The ability of cyclodextrin glucosyltransferase (CGTase) to use maltosyl fluoride (146) for the synthesis a-, p-, and y-cyclodextrins has been known for more than a decade [65],and more recent applications for the chemoenzymatic synthesis of regioselectively substituted cyclodextrins have also been applied [66, 671. In this fashion, the self condensation of 4-thio-a-maltosyl fluoride (147) in the presence of pure CGTases has afforded the hemithiocellodextrins (148 n = 4 (14%), n = 5 (l6%), n = 6 (16%)) [8] (Scheme 30). Michael Addition to Unsaturated Acceptors A recent application of Michael addition of sugar thiols to levoglucosenone (149) has been reported [6, 68, 691. The addition of the thiols 9 and 34 to levoglucosenone (149) resulted in the synthesis of the anhydrides 150 and 151. These were stereoselectively reduced, subjected to acetolysis conditions, and deprotected to yield the 3-deoxydisaccharides 152 and 153 in moderate yield (approximately 60% overall) [69]. The 2-acetamido-2-deoxyglucose derivative 154 was achieved by oximation of 151 which gave 155 before reduction and ring opening (Scheme 31).

21 Synthesis of Oligosuccharide Mimics: S-Analogs

550

OH

OH

OH

145

HO

HO

146X=O 147X=S

148

Scheme 30. (i) Cellulases (CH3CN/acetate buffer 0.05 M, pH 5.0); (ii) CGTases (phosphate buffer 0.2 M, pH 6.5).

Solid-Support Synthesis

Despite the popularity of the triflate displacement method, it does have one drawback-when the acceptor molecule bears an aglycone of the p-D configuration, H5 and H3 abstractions are known to occur. The yield of the desired coupled product and that resulting from elimination vary depending on the reaction conditions and the OH-6 and OH-3 protecting groups of the acceptor [7, 55, 701. For acceptors bearing a P-aglycone, the recent application of solid-support synthesis could be an important improvement [71]. The key feature of such a synthetic scheme is the use of bound and deprotected nucleophilic sugar 1-thiolates. Subsequent condensation with triflates gives the expected compounds in good to high yields. In the case where the aglycone of the acceptor molecule is in the p configuration, the usual by-products from the condensation are negated by the flushing the support after glycosylation, and the thiooligosaccharide is the only product observed when cleavage and isolation have been effected. The suitability of this method for the construction of trisaccharides is demonstrated below (Scheme 32).

551

21.5 Establishment of 1,S-Thio Linkages

9-

i

+

0

149

+ 34

. i * Me@OA~

HR -

SI

O ,H

ii-iv

or vi,iii,iv

1 OH

HO

AcO

v(

151 R = 0 155R=NOAc

153 R = O H 154 R = NHAc

Scheme 31. (i) Catalytic Et3N (benzene); (ii) L-selectride@ and Ac2O (pyridine); (iii) BF,.OEt2 (AczO); (iv) Et3N (MeOH/H20); (v) NHzOH.HC1 (EtOH/pyridine); (vi) 9-BBN (H202, NaOH) then Ac2O (pyridine).

21.5 Establishment of 1,3-Thio Linkages 21.5.1 1,3-Thiodisaccharides Conventional Methods Sp~2displacement of the activated allofuranose 160 has afforded the a- and 8-1,3thiodisaccharides of glucose and xylose [35, 37, 721. The donors (9, 23 or 162, 161) have been condensed with the triflate 160 thereby generating the thiodisaccharides 163 (84%), 164 (94%), and 165 (54%), respectively (Scheme 33). Deprotection then afforded the thiolaminaribiose 166, thionigerose 167 and, after standard procedure, the ~-1,3-thioxylobiose168. Cyclic Sulfamidate and Aziridine The sequence a-~-fuc( 1j 3 ) - ~ - G l c N H A cis a common constituent of Lewisxbearing glycoconjugates and the establishment of non-hydrolyzable analogs could prove invaluable for several biological processes. To this end the cyclic sulfamidate activating group has also been targeted for nucleophilic attack and thence genera-

552

21 Synthesis of Oligosaccharide Mimics: S-Analogs

Tfos

OTr -@ HO HO*sNa

+

BzO

SSEt

OBz 157

OH 156

OTr -@ H HO O - t e S S E t

OH

0 Bz

158

I H HO O & g e OH ; &

ii,iii,ii then 157, i,iW

OH

OH

SSEt

159 Scheme 32. (i) Crown ether 15-5 (THF); (ii) NaOMe (MeOH/THF); (iii) DTT (MeOH/THF/ Et3N); (iv) TFA (CH2C12).

tion of a 1,3-thio linkage to 2-acetamido-2-deoxyglycosides [73, 741 (Scheme 34). The 1-thiofucose 34 has been rapidly condensed with the acceptor 169, generating the expected disaccharide 170 (50% yield). The deprotection steps gave 171 in good overall yield. The reactive N-tosyl aziridine 172 has also been condensed with the 1-thiol fucose 34 to yield a mixture of the (11 2 ) and (1 '3) thiodisaccharides (2 : 1) [7]from which the desired compound (173) was isolated in 45% overall yield (Scheme 34). 21.5.2 1,3-Thiooligosaccharides

31,3rr-Dithiolaminaritriose (174) has been conveniently prepared (albeit in only 15% overall yield) from the bromide 175 which is derived from 166 [35]. Nucleophilic displacement on 175 with the sodium salt of 176 gave the tri-

21.6 Establishment of1,2-Thio Linkages

553

OH

SR

HO

162R=H i,V

164

I6O

AcO

\ SH

vi

167

Me

~

/ 'Me

OAc

9 R = C H OAc 161 R = d

163 R = CH~OAC 165 R = H

vii,i or viii,ix, vii,iii,i

166 R = CHpOH 168R=H

Scheme 33. (i) NaOMe (MeOH); (ii) HMPA; (iii) AczO (pyridine);(iv) NaH (THF), then (DMF); (v) (80% AcOH); (vi) NaH, Kript0fix~21(THF); (vii) (60% AcOH); (viii) (50% AcOH); (ix)NaI04 then NaBH4 (EtOH/H20).

saccharide 177 in 30% yield. The usual manipulations gave the free sugar (174; Scheme 35).

21.6 Establishment of 1,ZThio Linkages 21.6.1 1,2-Thiodisaccharides 1-Thiosucrose, the true analog of sucrose, was first obtained in 1984 [23] by Lewis acid catalysed glycosylation using tetra-0-acetyl- 1-thio-a-D-glucose (162) (not

554

21 Synthesis of Oligosuccharide Mimics: S-Analogs

i ii OR' OAc AcO

34 170 R' = OBn, R2,R3= PhCH, R4 = Ac

HO

172

iii,k

Ts

E 171 R' = OBn, R2 = R3 = R4 = H 173 R' = Pr, R2 = R3 = R4 = H

Scheme 34. (i) NaH (DMF); (ii) HzS04 (HzO/THF); (iii) H+ (MeOH); (iv) NaOMe (MeOH); (v) AczO, DMAP (pyridine); (vi) Pd/C, Hz (MeOH); (vii) Na (NH3(llq)/THF).

kO*:c AcO

OAc

175

Br

) *HOa:H i,ii

HO

OH

SR

HO

Me 0

M e x o &-bM ;e o

177 R =E x z b ;,s-hle o

1 iii

176 174R= HO

OH

Scheme 35. (i) Na (MeOH); (ii) crown ether 15-5 (THF); (iii) (60% AcOH).

shown). The reaction proceeded in very good yield (70%) but the isolation of pure 1-thiosucrose was laborious. Interestingly, the reverse condensation did not afford the desired compound. Conventional Methods Most 1,2-thio linkages have been established via S N displacement ~ of a triflate of either glucosyl or mannosyl acceptor molecules with 1-thiolates.

21.6 Establishment of 1,2-Thio Linkages

R +

AcO

162

i-iii

3

R

-

-c

OAc

R

q R1

R’ R2 178 H.OH 181 OAC H 183 OAll H 185 OBn H

555

R3 OH OAc OAc OH

179

79

+

i-iii

=

R’ R2 R3 R4 R’ 180 H,OH OH CH20H R2 182 184 OAc OAll H OAc EH20Ac CH20Ac

RkR+ R3R&s

R3

186 H,OH

OH

Scheme 36. (i) NaOMe (MeOH); (ii) (DMF) or (HMPA); (iii) AczO (pyridine)

There has recently been interest in synthesizing thio analogs of kojibiose, a known a-glucosidase inhibitor. The synthesis of thiokojibiose (178) was first achieved in very good yield (87%) by condensation of the triflate 179 with 1-thio-a-D-glucose (162) [27] (Scheme 36). Thiosophorose (180) was similarly produced from the pthiol 9 but when it was observed that deprotection of the peracetylated products of condensation (181 and 182) under Zemplen conditions was somewhat problematic, the easily removed allyl group was introduced at the anomeric position thus affording the glycosides 183 and 184. Shortly thereafter the acetylated 181 was converted to the benzyl kojibioside (185) by conventional methods [37]. A similar method (not shown) was used during the synthesis of the 2-thioxylobiose (186) [72] (Scheme 36). Other Approaches Pinto and co-workers [75] have chosen a more traditional approach toward establishing 1,2-thio linkages and ultimately the production of thiokojibiosides and related derivatives. The thiol 187 was coupled to the p-trichloroacetimidate 188 thereby yielding the protected allyl kojibioside 189 [75] (Scheme 37). Interestingly, when the w-trichloroacetimidate 190 was used as donor the thiosophorose derivative 191 was obtained as the major product (191 : 192, 2.3: 1 ) in a reaction proceeding via the stable orthoester 193. In addition, the thiol 194 was condensed with the 5-thiotrichloroacetimidate donor 195 to yield a mixture of compounds from which the dithiodisaccharide 196 was isolated in good yield (approximately 70%) [76] (Scheme 37). A further paper by the same authors reports the synthesis of the dithiodimannoside 197 from the appropriate precursors; the thiol 198 was condensed with 199 to afford 200 [77] in modest yield (37%) (Scheme 38).

556

R

21 Synthesis of Oligosuccharide Mimics: S-Analogs

R 2O

:

:

OR’

k

+

-

187

+

190

OR’

0TCC’3 NH

OR

SH

anomer 188 p 190 a 195 a

187 R’ = All, R2 = Bn 194 R’ = Pr, R2 = Bz

R20

i

%O :R

X 0 0 S

X 189 0 192 0 196 S

R Bn Ac Ac

R’ R2 R3 All All Pr

Bn Bn

Bn Ac

H

H

-

-

ON1

i

OAc 193

191

Scheme 37. (i) TESOTf (CH2C12).

phTq +

OMe

AcO

AcO A

c

t

q

Rq; Rzr% K i

3I‘ NH

R’ 0

198

Scheme 38.

(1)

TESOTf (CH2C12); (ii) HOAc/H,O; (iii) NaOMe (MeOH).

2,3-Epoxides have also been targeted for nucleophilic attack. Ring opening of the talopyranose 201 with a-fucosyl thiol (34) afforded a mixture of fucosyianhydrogalactoside and anhydroidosides (not shown). Separation and further reaction of the former compound gave 202 in moderate overall yield (50%) [7] (Scheme 39).

OMe

557

21.7 Establishment oj1,l-Thio Linkages

cvi

+

&d

HO

Aces

OAll

AcO

*

OAc AcO

Ms$e

OAc

OAc

AcO

202

34

201

Scheme 39. (i) NaOMe (MeOH); (ii) Aq0-DMAP (pyridine); (iii) H*SOd/HOAc (AczO); (iv) N2H4.AcOH (DMF); (v) CC13CN (DBU/CH2C12); (vi) AIlOH/BF3.OEtz (CH2C12).

21.7 Establishment of 1,l-Thio Linkages The syntheses of a,a-thiotrehalose (203) and a-D-glucopyranosyl 1-thio-a-Dmannopyranose (204) as substrate analogs for trehalase were achieved more than twenty years ago [ 141. This study was the first to note that the condensation of a thiolate or a glycosyl halide proceeds in HMPA without any neighboring group participation (Scheme 40). Na2S

Z

i,iii CI

AGO% AcO

82 OAC

W

o

H OH

HO

\-//

e

OH ‘s

1, 111

203

Y+ 60

82

57

SNa

ii,iii

Hr-HO

SH

Scheme 40. (i) HMPA (ii) K2CO3 (HMPA/H2O); (iii) NaOMe (MeOH).

&OH OH

s 204

OH

558

21 Synthesis of Oligosaccharide Mimics: S-Analogs

Hydrogen sulfide reacted with D-glucose (not shown) in hydrogen fluoride solution to yield predominantly compound 203 [78].

21.8 Establishment of Mixed Thio Linkages Ganglioside and Lewis A epitopes in which some of glycosidic oxygens have been substituted by sulfur have recently been obtained [38, 791. The thio-linked sialyl Lewis' analog and its precursor, the thio-linked ganglioside GM3 epitope have also been described [20, SO]. The coupling of the thiodisaccharide precursors 205 and 206 using well established methods has afforded the expected tetrasaccharide 207 in good yield. The thiolate 208 plays a pivotal role in this synthesis (Scheme 41). The branched sulfur-linked tetrasaccharides 211 and 212 have been conveniently prepared by s N 2 displacement of the triflate [35, 811 group of 160 or the iodide 20 with the sodium salt of the trisaccharide 213 (Scheme 42).

21.9 Thiooligosaccharides and Proteins Thioglycosides, and more commonly thiooligosaccharides, have played an important role in probing natural oligo- and polysaccharide recognition by proteins. Most of these studies have been performed during the last five years. These non-natural substrates have been used to define protein function often by intercepting the very mechanism by which they act, because these compounds are resistant to hydrolysis by O-glycosylhydrolases. Notably, there is only one class of naturally occurring thioglycoside, the glucosinolates, which are degraded under the action of myrosinases [82]. This enzyme is the only S-glucosidase known in nature, and recent studies reveal the remarkable similarity between this protein and the cyanogenic P-glucosidase [83].

21.9.1 The Conformation of Thiooligosaccharides in Solution The first step in carbohydrate-protein interaction is the recognition; this is dependent on the overall conformation of the oligosaccharide. It is, therefore, important to determine how closely these non-natural substrate analogs represent the natural compounds in aqueous solution. This can be achieved by a combination of NMR spectroscopy, molecular mechanics calculations, and X-ray crystallography. Conformational assessments of 4-thiomaltose [84-861, 4-thiogalabiose [48], a - ~ - 3 thioFuc( 143)GlcNHAc [ 871, a,a-thiotrehalose [88] and thiogangliosides [ 891 have all resulted in a consistent finding-the thio linkage provides a high degree of flex-

559

21.9 Thiooligosaccharides and Proteins

A AcHN cAcO"' o w

C

0

2

M

x

+

e

+ Me@

AcO

HS

O

kBzOo

m

AcO

y "J

209

208

OMPM OAc

210

v-xi

S(CH2)6CH3 AcHN

AcO

AcO

OAC

ACO

OAc AcO

205

206

207

Scheme 41. (i) NaH, Kryptofixm21(THF); (ii) TFA (H,O); AczO (pyridine); (iii) TBAF (THF); AczO (pyridine); (iv) HBr (AcOH); (v) TMSOTf (CHzC12); (vi) CAN (CH3CN/H20) then Ac20 (pyridine); (vii) EtSH, p-TsOH, then BzCN, NEtl; (viii) Tf20 (pyridine), KSAc (THF); (ix) TBAF (THF), then CC13CN, DBU; (x) HSC7H15, BF3.0Et2; (xi) NzH4.HOAc (DMF); (xii) NaH (DMF); (xiii) NaOMe (MeOH) then 0.2 M KOH.

ibility between sugar units and the thiooligosaccharides present a conformational equilibrium in solution between at least two major populations, whereas in the 0series only one rotamer is heavily populated. It is, however, possible to lock these substrate analogs by attachment of alkyl groups near the thio linkage into the rigid conformation required [90].

21 Synthesis of Oligosaccharide Mimics: S-Analogs

560

i-iii

160

211

SH

AcO

OAc

OAc

213

\

i,iii

O .H% &: ;& H O H HO OH

212

ko%OA~ AcO

HO HOo&

OAC

20 Scheme 42. (i) NaH (THF), then (DMF); (ii) 90% TFA (H20), then Ac20 (pyridine); (iii) NaOMe (MeOH).

21.9.2 Enzyme-Substrate Interactions The elucidation of structure-function relationships to facilitate the design and production of enzyme variants with altered properties has mainly been achieved by solution and X-ray studies of protein-substrate interactions. This section deals only with compounds previously described. a-Glucan-Active Enzymes

a-Amylases catalyze the hydrolysis of a-(1,4) glycosidic linkages of starch components. These enzymes are members of the glycosyl hydrolase family 13 [91], which also contains cyclodextrin gluconotransferases (CGTases) and pullulanases. The structures of pig pancreatic a-amylase and inactive CGTase in complexes with substrate analogs 138, Scheme 29, and 50, Scheme 15, respectively, have been recently solved [2, 921. 4-Thiomaltosyl fluoride (147, Scheme 30) was recognized as a substrate by CGTases and led to hemithiocyclodextrins in good yield [8].

H

21.9 Thiooligosaccharides and Proteins

561

Most fungal glucoamylases have a raw starch-binding domain (SBD) which is distinct from the catalytic domain. To obtain a better understanding of the role of this domain during the enzymatic process the inhibitors 43, 46, and 47 (Scheme 13) were used [40, 411. The thermodynamics of binding of compounds 78-81 (Scheme 16) to the two domains of glucoamylase GI from A . niger shows that the catalytic and starchbinding sites are in close proximity in space 1931. The hydrodynamic dimensions of GI alone and in interactions with these heterobidentate ligands were determined by quasi elastic light-scattering experiments [46]. From these results it seems that the presence of the bound bifunctional ligands stabilizes a more compact conformation of the enzyme. The motion between the two domains might achieve the processivity of this exo-enzyme which releases P-D-glucose units from the non-reducing ends of starch. Several ganglioside analogs containing an a-thiosialic acid residue, derived from compound 32 (Scheme 10) have been found to be potent inhibitors of sialidase activities of different subtypes of the influenza virus [96]. Disaccharide structures containing the a-thio-L-fucosyl residue-36 (Scheme 1 l), 119 (Scheme 25), 173 (Scheme 34), and 202 (Schcmc 39) -arc all inhibitors of a - ~ fucosidases from bovine kidney and epididymis with K, values of 0.65-5 mM [7]. Several 1,2-cis-thiodisaccharideswere evaluated for their capacity to inhibit HIVinduced cell killing and virus production in LEM or MT-2 cells. All the compounds were inactive except for the thiokojibiose octaacetate 181 (Scheme 36), but none was evaluated as an inhibitor of glucosidases I or I1 [ 3 7 ] . P-Glucan-Active Enzymes

Cellulases (cellobiohydrolases and endoglucanases) are glycosyl hydrolases able to cleave the p-1,4-linkages of cellulose. The catalytic domains of cellulases and related xylanases have been identified in 14 of the 63 families of homologous folds of all glycosyl hydrolases [91]. The structure of the endocellulase EGI from Fusarium oxysporum (family 7), complexed to the non-hydrolyzable thiocellopentaoside 129 (Scheme 26) has been determined by X-ray crystallography at 2.7 A resolution [94]. This study shows for the first time the same oligosaccharide molecule spanning the point of enzymatic cleavage and the distortion of the sugar residue in the - 1 subsite. The crystal structure of a processive endo-cellulase CelF of Clostridium cellulolyticum in complex with two molecules of 4"-thiocellotetraose 131 (Scheme 27) at 2.0A resolution has very recently been solved [95]. One inhibitor molecule was enclosed in a tunnel and the other was bound in an open architecture. They constitute the two parts of the active site. The P-thiocellobiosyl fluoride 144 (Scheme 30) was recognized as substrate by cellulases and afforded hemithiocellodextrins in good yield [64]. The usefulness of 4-thiochitobiose 114 (Scheme 24), 4,4"-dithiochitotriose 128 (Scheme 26), and their glycosides has been demonstrated in the characterization of the transport processes, the screening, isolation and kinetic study of the chitinolytic system of Vibriofurnissii [ 971. The free compound derived from the trisaccharide 49 (Scheme 14) was kinetically

562

21 Synthesis of Oligosacchavide Mimics: S-Analogs

evaluated as a substrate for N-acetylglucosaminyltransferase-V and found to be an acceptor with threefold increase in V,,, but higher K, value (330 PM) than the natural acceptor [34].

21.9.3 Lectin-Ligand Interactions

The in vitro flocculation of the yeast Kluyveromyces bulgavicus was inhibited by the branched thiocyclodextrin 64 (Scheme 15) with the same efficiency as the best known ligand [45]. On the other hand, the 4-thiogalabioside 90 (Scheme 18) has been shown to be ten times less recognized than its oxygen analog by a bacterial pilus adhesin [48].

21.10 Conclusion We have presented an exhaustive and up to date review on the synthesis of thiooligosaccharides which can be used to probe interactions with proteins. The ease of synthesis of this class of molecule should furnish new tools for glycobiochemists and glycobiologists.

Acknowledgments

Our own work described in this review has been supported by CNRS and several EU biotechnology programs (BI02 CT94 3008, 3018, B104 CT97 2303 and B104 CT98 0022).

References 1. Q. Wang, R. W. Graham, D. Trimbur, R. A. J. Warren, S. G. Withers, J. Am. Chem. Soc. 1994, 116, 11594-11595. 2. A. K. Schmidt, S. Cottaz, H. Driguez, G. E. Schultz, Biochemistry, 1998,37, 5909-5915. 3. J. M. Beau, T. Gallagher in Top. Curr. Cliem. 1997, 187, 1-54. 4. A. de Raadt, C. W. Ekhart, M. Ebner, A. E. Stiitz in Top. Curr. Chem. 1997, 187, 157-186. 5. H. Driguez in Top. Curr. Chem. 1997, 187, 86-116.

6. Z. J. Witczak, R. Chhabra, H. Chen, X. Q. Xie, Curbohydr. Res. 1997, 301, 167-175. 7. H. Hashimoto, K. Shimada, S. Horito, Tetrahedron: Asymm. 1994, 5, 2351-2366. 8. L. Bornaghi, J. P. Utille, E. D. Rekai', J. M. Mallet, P. Sinay, H. Driguez, Curbohydr. Res. 1997, 305, 561- 568, and references cited therein. 9. D. Horton, D. H. Huston, Adu. Curbohydr. Chem. 1963, 18, 123-199, and references cited therein. 10. D. Horton, Methods in Carhohydr. Chem. 1963, 2, 433-437.

References

563

11. W. K. C . Park, S. J. Meunier, D. Zanini, R. Roy, Curbohydr. Lett. 1995, I , 179-184. 12. J. Defaye, H. Driguez, E. Ohleyer, C. Orgeret, C. Viet, Curbohydr. Res. 1984, 130, 317-321. 13. M. Sakata, M. Haga, S. Tejima, M. Akagi, Chern. Pharm. Bull. 1964, 12, 652-656. 14. M. Blanc-Muesser, J. Defaye, H. Driguez, Carbohydr. Res. 1978, 67, 305-328. 15. M. Blanc-Muesser, J. Defaye, H. Driguez, J. Chem. Soc. Perkin I , 1982, 15-18. 16. A. Gadelle, J. Defaye, C . Pedersen, Carbohydr. Res. 1990, 200, 497-498. 17. M. Akagi, S. Tejima, M. Haga, Chem. Phurrn. Bull. 1962,10, 562-566. 18. R. L. Whistler, W. C . Lake, Methods in Cavbohydr. Chem. 1972, 6, 286-291. 19. S. D. Gero, R. D. Guthrie, J. Chem. Soc. ( C ) 1967, 1761-1762. 20. T. Eisele, A. Toepfer, G. Kretzschmar, R. R. Schmidt, Tetrahedron Lett. 1996, 37, 1389-1392. 21. J. Kocourek, Curbohydr. Res. 1967, 3, 502-505. 22. R. J. Ferrier, R. H. Furneaux, Curbohydr. Res. 1977, 57, 73-83. 23. J. Defaye, H. Driguez, S. Poncet, R. Chambert, M. F. Petit-Glatron, Carbohydr. Res. 1984, 130, 299-3 15. 24. A. Hasegawa, J. Nakamura, M. Kiso, J . Curbohydr. Chem. 1986, 5, 11-19. 25. S. Cottaz, P. Rollin, H. Driguez, Curbohydr. Res. 1997,298, 127-130. 26. S. Bennett, M. von Itzstein, M. J. Kiefel, Curbohydr. Res. 1994, 259, 293-299. 27. J. Defaye, J. M. Guillot, Carbohydr. Res. 1994, 253, 185-194. 28. M. Blanc-Muesser, H. Driguez, J. Chem. SOC.,Perkin Trans. I , 1988, 334553351, 29. C. Simiand, E. Samain, 0. R. Martin, H. Driguez, Curbohydr. Res. 1995, 267, 1-15. 30. D. H. Hutson, J. Chem. Soc. ( C ) 1967,442-444.

31. J. A. Calvo-Asin, F. G. Calvo-Flores, J. M. Exposito-Lopez, F. Hernandez-Mateo, J. J. Garcia- Lopez, J. Isac-Garcia, F. Santoyo-Gonzalez, A. Vargas-Berenguel, J. Chem. Soc. Perkin Trans. I , 1997, 1079-1081. 32. H. Driguez, J. C. McAuliffe, R. V. Stick, D. M. G. Tilbrook, S. J. Williams, Aust. J. Chem. 1996,4Y, 343-348. 33. I. Lundt, B. Skelbaek-Pederson, Actu. Chem. Scand. Sect. B, 1981,35, 637-642. 34. P. P. Lu, 0. Hindsgaul, H. Li, M. Palcic, Can. J. Chem. 1997, 75, 790-800. 35. M. 0. Countour-Galcera, J. M. Guilliot, C . Ortiz-Mellet, F. Plieger-Carrara, J. Defaye, J. Gelas, Carbohydr. Res. 1996,281, 99-1 18. 36. W. Boos, P. Schaedel, K. Wallenfels, Eur. J. Biochem. 1967, I , 382-394. 37. R. N. Comber, J. D. Friedrich, D. A. Dunshee, S. L. Petty, J. A. Secrist 111, Curbohydr. Res. 1994,262, 245-255. 38. A. Hagesawa, T. Terada, H. Ogawa, M. Kiso, J. Curbohydr. Chem. 1992, 11, 319-331. 39. S. Cottaz, H. Driguez, Synthesis, 1989, 10, 755-758. 40. S. Cottaz, H. Driguez, B. Svensson, Carbohydr. Res. 1992, 228, 299-305. 41. C. Apparu, H. Driguez, G. Williamson, B. Svensson, Carbohydr. Res. 1995,227, 31 3-320. 42. C. Lancelon-Pin, H. Driguez, Tetrahedron Lett. 1992,33, 3125-3128. 43. J. Defaye, A. Gadelle, A. Guiller, R. Darcy, T. O’Sullivan, Curbohydr. Res. 1989, 192, 251258. 44. B. Brady, R. Darcy, J. Carhohydr. Res. 1998, 309, 237-241. 45. L. de Robertis, C. Lancelon-Pin, H. Driguez, F. Attioui, R. Bonaly, A. Marsura, Biorg. Med. Chem. Lett. 1994, 4 , 1127-1 130. 46. N. Payre, S. Cottaz, C . Boisset, R. Borsali, B. Svensson, B. Henrissat, H. Driguez, Angew. Chern. Int. Ed. 1999, 38, 974-977. 47. L. A. Reed, L. Goodman, Curbohydr. Rex 1981, 94, 91-99. 48. U. Nilsson, R. Johansson, G. Magnusson, Chem. Eur. J. 1996, 2, 295-302. 49. L. X. Wang, N. Sakairi, H. Kuzuhara, J. Chem. Soc. Perkins Trans. 1, 1990, 1677-1682. SO. D. Rho, M. Desrochers, L. Jurasek, H. Driguez, J. Defaye, J. Bacteriol. 1982, 149, 47-53. 51. C . Orgeret, E. Seillier, C. Gautier, J. Defaye, H. Driguez, Curbohydr. Res. 1992, 224, 29-40. 52. B. Albrecht, U. Piitz, G. Schwartzmann, Curbohydr. Res. 1995, 276, 289-308. 53. J. Defaye, H. Driguez, M. John, J. Schmidt, E. Ohleyer, Curbohydr. Res. 1985, 139, 123-132. 54. M. McDonough, R. V. Stick, D. M. G. Tilbrook, Aust. J. Chem. 1999,52, 143-147. 55. L. X. Wang, Y. C. Lee, J. Chem. Soc. Perkin Trans. 1, 1996, 581-591. 56. C. Schou, G . Rasmussen, M. Schulein B. Henrissat, H. Driguez, J. Curhohydr. Chem. 1993, 12, 743-752.

564

21 Synthesis of Oligosaccharide Mimics: S-Analogs

57. V. Moreau, J. L. Viladot, E. Samain, A. Planas, H. Driguez, Biorg. Med. Chem. 1996,4, 18491855. 58. C. Reverbel-Leroy, G. Parsiegla, V. Moreau, M. Juy, C. Tardif, H. Driguez, J.-P. Belaich, R. Haser, Actu Crystallogr. Sect. D,1998, 54, 114-118. 59. M. Blanc-Muesser, J. Defaye, H. Driguez, G. Marchis-Mouren, C. Seigner, J. Chem. Soc. Perkin Trans. 1, 1984, 1885-1889. 60. M. Blanc-Muesser, H. Driguez, B. Joseph, M. C. Viaud, P. Rollin, Tetrahedron Lett. 1990, 31, 3867-3868. 61. M. Blanc-Muesser, L. Vigne, H. Driguez, Tetrahedron Lett. 1990, 31, 3869-3870. 62. M. Blanc-Muesser, L. Vigne, H. Driguez, J. Lehmann, J. Steck, K. Urbahns, Carbohydr. Res. 1992,224, 59-71. 63. M. Blanc-Muesser, H. Driguez, unpublished results. 64. V. Moreau, H. Driguez, J. Chem. Soc. Perkin Trans. I , 1996, 525-527. 66. E. J. Hehre, K. Mizokami, S. Kitahata, Denpun Kugaku. 1983, 30, 76-82. 66. S. Cottaz, C. Apparu, H. Driguez, J. Chem. Soc. Perkin Trans. 1, 1991, 2235-2241. 67. C. Apparu, S. Cottaz, G. Bosso, H. Driguez, Carbohydr. Lett. 1995, 1, 349-352. 68. B. Becker, J. Thimm, J. Thiem, J. Carbohydr. Chem. 1996,15, 1179-1181. 69. Z. J. Witczak, J. Sun, R. Mielguj, Biorg. Med. Chem. Lett., 1995, 5, 2169-2174. 70. V. Moreau, J. Ch. Norrild, H. Driguez, Carbohydr. Res. 1997, 300, 271-277. 71. G. Hummel, 0. Hindsgaul, Angew. Chem. Int. Ed. 1999,38, 1782-1784. 72. J. Defaye, J. M. Guillot, P. Biely, M. Vrianka, Curbohydr. Res. 1992, 228, 47-64. 73. B. Aguilera, A. Fernadez-Mayoralas, Chem. Commun. 1996, 127-128. 74. B. Aguilera, A. Fernadez-Mayoralas, C. Jaramille, Tetrahedron, 1997, 53, 5863-5876. 75. J. S. Andrews, B. M. Pinto, Carbohydr. Res. 1995, 270, 51-62. 76. J. S. Andrews, B. D. Johnston, B. M. Pinto, Curbohydr. Res. 1998, 310, 27-33. 77. B. D. Johnston, B. M. Pinto, Carbohydr. Rex 1998, 310, 17-25. 78. J. Defaye, A. Gadelle, C. Pedersen, Curbohydr. Res. 1991,217, 51-58. 79. T. Eisele, R. Windmiiller, R. R. Schmidt, Curbohydr. Res. 1998,306, 81-91. 80. T. Eisele, R. R. Schmidt, Liebigs AnnlRecueil. 1997, 865-872. 81. M. 0. Contour-Galcera, Y. Ding, C. Ortiz-Mellet, J. Defaye, Carbohydr. Res. 1996, 281, 119128. 82. G. R. Fenwick, R. K. Heaney, W. J. Mullin. CRC Crit. Rev. Food. Sci. Nutr. 1983, 18, 123201. 83. W. P. Burmeister, S. Cottaz, H. Driguez, R. Iori, S. Palmieri, B. Henrissat, Structure, 1997, 5, 663-675. 84. S. Perez, C. Vegelati, Actu Crystullogr. Sect. B, 1984, 40, 294-299. 85. K. Mazeau, I. Tvaroskd, Curbohydr. Res. 1992,225, 27-41. 86. K. Bock, J. 0. Duus, J. Refn, Curbohydr. Res. 1994,253, 51-67. 87. B. Aguilera, J. Jimenez-Barbero, A. Fernandez-Mayoralas, Carbohydr. Res. 1998, 308, 19-27. 88. K. Bock, J. Defaye, H. Driguez, E. Bar-Guilloux, Eur. J. Biochem. 1983, 131, 595-600. 89. A. Geyer, G. Hummel, T. Eisele, S. Reinhardt, R. R. Schmidt, Chem. Eur. J. 1996,2, 981-988. 90. S. Sabesan (du Pont de Nemours, E. I. ), US 5489675A6, 1996 (Chem. Ahstr., 1997, 124, 343979). 91. B. Henrissat, Biochem. J. 1991,280, 309-316. 92. M. Quian, S. Spinelli, H. Driguez, F. Payan, Protein Sci. 1997, 6, 2285-2296. 93. B. W. Sigurskjold, T. Christensen, N. Payre, S. Cottaz, H. Driguez, B. Svensson, Biochemistry, 1998,37, 10446-10452. 94. G. Sulzenbacher, H. Driguez, B. Henrissat, M. Schiilein, G. J. Davies, Biochemistry, 1996, 35, 15280-15287. 95. G. Parsiegla, M. Juy, C. Reverbel-Leroy, C. Tardif, J. P. Belakh, H. Driguez, R. Haser, EMBO J., 1998, 17, 5551-5562. 96. Y. Suzuki, K. Sato, M. Kiso, A. Hasegawa, Glycoconj. J . 1990, 7, 349-356. 97. L-X. Wang, N. 0. Keyhani, S. Roseman, Y. C. Lee, Glycobiology, 1997, 7, 855-860, and references cited therein.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

22 Saccharide-Peptide Hybrids Hans Peter Wessel

22.1 Introduction The term ‘saccharide-peptide hybrids’ (SPHs) encompasses sugar molecules linked by an amide group, e.y. a disaccharide with the interglycosidic oxygen formally replaced by an amide group (Scheme 1).

Scheme 1.

For oligomers, the monomer units are thus carbohydrate amino acids (CAAs). SPHs are interesting carbohydrate mimetics-although the linking unit is modified, typical carbohydrate epitopes are maintained. For the oligomerization process the synthetic advantage is obvious-the amide linkage is readily formed and, in contrast with glycosylations, avoids stereochemical complications because no anomers are formed. Another advantage is that the amide coupling reactions can be conducted without protection of the hydroxyl groups because the amino group is more nucleophilic than the hydroxyl groups. While solid-phase glycosylations are still in their infancy, amide bond formation by solid-phase chemistry has been well established for a considerable time. SPHs are therefore amenable to the combinatorial approaches which have become an indispensable tool in medicinal chemistry. With advances in glycobiology and glycochemistry interest in SPHs is increasing. The amide-linkage might render these carbohydrate mimetics more stable than oligosaccharides under physiological conditions, particularly towards glycosidases. The nomenclature of SPHs has not been used consistently. We prefer the term ‘saccharide-peptide hybrids’ [ 11 for reasons of clarity. As an abbreviated form, the

566

22 Saccharide-Peptide Hybrids

term ‘peptidosaccharides’ [2] has also been coined. The term ‘carbopeptoid’ [3] for an SPH is questionable because a ‘peptoid’ [4] is clearly defined as a non-chiral peptide mimetic with an N-substituted glycin as amino acid replacement. In this connection the term ‘glycopeptoid’ [5] for a peptoid carrying a carbohydrate residue has also been introduced. In the expression ‘glycotides’ [6] the suffix is not clearly connected with peptides. Alternatively, and in keeping with the hybrid character, SPHs can be viewed as peptide mimetics. One feature of oligomeric SPHs is their conformational rigidity because of the amide bond. CAAs have therefore been employed as turn mimetics in oligopeptides [7]. This type of molecule would fall into the class of ‘carbohydrate amino acid conjugates’, for which the expression ‘saccharopeptides’ has also recently been used [8].

22.2 Carbohydrate Amino Acids 22.2.1 Natural Carbohydrate Amino Acids Some carbohydrate amino acids occur in Nature as important construction elements. One example is muramic acid, R-2-amino-3-0-( 1-carboxyethyl)-2-deoxy-~-glucose (l),occurring in the peptidoglycan portion of Gram-negative bacterial cell walls in the form of the muramyl dipeptide, N-acetylmuramyl-L-alanyl-D-isoglutamine (Scheme 2) [9]. Glycosaminuronic acids, such as 2-acetamido-2-deoxyglucuronic acid, have also been found in bacterial cell walls [lo]; its N-acetylalanine analog [ 111 2-acetamido-2-deoxygalacturonicacid was found to be the main constituent of the Vi-antigen of E. coli and other bacteria [12]. Derivatives of glucosaminuronic acid were also detected in the cancomycin family of antibiotics [13]. Sialic acids are important constituents of glycoconjugates, often located peripherally on glycoproteins. This family of natural CAAs consists of N- and 0-acyl derivatives of neuraminic acid 2; the main substituents on nitrogen are the N-acetyl and N-glycolyl groups [ 141. Interesting natural CAAs were found in nucleoside antibiotics (for a review see Ref. [ 151). A furanoid amino uronic acid, 5-amino-5-deoxy-~-~-allofuranuronic acid, occurs in the polyoxins [16], e.g. polyoxin J (3) and in the nikkomycins [ 171. The polyoxins also contain a C-5 open-chain 2-amino-2-deoxy sugar acid, a derivative of 2-amino-2-deoxy-~-xylonic acid. A branched chain 6-amino-6deoxyheptopyranuronic acid is the central moiety of amipurimycin (4) and in the related miharamycins [ 18-20]. Two different 3-amino-3-deoxy uronic acids, acid and 3-amino-3,4-dideoxyderivatives of 3-amino-3-deoxy-~-gulopyranuronic D-xylohexopyranuronic acid, were found in ezomycin A1 (5) [211. 4-Amino-4deoxy-glucopyranuronic acid is the carbohydrate moiety in the aspiculamycin/ gougerotin nucleosides [22, 231.

22.2 Carbohydrate Amino Acids

567

HO W H2NC

O

O

H

HO

Me

1

2

3 Polyoxin J

Holln-.
4 Amipurimycin COOH

H2N

5 Ezomycin A,

I NHZ

Scheme 2.

22.2.2 Synthetic Carbohydrate Amino Acids In this chapter some synthetic approaches to CAAs will be described. The most straightforward route to CAAs is oxidation of an amino sugar, with the obvious limitation that few amino sugars, e.9. glucosamine, are readily available. Thus, Heyns and Paulsen [24], back in 1955, described the synthesis of glucosaminuronic acid by catalytic oxidation of the primary hydroxyl group, the method was later improved by Weidmann and Zimmermann [25]. Alternatively, hypochlorite oxidation in the presence of TEMPO can be employed (Scheme 3) to obtain CAA 7 in three steps (40%) [26]. Nucleoside analogs based on glucosaminuronic acid were also prepared in this fashion [27]. Other approaches have been to attach acid residues as substituents. Examples are a CAA 10, derived from activated chitin in four steps (26%) [28], or the demethyl

568

22 Saccharide-Peptide Hybrids

z

- E%

?

HO

OH

11

HO

NH,.HCI

HyOBn

HyOBn

Z

GlcN

Chit1n-H...

HOOC Ho-i

2

i

7

6

0-CHz-COOH O & o $ A

H

0-CH2-COOH k H

0

2 ACO%

[HO

6-

8

NH,.HCI

n

HzN Br 10

9

.HBr

vii

Vi

BuO'OC

HOOC

Z

11 GlcN

Z

12

- L&&cN ...

ix, x, xi

Vlll

AcO

H~&&cooH HO

NPhth

13 GlcNAc-

0-CH,-COOH

AcO

HO

NHAc

Z

xii, xiii

yo.0*

NH2.HCI

-

14

xiv, xv

OH NHBoc

Me0

15

OMe OMe

l6

Scheme 3. (i) ZCI, NaOH; BnOH, Tf20, 0°C to room temp., 4 days, 60%; (ii) NaOC1, TEMPO, KBr, Bu4NC1, HzO, CH2C12, O T , 1.5 h, 67%; (iii) Activation to alkali chitin; ClCHzCOOH, (CH3)2CHOH,room temp.; (iv) lysozyme, 45 "C, 3 days; 2 M HCl, reflux, 5 h, 70%; (v) AcBr, 55 "C, 90 min, 37%; (vi) PhCHO, ZnC12, room temp., 85%; BrCH*COO'Bu, KOH, dioxane, 55 "C, 2 h, 91%; (vii) KOCMe3, THF, H20, O"C, 2 min, 95%; (viii) four steps from GlcN [29]; (ix) 30% HBrAcOH, 0 ° C to room temp., 3 h, 85%; (x) Dowex 50W-X8 [HC], MeOH, 80"C, 16 h, 97%; (xi) LiOH, MeOH/H20 3: 1, 60"C, 16 h; 3 M HCI, reflux., 3 h, 95%; (xii) KOH, MeI, 18-crown-6, DMF, room temp., 16 h, 65%; (xiii) HCl, then di-'Bu-dicarbonate, NEt3, CHC13, room temp., 24 h, 90%; (xiv) PCC, 3 L% molecular sieve, CH2C12, room temp., 5 h, 85%; (xv) KOH, MeI, 18-crown-6, THF, room temp., 3 h; then 4 M HC1, room temp., 16 h, 49%.

22.2 Carbohydrate Amino Acids

569

analog 12 of muramic acid (five steps from glucosamine; 44%) [ 11. By carboxylation of the anomeric center of glucosamine via formation of an anomeric nitrile [29], the heptonic acid derivative 14 was synthesized in seven steps (36%). A longer way to an analogous heptonic acid derivative was chlorination followed by tin activation, lithiation, and carboxylation with carbon dioxide at the anomeric center of a protected glucosamine derivative. Direct PCC oxidation of the anomeric center of a protected glucosamine furnished the open-chain carbohydrate-derived a-amino acid derivative 16 (five steps from N-acetylglucosamine, 24%). The second obvious preparation of CAAs is the amination of a carbohydrate acid. Glucuronic acid was used to introduce an amino function at the anomeric center via an azide delivering a CAA building block on the solid phase in only four steps and excellent yield, giving the &-aminoacid 18 (Scheme 4) [30]. Kessler and colleagues [7] described a similar sequence from glucuronolactone. Also starting with glucuronolactone, the P-glycoside of glucosaminuronic acid was synthesized via glucal formation followed by azidonitration (12 steps, 6% yield) [31]. This approach is less efficient than that starting with glucosamine as described above, in comparison, a simple p-glycosidic analog was prepared in four steps and 49% yield [7].A furanoid CAA was prepared from D-galactonolactone (Scheme 4)-an azide was introduced at the primary position and reaction with triflic anhydride in pyridine, then with methanol, unexpectedly gave two products epimeric at C-5. The diastereomeric mixture was only separable after isopropylidenation to 22 and 23. The free carboxylates, e.g. 24 (18%), or amines were prepared to yield building blocks ready for amide coupling [32]. Although the number of carbohydrate acid derivatives available in reasonable quantity and price to be used as starting materials is limited, it should be pointed out that occasionally conversions to other configurations are straightforward. Thus, the ~-idufuranurono-3,6-lactone 25 is readily available after isopropylidene protection and inversion at C-5 [33]. This lactone was employed to synthesize the piperidine-type CAA 27 with a secondary amine as outlined in Scheme 4 [34]. Often CAAs have been synthesized from sugars, and both amino and carboxyl functionalities have been introduced by using a combination of methodologies described above, or others. Usually the amino function was introduced via an azide. Nitriles are particularly versatile functional groups which can be reduced to amines or hydrolyzed to carboxylic acids [35, 361. An obvious site for the modification of sugars is at the reactive anomeric center. As an alternative to anomeric nitrile chemistry, an anomeric aminomethyl group was introduced via nitromethane condensation (Scheme 5 ) to yield CAA 30 (six steps from glucose, 12%) [37]. Another method used for the introduction of a carboxylic acid was C-allylation followed by oxidative cleavage of the double bond, e.y. CAA 33 was synthesized from peracetylated ribofuranose (six steps, 9Yn) [ 61. Gurjar et al. converted a C-ally1 glycoside to a CAA by hydroxylation of the double bond followed by Jones oxidation [38]. An acid function was also introduced by reacting the free anomeric center with methyl(tripheny1phosphoranylidene) acetate in a Wittig reaction and followed by conversion to a C-glycoside by intramolecular Michael addition, CJ Scheme 5 for the preparation of 36 from 2,3 : 5,6-di-O-isopropylidene-~-mannofuranose (34) [6]. A disadvantage of this

570

22 Saccharide-Peptide Hybrids

HOOC

i, ii

HO OH

OH

HOOC N *oA c AcO

iii, iv

3

OAc

GlcA

OAc

17

19

20

18

21

1+

vii

N3\ + r p r o p

O

25

26

x

23

o

27

Scheme 4. (i) AQO, 12, 0°C to room temp., 2 h; (ii) TMSN3, SnC14, CH2C12, room temp., 16 h, 73%; (iii) 2-chlorotrityl-linked resin, DIPEA, CH2C12, room temp., 1 h, then MeOH; (iv) HS(CHl)zSH, NEt3, room temp., 10 h; (v) 61%; (vi) TfiO, Py, AcOEt, then MeOH, 46%; (vii) MezCHOH, HCI, then Me2CO; (viii) H2, Pd/C, 22 (65%), 23 (23%); (ix) TfiO, Py, CH2C12, -2O"C, then NaN3, DMF, -10 to -2O"C, 1 h; (x) H2, 10% Pd/C, AcOEt, then ZC1, NaHC03, AcOEt/HzO; (xi) CF3COOH-H20, room temp., then HI, Pd black, H20-AcOH 9 : 1 , 4 days, 44% from 25.

approach is the formation of a mixture of both anomeric C-glycosides; these were separated only after azide formation. Similarly, a Wittig reaction was applied to a 3-ketofuranose to introduce a carboxylate into a branched sugar [6]. Dondoni et al. [39] used their anomeric thiazole approach to furnish a-amino acids with the anomeric center as the a-carbon atom, see Scheme 5 for the synthesis of CAA 39. The coupling of a protected serine aldehyde, in which the amino acid functionalities are preformed, to a substituted furan was the key step of the preparation of

571

22.2 Carbohydrate Amino Acids

DGlc

-

-

iii, iv, v

HO HO OH

NO2

MeOOC

HOOC

L*OH HO OH

NHZ

H O HO

29

28

31

33

-

35

'COOMe

36

COOMe XVII-xx

37

NH2

30

32

34

OH

38

xxi

39

Scheme 5. (i) CH3N02, NaOMe; (ii) H+, H20; (iii) H2, Pd/C, ZC1; (iv) 0 2 , Pt/C; (v) MeOH, DCC-DMAP; (vi) NaOH, H2, Pd/C, 120/0; (vii) TMSAlI, TMSOTf, MeCN, 0°C to room temp., 6 h, 67%; (viii) LiOMe, MeOH, room temp., 3 h, 98%; (ix) TsC1, Py, room temp., 2 days, 53%; (x) NaN3, Bu4NI, DMF, llO"C, 4 days; (xi) AczO, NEt3, CH2C12, room temp., 1 day, 290/0; (xii) NaI04, RuC13, MeCN, CCld, H20, room temp., 15 min, 93%. (xiii) Ph3P=CHCOOMe, MeCN, reflux, 14 h, 90%; (xiv) AcOH, HzO, room temp., 16 h, 85%; (xv) MsC1, Py, DMAP, 0°C to room temp., 22 h, 76%; (xvi) NaN3, DMF, 70"C, 8 days, 84%, no yield given for the separation of anomers; (xvii) TMSN3, TMSOTf, CH2C12,4 A molecular sieve, room temp. 20 min, 84%; (xviii) MeOTf, MeCN, 4 A molecular sieve, room temp. 15 min, then NaBH4, room temp., 10 min; (xix) HgC12, MeCN/H20 10: 1, room temp.. 15 min; (xx) AgN03, aq. NaOH, THF, room temp., 36 h, then CH2N2, MeOH/Et20, O"C, 20 min, 53%; (xxi) H2, Pd/C, 'BuOH/H20 9: 1, room temp., 1 h, quant.

572

40

22 Saccharide-Peptide Hybrids

41

42

Scheme 6. (i) Baker’s yeast, 78%; (ii) LiOH, THF, 92%.

the 6-amino-6-deoxyheptopyranuronic acid of amipurimycin (4) [ 191. An interesting approach to a CAA from a non-carbohydrate starting material is by asymmetric synthesis using baker’s yeast. Keto-piperidine carboxylate 40, which is available by cyclization of an a-diazo P-keto ester precursor [40], was reduced to 41 [41] and, after ester cleavage, employed as carbohydrate mimetic 42 (Scheme 6) [42]. A dihydroxylated piperidine carboxylic acid mimicking galactose was prepared recently by syn-hydrogenation of an unsaturated precursor [43].

22.3 Amide-Linked Carbohydrate Polymers The condensation of carbohydrate amino acids was first performed in a polymerization mode, furnishing mixtures of compounds with different degrees of polymerization. Various techniques have been employed. Condensation of the ester 43 in methanolic sodium methoxide furnished a water-soluble polymer 44 which was not further characterized (Scheme 7) [44]. Polymerization of CAA 10 after activation of the acid with diphenylphosphoryl azide (DPPA) resulted in a chloroformsoluble polymer 45. Deacetylation and bromide hydrolysis with sodium methoxide in methanol gave a water-soluble polymer 46 with a free anomeric center readily reducing Fehling solution [28]. The estimated molecular weight was 15 kD as judged by SDS polyacrylamide electrophoresis and also by gel filtration on Sephadex8 G50. Similarly, an amphiphilic alkyl glycoside of glucosaminuronic acid was polymerized with DPPA to result in a degree of polymerization of only 2-13 as judged by MALDI-TOF mass spectrometry [31]. In another approach, the CAA 16 was activated with trichloromethyl orthochloroformate to give the isolable 1,3oxazolidine 47. This intermediate, on treatment with triethylamine in DMF, furnished a polymer 48 with an average molecular weight of 10 kD as judged by gel permeation chromatography [45]. At the structural limit of a CAA, the bicyclic oxalactam 8-oxa-6-azabicyclo[3.2.l]octan-7-one was polymerized in the presence of basic catalysts to give amino-‘uronic acid’ polymers devoid of hydroxyl groups [46]. Whereas the above examples start from CAAs, the construction of amide-linked polymers was also possible in a Nylon type condensation using sugar diamines derived from anhydroalditols [47] or hexoses [48] and dicarboxylic acids as building blocks. For example, methyl 2,6-diamino-2,6-dideoxy-a-~-glucopyranoside (49) was

22.3 Amide-Linked Curhohydrufe Polyrnevs

-

573

i

H2N& HO

COOMe OH

Me n

OH

43

44

..

45

Me0

46

iv

p::

OMe

6Me OMe 0

OMe

16

hH;j 47

vi T

Ho% HO CH,

49

48

HO

n

Scheme 7. (i) NaOMe, MeOH, 10O"C, 4 h, autoclave; (ii) PhZP(O)N3, NEt3, DMSO, 20°C; 4 days, 44%; (iii) NaOMe, MeOH, room temp., 30 min; (iv) C13CO-COC1, THF, 5 5 " C , 4 h, 95%; (v) NEt,, DMF, room temp., 2 days, precipitate. with EtzO; (vi) ClOC-R-COCl, Na2C03, H2OCC14, room temp., 2h.

coupled to aliphatic or aromatic diacids in an interfacial polycondensation [48]. Molecular weights of polymers 50 were in the range of 20 kD for aromatic spacers. The products were structurally not uniform, because 2-6, 2-2, and 6-6 linked polymers could be formed. This field of carbohydrate-derived polyamides has been expertly reviewed [49].

574

22 Succhuride-Peptide Hybrids

22.4 Amide-Linked Carbohydrate Oligomers 22.4.1 Solution Synthesis

A first synthesis of a saccharide-peptide hybrid in a directed synthesis was performed by Yoshimura et al. in 1976 [50]. The CAA 51 was activated with cyclohexylcarbodiimide and linked to amino sugar 52 to result in dimer 53. This glucuronic acid amide was less soluble in water than the corresponding mannuronic acid derivative. Dimer 53 could be converted, after removal of the anomeric protecting group by hydrogenation, to the amide-linked alditol 54 (Scheme 8) by reduction with sodium borohydride. This publication was not followed up at the time until the interest in carbohydrate mimetics rose [51]. The first SPH higher oligosaccharide mimetic was built up in a [ 2 + 2 ] block synthesis by Wessel et al. in 1995 [ 11. The benzyloxycarbonyl (Z) group was employed as the temporary amine protective group, and carboxylic acids were protected as tert-butyl esters. The CAAs of type 12 were coupled in very good yields (84-87%) using mixed anhydrides, and the dimeric building block 56 was activated with 2-chloro-4,6-dimethoxy-l,3,5-triazine (CDMT). Notably, this methodology enabled the coupling of a building block with four unprotected hydroxyl groups to give tetramer 57. In recent years different peptide coupling techniques have been used to assemble oligomers. Thus, McDevitt and Lansbury [6] coupled their protected CAAs with 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDCI) in dichloromethane in a stepwise fashion to result in trimer 59 and a number of dimeric SPHs. The amino function was protected as an azide. For the construction of a small carbohydrate mimetic library the group also prepared ‘spaced sugars’ [ 521 such as 60 on an aromatic template. For a better overview, the various coupling techniques are summarized in Table 1. Sialic acid derivatives have received particular attention. Sabesan [2] prepared an isosteric SPH analog 63 of the a-~-NeuAc(2-6)-P-~-Galp disaccharide, the terminal unit of numerous glycoproteins and glycolipids and the receptor ligand for influenza virus hemagglutinin and substrate for neuraminidase. The neuraminidase activity of 63 and analogs have not yet been reported. Gervay et al. described amino acid conjugates of neuraminic acid [6I], but also an amide-linked dimeric neuraminic acid 64 [55](Scheme 9). In this case the hydroxyl groups of the acid component had to be protected to avoid lactonization. The spaced neuraminic acid oligosaccharide mimetics 65 and 66 were synthesized by use of a glycine spacer. The Ishikawa group [ 621 also reported amide-linked oligomers with alternating CAAs and amino acids. The oligomer 67 was constructed with a BOP coupling approach as described above (Table 1, entry 6). For the synthesis of SPH 68, EDCI in combination with HOBt turned out to be advantageous coupling conditions. It is worth noting that the coupling yields for the higher oligomers did not decrease (Table 1, entry 11). This is a clear advantage of the amide coupling and in contrast with many oligosaccharide syntheses where yields are often getting lower for higher oligomers. The same coupling technique was employed by Fleet and colleagues [57]in the synthesis of a furanoid tetramer 69

22.4 Amide-Linked Carbohydrate Oligomers

-

HOOC Ho-i

+

HO HO

515

HO

H/NoBn Ad

51

52 53 ii, iii

I

Ac

,OAc

55

0

+

0 0

56

57

O

58 R = N , Ms

59 R =

MeOOC COOMe

H

60

Scheme 8. (i) DCC, Py-NEt3, room temp., 5 days, 710/0;(ii) Hz, Pd/C, HlO, HOAc, 80%; (iii) NaBH4, Hz0, room temp., 2 h, 99%; (iv) CDMT, N-methylmorpholine, DMF, 50"C, 18 h, then Ac20, Py, 400/; (v) NaOMe, dioxane-MeOH, room temp., 2.5 h, 81Y0.

516

22 Saccharide-Peptide Hybrids

0

OH H

HO

O

b N NHBoc 0

H

NHBoc

61 H HO O

AcOyOAc

OH

W NH I

COOH

BocHN AcO

64

AcHN

AcO

AcO

,OTr

68

Scheme 9.

22.4 Amide-Linked Carbohydrate Oliyomers

577

N3qE*~e 69

AcO*#

OAc

AcO

OAc

2

AcO

OAc

BOC

BOC

71

72

I

Me0

O(CH2)&OOMe

HO

bH

73

Scheme 10.

(Scheme 10). Even an octamer 70 could be assembled in very good yield in a block synthesis (Table 1, entry 13) [58]. By use of pyrrolidine carbohydrate mimetics and a 6-aminoglucose or glucuronic acid building block it was demonstrated that isomeric disaccharide mimetics can be prepared by formation of amide- and inverse amide-linked analogs such as 71 and 72 1591. SPH 73 and several analogs with different uronic acid moieties were prepaied- to serve as starting materials for higher oligosaccharide mimetics [60, 631.

578

22 Saccharide-Peptide Hybrids

Table 1. Conditions for coupling reactions of CAAs. Entry 1 2 3 4

5

6

Coupling conditions Dicyclohexylcarbodiimide, Py-NEt3, room temp., 5 days ClCOO'Bu, NEt3, THF-MeCN, room temp., 12 h; then AczO, Py CDMT, N-methylmorpholine, DMF, 50"C,18 h; then AczO, Py 2.5 equiv. acid, EDCI, NEt3, room temp., 24 h Acid chloride, NEt3, CH2C12 1.2 equiv. acid, BOP, DIPEA, DMF, room temp., 16 h

7

1.2 equiv. acid, DEPC, NEt3, DMF, 0 ° C to room temp.

8 9

CBMIT, MeCN, room temp., 4 days BOP, DIPEA, DMF, room temp., 48 h; then Ac20, Py BOP, HOBt, DIPEA, 1.2 equiv. amine, CH2C12-DMF, room temp., 48 h, then Ac20, PY BOP, DIPEA, CH2C12, room temp., 1 h EDCI, HOBt, THF, room temp., 10 h

10

11

12 13 14 15

EDCI, HOBt, DIPEA, DMF, room temp.; then Ac20, Py EDCI, HOBt, DIPEA, CH2C12 DCC, HOBt, NEt3, THF, 0 ° C to room temp., overnight HBPyU, NEt3, DMF, room temp., overnight

Block size

Yield ("YO)

Structure (Scheme)

71

53 (8)

? 90 85 81 45 27

61 (9)

Ref.

62 (9) 63 (9) 64 (9)

73 81 84 78

68 (9)

55

69 (10)

75 62

70 (10) 71 (10)

71

73 (10)

Abbreviations: CDMT = 2-chloro-4,6-dimethoxy-l,3,5-triazine; EDCI = l-ethyl-3-(3-dimethylaminopropy1)carbodiimide; BOP = benzotriazol-l-yloxy-tris-(dimethylamino)phosphonium hexafluorophosphate; DIPEA = diisopropylethylamine; DEPC = diethylphosphoryl cyanide; CBMIT = 1,l-carbonylbis(3-methylimidazoliumtriflate; HOBt = 1-hydroxybenzotriazole hydrate; HBPyU = 0-(benzotriazole-1 -yl)-N,N,N,N-bis(tetramethy1ene)uronium hexafluorophosphate.

22.4.2 Solid-Phase Synthesis With a view to application in combinatorial chemistry and to reduce solubility problems, Miiller et al. [26] developed a solid-phase approach to SPHs in 1995. Uronic acid 7 was coupled to benzhydrylamine polystyrene functionalized with

22.4 Amide-Linked Carbohydrate Oligomers

579

an amide linker to result in 74. The N-protection of choice was the fluoren-9ylmethoxycarbonyl (=Fmoc) group. Uronic acids were activated in situ with TATU in the presence of Hunig’s base, these coupling conditions did not require the protection of hydroxyl groups. The couplings were quantitative (Scheme 11), and the crude product after detachment from the resin was pure by NMR. Three coupling cycles and cleavage from the resin furnished the homotetramer 75. A similar strategy was adopted to link glucosaminuronic acids carrying a nucleo-base at the anomeric center [64]. The authors used 0-silyl protected CAA monomers and the TATU-related hexafluorophosphate ‘HATU’ (Scheme 11) as activating agent to generate amide-linked nucleic acid analogs as antisense agents. Oligomers up to 14mer 76 were prepared. The solid-phase-attached and O-protected CAA 18 (Scheme 4) was extended by azide-protected uronic acid 17 to give the dimeric SPH 77 (Scheme 11) which was further elongated with amino acids (see Chapter 28) [30]. With 3.6 equivalents of acid using Knorr activation and 25 equivalents of reducing agent (propane-l,3thiol) for the azide, nearly quantitative yields were reached for the dimer. In contrast, Long et al. pointed out that neither coupling nor azide reduction on the solid phase proceeded to completion when homologating their solid-phase-bound monomer 78 when using a diimide coupling. Even for the trimer 79 low yields were obtained (Scheme 11) [65]. With the piperidine carboxylic acid 42 and two related unhydroxylated amino acids (Boc-4-piperidine carboxylic acid and Boc-R-pipecolic acid) a small library was constructed on a 4-methyl benzhydrylamine resin by use of HATU activation in DMF-dichloromethane 1 : 1. N-Fmoc-protected neuraminic acid coupled to rink resin via an E-aminocaproic acid linker was homologated up to an octamer 80 (Scheme 11), and the low yields experienced during solution synthesis [ 551 were overcome; oligomer yields ranged from 44&55%(1.4 equiv. acid-more acid was added when required, BOP, DIPEA, NMP) [66]. 22.4.3 Biological Activity The main reason for the recent interest in SPHs is the topicality of carbohydrate mimetics, the attempt to mimic the biological function of carbohydrates. The advantage of SPHs would be the access in a synthetically facilitated fashion, and stability towards glycosidases would be expected. The oligomers 57 (Scheme 8) and 75 (Scheme 11) were sulfated to be investigated as heparinoid mimetics (unpublished data). A sulfated derivative of 61 (Scheme 9) fully blocked syncytium formation caused by HIV infection to CD4 cells at 50 J ~ Mconcentration [53], and the persulfated derivative of tetramer 62 (Scheme 9) had micromolar activity in the protection of MT2 cells from HIV infection [54]. The oligomeric conjugate 67 (Scheme 9) was evaluated in highly metastatic cell lines; inhibition of cell adhesion (Chemotaxis) and invasion were measured. In a chemotactic assay in which CD44-expressing lung carcinoma cells adhere to vitronectin, this pentamer with L-aspartate spacer and the analogous trimer with a

580

22 Saccharide-Peptide Hybrids

H2$f+ H

-

HNOBn

i,ii

7 -

74

H i s s ) HN OBn

Ho* HO

Fmoc

HNOBn

75

HO% HO H2N

HO & :H 2 HO

HNwBasi& -

...

111

OBn

HN

18

HO

76 n = 1 2 Base = Thymine or cytosine

HO

iv

Base

OAc

H2N

OH

H

AcO 2

OAc N

AcOe

78

OAcw

;

OAc p NH2

79

Hd

"

*'"

' 6

80

Scheme 11. (i) one cycle: 2 equiv. acid, 0-(7-azabenzotriazol-l-yl)-l, 1,3,3-tetramethyluronium tetrafluoroborate (TATU), DIPEA, DMF, room temp., 45 min, quant., then 20% piperidine, DMF, 2 x 7 min; (ii) TFA, CHzC12, room temp., 2 x 2 h; (iii) one cycle: 2 x 5 equiv. acid, G ( 7 azabenzotriazol-l-yl)-l,1,3,3-tetramethyluronium hexafluorophosphate (after structure revision: N[(dimethylamino)-lH-1,2,3-triazolo-[4,5-b]pyridin-l-ylmethylene]-N-methylmethanaminium hexafluorophosphate N-oxide [67], HATU), DIPEA, DMF, room temp., 20 min, 65-1000/0; (iv) 3.6 equiv. 17, 0-(benzotriazol-l-yl)-l, 1,3,3-tetramethyluroniumhexafluorophosphate (HBTU), HOBt, PIPEA, DMF, room temp., 1 h, quant.: (v) one cycle: diisopropylcarbodiimide, HOBt, DMF; dithiothreitol, DIPEA, DMF, 50 "C, 30%, lower yield for trimer; (vi) AczO, Fy;50% TFA-CH2Clz.

22.4 Amide-Linked Carbohydrate Oligomers

58 1

HOOC HO NHCO-Leu-Phe-Gly-Gl y -Tyr-H OH

81

HO OH HO HO

HO HO

82

HO HO

OH OH

83

Me?

Scheme 12.

D-aspartate spacer inhibited adhesion by ca 50% at 10 p~ [62]. In the tumor cell invasion assay, 67 slightly inhibited the invasion of A549 cells through an artificial membrane, by 64% at 10 p ~ . On the basis of the notion that the 6-glucuronide of morphine is distinctly more active in analgesia than morphine itself, glucuronide analogs of enkephalins were targeted, and the peptide conjugated SPH 81 (Scheme 12) was synthesized from 77 [301. In a mouse deferens assay, the sugar dimer-modified leucin enkephalinamide 81 was twice as active as the Leu-enkephalinamide standard. As an exact analog of the phytoalexin elicitor branched heptasaccharide 82 the branched SPH 83 was devised with four amide bonds replacing the 1 + 6 linkages of the pentasaccharide backbone and thus producing the same length. No elicitor activity was detected for 83, possibly because of the reduced flexibility or different conformation of the mimetic structure [56]. Mimetics of the sialyl Lewis' and sialyl Lewisa oligosaccharides have recently found increased attention. Baisch and

582

22 Saccharide-Peptide Hybrids

Ohrlein have prepared mimetics in which the acetamido group of the central glucosamine was replaced by methyl glycopyranosiduronamides [60, 631. Disaccharide mimetics of type 73 were accepted as substrates by the enzymes 2,3-sialyltransferase, p(1 + 4)galactosyltransferase and fucosyltransferases I11 and IV, so enzymatic synthesis of sialyl LewisXand sialyl Lewisa mimetics such 84 (Scheme 12) was feasible. It will be interesting to see the biological activity of these mimetics. NucleoSPHs such as 76 with amide replacement of the usual phosphodiester linkage were synthesized as antisense agents. Judged by melting temperatures and thermodynamic constants some binding properties of DNA and RNA could be mimicked ~41. A library of piperidine-based SPH trimers incorporating monomer 42 (Scheme 6) was tested for glycosidase activity. No inhibition was observed for yeast a-glucosidase and isomaltase, almond P-glucosidase, human placenta a-fucosidase, snail pmannosidase, and E. coli P-galactosidase, possibly because of the small number of hydroxyl groups present. 22.4.4 Conformational Properties In their pioneering work, Kessler and coworkers described the conformational role of CAAs in peptide strands, fixing a certain conformation or inducing turns [7]. Glucosaminuronic acids as dipeptide isosteres constrain the conformation of linear peptides. 1-Aminoglucuronic acids (cf 18, Scheme 4) would be p-turn mimetics, 2,6anhydro-7-amino-7-deoxyheptonic acids (cf: 30, Scheme 5) flexible p-turn mimetics. 2,6-Anhydro-3-amino-3-deoxyheptonic acids ( c j 14, Scheme 3) might serve as yturn mimetics, and hydroxylated pipecolic acid derivatives (cf 27, Scheme 4) can be regarded as homoproline derivatives. Thus, in this view the CAAs are seen as peptide mimetics. The conformation of the oligomers formed from CAAs and D- or Laspartic acid (67, Scheme 9) were, according to calculations, distinct and determined by the interaction of the CAA 3-OH group and the a-carboxylic moiety of the aspartic acid residues (cf: 85, Scheme 13) [62]. The other conformational determinant would be the positioning of the anomeric amide with the carbonyl group avoiding the vicinity to the anomeric oxygen and favoring H-bond formation with the CAA 2-OH group. The resulting two different conformations for the D- and Laspartic acid series would be in keeping with the different biological activity of both series. From the properties of the nucleo-CAAs like 76 and their RNA binding consistent with a Watson-Crick binding model, it follows that preferred conformations were adopted [64]. The SPH pentamer backbone of 83 (Scheme 12) was found to be more rigid, according to force field calculations for hydrated oligosaccharides, than the corresponding oligosaccharide 82, thus explaining the loss of biological activity [56]. The N-acetylneuraminic acid oligomers prepared by Gervay’s group [66] possibly occur in helical conformations according to circular dichroism data and NMR hydrogen-deuterium exchange experiments, the first ordered structure was observed for a tetramer. The C-glycosidic P-D-arabinofuranose oligomer 69 (Scheme 10) and the corresponding hexamer described by Fleet and colleagues were found to form p-turn-

22.4 Amide-Linked Carbohydrate Oliyomers

583

85 For D-aspartic acid: R’ = H, R2 = NH-R For L-aspartic acid: R2 = H, R’ = NH-R

86

BAC

Scheme 13.

type structures stabilized by hydrogen-bonds of neighboring units [ 681. As shown for the analogous terminal amide 86 (Scheme 13), a clear secondary structure was already visible for a trimer [65].The tetramer 87, 3,4-isopropylidene acetalprotected and with regard to the backbone isomeric with 69, and its cyclohexylidene acetal-protected enantiomer had similar p-turn structures [ 581. In contrast, the CI-Darabinofuranose analogs with the C-2/C-5 substituents of the furanose rings in the trans position showed no indication of secondary structure [69]. The conformation of the SPH 70 (Scheme 10) and the tetramer substructures, also with the C-2/C-5 trans substituted tetrahydrofuran monomers, differed from the previous cases, and the NMR data were best interpreted in terms of a left-handed a-helix [58]. In conclusion, it seems that some SPHs as new unnatural oligomers [70] adopt specific compact conformations and can thus be regarded in Gellman’s terminology as ‘foldamers’ [71]. References 1. H. P. Wessel, C. M. Mitchell, C. M. Lobato, G. Schmid, Angew. Chem. Znt. Ed. Engl. 1995, 34, 2712-2713. 2. S. Sabesan, Tetrahedron Lett. 1997, 38, 3121-3130. 3 . K. C. Nicolaou, H. Florke, M. G. Egan, T. Barth, V. A. Estevez, Tetrahedron Lett. 1995, 36, 1775-1778. 4. R. J. Simon, R. S. Kania, R. N. Zuckermann, V. D. Huebner, D. A. Jewell, S. Banville, S. Ng, L. Wang, S. Rosenberg, C. K. Marlowe, D. C. Spellmeyer, R. Tan, A. D. Frankel, D. V. Santi, F. E. Cohen, P. A. Bartlett, Proc. Natl. Acud. Sci. USA 1992, 89, 9367-9371. 5. U. K. Saha, R. Roy, Tetruhedron Lett. 1995, 36, 3635-3638. 6. J. P. McDevitt, P. T. Lansbury Jr., J. Am. Chem. Soc. 1996, 118, 3818-3128.

584

22 Saccharide-Peptide Hybrids

7. E. Graf von Roedern, E. Lohof, G. Hessler, M. Hoffmann, H. Kessler, J. Am. Chem. Soc. 1996,118, 10156-10167. 8. P. S. Ramamoorthy, J. Gervay, J. Org. Chem. 1997, 62, 7801-7805. 9. G. Baschang, Tetrahedron 1989, 45, 633 1-6360. 10. A. R. Williamson, S. Zamenhof, J. Biol. Chem. 1963,238, 2255-2258. 11. S. Hanessian, T. H. Haskell, J. Biol. Chem. 1964,239, 2758-2764. 12. K. Heyns, G. Kiessling, W. Lindenberg, H. Paulsen, M. E. Webster, Chem. Ber. 1959, 92, 2435-2438. 13. J. P. Waltho, D. H. Williams, E. Selva, P. Ferrari, J. Chem. Soc. Perkin Trans. 1 1987, 21032107. 14. R. Schauer, Adv. Carbohydr. Chem. Biochem. 1982, 40, 131. 15. S. Knapp, Chem. Rev. 1995, 95, 1859-1876. 16. K. Isono, K. Asahi, S. Suzuki, J. Am. Chem. SOC. 1969, 91, 7490-7505. 17. W. A. Konig, H. Hahn, R. Rathmann, W. Hass, A. Keckeisen, H. Hagenmaier, C. Bormann, W. Dehler, R. Kurth, H. Zahner, Liebigs Ann. Chem. 1986,407-421. 18. P. Garner, in Atta-ur-Rahman (Ed.): Studies in Natural Product Chemistry, Vol. I , Elsevier, Amsterdam 1988, p. 397-434. 19. G. Casiraghi, L. Colombo, G. Rassu, P. Spanu, J. Org. Chem. 1991, 56, 6523-6527. 20. A. J. Fairbanks, P. Sinay, Synlett 1995, 277-279. 21. S. Knapp, C. Jaramillo, B. Freeman, J. Org. Chem. 1994,59,4800-4804. 22. J. J. Fox, Y. Kuwada, K. A. Watanabe, Tetrahedron Lett. 1968, 57, 6029-6032. 23. F. W. Lichtenthaler, T. Morino, H. M. Menzel, Tetrahedron Lett. 1975, 9, 665-668. 24. K. Heyns, H. Paulsen, Chem. Ber. 1955, 88, 188-195. 25. H. Weidmann, H. K. Zimmerman jr., Justus Liebigs Ann. Chem. 1961, 639, 198-203. 26. C. Muller, E. Kitas, H. P. Wessel, J. Chem. Soc., Chem. Commun. 1995, 2425-2426. 27. R. A. Goodnow Jr., A.-R. Richou, S. Tam, Tetrahedron Lett. 1997, 38, 3195-3198. 28. S. Tokura, Y. Ikeuchi, S.4. Nishimura, N. Nishi, Int. J. Biol. Macromol. 1983, 5, 249. 29. R. W. Myers, Y. C. Lee, Curbohydr. Res. 1984, 132, 61-82. 30. B. Drouillat, B. Kellam, G. Dekany, M. S. Starr, I. Toth, Bioorg. Med. Chem. Lett. 1997, 7, 2241-2250. 31. S.-I. Nishimura, S. Nomura, K. Yamada, J. Chem. Soc., Chem. Commun. 1998, 617-618. 32. D. D. Long, N. L. Hungerford, M. D. Smith, D. E. A. Brittain, D. G. Marquess, T. D. W. Claridge, G. W. J. Fleet, Tetrahedron Lett. 1999, 40, 2195-2198. 33. R. Csuk, H. Honig, J. Nimpf, H. Weidmann, Tetrahedron Lett. 1980,21, 2135-2136. 34. B. P. Bashyal, H.-F. Chow, G. W. J. Fleet, Tetrahedron Lett. 1986, 27, 3205-3208. 35. E.-F. Fuchs, J. Lehmann, Chem. Ber. 1975, 108, 2254-2260. 36. H. S. Overkleeft, S. H. L. Verhelst, E. Pietermann, N. J. Meeuvenoord, M. Overhand, L. H. Cohen, G. A. van der Marel, J. H. van Boom, Tetrahedron Lett. 1999, 40, 4103-4106. 37. E. Graf von Roedern, H. Kessler, Angew. Chem. 1994, 106, 684-686. 38. M. K. Gurjar, A. S. Mainkar, M. Syamala, Tetrahedron Asymm. 1993, 4, 2343-2346. 39. A. Dondoni, M.-C. Scherrmann, A. Marra, J.-L. Delepine, J. Org. Chem. 1994,59, 7517-7520. 40. M. P. Moyer, P. L. Feldman, H. Rapoport, J. Org. Chem. 1985,50, 5223-5230. 41. D. W. Knight, N. Lewis, A. C. Share, Tetrahedron Asymm. 1993, 4, 625-628. 42. E. Byrgesen, J. Nielsen, M. Willert, M. Bols, Tetrahedron Lett. 1997, 38, 5697-5700. 43. P. Bach, A. Lohse, M. Bols, Tetrahedron Lett. 1999, 40, 367-370. 44. E.-F. Fuchs, J. Lehmann, Carbohydr. Res. 1976, 49, 267-273. 45. M. d. G. Garcia Martin, M. Violante de Paz Bafiez, J. A. Galbis Perez, Carbohydr. Res. 1993, 240, 301-305. 46. H. Sumitorno, K. Hashimoto, Macromolecules 1977, 10, 1327-1331. 47. J. Thiem, F. Bachmann, Makromol. Chem. 1991, 192, 2163-2182. 48. J. Thiem, F. Bachmann, Makromol. Chem. 1993, 194, 1035-1057. 49. J. Thiem, F. Bachmann, Trends Polym. Sci. 1994,2,425-432. 50. J. Yoshimura, H. Ando, T. Sato, S. Tsuchida, H. Hashimoto, Bull. Chem. Soc. Jpn. 1976, 49, 2511-2514. 51. Y. Chapleur, Carbohydrate Mimics: Concepts and Methods, Verlag Chemie, Weinheim 1998. 52. H. P. Wessel, Topics Curr. Chem. 1997, 187, 215- 239.

References

585

53. Y. Suhara, J. E. K. Hildreth, Y. Ichikawa, Tetrahedron Lett. 1996, 37, 1575-1578. 54. Y. Suhara, M. Ichikawa, J. E. K. Hildreth, Y. Ichikawa, Tetrahedron Lett. 1996, 37, 25492552. 55. J. Gervay, T. M. Flaherty, C. Nguyen, Tetrahedron Lett. 1997, 38, 1493-1496. 56. C. M. Timmers, J. J. Turner, C. M. Ward, G. A. van der Marel, M. L. C. E. Kouwijzer, P. D. J. Grootenhuis, J. H. van Boom, Chem. Eur. J. 1997,3, 920-929. 57. M. D. Smith, D. D. Long, D. G. Marquess, T. D. Mi. Claridge, G. W. J. Fleet, Chem. Cornmun. 1998,2039-2040. 58. T. D. W. Claridge, D. D. Long, N. L. Hungerford, R. T. Aplin, M. D. Smith, D. G . Marquess, G. W. J. Fleet, Tetrahedron Lett. 1999, 40, 2199-2002. 59. I. McCort, A. Dureault, J.-C. Depezay, Tetrahedron Lett. 1998, 39, 4463-4466. 60. G. Baisch, R. Ohrlein, Bioorg. Med. Chem. 1998, 6, 1673-1782. 61. J. Gervay, P. S. Ramamoorthy, N. N. Mamuya, Tetrahedron 1997, 53, 11039-11048. 62. Y. Suhara, M. Izumi, M. Ichikawa, M. B. Penno, Y. Ichikawa, Tetrahedron Lett. 1997, 38, 7167-71 70. 63. G. Baisch, R. Ohrlein, Carbohydr. Res. 1998, 312, 61-12. 64. R. A. Goodnow Jr., S. Tam, D. L. Pruess, W. W. McCornas, Tetrahedron Lett. 1997,38, 31993202. 65. D. D. Long, M. D. Smith, D. G. Marquess, T. D. W. Claridge, G. W. J. Fleet, Tetrahedron Lett. 1998, 39, 9293-9296. 66. L. Szabo, B. L. Smith, K. D. McReynolds, A. L. Parrill, E. R. Morris, J. Gervay, J. Org. Chem. 1998,63, 1074-1078. 67. I. Abdelmoty, F. Albericio, L. A. Carpino, B. M. Foxman, S. A. Kates, Letters in Peptide Science 1994, 1 , 57. 68. M. D. Smith, T. D. W. Claridge, G. E. Tranter, M. S. P. Sansom, G. W. J. Fleet, Chem. Cornmun. 1998, 2041-2042. 69. M. D. Smith, D. D. Long, A. Martin, D. G. Marquess, T. D. W. Claridge, G. W. J. Fleet, Tetrahedron Lett. 1999,40,2191-2194. 70. M. J. Soth, J. S. Nowick, Curr. Opin. Chem. Bid. 1997, 1, 120-129. 71. S. H. Gellman, Acc. Chem. Res. 1998, 31, 173-180.

Part I Volume 2

I11

Enzymatic Synthesis of Glycosides

Carbohydrates in Chemistry and Biology Editor Beat Ernst, Gerald W. Hart, Pierre Sinay copyright OWILEY-VCH Verlag GmbH, 2000

23 On the Origin of Oligosaccharide SpeciesGlycosyltransferases in Action Dirk H. van den Eijnden

23.1 Introduction Glycoproteins and glycolipids, jointly called glycoconjugates, are macromolecules essential to eukaryotic life. The oligosaccharide chains carried by glycoconjugates often contain biological intelligence which is encoded by the oligosaccharide structure. These chains thus confer specific biological functions on the proteins and lipids carrying them. When embedded in the cellular plasma membrane they decorate the cell surface with specific oligosaccharide structures which can be crucial to cell function. The synthesis of protein- and lipid-linked oligosaccharides proceeds by the sequential addition of monosaccharide building blocks to a growing chain that is covalently linked to the protein or lipid aglycone. In a similar way oligosaccharides present in some secretions (e.g. milk) are built up from a single sugar (glucose). Unlike the processes of protein and nucleic acid synthesis, in which the order of attachment of amino acids and nucleotides is read from a nucleic acid matrix, glycosylation is a non-template-directed process. This enables the formation of branches which are common to oligosaccharides but are not found in nucleic acids and proteins. The enzymes that are responsible for the assembly of the oligosaccharides on glycoconjugates are known as glycosyltransferases (in the early stages of N-glycosylation glycosidases also play a role-see later in this chapter). Being proteins, glycosyltransferases are primary products of their respective genes and, therefore, the oligosaccharide chains formed by their action can be considered as secondary gene products. Glycosyltransferases typically catalyze the following general reaction: nucleoside-P-Psugar

+ acceptor + sugar-acceptor + nucleoside-P-P

In this reaction nucleotideesugars are the donor substrates (for some glycosyltransferases involved in the early stages of protein N-glycosylation a sugarphosphate-dolichol is the donor molecule; see later). In sequential reactions mono-

590

23 On the Origin of Oligosaccharide Species-Glycosyltransferases in Action

saccharides are attached, one at a time, through glycosidic linkages at nonreducing ends and at branching points in a specific order; the sugar-acceptor product of one reaction is the acceptor substrate in the next. In vitro glycosyltransferases generally have optimum activity at a pH of ca 7.0. In most instances Mn2+ ions are required for enzymatic activity (except for sialyltransferases and some p6-N-acetylglucosaminyltransferases which are cation-independent). Because glycosyltransferases are generally membrane-associated proteins, determination of their activity in vitro requires the solubilization of the enzymes with detergents. Some glycosyltransferases, however, occur in a soluble form in body fluids, probably as a result of proteolytic release of the catalytically active part of the enzyme. Glycosyltransferases of eukaryotic organisms have high specificity for the sugar they transfer and the linkage they establish. This has enabled their classification in groups, such as the sialyltransferases which all exclusively utilize CMP-sialic acid as a donor substrate, and the galactosyltransferases which highly prefer UDP-Gal as a donor, with subgroups based on the linkage that is established. In many (but not all) instances the molecular cloning of the cDNAs of these enzymes has provided molecular support for this classification (see Section 23.17, Glycosyltransferase families). Glycosyltransferases are, furthermore, generally specific for the accepting sugar and its anomeric configuration, the linkage by which this sugar is linked to or is substituted by another sugar residue, and the nature of these other residues. Sometimes the entire underlying oligosaccharide structure and possibly its spatial conformation seems to be recognized. It also has been suggested that some glycosyltransferases recognize specific peptide motifs of the polypeptide aglycone, which could be one explanation for site specific glycosylation patterns on glycoproteins with more than one glycosylation site. Very many different glycosidic linkages are known to occur in glycoconjugatelinked and free oligosaccharides. The formation of each linkage is catalyzed by at least one specific glycosyltransferase. As mentioned above, such an enzyme is generally not capable of catalyzing the formation of another linkage-type. A notable exception to the ‘one enzyme-one linkage’ concept [ 11, however, is the Lewis blood group a3/4-fucosyltransferase, which can establish Fuc(a1-4)[Gal( Pl-3)IGlcNAc and Fuc(a1-3)[Gal(P1-4)]GlcNAc linkages in vivo and in vitro [2-41. Also other glycosyltransferases, particularly when the substrate concentrations used are unphysiologically high, can establish more than one linkage by transferring to another acceptor sugar [5] or by transferring a monosaccharide from a different sugardonor substrate (donor promiscuity) [6-111. Even though glycosylation is a non-template directed process, oligosaccharide structures are generally produced with high fidelity. Firstly, the glycosylation potential of a cell (expression levels of the different glycosyltransferases; subcellular localization of the enzymes; supply of donor substrates) determines which pathways potentially can be followed. It should be noted that this potential is not constant but may vary with cell status (normal versus malignant), with the developmental stage of the cell, and with the environment wherein the cell grows (e.g. energy supply). Furthermore the specificities of the glycosyltransferases involved play a major controlling role in the formation of the correct final product structure(s). In such a process, however, it is inevitable that at intersections in the metabolic pathway,

23.2 Protein N-Glycosylation

59 1

where enzymes compete with one another for common intermediate substrate structures, various directions can be taken. In practice this leads to the synthesis of an array of related oligosaccharide structures rather than of one single structure. Such variations in oligosaccharide structure at single glycosylation sites is a common feature of glycoproteins and is known as microheterogeneity. It causes glycoproteins to occur in different glycoforms, which are oligosaccharide variants of the same polypeptide. Many excellent reviews have described the topography, mechanisms, and enzymology of protein and lipid glycosylation [12-25). This chapter gives a concise introduction to the enzymatic properties and molecular relationships of glycosyltransferases with emphasis on protein N-glycosylation and the corresponding pathways of oligosaccharide assembly, in particular the late stages. Where appropriate, however, the glycosyltransferases involved in protein O-glycosylation, biosynthesis of the core of proteoglycans, and the glycosylation of sphingoglycolipids are also discussed.

23.2 Protein N-Glycosylation: Pre-assembly of Oligosaccharide-PPDolichol and en bloc Transfer Protein N-glycosylation is a process that has been highly conserved during evolution. In this process several stages can be distinguished; the early ones are shared by all organisms from yeasts to mammals. The initial steps, in which an oligosaccharide precursor structure is synthesized on a dolichol-diphosphate (Dol-PP) lipid carrier, occur in the endoplasmic reticulum (ER). The most simple picture that complies best with current knowledge [ 12, 15, 181 is that depicted in Figure 1. Several glycosyltransferases, embedded in the membranes of the ER with their catalytic domain facing the cytosol of the cell, transfer GlcNAc-P, GlcNAc, and Man to Dol-P from UDP-GlcNAc and GDP-Man, respectively, to yield a DolPP linked Man~GlcNAc2oligosaccharide. The molecule then flip-flops over the membrane resulting in a luminal exposed oligosaccharide structure. Subsequently four additional Man and three Glc residues yielding Glc3Man~GlcNAc2-PP-Dol are attached by glycosyltransferases that have their catalytic domain facing the lumen. The latter enzymes are unusual in that they utilize Man-P-Do1 and Glc-PDo1 rather than GDP-Man and UDP-Glc as a sugar donor. The Dol-P linked sugars are synthesized from the nucleotide-sugars at the cytosolic face of the ER, whereafter they flip-flop over the membrane to become utilizable for the luminal glycosyltransferases. In the next stage of N-glycosylation this pre-assembled oligosaccharide is transferred en bloc to an Asn-X-Ser/Thr sequon (where X can be any amino acid except Pro or Asp) in the nascent polypeptide chain of a glycoprotein being synthesized at the ribosome (Figures 1 and 2). This co-translational step is catalyzed by the oligosaccharyltransferase (OST) complex which consists of at least three proteins [26, 271 that seem to be intimately associated with membrane-bound ribosomes [ 191.

592

23 On the Origin of Oligosaccharide Species-Glycosyltransferases in Action

Figure 1. Initial steps in protein N-glycosylation. Pre-assembly of oligosaccharide-P-P-Do1 and en bloc transfer to the protein in the endoplasmic reticulum. Dol, dolichol; OST, oligosaccharyltransferase; P, phosphate; W , GlcNAc; 0, Man; A, Glc.

23.3 Trimming of the Polypeptide-Bound Oligosaccharide Once the preformed oligosaccharide is attached to the polypeptide it is extensively trimmed before it is extended with more sugars. Three Glc and four Man residues are removed from the polypeptide-bound oligosaccharide to yield the common intermediate MansGlcNAcz structure (Figure 2). The first Glc (a2-linked) is removed by ER glucosidase I that might already act while the glycosylated polypeptide chain is still being synthesized on the ribosome. The other two Glc residues (that are a3linked) are removed by ER glucosidase I1 (reviewed in [ 19, 231). The catalytic sites of both enzymes face the lumen of the ER in which the glycosylated polypeptide resides after its vectorial translocation over the ER membrane. The first (a2linked) Man may be removed in the ER by ER mannosidase to yield a unique MansGlcNAq structure after which the glycoprotein is translocated to the cisGolgi compartment (Figure 2) [19, 281. Alternatively, the first Man residue is removed after this translocation by Golgi-localized (u2-)mannosidase IA and IB. The specificity of these enzymes differs from that of the ER mannosidase, and they yield the two other possible MansGlcNAcz isomeric structures. Subsequent repeated actions of mannosidase IA and/or IB on either MansGlcNAcz structure results in the aforementioned single intermediate MansGlcNAcl structure [28]. In yet another

23.4 Folu'ing and Quality Control

RER

593

GOLGI CIS

MEDIAL

TRANS

Figure 2. Topography of reactions involved in N-glycosylation in the rough endoplasmic reticulum and the Golgi membranes. Note that the boxed structure remains untouched during the entire process. The pathway leading to a di-antennary disialylated oligosaccharide is shown. Dol, dolichol; OST, oligosaccharyltransferase; P, phosphate; , GlcNAc; 0, Man; A , Glc; 0 , Gal; V , Fuc; 4, NeuAc.

pathway the last Glc and the Man residue to which it is attached are removed as Glc(u1-3)Man in one step by an ER localized endo-mannosidase yielding a Mans GlcNAcz structure that is processed further in the cis-Golgi [29].

23.4 Folding and Quality Control Although the synthesis of the Glc3 Mans GlcNAcz oligosaccharide followed by its degradation to a MansGlcNAcz structure might seem a waste of energy, one important role of the oligosaccharide is that it aids the maturation of the polypeptide chain to a functional protein [30-321. After the removal of the first two Glc residues by ER glucosidase I and I1 the glycosylated polypeptide is released from the ribosome. In the lumen of the ER it then associates with membrane bound calnexin or soluble calreticulin, chaperones which recognize the monoglucosylated oligosaccharide in a lectin like fashion. While transiently bound the polypeptide is as-

594

23 On the Origin of Oligosaccharide Species-Glycosyltransferases in Action

sisted by these chaperones to fold properly and, for oligomeric glycoproteins, to assemble in oligomers. At this stage also disulfide bridges are formed. After subsequent removal of the third Glc from the oligosaccharide by glucosidase I1 the glycoprotein is no longer bound and is released, and the oligosaccharide is further processed on the glycoprotein’s journey via the Golgi to its cellular destination. In case the process of correct folding or assembly fails, however, the deglucosylated polypeptide is retained in the ER where it can be re-glucosylated by a glycoprotein glucosyltransferase specific for improperly folded glycoproteins [ 33, 341. Re-glucosylated proteins can be bound again by calnexin or calreticulin and so have another chance of folding correctly. The glycoprotein glucosyltransferase (together with the chaperones) thus functions in the control mechanism by which only properly folded and assembled glycoproteins can leave the ER.

23.5 Committed Steps in the Formation of Complex-Type Oligosaccharide Chains and Branching After the trimming reactions in the ER and the cis-Golgi the glycoprotein containing the resulting MansGlcNAcz oligosaccharide is translocated to the medial-Golgi compartment (Figures 2 and 3 ) . Here the first committed step in the formation of

Window of action

GlcNAcT 111 GlcNAcT IV & V

tri- and tetra-antennary

Figure 3. Committed steps in the synthesis of complex-type N-linked oligosaccharides. Action of GlcNAcT I on the oligomannoside structure opens the window for several possible reactions as catalyzed by a-mannosidase I1 (Man-ase II), GlcNAcT 11-V, and core-specific ah-FucT. The window is closed by the action of p - Gal T as shown for a di-antennary structure. For key to the symbols see legend to Figure 2.

23.5 Committed Steps in the Formation of Complex-Type Oligosaccharide Chains V

GlcNAcBl

II

GlcNAc~l~SManal

595

I6 6' 111

G l c N A c ~ l 4 M a n ~ l - r 4 G l c N A c ~ l - + 4 G l c N A-Asn c~1

I

GlcNAc~l-.PManulr3

IV

GlcNAcPl

r4

Figure 4. Structural representation of the linkages formed by GlcNAcT ILV. The Roman number refers to the GlcNAcT involved.

complex-type oligosaccharides takes place by the addition of a peripheral GlcNAc residue in ((31-2) linkage to the Man(a1-3)Manp site as catalyzed by GlcNAcT I to form the first branch [13, 17, 201. This GlcNAc functions as a green traffic light for several possible reactions catalyzed by a-mannosidase 11, GlcNAcT 11, 111, IV and V (Figure 4), and N-linked core-specific a6-FucT that catalyzes reactions leading to di-, tri- and tetra-antennary oligosaccharides, hybrid and bisected structures, and core-fucosylated structures [13, 171 (Figure 3). It is at this stage of oligosaccharide processing that there is an extensive regulation at the substrate level. Divergence in the pathway occurs here in particular and structures with varying number of branches (a branch or antenna is build up from a GlcNAc residue as introduced by GlcNAcT I, 11, IV or V) might be formed [ 13,211. Evidently GlcNAcT 11, yielding a di-antennary oligosaccharide, can only act after a-mannosidase I1 has trimmed two additional Man residues from the Man(a16)ManP branch. When GlcNAcT 111 acts before a-mannosidase I1 the oligosaccharide is frozen into a hybrid structure (combining the features of a complex-type structure at the Man(al-3)Manp branch and an oligomannosidic-type structure at the Man(a16)Manp branch) because a-mannosidase I1 is blocked by the bisecting GlcNAc residue attached by GlcNAcT I11 action. When GlcNAcT I11 acts after GlcNAcT I1 a so-called bisected di-antennary oligosaccharide is formed. Action of GlcNAcT IV and V leads to the formation of tri- and tetra-antennary oligosaccharides and these might also become bisected by the subsequent action of GlcNAcT 111. The other route, in which the order of action of these GlcNAcTs is reversed, is impossible because GlcNAcT 11, IV and V do not act on structures containing a bisected GlcNAc residue. A bisecting GlcNAc also inhibits core 1x6-fucosylation. It should be mentioned here that in plants and insects a3-fucosylation of the core GlcNAc might occur as catalyzed by a core-specific a3-FucT [35, 36, 1881. In addition, in plants an 1x2-xylosyltransferase has been described that can introduce an aXyl residue to OH-2 of the (3-Man of the core [37].All these reactions can no longer occur after in a second committed step of complex-type oligosaccharide formation, by P4-galactosyltransferase action, a Gal residue is attached in (81-4) linkage to the GlcNAc((31-2)Man(al-3)ManP branch (to which the first Gal preferentially is in-

596

23 On the Origin of Oligosaccharide Species-Glycosyltransferases in Action

troduced [38, 391) to yield a Gal@-4)GlcNAc (lacNAc) terminus [13, 17, 211. This Gal residue thus functions as a red light and closes the stage of oligosaccharide branching and modification of the core (Figure 3). Occasionally a GalNAc rather to yield a terthan a Gal is introduced by a 1)4-N-acetylgalactosaminyltransferase minal GalNAc(pl-4)GlcNAc ( N ,N’-diacetyllactosediamine, 1acdiNAc) unit [40] (see also Section 23.15-The lacdiNAc Pathway of Complex-Type Oligosaccharide Synthesis).

23.6 Topology of the Reaction Catalyzed by a Typical GlcNAcT All eukaryotic Golgi glycosyltransferases are typically membrane-bound enzymes with a type I1 membrane topology (cytosolic NHz-terminus) as has become apparent from analysis of the deduced amino acid sequences of cloned enzymes [41, 421. GlcNAcT I, 11, 111, IV and V form no exceptions in this respect [43-491. Most glycosyltransferase polypeptides consist of ca 325-425 amino acids and have a very similar domain structure-a short cytoplasmic tail ( ~ 5 - 3 0amino acids) attached to a transmembrane domain (x15-20 amino acids), which functions as a signalanchor sequence by which the enzyme is targeted to and anchored in the Golgi membrane. This is followed by a so-called stem region ( ~ 3 5 - 4 5amino acids) and a large catalytic domain (approximately 300 amino acids, containing the COOH terminus) extending into the lumen of the Golgi (Figure 5). The immature protein-linked oligosaccharide chains are also exposed into the Golgi lumen during the glycoprotein’s traverse through the cellular endo membranes. UDP-GlcNAc is thus utilized inside the Golgi lumen. The synthesis of this substrate by the enzyme UDP-GlcNAc pyrophosphorylase, however, occurs in the cytosol and, as nucleotide-sugars cannot freely diffuse through the Golgi membrane, it has to be transported across this barrier. This transport is facilitated by a specific UDP-GlcNAc transporter by a mechanism in which UDP-GlcNAc is exchanged for UMP that is formed from UDP, the other reaction product of the GlcNAc-transfer reaction [25] (Figure 5). Similar specific transporters for CMPNeuAc, UDP-Gal, GDP-Man, and UDP-GalNAc have also been demonstrated and cloned [25, 501, and aberrant glycosylation patterns are, indeed, observed for cells with defects in transporter activity [25]. As yet, however, it is unclear if and to what extent the import of nucleotide-sugar donor-substrates might be limiting and can thus contribute to the regulation of protein glycosylation.

23.7 Elongation and Termination Reactions in the trans-Golgi By the time the first Gal residue is introduced to the oligosaccharide chain, the glycoprotein has reached the trans-Golgi in which the elongation and termination

23.7 Elongation and Termination Reactions in the trans-Golgi

591

Glycoprotein

-

UTP + m ~ l c ~ ~ c ~ 1 - p UDP--/+ Pyrophosphorylase

PPi

Figure 5. Organization of a typical glycosyltransferase in the Golgi membrane and topology of the reaction catalyzed as exemplified for a GlcNAcT. Note that the nucleotide sugar donor substrate is synthesized in the cytosol. Before it can be utilized by the glycosyltransferase it has to be translocated in exchange for a nucleotide monophosphate into the lumen of the Golgi by a specific transporter. Sugar transfer takes place to the oligosaccharide chain of a glycoprotein during its traverse through the secretory system. TMD, trans-membrane domain.

reactions occur (Figures 2 and 6). As mentioned above, P4-galactosylation disables the action of a-mannosidase 11, GlcNAcT I-V, and N-linked core-specific a6-FucT. In contrast, it is the green traffic light for many glycosyltransferases which all act on an accepting lacNAc unit. The action of these enzymes yields the mature oligosaccharide structures. Here additional diversity (leading to microheterogeneity and different glycoforms of a glycoprotein; see Section 23.1, Introduction) is introduced because these enzymes potentially compete for common Gal@1-4)GlcNAc acceptor sites on the protein-linked oligosaccharides. Six of the most common reactions, as catalyzed by a6-NeuAcT, a3-NeuAcT, a3-GalT, a3-FucT, a2-FucT, and P3-GlcNAcT, respectively, are shown in Figure 6 for a di-antennary glycan as acceptor. a6-NeuAcT, a3-NeuAcT, a3-GalT, and a2-FucT directly cap the Gal of the lacNAc units by introduction of a terminal sugar, whereas a3-FucT adds a fucose to the GlcNAc of these units. Introduction of a P3-GlcNAc residue to a lacNAc unit by P3-GlcNAcT enables a second P4-galactosylation reaction (Figure 6). In this concerted action of P3-GlcNAcT and b4-GalT polylactosaminoglycan extensions are formed which also may be sialylated, fucosylated, or a-galactosylated. These reactions can in principle take place with any terminally exposed lacNAc unit irrespective of whether it is presented on di-, tri- or tetra-antennary, hybrid, or bisected N-linked oligosaccharides or on 0-linked oligosaccharides (see later).

598

23 On the Origin of Oligosaccharide Species-Glycosyltrnnsferases in Action

Window of action

ta

NeuAca2->6 O+

1

Figure 6. Elongation and termination reactions as enabled by the action of p4-GalT. P4-Galactosylation blocks the action of the branching GlcNAcTs, Man-ase I1 and core a6-FucT (c.J Figures 2 and 3). By way of contrast, it opens the window of action for the elongating (polylactosaminoglycan-forming) and terminating enzymes shown. The reactions cannot only occur with the di-antennary substrate shown, but also with tri- and tetra-antennary, bisected, and hybrid structures.

Because of the substrate specificities of each of these enzymes, however, not all of the theoretically possible product structures are formed in nature in detectable amounts. By way of contrast, such structures may be formed in vitro where very active (recombinant) enzyme preparations can be used. In the next paragraphs the specificities of the enzymes are further described.

23.8 Activity with Branched Substrates Not all N-linked oligosaccharides with terminal lacNAc units are acted upon by a6NeuAcT, a3-NeuAcT, a3-GalT, a3-FucT, and p3-GlcNAcT at the same rate. Substrate specificity studies have shown that a6-NeuAcT (bovine colostrum; corresponds to ST6Gal I described below in Section 23.18) much prefers di-antennary substrates and is less active with tri- (having an additional Gal@-4)GlcNAc p4linked at the Man(a1-3)ManD branch as initiated by GlcNAcT IV action), tri'(having an additional Gal@-4)GlcNAc p6-linked at the Man(a1-6)Manp branch as initiated by GlcNAcT V action), tetra- (having both additional G a l @ -

23.8 Activity with Branched Substrates

599

300 h

E

200

h .tl

150

> .c

0 (D

$ .-

100

c (D

-a 50 0 06-NeuAcT 03-NeuAcT 03-GalT

03-FucT B3-GlcNAcT

Figure 7. Relative activities of u6-NeuAcT, a3-NeuAcT, a3-GalT, a3-FucT, and P3-GlcNAcT with structures that form part of branched, N-linked oligosaccharides. The acceptor oligosaccharides are represented by symbolic structures the branch linkages of which correspond to those in the structures shown in Figures 4 and 8.

4)GlcNAc units), and bisected di-(GlcNAcT 111 action) antennary oligosaccharides [51] (Figure 7). By way of contrast, a3-NeuAcT (human placenta; high preference for type-2 chain acceptors; corresponds to ST3Gal IV) prefers the tri-antennary substrate whereas a bisected acceptor is a poorer substrate [52]. Very similar specificity is shown by 1x3-FucT VI from human milk (D. H. Van den Eijnden and W. E. C. M. Schiphorst, unpublished results). a3-GalT (calf thymus) is relatively indifferent to the branching of the oligosaccharide substrates, but also acts at a lower rate on a bisected di-antenna [53]. Finally P3-GlcNAcT (Novikoff tumor cell ascites fluid) has the most unusual behavior of these enzymes, because it strongly prefers structures with a Gal(P1-4)GlcNAc(Pl-6)Man sequence as occurs in tri'- and tetra-antennary oligosaccharides [ 541. Consequently, changes in the activities of GlcNAcT 111, IV and V, yielding a differently branched substrate population for a6-NeuAcT, a3-NeuAcT, a3-GalT, a3-FucT, and p3-GlcNAcT, by themselves will result in quantitative changes in the terminal structures on N-linked oligosaccharides. Increased GlcNAcT I11 activity will result in suppression of the elongation and termination reactions. Elevated activity of GlcNAcT IV will result in a shift from a6-sialylation to a3-sialylation [ 551 and a3-fucosylation (and consequently in increased expression of Lewis' and sialyl-Lewis", see later), whereas increased GlcNAcT V activity will promote polylactosaminoglycan formation and will further suppress a6-sialylation. Elevated activities of the branching GlcNAcTs (111, IV, and V) have been reported [56-601 or have been concluded [61] to occur in cases of malignant transformation. Such changes thus form one cause of the altered terminal

600

23 On the Origin of Oligosuccharide Species- Glycosyltrunsferases in Action

glycosylation of N-linked glycans which is often observed in malignant cells. These effects might be further amplified by changes in the activities of the glycosyltransferases acting on the lacNAc termini [ 581.

23.9 Branch Specificity The different IacNAc termini on branched N-linked oligosaccharides are not acted upon at the same rates by all enzymes. The preference of an enzyme for a lacNAc unit on one particular branch has been termed ‘branch specificity’. A summary of the branch specificities of four common glycosyltransferases acting on Gal(814)GlcNAcp termini is shown in Figure 8. In particular 1x6-NeuAcT shows a strong preference for the Gal( 1-4)GlcNAc( p 12)Man(al-3)Manp branch in di-, tri-, and tetra-antennary oligosaccharides [51, 62, 631. The Gal(pl-4)GlcNAc(pl-2)Man(al-6)Manp branch is a6-sialylated at least 10 times more slowly, whereas the Gal(~1-4)GlcNAc(~l-6)Man(ul-6)Man~ branch as occurs in tri’- and tetra-antennary oligosaccharides is very resistant to a6sialylation. Branch specificity, however, is only displayed by this sialyltransferase when the N-linked substrate contains at least one core GlcNAc residue and it has been proposed that the a6-NeuAcT recognizes this residue in a lectin like fashion aiding the correct positioning of the substrate on the enzyme [62]. By way of conbranch 4-5 trast u3-GalT prefers the Gal(~1-4)GlcNAc(~l-2)Man(al-6)Man~ A. a3-GalT

-D

a6-NeuAcT

-+

Ga Ipl-.4G I c N A c ~ l - D 2 M a n a l bvlanpl4GlcNAc~1-DR

r3 Gal~l-D4GlcNAc~l-r2Manal

B. 83-GlcNAcT

Gal/31-.4GlcNAcpl

-D

a3-GalT/p3-GlcNAcT

+

‘6 Gal~l-~4GlcNAc/31-~2Manal, \r

.36Man,814GlcNAc~l-4 a6-NeuAcT

-D

Gal~1-+4GlcNAc~l-D2Manal

u3-Neu AcT

-s

GaI/31-.4GlcNAc/31

r4

Figure 8. Branch specificities of Gal@-4)GlcNAc-specific enzymes with (A) an N-linked diantennary substrate and (B) a tetra-antennary substrate. The branches to which the enzymes preferentially introduce a sugar residue are indicated by 4. Note that a3-NeuAcT and P3-GlcNAcT do not show branch specificity with the di-antennary structure.

23.10 Essential Requirements for Activity with lacNAc

601

branch in a ditimes more than the Gal(~l-4)GlcNAc(~l-2)Man(al-3)Man~ antennary substrate [64]. The Gal(~1-4)GlcNAc(pl-2)Man(al-6)Man~branch seems to be even more preferred by a3-GalT when it is present in tri- and tetraantennary oligosaccharides [65]. Branch specificity is different yet again for a3NeuAcT (human placenta). Although it shows no preference for either branch in a di-antennary substrate [52] it prefers the Gal(P1-4)GlcNAc(Pl-4)Man(al-3)Man~ branch in tri- and tetra-antennary substrates although this sialyltransferase can also act on all other branches (M. Nemansky, C. H. Hokke and D. H. Van den Eijnden, unpublished results). The typical a3-/a6-sialylation patterns as occur on N-linked glycans [51] therefore seem to be the result of the differential branch specificities of both sialyltransferases. Like a3-NeuAcT, the P3-GlcNAcT does not show branch preference with di-antennary oligosaccharides [ 641. This enzyme, however, highly prefers the Gal( P 1-4)GlcNAc( pl-2)Man(wl-6)Manp and Gal(P1-4)GlcNAc( Pl 6)Man(al-6)Manp branches equally well to the other branches of tri’- and tetraantennary oligosaccharide structures [ 541.Because of the strongly increased activity of P3-GlcNAcT on such substrates (Figure 7) polylactosaminoglycan chain formation is highly favored on such oligosaccharides at the branches on the Man(a16)ManP unit. This further emphasizes the important role of GlcNAcT V (which initiates the formation of the Gal(P1-4)GlcNAc(~l-6)Man(al-6)Man~branch) in this process. -

23.10 Essential Requirements for Activity with LacNAc It will be clear from the previous paragraphs that the activity of the glycosyltransferases with each of the lacNAc units on branched N-linked oligosaccharide substrates can be strongly influenced by the structural environment (number and nature of neighboring branches, branch location) of the oligosaccharide substrate. In addition, studies with deoxygenated, substituted and otherwise modified IacNAc derivatives as acceptors for a6-NeuAcT, w3-NeuAcT (rat liver; preference for type 1 acceptors but also acting on type 2 structures; corresponds to ST3Gal I11 described below in Section 23.18, Siulyltransferase Family), a3-FucT and a3-GalT have revealed which modifications of the lacNAc acceptor site are allowed and which abolish the acceptor property for each of these enzymes [ 5 , 39, 66-72]. Obviously the OH group to which a sugar is to be attached by the glycosyltransferase is essential and should be unsubstituted. The further requirements for activity explain why some enzymes can still act after another glycosyltransferase has introduced a sugar to the lacNAc unit (e.g. a3-fucosylation is still possible after a3-sialylation), whereas other enzymes are mutually exclusive (e.g. 1x6-sialylation precludes a3fucosylation vice versa). On the other hand not all product structures of reactions that are possible in vitro are found in nature (e.g. no a6-sialylated form of the blood group H type 2 oligosaccharide has been reported even though a6-NeuAcT can act on Fuc(al-2)Gal(Pl-4)GlcNAc). A summary of the requirements for activity of these enzymes with lacNAc is given in Figure 9.

602

23 On the Origin of Oligosaccharide Species-Glycosyltvansferases in Action

no aGal

no Fuc

NHAc

fmo* m -

OH

OH

OH no Neu,Ac fFuc no atial or NHAc r10PGlcNAc

-

OH

OH

f Fuc

no Fuc no NHAc

a6-sialyltransferase

f Fuc

a3-sial yltransferase

Hoeo no SO; no Fuc

HO

fNeuAc

aGal

.O=OH

OH +Fuc

or N H A ~

NHAc

f

so;

f PGlcNAc

a3-fucosyltransferase

OH

f no Fuc no NHAc

a3-galactosyltransferase

Figure 9. Summary of the structural requirements of a6-NeuAcT, a3-NeuAcT, a3-FucT and a3GalT for activity with lacNAc and its derivatives. Essentially required OH groups are indicated by an arrow (site of sugar attachment) and by a circle. The latter OH groups are required for enzyme binding and catalysis (key polar groups). OH groups that can be deoxygenated without complete loss of activity but cannot be substituted by another sugar are indicated by a square. Other OH groups may be modified (deoxygenated, substituted, and/or replaced by an N-acetyl group) with only limited effect on the activity of the enzyme. All enzymes accept the replacement of the NHAc group of the GlcNAc residue by an OH moiety but are (much) less active with the resulting compound (lactose).

In particular a3-FucT seems to be the enzyme that is most easily satisfied, which explains why in metabolic routes a3-fucosylation is generally a late step. It clearly seems that each of the enzymes recognizes a lacNAc acceptor site in a unique way, suggesting that these enzymes do not share a common lacNAc recognition sequence motif. Indeed it has not yet been possible to identify such a motif in the deduced amino acid sequences of cloned lacNAc-specific enzymes.

23.11 Further Terminal Reactions in Complex-Type Oligosaccharide Synthesis In addition to the glycosyltransferases mentioned in the previous paragraphs several other enzymes are known that can act on lacNAc units but catalyze less common reactions. Among these are a p3-GlcAT [73], a 3’-sulfotransferase (to Gal) [74], and

23.12 Specijic ModGcations of Polylactosaminoglycans

603

a 6-sulfotransferase (to GlcNAc) [75], a P6-GlcNAcT [76], and a P3-GalNAcT [77] (Table 1). Furthermore, an a4-GalT acting on the lacNAc terminus of lacto-N-neotetraosylceramide [ 781 has been reported. Very recently a novel a4-GlcNAcT was cloned that preferentially acts on lacNAc present on O-linked core 2 [79]. After a6sialylation or a3-fucosylation of lacNAc generally no further modifications are possible. The only exception known so far is the action of schistosomal a2-fucosyltransferase on the Lewis’ structure to form Gal(P1-4)[Fuc(al-2)Fuc(al-3)]GlcNAcP-R (difucosyllacNAc) [ 801. Action of one of the other enzymes, however, often enables additional reactions to occur. This is particularly true of a2-fucosylation or a3-sialylation (Table I). The first reaction yields the H-type 2 antigen that not only can be converted into the LewisY determinant by the action of a3-FucT7 but also into the blood group A or B antigenic structures by the action of the Aenzyme (1x3-GalNAcT)and B-enzyme (a3-GalT), respectively [24, 8 11. a3-Sialylation in turn yields the structure that is intermediate in the synthesis of the sialylLewis’ determinant by a3-FucT action, or polysialic acid through the action of one of the two known polysialyltransferases [82], or the Sda antigen by the action of a P4-GalNAcT [83] (Table 1). For reasons of completeness the reactions catalyzed by three different sulfotransferases are also included in the figure. On the basis of the sequences of the two sulfotransferases of this group that have been cloned so far it seems that these enzymes have a domain structure that strongly resembles that of a Golgi glycosyltransferase [75, 841 (c.$ Figure 5).

23.12 Specific Modifications of Polylactosaminoglycans

The terminal lacNAc unit on polylactosaminoglycan chains can be capped like any other terminal lacNAc unit by the action of the glycosyltransferases described in the previous paragraphs, albeit that some of these enzymes act only at low rates on these glycans (e.g. a6-NeuAcT [55]).In addition to these reactions, however, several specific reactions can occur in which sugars are introduced to internal Gal or GlcNAc residues. Linear polylactosaminoglycan chains as synthesized by the concerted action of P3-GlcNAcT and B4-GalT (Figure 6) show blood group i-activity [24, 811. They may be converted to branched, blood group I-active structures by the action of a P6-GlcNAcT and subsequent B4-galactosylation yielding a Gal(P14)GlcNAc(al-3)[Gal( B1-4)GlcNAc( Pl-6)IGal branching point [85, 861. Two different type of branching P6-GlcNAcTs have been described, one of which acts only on penultimate Gal residues capped by a single B3-linked sugar (GlcNAc or Gal) (‘predistally acting’) [87, 881, the other acting on any Gal residue along the linear chain except the terminal or penultimate (‘centrally acting’) [89, 901. In a recent study, however, a recombinant B6-GlcNAcT was described which has both these activities [91]. Earlier a P6-GlcNAcT acting on terminal Gal residues (‘terminally acting’) has been reported [76]. A P3-GlcNAcT acting on resulting

Sda Lewis'

a-galactosyl-Lewisx

ti-type 2

HNK-1

Gal/31-.4[Fucal -ZFucul-.31GlcNAc/3-R

Galul-.3Ga1j1-.4[Fucu1-b3]GlcNAcj-R

Fucal-2Gal/31-4[Fucal-.31GlcNAc/3-R Fucul-.2 [GalNAcal -31Gala1-4G IcNAcB-R Fucal-.2[Galal-b3]Gal/3l-b4GlcNAc/3-R S04--3GlcA/31-3Gal/31+4G Ic NAcp-R S04--3Gal/31-.4IFucal-.31GlcNAc~-R (NeuAco2-3) ~Gal~1-.4[Fucal-.31[SO~-6]GlcNAc/3-R

? ? ?

7

Gal/314[Fucal-.3lGlcNAc/3-R

Galo1-.3Gal/31+4G IcN Acp-R

Fucul-2Gal/31-.4GlcNAc/3-R

GlcA/31-.BGal/31-.4GlcNAcf?-R

SO~-3Gal/31-.4GlcNAc/3-R

Gal/314[SO~-GIGlcNAc/3-R

GlcNAcfll -.GGal/31-4GlcNAc/3-R

GalNAc/31+3Ga1/31 -4GlcNAcB-R

Gala1-+4Gal/3l-b4GlcNAc/3-R

GIcNAcal--4Gal/31+4GlcNAc/3-R

LewisY blood group B

blood group A

PI

6-sulfo-(sialyl-)LewisX

3'-sulfo-Lewis'

-b

difucosyllacNAc

polysialic acid

NeuAca2-.3[GalNAc/31-.41Ga1/314GlcNAc/3-R

+

sialyl-Lewis'

NeuAca2-.3Gal/31-b4GlcNAc/3-R

(NeuAca2-.8),NeuAca2-3Gal/31-.4GlcNAc~-R

no further reaction

NeuAca2+6Gal/3l-b4GlcNAc/3-R

Name of (antigenic) structure

NeuAca2~3Gal/31-.4[Fucal-.31GlcNAc/3-R

Final structure

Primary product structure

Table 1. Modifications of lacNAc as shown by primary product structures and further terminal reactions in complex-type oligosaccharide synthesis. After the introduction of a sugar to lacNAc further additions are often, but not always, possible.

23.12 Specific Modijcations of Polylactosaminoylycans

605

-

G-GN-G-GN-G-GN LAG

a3-FT

G-GN-G-GNG-GN-

LeX

a3-ST

a3-FT

SA-GGN-G-GN-G-GN-

\

G-GN-G-GN-G-GNF F

, /

J’-sialyl-LAG,

/

i n t e T l dimeric-LeX a3-ST I

-

SAG-GN-G-G N-G-GN

k 7 F 4

-

G-GN-G-GN-G-GN F F F

sialyl-LeX

F

L

.$

-

SA-G-G N-G -GN-G-GN F F

frimeric-LeX

VIM-2/internal dimeric-LeX

/

\ a3-FT

4

\r SA-G-GN-G-GN-G-GN F F F

-

sialyl-trimeric-Le“

Figure 10. Cooperative action of a3-NeuAcT (a3-ST, dashed arrow), a3-FucT IV (myeloid enzyme; u3-FT, thin solid arrow), and a3-FucT VII (w3-FT, leukocyte enzyme; bold solid arrow) in the synthesis of sialyl-oligomericLewisXand related structures on polylactosaminoglycan chains. F, Fuc; G, Gal; GN, GlcNAc; LAG, polylactosaminoglycan; SA, NeuAc.

GlcNAc(p1-6)Galpl-R to yield a GlcNAc(al-3)[GlcNAc( p1-6)]GalpI-R branching point, however, has not yet been described. (See also Section 23.23, p6-Nacetylylucosaminyltransferase family). Another specific modification of polylactosaminoglycan chains is the introduction of a3-Fuc residues to non-terminal GlcNAcs along the chain to yield a (sialyl-) oligomeric-Lewis” structure. In leukocytes the formation of this structure, which is a ligand for E- and P-selectin, is catalyzed by two different a3-FucTs, the myeloid FucT (FucT IV) and the leukocyte FucT (FucT VII) [92, 931 (Figure 10) (see also Section 23.20, a3/4-Fucosyltransferase Family below). p-GlcNAc cannot be introduced to position C-6 of Gal residues that form part of the Lewis’ structure [90]. In turn, as the OH at C-6 of Gal in lacNAc is a key polar group for a3-FucTs [ 691, a3-fucosylation of GlcNAc residues underlying blood group 1 branching points is impossible. It thus seems that the formation of oligomeric LewisX and of blood group I branching points are competing processes. Branched polylactosaminoglycans can, therefore, only carry LewisXgroups at lacNAc units that are not engaged in a branch [94]. Similar to the P6-GlcNAcTs, different a3-FucTs show differential preferences for the penultimate and internal acceptor sites along the polylactosaminoglycan acceptor chain (“site specificity”). This specificity is further detailed under a3/4-Fucosyltransferase family described below.

606

23 On the Origin of Oligosaccharide Species-Glycosyltransferases in Action

23.13 The Invariable Core of N-linked Oligosaccharide Chains, and Site- and Protein-Specific Processing

Mature N-linked oligosaccharides all have the same basic core consisting of three Man and two GlcNAc residues. This core structure is already present in the preformed dolicholLPP-linked oligosaccharide precursor and remains untouched throughout the entire processing after its en bloc transfer to the polypeptide chain (Figure 2). Nevertheless, the oligosaccharide chains on different glycoproteins synthesized in one particular tissue or cell type are often structurally different. Apparently, the protein-linked chains can be processed to different extents. When, for some reason, GlcNAcT I fails to act or when not all a2-linked Man residues are removed by a-mannosidase IA or IB, the oligosaccharide will remain as an oligomannosidic-type structure. Such a chain can occur at one Asn of a polypeptide even though the oligosaccharides at other Asn residues of the same glycoprotein have been processed to complex-type structures. Many examples of glycoproteins with this characteristic are known (e.g. tissue plasminogen activator [95]; glycodelin [96]). A differential accessibility of the various oligosaccharide chains to the processing enzymes as dictated by the polypeptide environment of the individual chains has been proposed to account for this feature [97]. However, the variations in structure at a single glycosylation site (microheterogeneity) rather seem to be determined stochastically. In other instances the N-linked oligosaccharides at different Asn residues differ in their amount of branching (e.g. human a1-acid glycoprotein [98]) or in the presence or absence of a bisecting GlcNAc residue (human myeloma IgGl [99]). A hypothetical explanation of the latter differences might be that the polypeptide can interact with the oligosaccharide in a way that it is locked in a conformational state that is not acted upon by certain GlcNAcTs (c.J: [991). Finally, two classes of glycoprotein have been described the N-linked chains of which have a specific character because of the recognition of a peptide motif by a sugar-transferring enzyme. The first class is formed by the lysosomal enzymes. When aZlinked Man residues are still present in the oligosaccharide the structure is phosphorylated by the sequential action of an N-acetylglucosamine phosphotransferase and a phosphodiesterase resulting in the formation of Man 6-P determinants [loo] (Figure 2). This structural element mediates the targeting of the lysosomal enzyme to the lysosome through its interaction with one of the two Man 6-P receptors [ 1001. Only lysosomal enzymes become phosphorylated because the N-acetylglucosamine phosphotransferase specifically recognizes a peptide sequence that is uniquely present in these enzymes. The second class comprises the pituitary glycohormones which are rich in complex-type glycans based on GalNAc(014)GlcNAc (lacdiNAc) rather than Gal(fl1-4)GlcNAc (lacNAc). It has been proposed that the pathway is directed toward the synthesis of such oligosaccharide chains by a hormone-specific N-acetylgalactosaminyltransferase that recognizes a peptide determinant present in the polypeptide of these hormones [ 1011.

23.15 The IacdiNAc Pathway of Complex-Type Oligosaccharide Synthesis

601

23.14 Synthesis of Type 1 (Gal(p1-3)GlcNAcp-R) versus Type 2 (Gal(p1-4)GlcNAcP-R) Chains Most complex-type chains are based on lacNAc and are referred to as type 2 chains. In some instances, however, protein-linked chains are based on lacto-N-biose (Gal(pl-3)GlcNAc) and are referred to as type-1 chains. Notable examples of glycoproteins carrying such chains are bovine fetuin [ 1021 and rat a1-acid glycoprotein [ 1031. Although glycoproteins with this type of chain are rather rare, a major fraction of human milk oligosaccharides is based on the type-1 chain compound lacto[104]. For a long time the only N-tetraose (Gal(~1-3)GlcNAc(a1-3)Gal(~1-4)Glc) p3-GalT involved in the synthesis of type-1 chains was that described in porcine trachea [105]; this activity was not reported in tissues such as mammary gland. Recently the cloning of four human GlcNAcp-R-specific p3-GalTs [ 106-1091 and the mouse orthologs of three of these [110] has been reported. Indeed these 03GalTs seem to have a rather restricted tissue expression. Like lacNAc termini, type-1 structures may be fucosylated by a2-FucT and a4FucT (Lewis enzyme, FucT 111) to yield the H-type-l and the Lewisa structures, respectively [24, 811. In turn the H-type-1 structure can be converted into a Lewisb, or a blood group A or B active compound, by the action of a4-FucT, and blood group A- (a3-GalNAcT) or B-enzyme (a3-GalT), respectively [24, 8 I]. Alternatively, type-1 structures can be disialylated by the concerted action of a a3-NeuAcT and a a6-NeuAcT to yield the terminal NeuAc(a2-3)Gal(pl-3)[ NeuAc(a2-6)]GlcNAc structure [ 111, 1 121.

23.15 The LacdiNAc Pathway of Complex-Type Oligosaccharide Synthesis As another alternative to type-2 chain complex-type oligosaccharide formation a GalNAc can be introduced to accepting GlcNAc residues yielding a GalNAc(p14)GlcNAc (lacdiNAc) unit [40]. In particular bovine milk glycoproteins carry oligosaccharide chains which are rich in such units, but they also occur on many other glycoproteins including several of human origin (reviewed in Ref. [40]). As mentioned above it has been proposed that a glycoprotein hormone-specific 84GalNAcT exists which specifically recognizes a peptide determinant present in the polypeptide chain of these glycoproteins [ 1011. In schistosomes [ 113, 1141, snail [115], and bovine mammary gland [116], however, a P4-GalNAcT has been described which is hormone-non-specific and strongly resembles P4-GalT in acceptor properties. It has been proposed that this 04-GalNAcT controls the lacdiNAc pathway of complex-type oligosaccharide synthesis [40]. Several enzymes that act on lacNAc can also act on lacdiNAc, in conformity with their essential requirements for activity described above. a6-NeuAcT and a3-

608

23 On the Origin of Oligosaccharide Species- Glycosyltransferases in Action

FucT can catalyze the formation of NeuAc(a2-6)GalNAc( 01-4)GlcNAc [66] and GalNAc(~l-4)[Fuc(al-3)]GlcNAc (lacdiNAc variant of Lewis") [67], respectively. By way of contrast a3-NeuAcT and a3-GalT do not act on GalNAc(P1-4)GlcNAc [ 1171 and P3-GlcNAcT only acts with very low activity [40]. Furthermore a sulfotransferase catalyzing the transfer of sulfate to OH-4 of GalNAc has been identified [118], as has a p3-GalT catalyzing the formation of Gal~l-GalNAc(~l-4)G1cNAc [119].

23.16 Protein 0-Glycosylation In contrast with N-glycosylation, formation of 0-linked chains as occurs on mucins is initiated by the addition of a single a-GalNAc residue (rather than of a preformed oligosaccharide) to the OH-group of a Ser or Thr in the polypeptide chain. It is believed that this step occurs in the cis-Golgi [22].Additions to this GalNAc residue of Gal in (P1-3)-, GlcNAc in (pl-3)- and (p1-6)-, and GalNAc in (al-3)-linkage might subsequently occur in the medial-Golgi which leads to different core structures (Figure 11). These reactions are all catalyzed by glycosyltransferases specific for 0-linked chains. Also, direct addition of NeuAc to GalNAc and to the core 1 disaccharide (Gal@-3)GalNAc) can occur as catalyzed by NeuAcTs that are highly specific for these 0-linked sugars (Figure 11, right). In the trans-Golgi terminal GlcNAc residues on the different core structures can be further acted upon by P4-GalT yielding 0-linked complex-type chains. Polylactosaminoglycans can also be formed on these chains by the concerted action of P3-GlcNAcT and P4-GalT. The chains can be completed by addition of NeuAc, Fuc and sulfate residues as catalyzed by the same sialyltransferases, fucosyltransferases, and sulfotransferases that act in the synthesis of N-linked chains. Although not all glycosyltransferases acting on lacNAc in Nlinked chains (see Figure 6) act with the same efficiency on lacNAc in O-linked chains (e.g. a6-NeuAcT does not act well on Gal(~1-4)GlcNAc(~1-6)[Gal(~l-3)]GalNAca-0-R) elongation and termination of the complex-type 0-linked structures follows in principle the same pathways as for N-linked chains and are catalyzed by the same enzymes that act in the terminal stages of N-glycosylation. Thus on 0-linked chains blood group determinants such as (sialyl-)Lewis",H(O), A, B, Sda, i, I, and type-1 chain-based H, Lewisa and Lewisb can be formed in an analogous way [22].

23.17 Glycosyltransferase Families Before the days of molecular cloning of glycosyltransferases the classification of these enzymes was based on common nucleotide-sugar donor substrate usage with sub-classification based on the linkage formed and the acceptor sugar used.

23.17 Glycosyltransferase Families

609

GOLGI MEDIAL GalNAca-3GalNAca-R core 5

GlcNAc p-3GalNAca-R core 3

CIS

TRANS

R (= Ser/Thr-polypeptide)

Gal p-3GalNAca-RF core 1

Galp-3GalNAca-R

GlcNAcp-3GalNAca-R core 4

NeuAc a-3Galp-3GalNAca-R

GlcNAcp 6 GlcNAcp-3Galp-3GalNAca-R

6 GlcNAcp-3GalP-3GalNAca-R

GlcNAcp 6 NeuAc a-3Galp-3GalNAca-R

Figure 11. Major pathways in 0-linked o~igosdcchdridesynthesis. The reactions leading to the sialylated and the different core structures are depicted. Once a GlcNAc residue is attached it can be acted upon by P4-GalT to form a lacNAc unit, which can be further modified as outlined in Figures 6 and 10, and Table 1.

Thus for a long time families of sialyltransferases, fucosyltransferases, galactosyltransferases, etc. have been distinguished. During the past 14 years the cloning of the cDNAs of over 50 glycosyltransferases in eukaryotes has been achieved, (A glance at the web site www.vei.co.uk/tgn/gt-guide.htm.gives an impression of the vast number of glycosyltransferases that have been cloned. Note, however, that the list has not been updated recently.) The structural information obtained not only led to the insight that all of these enzymes share a common domain structure (Figure 5), but also has provided a rationale for the aforementioned classification system. On the basis of sequence similarities at least 11 different glycosyltransferase families can now be identified (Table 2). Members within a family have extensive sequence identity at the amino acid level, which is often confined to specific boxes, and have one or more enzymatic properties in common. In the following paragraphs these glycosyltransferase families are discussed in more detail.

610

23 On the Origin of Oligosaccharide Species-Glycosyltransferases in Action

Table 2. Glycosyltransferase families. Transferdse family

Linkage-type established

Sialyl

17 a3-/a6-/a8-NeuAc to Gal(NAc)/NeuAc 2 a2-Fuc to Gal a 3 - l a 4 - F ~to Glc(NAc) 6

a2-Fucosyl a3/4-Fucosyl

Number of Notes enzymes with sequence similarity

a3-Galactosyl/Na3-Gal(NAc) to Acetylgalactosaminyl Gal(NAc)

4

p4-NAcetylglucosaminyl

P4-GlcNAc to Man

2

P6-NAcetylglucosaminyl

P6-GlcNAc to Gal(NAc)

3

a-GalNAc to Ser/Thr a-NAcetylgalactosaminyl P4-GalNAc to P4-NAcetylgalactosaminyl [NeuAca3]Gal fi4-Galactosyl P4-Gal/p4-GlcNAc to Glc(NAc)/Xyl

7

P3-Galactosyl

P3-Gal to GlcNAc or GalNAc

5

P3-Glucuronyl

P3-GlcA to Gal

3

2

8

All contain sialyl motif L and S H- and Se-enzyme FucT 111-VII and IX; a3-fucosyl motif A- and (pseudo) Benzyme, Forssman glycolipid synthase GlcNAcT IV is the only N-glycan branching GlcNAcT known so far that is not a “stand alone” 0-linked core 2 and 4 synthases and Ienzyme (GlcNAcT-V is NOT a member) Enzymes have different peptide specificities Sda- and GM2-synthase a-lactalbumin-sensitive and insensitive p4GalTs and chitobiosynthase; all share several P4-galactosyl motifs Type 1 chain synthases and ganglioside GM I -synthase (Core 1 synthase is NOT a member) Involved in HNK-1 epitope and proteoglycan core synthesis

23.18 Sialyltransferase Family All members of the sialyltransferase family share a highly conserved so called ‘sialyl motif’ (designated L for large) in their primary amino acid sequence; this motif has been found to participate in the binding of the common donor substrate CMPNeuAc [120]. In addition to this sialyl motif L a smaller more C-terminally located

23.19 a2-Fucosyltransferase Family

6 11

common sequence motif has been found; this is designated sialyl motif S (S for small) [121]. Once these motifs had been identified they were exploited to clone novel sialyltransferases by homology and today the cDNAs of seventeen of such enzymes have been identified (reviewed in Refs. [ 1221 and [ 1231). A unique feature of this family is that it comprises enzymes that establish different linkages ((a2-3)-, (a2-6)- or (a2-8)-) and use different acceptor sugars (Gal, GalNAc, or NeuAc). This is quite different from, for instance, the galactosyltransferases where enzymes with different linkage and acceptor specificity have very little sequence similarity and consequently are grouped in separate families (see below). Members of the sialyltransferase family are now commonly sub-classified on the basis of the linkage they establish and the accepting sugar [123]. There are six a3NeuAcTs that act on terminal Gal residues (ST3Gal I-VI). Best substrate structures for these enzymes are ST3Gal I, Gal(f3-3)GalNAca (protein linked); ST3Gal 11, Gal(plL3)GalNAcp (lipid linked); ST3Gal 111, Gal(p1-3)GlcNAcp; ST3Gal IV, Gal(~1-4)GlcNAc~ (protein linked); ST3Gal V, Gal(P1-4)Glcpl-Cer, and ST3Gal VI, Gal(~1-4)GlcNAc~ (lipid or protein linked), respectively [ 122, 124-1261. Only one a6-NeuAcT occurs (ST6Gal I); this specifically acts on Galp-4GlcNAcp (Fig. 6, Table 1) and GalNAcp-4GlcNAcp [66] but not on type 1 chains or O-linked aGalNAc and core 1. Five a6-NeuAcTs (ST6GalNAc I-V) have been cloned that use a GalNAc residue in different structures as accepting sugar (see also Fig. 11). ST6GalNAc I acts on O-linked GalNAca-, Gal(pl-3)GalNAca, and NeuAc(a23)Gal@-3)GalNAca- but requires a protein aglycone; ST6GalNAc I1 acts on Galpl-3GalNAca, and NeuAca(2-3)Galp1-3GalNAca- but not on GalNAca; ST6GalNAc 111, -1V and -V all three act on O-linked as well as lipid linked NeuAca(2-3)Gal(Pl-3)GalNAca/p [ 122, 127-129, 1891. The latter three enzymes tolerate aromatic aglycons, but do not act on non-sialylated substrate structures. ST6GalNAc I11 has no pronounced preference for either acceptor structure, but ST6GalNAc IV prefers O-linked acceptor chains, while ST6GaINAc V acts with preference on lipid linked substrates. In contrast to ST6GalNAc I-IV, which seem to be involved in the sialylation of O-glycans, ST6GalNAc V has been suggested to function in the synthesis of ganglioside GD1,. Five a8-NeuAcTs (ST8Sia I-V) are known of which ST8Sia I1 and IV are involved in the synthesis of protein-linked polysialic acid chains, whereas ST8Sia I, I11 and V seem to function in the synthesis of NeuAc(a2-8)NeuAca and NeuAc(a2-8)NeuAc(a2-8)NeuAca sequences occurring on gangliosides such as GD3, GTl,, GT3, and GQlb [122].

23.19 a2-Fucosyltransferase Family Two human a2-FucTs have been cloned [ 130, 1311. One enzyme, occurring in serum and plasma, is the product of the H gene and controls the synthesis of the blood group H-antigen (Fuc(al-2)Galp), which is the precursor of the A- and B-antigens, on erythrocytes. The other a2-FucT occurs in milk, saliva, and other secretions, and is the product of the Secretor (Se) gene. It determines the expression of soluble H(and A- and B-) antigen in secretions and the Lewisb (Fuc(al-2)Gal(~1-3)[Fuc(al4)IGlcNAc) blood group antigens on red cells [24, 811. The enzymes have 68%

612

23 On the Origin of Oligosuccharide Species-Giycosyltrunsferuses in Action

identity at the amino acid level. They both catalyze the formation of a Fuc(u12)Galp linkage, but have different acceptor specificities. Whereas the H enzyme highly prefers type-2 chain acceptors and efficiently acts on Calp-phenyl, the secretor enzyme has a preference for type-1 chain-based acceptors and for Gal(p13)GalNAc [ 131- 1331. The latter enzyme also has much lower affinity for the donor substrate GDP-Fuc than the H enzyme [131, 1331. In rabbit three a2-FucTs occur; one resembles the human H enzyme in molecular and enzymatic properties, the others are molecularly similar to, and have the enzymatic characteristics of, the secretor enzyme [ 1341.

23.20 a3/4-Fucosyltransferase Family Within the a3/4-fucosyltransferase family six different members have been identified. They are numbered in the order they were cloned FucT 111-VII and IX. Human FucT 111 (Lewis enzyme), V and VI (liver/plasma enzyme) are highly related enzymes with >90% identity at the amino acid level in their catalytic domain; their genes are all located on human chromosome 19 [135]. By contrast, FucT IV (myeloid enzyme, chromosome 11) and FucT VII (leukocyte enzyme, chromosome 9) have a more distal relationship to each other and to the other members of this family [135, 1361. Recently a novel member, FucT IX, which has highest homology to FucT IV, was cloned from a mouse library [ 1371. Recently the human ortholog of FucT IX was described [ 1381. A highly conserved common motif consisting of some 10-15 amino acids in the primary structures of these enzymes could be identified upon the cloning of a bacterial (Helicobacter pylori) a3-FucT [ 139, 1401. The members of this transferase family all utilize GDP-Fuc as a common donor but it remains to be established whether this ‘u3-fucosyl motif’ is somehow involved in the binding of this nucleotide-sugar. All members of this family use a substituted GlcNAc or Glc as accepting sugar and all but one establish a Fuc(al-3)-linkage. The exception is formed by FucT I11 which preferentially acts on type 1 acceptors yielding the Lewisa structure in which the Fuc is linked (al-4) (Gal(~1-3)[Fuc(al-4)]GlcNAc), whereas the other a3-FucTs exclusively act on type-2 chain-based acceptors yielding LewisX blood group determinant-based structures. FucT I11 can, however, also support the synthesis of LewisXby establishing a Fuc(ul-3)-linkage both in vitro and in vivo [2-41. This enzyme is still the only exception to the one linkage-one enzyme concept mentioned in the introduction. Recently it was reported that a single amino acid in the hypervariable stem domain of FucT I11 determines the preference of this enzyme for type-1 structures [ 1411. Preferred substrate structures for the members of this family are: FucT 111, Gal@-3)GlcNAcp-R; FucT IV, Gal(p1-4)GlcNAcpR; FucT V and VI, both Gal(P1-4)GlcNAcP-R and NeuAc(a2-3)Gal(p1-4)GlcNAcp-R; FucT VII, NeuAc(a2-3)Gal( p 1-4)GlcNAcp-R, and FucT IX, Gal(p1-4)GlcNAcp-R, respectively [69, 92, 93, 137, 1421. FucT 111-VI also can act with varying efficiencies on lactose (Gal(p1-4)Glc) and H-type-2 acceptors such as Fuc(al-2)Gal( p1-4)Glc(NAcp-R). With oligomers of lacNAc (polylactosaminoglycans), wherein Fuc may be introduced to penultimate and to internal GlcNAc

23.22 ~6-N-Acetyl~lucosuminyltvclnsferuse Family

6 13

residues, a3-FucTs show “site specificity”. Whereas FucT IV highly prefers internal GlcNAc residues, FucT 1X acts preferentially on the penultimate GlcNAc [92, 1901. FucT VII acts specifically on penultimate GlcNAc residues too, but requires that the lacNAc unit comprising this GlcNAc carries a terminal (a2-3)-linked NeuAc residue [92, 931 (Fig. 10). By contrast FucT VI (from human milk) can introduce an a3-Fuc to any GlcNAc in such acceptors whether the chain is a3-sialylated or not [143, 1441.

23.21 a3-Galactosyl/N-Acetylgalactosaminyltransferase (Histo-Blood Group ABO) Family This family, consisting of four members, is heterogeneous in that two member enzymes preferentially utilize UDP-Gal whereas the two others use UDP-GalNAc. Furthermore, one GalNAcT (blood group A transferase) and one GalT (blood group B transferase) act exclusively on substrates with Fuc(a1-2)GalP- termini (blood group H structure) to yield the blood group A and B epitopes, respectively [7, 1451 (Table 1). By contrast the other GalT (‘pseudo B-enzyme’, for which the enzymatic properties of a3-GalT have been described in detail above) [146, 1471 acts exclusively on non-fucosylated structures, preferentially on Gal( pl-4)GlcNAc, whereas the other GalNAcT acts on P-linked GalNAc in globoside (GalNAc(P13)Gal(al-4)Gal(~l-4)GlcPl-Cer) to yield the Forssman glycolipid (GalNAc(a13)GalNAc(p1-3)Gal(al-4)Gal( B1-4)Glcpl -Cer) [ 1481. The single common enzymatic property of these enzymes is that they all establish a (al-3)-linkage. Nevertheless the primary structures of these enzymes have extensive sequence similarities. This is particularly true for the human blood group A and B transferases which differ in four amino acids only [7]. It has long been known that for these enzymes sugar-donor promiscuity is relatively high [ 1491. By mutating each of the four amino acids in these blood group transferases it has been shown that two of these amino acids are crucial in determining the nucleotide-sugar specificity [7].

23.22 P4-N-Acetylglucosaminyltransferase(IV)Family The individual GlcNAcTs involved in the branching of N-linked glycans (Fig. 4) do not generally seem to belong to a glycosyltransferase family (see Glycosyltransferuses standing alone below). An exception is human GlcNAcT IV that has been described to occur in two isoforms, which show 62% identity at the amino acid level [191]. So far it is not clear whether these two forms have different specificities.

23.23 P6-N-AcetylglucosaminyltransferaseFamily This family comprises three members; all occur in man. The enzyme that was cloned first was identified as an 0-linked core 2 synthase [150] ( c . j Figure 11). It

614

23 On the Origin of Oligosaccharide Species-Glycosyltransferases in Action

acts on Gal(pl-3)GalNAca-pNP, but not on GlcNAc(p1-3)GalNAca-pNP or structures that form part of linear polylactosaminoglycans [91, 1511. A structurally related glycosyltransferase has been shown to control the synthesis of blood group I active branching points in polylactosaminoglycans [ 1521. In vitro this enzyme has been reported to act on acceptors that form part of linear polylactosaminoglycans, either as a centrally acting p6-GlcNAcT [90] or a centrally and predistally acting enzyme [91], but not on Gal(p1-3)GalNAca-pNP or GlcNAc(P1-3)GalNAcapNP. A third member of this family was found to act preferentially on Gal(p1-3)GalNAca-pNP, but also on GlcNAc(pl-3)GalNAca-pNP and thus also has 0linked core 4 synthase activity [91, 1511. A mouse homolog of the human core 2 synthase seems to be involved in the synthesis of a globo-type sphingoglycolipid core structure (GlcNAc(pl-6)[Gal(pl3)]GalNAc(~l-3)Gal(a1-4)Gal(~1-4)Glc~l-Cer) [ 1531, suggesting that core 2 synthase might have activity with acceptors having Gal(p1-3)GalNAca or Gal(p13)GalNAcp termini. This family might contain additional, not yet cloned enzymes which act with preference on penultimate (predistal) or terminal Gal residues (see also Section 23.12, Specific Modzjications of’Polylactosaminoyl~~cans).

23.24 Polypeptide N-Acetylgalactosaminyltransferase Family This family contains seven members cloned from human, rat, mouse, or bovine libraries [ 149-155, 192-1941. These enzymes (ppGalNAcTs), which catalyze the initiating step in 0-glycosylation (c.f Figure I 1), have overlapping yet distinctly different peptide-acceptor specificities [ 1 1, 156- 1581. With most peptide substrates ppGalNAcT 1 seems to be the most potent enzyme in terms of rate. Different relative rates, however, have been found for these enzyme with various peptides and some peptides are acted upon by one enzyme but not by others, and vice versa. In general all enzymes seem to prefer Thr residues to Ser in most peptide substrates. ppGalNAcT 1 and 2 introduce GalNAc residues to two Thr and one Ser position in a peptide (DTRPAPGSTAPPAHGVTSAP, points of attachment in italics) which is a repeat of MUCI. Rates for ppGalNAcT 1-3 are, however, different for each of these attachment sites [ 111. With another peptide substrate (PTTTPLK, repeat of MUC2) it seemed that the number of GalNAc residues that can be incorporated by ppGalNAcT 1-3 varies from one to three and that the order of attachment is also different for each enzyme [159]. ppGalNAcT 4 and ppGalNAcT 5 seem to act on a more limited range of peptides than ppGalNAcT 1-3 [157]. On the other hand ppGalNAcT 4 seems to complement ppGalNAcT 1-3 in that it acts on a Thr and a Ser site in TAPPAHGVTSAPDTRPAPGSTAPPA (repeat of MUC1) [ 1581. ppGalNAcT 6 resembles ppGalNAcT 3 in acceptor preferences [ 1921. Interestingly, ppGalNAcT 7 shows exclusive specificity for partially GalNAc glycosylated substrate peptides resulting from prior action of other ppGalNAcTs [ 193, 1941. This shows that the initiating steps in 0-glycosylation occur in a coordinate fashion by different ppGalNAcTs. So far all ppGalNAcTs but one are specific for UDP-GalNAc as a donor substrate. Only ppGalNAcT 2 can also efficiently utilize UDP-Gal [ 1 11. Although the

23.25 ~4-Gal~ctosyltransjerase Family

6 15

overall sequence similarity of the enzymes is ca 45% [155], all contain a segment of 41 amino acids that is highly conserved in all ppGalNAcTs [160]. Interestingly, a box within this segment shows strong similarity with a conserved box found in the P4-galactosyltransferase family. It has been suggested that this segment is involved in UDP-sugar binding [ 1601.

23.25 P4-N-AcetylgalactosaminyltransferaseFamily Of this family one human and one murine member have been identified [83, 1611. Little information is available on the specificities of the recombinant forms of these GalNAcTs, but it has been demonstrated that the murine enzyme acts on 3'sialyllacNAc (but not on lacNAc) to yield the Sd" blood group-active structure GalNAc(P1-4)[NeuAc(u2-3)]Gal( pl-4)GlcNAc [83]. Transfection studies using the cDNA of the other enzyme showed that the human GalNAcT can support the synthesis of ganglioside GM2 from GM3, and GD2 from GD3 [161].Whether these enzymes have overlapping acceptor specificities is not known, but they share the common property of acting on a Gal residue bearing a sialic acid at its OH-3 and establishing a GalNAc(P1-4)[ NeuAc(tr2-3)IGal linkage.

23.26 P4-Galactosyltransferase Family This family consists of eight enzymes, seven vertebrate and one molluscan. All establish a (01-4) linkage and although the sequence similarity among them might be lower than 40%, so far all vertebrate members are true P4-GalTs acting on GlcNAc, Glc, or Xyl. The numbering of the P4-GalTs has been confusing. Originally they were numbered in order of their cloning, the system commonly used in other glycosyltransferase families (e.g. sialyltransferases and a3-fucosyltransferases). Nowadays they are numbered according to their similarity to P4-GalT I [162], the first mammalian glycosyltransferase ever cloned [ 163-1 651. Of all members P4-GalT I has by far the highest intrinsic activity. It can act on virtually any terminally exposed P-GlcNAc to form a Gal( plL4)GlcNAc unit (see also Section 23.5, Committed Steps in the Formation of Complex-Type Oligosaccharide Chains and Branching). Free GlcNAc and P-GlcNAc linked to an aromatic or fatty aglycone are excellent acceptors [166, 1671. Also GlcNAc in oligosaccharides, whether related to N-protein (GlcNAc linked to Man) or 0-protein (GlcNAc linked to GalNAc) linked glycans, or contained in a polylactosaminoglycan chain (GlcNAc linked to Gal), is an efficient acceptor site for P4-GalT I [38, 39, 85, 167-1691. P4-GalT I can, furthermore, act on GlcNAc(P1-3)Gal(~l4)GlcPl-Cer to yield Gal(~1-4)GlcNAc(~1-3)Gal(~l-4)Glc~l-Cer (paragloboside) [166, 1671. P4-GalT I thus can function in N-protein, 0-protein, and lipid glycosylation. For over 30 years it has been known that in the lactating mammary gland this enzyme can combine with the milk protein a-lactalbumin (a-LA) to form the lactose

6 16

23 On the Origin of Oligosaccharide Species- Glycosyltransferases in Action

synthase complex [170, 1711. a-LA stimulates (34-GalT I to act on Glc yielding lactose, at the same time inhibiting the action on GlcNAc. This enzyme, therefore, also serves a specific function in lactose synthesis. (34-GalT I1 and (34-GalT I11 have acceptor properties which resemble those of (34-GalT I but act much less efficiently [166]. Whereas (34-GalT I1 is similarly responsive to a-LA as (34-GalT I, (34-GalT I11 is not, however, induced by a-LA to act on Glc, and activity toward GlcNAc is not inhibited by this modifier protein [166]. (34-GalT IV is a poorly acting enzyme and no activity could be detected with glycoprotein acceptors [ 1721. It has, however, notable activity with glycolipid acceptors and can support the synthesis of paragloboside. a-LA does not significantly induce the enzyme to act on Glc, but rather stimulates activity toward free GlcNAc [172]. (34-GalT V, like (34-GalT I, can act on a variety of acceptors albeit at a relatively low rate [167, 1731. The enzyme is not induced by a-LA to act on Glc, but a-LA inhibits its action on free GlcNAc. The highest activity is shown toward acceptors containing a GlcNAc((3 1-6)GalNAca structural element, suggesting that this enzyme predominantly functions in the galactosylation of O-linked core 2- and core 6based structures [167]. A similar function has been ascribed to P4-GalT IV [174], but this contradictory finding is probably a consequence of erroneous numbering of the clones resulting from the confusion in the numbering system mentioned above. (34-GalT VI has been identified as a lactosylceramide synthase [ 1751. Whether it acts on substrates other than glucosylceramide has not been determined. Finally, the (34-GalT involved in the first step of the synthesis of the GlcA(P1-3)Gal((313)Gal((31-4)Xyl core region of proteoglycans has recently been cloned and seems to be another member of this family 1176, 1951. It is proposed to refer to this enzyme as (34-GalTVII. The only member of this family found in invertebrates has acceptor specificity resembling that of (34-GalT I. In stead of Gal, however, it transfers GlcNAc from the corresponding nucleotide sugar to form a GlcNAc((31-4)GlcNAc ( N ,N’diacetylchitobiose) unit and therefore is a (34-GlcNAcT [177, 1781. It has no structural relationship to any other GlcNAcT yet cloned. The enzyme is insensitive to aLA. A mutant of this (34-GlcNAcT, in which two out of three amino acid repeats are removed, shows acquired donor promiscuity toward UDP-GalNAc coupled to an overall higher kinetic efficiency [9]. Donor promiscuity toward UDP-GalNAc is also shown by bovine milk (34-GalT ((34-GalT I) [8]. An interesting common property of (34-GalT I, (34-GalT V, and (34-GlcNAcT is that their activity is highest with acceptors containing a ((31-6)-linked GlcNAc [85, 167, 173, 1781. With branched acceptors such as GlcNAc((3l-6)[GlcNAc((3l-3)]Gal ((34-GalT I [85], P4-GlcNAcT [ 1781) and GlcNAc((3l-6)[GlcNAc((31-2)]Man(al-6)Man(3-R ((34-GalT I 11691) they act with preference on the ((31-6)-linked GlcNAc. Despite the small amount of overall sequence similarity, several stretches of amino acids have been highly conserved in all members (‘P4-galactosyl motifs’) 1172, 175, 1791. Interestingly one conserved stretch is very similar to a box conserved in the sequences of the pp-GalNAcTs [ 1601. The amino acids in this box might, therefore, be involved in UDP-sugar binding, but apparently do not determine donor sugar specificity. It should, furthermore, be noted that there is very little sequence similarity between members of the a3-galactosyltransferase, (34-

23.28 Glycosyltransferases Standing Alone

6 17

galactosyltransferase, and 83-galactosyltransferase families and hence it seems that UDP-Gal is recognized in different ways within each of these families. Potential other members of the P4-galactosyltransferase family are the 04GalNAcTs that occur in snail [180] and bovine mammary gland [116]. These enzymes not only strikingly resemble P4-GalT I in acceptor requirements, but also are both induced by a-LA to act on Glc yielding the lactose analog GalNAc(pl-4)Glc, suggesting that they are molecularly related to the P4-GalTs. Attempts to clone the P4-GalNAcT have, however, failed so far.

23.27 P3-Galactosyltransferase Family This family comprises five different enzymes; all utilize UDP-Gal as donor substrate and all establish a Gal(PlL3)-linkage [ 106-1 101. Whereas four enzymes (P3-GalT 1-111 and V) act on P-linked GlcNAc yielding a type-1 disaccharide unit, P3-GalT IV acts on P-linked GalNAc as in ganglioside GM2 and thus rather functions in ganglioside biosynthesis. Only slight differences in acceptor specificities have so far been observed for the type-1 chain synthases, but it has been suggested that P3-GalT V is specifically involved in elongation of the O-linked core 3 [109]. It has been anticipated that the 03-GalT responsible for the synthesis of O-linked core 2 (Gal(~l-3)GalNAca-Ser/Thr) (Fig. 11) was also a member of this family. The recent cloning of this enzyme, however, revealed that it shows no homology with any other galactosyltransferase in the databases [ 1961.

23.28 P3-Glucuronyltransferase Family This family comprises enzymes which transfer GlcA from UDP-GlcA in P3-linkage to 0-galactosides. They are involved in the synthesis of either the precursor of the occurring on glycoHNK-1 epitope (S04--3GlcA( ~1-3)Gal(~1-4)GlcNAc~-R) proteins and glycolipids, or the GlcA(~l-3)Gal(~l-3)Gal(~l-4)Xyl core structure of proteoglycans. Two rat GlcATs have been cloned, GlcAT-P [181] and GlcATD [182]; these are 50% identical with each other at the amino acid level. Whereas GlcAT-P seems to act preferentially on protein-linked type-2 chains [ 1831, GlcATD has a broader specificity and acts on both type-1 and type-2 chains whether protein- or lipid-linked [182]. The third member of this family (GlcAT-I) has been cloned from a human library [184]. This enzyme specifically acts on G a l @ 3)Gal(P1-4)Xyl to form the proteoglycan core tetrasaccharide sequence, but does not act on Gal@-4)GlcNAcP-R.

23.29 Glycosyltransferases Standing Alone Although most glycosyltransferases can be grouped in a family, several important enzymes so far cannot. Well known examples are the GlcNAcTs involved in N-

6 18

23 On the Origin of Oligosaccharide Species-Glycosyltrunsferuses in Action

glycan branching (with the exception of GlcNAcT IV, see p4-N-acetylglucosaminyltransferase ( I V ) family). The cloning of these enzymes has been described: GlcNAcT I [43, 441, GlcNAcT I1 [45], GlcNAcT 111[46], and GlcNAcT V [48, 491. These enzymes are molecularly unrelated, either to each other or to any other cloned GlcNAcT. Also the a6-FucT is involved in N-glycan core fucosylation is an enzyme that does not belong to a family [185]. Finally the cloning of two p3GlcNAcTs, both of which might be involved in polylactosaminoglycan synthesis, has revealed that these enzymes are not related to each other [86, 1861. It can, however, not be excluded that in the future molecular relatives of each of these enzymes will be discovered.

23.29 Concluding Remarks In this chapter the pathways leading to the various oligosaccharide structures occurring on soluble and cell-surface-bound glycoconjugates have been described, and the specificities of the glycosyltransferases which control the glycosylation reactions have been discussed. As a result of the molecular cloning of a vast number of these enzymes it has become evident that many can be grouped into families. It also has become clear that the same reaction can be catalyzed by related, yet different enzymes. Tissue-specific expression of these enzymes is, therefore, a way in which oligosaccharide biosynthesis might be fine-tuned. In addition to glycosyltransferases many other enzymes are involved in the process of glycosylation (processing glycosidases involved in the early stages of Nglycosylation; transporters carrying the nucleotide sugars across the intracellular membranes; pyrophosphorylases catalyzing the synthesis of these nucleotide sugars; enzymes of the intermediary carbohydrate metabolism, and epimerases, mutases, transaminases, acetylases, dehydrogenases, decarboxylases, dehydratases, etc., yielding the various precursor sugars) and the number of these ‘glycosylation enzymes’ might easily exceed 500. Within a few years the entire human genome will have been sequenced and the functional genes will become known. Some 3-5% of these genes can be estimated to be somehow involved in the process of glycosylation [ 1871. It will be a challenge for the future to identify these ‘glycosylation genes’ and to find the function of their individual protein products.

Note added in proof Very recently two additional glycosyltransferases were cloned. The first one is a sialyltransferase (ST6GalNAc VI) that acts on gangliosides but not on glycoproteins. It seems to function in the synthesis of ganglioside GDla, GQlba and GTlaa (T. Okajima, H.-H. Chen, H. Ito, M. Kiso, T. Tai, K. Furukawa, T. Urano and K. Furukawa, J. Biol. Chem., 2000, 275, 6717-6723). The other enzyme is an additional B6-GlcNAcT that seems to function in the synthesis of the 0-linked core

References

6 19

2 structure (T. Schwientek, J.-C. Yeh, S. B. Levery, B. Keck, G. Merkx, A. Geurts van Kessel, M. Fukuda and H. Clausen, J. Biol. Chem., 2000,275, 11106-1 11 13). Consequently the number of cloned sialyltransferases counts 18 and of cloned p6GlcNAcTs 4. References 1. A. Hagopian and E. H. Eylar, Arch. Biochem. Biophys., 1968, 128, 422-433. 2. J. F. Kukowska Latallo, R. D. Larsen, R. P. Nair and J. B. Lowe, Genes & Dev., 1990, 4, 1288- 1303. 3. J. B. Lowe, J. F. Kukowska Latallo, R. P. Nair, R. D. Larsen, R. M. Marks, B. A. Macher, R. J. Kelly and L. K. Ernst, J. Biol. Chem., 1991,266, 17467-17477. 4. E. Grabenhorst, M. Nimtz, J. Costa and H. S. Conradt, J. Biol. Chem., 1998, 273; 3098530994. 5. C . H. Hokke, J. G. Van der Ven, J. P. Kamerling and J. F. G. Vliegenthart, Glycoconjugate J., 1993, 10, 82-90. 6. A. D. Yates, P. Greenwell and W. M. Watkins, Biochem. Soc. Trans., 1983, 11, 300-301. 7. F. Yamamoto and S. Hakomori, J. B i d Chem., 1990,265, 19257-19262. 8. M. M. Palcic and 0 . Hindsgaul, Glycohiology, 1991, I , 205-209. 9. H. Bakker, A. van Tetering, M. Agterberg, A. B. Smit, D. H. Van den Eijnden and I. van Die, J. Biol. Chem., 1997,272, 18580-18585. 10. H. Kitagawa, H. Shimakawa and K. Sugahara, J. Biol. Chem., 1999,274, 13933-13937. 11. H. H. Wandall, H. Hassan, E. Mirgorodskaya, A. K. Kristensen, P. Roepstorff, E. P. Bennett, P. A. Nielsen, M. A. Hollingsworth, J. Burchell, J. Taylorpapadimitriou and H. Clausen, J. Biol. Chem., 1997,272, 23503-235 14. 12. R. Kornfeld and S. Kornfeld, Annu. Rev. Biochem., 1985, 54, 631-664. 13. H. Schachter, Glycobiology, 1991, I , 453-461. 14. S. C. Basu, Glycobiology, 1991, 1 , 469-475. 15. C. Abeijon and C. B. Hirschberg, TIBS, 1992, 17, 32-36. 16. G. van Echten and K. Sandhoff, J. Biol. Chem., 1993,268, 5341-5344. 17. Schachter H., in: Molecular Glycobiology, eds. M. Fukuda and 0. Hindsgaul (IRL Press, Oxford, 1994) p. 88. 18. F. W. Hemming in Glycoproteins (J. Montreuil, H. Schachter and J. F. G. Vliegenthart, eds.), Elsevier, Amsterdam, 1995, pp. 127-143. 19. A. Verbert in Glycoproteins (J. Montreuil, H. Schachter and J. F. G. Vliegenthart, eds.), Elsevier, Amsterdam, 1995, pp. 145-1 52. 20. H. Schachter in Glycoproteins (J. Montreuil, H. Schachter and J. F. G. Vliegenthart, eds.), Elsevier, Amsterdam, 1995, pp. 153-199. 21. H. Schachter in Glycoproteins (J. Montreuil, H. Schachter and J. F. G. Vliegenthart, eds.), Elsevier, Amsterdam, 1995, pp. 28 1-286. 22. I. Brockhausen, in Glycoproteins (J. Montreuil, H. Schachter and J. F. G. Vliegenthart, eds.), Elsevier, Amsterdam, 1995, pp. 201-259. 23. J. Roth, in Glycoproteins (J. Montreuil, H. Schachter and J. F. G. Vliegenthart, eds.), Elsevier, Amsterdam, 1995, pp. 287-312. 24. W. M. Watkins in Glycoproteins (J. Montreuil, H. Schachter and J. F. G. Vliegenthart, eds.), Elsevier, Amsterdam, 1995, pp. 313-339. 25. C. B. Hirschberg, P. W. Robbins and C. Abeijon, Annu. Rev. Biochem., 1998, 67, 49-69. 26. D. J. Kelleher, G. Greibich and R. Gilmore, Cell, 1992, 69, 55-65. 27. D. J. Kelleher and R. Gilmore, J. Biol. Chem., 1994,269, 12908-12917. 28. A. Lal, P. Pang, S. Kalelkar, P. A. Romero, A. Herscovics and K. W. Moreman, Glycobiology, 1998, 8, R 4. 29. S. Weng and R. G. Spiro, Glycobiology, 1996, 6, 861-868. 30. W. J. Ou, P. H. Cameron, D. Y . Thomas and J. J. Bergeron, Nature, 1993, 364, 771-776. 31. C. Hammond, I. Braakman and A. Helenius, Proc. Natl. Acad. Sci. USA, 1994,9I, 913-917. 32. C. Hammond and A. Helenius, Curr. Opin. Cell Biol., 1995, 7, 523-529.

620

23 On the Origin of Oligosuccharide Species-GlycosyltrLinsferuses in Action

33. C. Labriola, J. J. Cazzulo and A. J. Parodi, J. Cell B i d , 1995, 130, 771-779. 34. M. Sousa and A. J. Parodi, EMBO J., 1995, 14,4196-4203. 35. E. Staudacher, F. Altmann, J. Glossl, L. Marz, H. Schachter, J. P. Kamerling, K. Hird and J. F. G. Vliegenthart, Eur. J. Biochem., 1991, 199, 745-751. 36. H. Leiter, J. Mucha, E. Staudacher, R. Grimm, J. Glossl and F. Altmann, J. Biol. Chem., 1999,274, 21830--21839. 37. Y. C. Zeng, G. Bannon, V. H. Thomas, K. Rice, R. Drake and A. Elbein, J. Biol. Chem., 1997,272, 31340-31347. 38. M. R. Paquet, S. Narasimhan, H. Schachter and M. A. Moscarello, J. Bid. Chem., 1984,259, 4716-4721. 39. W. M. Blanken, A. van Vliet and D. H. Van den Eijnden, J. Biol. Chem., 1984,259, 1513115135. 40. D. H. Van den Eijnden, H. Bakker, A. P. Neeleman, 1. M. Van den Nieuwenhof and I. van Die, Biochem. Soc. Truns., 1997, 25, 887-893. 41. J. C. Paulson and K. J. Colley, J. Biol. Clzem., 1989, 264, 17615- 17618. 42. D. H. Joziasse, Glycobiology, 1992, 2, 271-277. 43. R. Kumar, J. Yang, R. D. Larsen and P. Stanley, Proc. Nutl. A c d . Sri. U S A , 1990,87, 99489952. 44. M. Sarkar, E. Hull, Y. Nishikawa, R. J. Simpson, R. L. Moritz, R. Dunn and H. Schachter, Proc. Nutl. Acud. Sci. USA, 1991, 88, 234-238. 45. G. A. F. Dagostaro, A. Zingoni, R. L. Moritz, R. J. Simpson, H. Schachter and B. Bendiak, J. Biol. Chem., 1995,270, 15211-15221. 46. A. Nishikawa, Y. Ihara, M. Hatakeyama, K. Kangawa and N. Taniguchi, J. Biol. Chem., 1992,267, 18199-18204. 47. M. T. Minowa, S. Oguri, A. Yoshida, T. Hard, A. Iwamatsu, H. Ikenaga and M. Takeuchi, J. Biol. Chem., 1998,273, 11556-11562. 48. M. G. Shoreibah, G. S. Perng, B. Adler, J. Weinstein, R. Basu, R. Cupples, D. Wen, J. K. Browne, P. Buckhaults, N. Fregien and M. Pierce, J. Biol. Chem., 1993, 268, 15381-15385. 49. H. Saito, A. Nishikawa, J. Gu, Y. Ihara, H. Soejima, Y. Wada, C. Sekiya, N. Niikawa and N. Taniguchi, Biochem. Biophys. Rex Comrnun., 1994, 198, 318-327. 50. L. Puglielli, E. C. Mandon, D. M. Rancour, A. K. Menon and C. 9 . Hirschberg, J. Bid. Chem.. 1999.274.4414-4479. 51. D. H.’Joziasse, W. E. C. M. Schiphorst, D. H. Van den Eijnden, J. A. Van Kuik, H. van Halbeek and J. F. G. Vliegenthart, J. Biol. Chem., 1987; 262, 2025-2033. 52., M. Nemansky and D. H. Van den Eijnden, Gl~vcoconjugateJ., 1993, 10, 99-108. 53. W. M. Blanken and D. H. Van den Eijnden, J. Biol. Chem.: 1985,260, 12927-12934. 54. D. H. Van den Eijnden, A. H. L. Koenderman and W. E. C. M. Schiphorst, J. Biol. Chem., 1988,263, 12461- 12471. 55. M. Nemansky, W. E. C. M. Schiphorst and D. H. Van den Eijnden, FEBS Lett., 1995, 363, 280-284. 56. K. Yamashita, Y. Tachibana, T. Ohkura and A. Kobata, J. Bid. Chem., 1985,260,3963-3969. 57. S. Narasimhan, H. Schachter and S. Rajalakshmi, J. Biol. Chem., 1988, 263, 1273-1281. 58. E. W. Easton, J. G. Bolscher and D. H. Van den Eijnden, J. Bid. Chem., 1991, 266, 2167421680. 59. E. W. Easton, I. Blokland, A. A. Geldof, B. R. Rao and D. H. Van den Eijnden, FEBS Lett., 1992,308, 46-49. 60. M. Pierce, P. Buckhaults, L. Chen and N. Fregien, Clycoconjugute J . , 1997, 14, 623-630. 61. A. Kobata, Glycoconjugute J.? 1998, 15, 323-331. 62. D. H. Joziasse, W. E. C. M. Schiphorst, D. H. Van den Eijnden, J. A. Van Kuik, H. van Halbeek and J. F. G. Vliegenthart, J. Biol. Chem., 1985,260, 714-719. 63. M. Nemansky, H. T. Edzes, R. A. Wijnands and D. H. Van den Eijnden, Glycobiology, 1992, 2, 109-117. 64. D. H. Van den Eijnden, D. H. Joziasse, A. H. L. Koenderman, W. M. Blanken, W. E. C. M. Schiphorst and P. L. Koppen, in Proc. VIIIth Int. Symp. Glycoconjugates. (E. A. Davidson, J. C. Williams and N. M. Di Ferrante, eds.), Praeger, New York, 1985, Vol. I, pp. 285-286. 65. M. J. Elices and I. J. Goldstein, J. Biol. Chem., 1989,264, 1375-1380. 66. M. Nemansky and D. H. Van den Eijnden, Biochem. J., 1992,287, 31 1-316.

References

621

67. A. A. Bergwerff, J. A. Van Kuik, W. E. C. M. Schiphorst, C. A. M. Koeleman, D. H. Van den Eijnden, J. P. Kamerling and J. F. G. Vliegenthart, FEBS Lett., 1993, 334, 133-138. 68. K. B. Wlasichuk, M. A. Kashem, P. V. Nikrad, P. Bird, C. Jiang and A. P. Venot, J. Biol. Cheni., 1993,268, 13971-13977. 69. T. De Vries, C. A. Srnka, M. M. Palcic, S. J. Swiedler, D. H. Van den Eijnden and B. A. Macher, J. Biol. Chem., 1995,270, 8712-8722. 70. K. Sujino, C. Malet, 0. Hindsgaul and M. M. Palcic, Curbohydr. Rex, 1997, 305, 483-489. 71. J. Rabina, J. Natunen, R. Niemela, H. Salminen, K. Ilves, 0. Aitio, H. Maaheimo, J. Helin and 0. Renkonen, Curbohydr. Res., 1997,305,491-499. 72. C. L. M. Stults, B. A. Macher, R. Bhatti, 0. P. Srivastava and 0. Hindsgaul, Glycobiology, 1999, 9, 661-668. 73. S. Oka, K. Terayama, C. Kawashima and T. Kawasaki, J. Biol. Cheni., 1992, 267, 2271122714. 74. Y. Kato and R. G. Spiro, J. Bid. Chem., 1989, 264, 3364-3371. 75. K. Uchimura, H. Muramatsu, T. Kaname, H. Ogawa, T. Yamakawa, Q. W. Fan, C. Mitsuoka, R. Kannagi, 0. Habuchi, I. Yokoyama, K. Yamamura, T. Ozaki, A. Nakagawara, K. Kadomatsu and T. Muramatsu, J. Biochem. Tokyo, 1998, 124, 670-678. 76. D. H. Van den Eijnden, H. Winterwerp, P. Smeeman and W. E. C. M. Schiphorst, J. Bid. Chem., 1983,258, 3435-3437. 77. A. Takeya, 0. Hosomi, N. Shimoda and S. Yazawa, J. Biochern. Tokyo, 1992, 112; 389-395. 78. P. Bailly, F. Piller, B. Gillard, A. Veyrieres, D. Marcus and J. P. Cartron, Curbohydr. Rex, 1992,228, 277-287. 79. J. Nakayama, J.-C. Yeh, A. K. Misra, S. Ito, T. Katsuyama and M. Fukuda, Proc. Nut/. Acud. Sci. USA, 1999, 96, 8991 -8996. 80. C. H. Hokke, A. P. Neeleman, C. A. M. Koeleman and D. H. Van den Eijnden, Glycobiology, 1998,8, 393-406. 81. W. M. Watkins, P. Greenwell, A. D. Yates and P. H. Johnson, Biochimie, 1988, 70, 1597-161 1. 82. K. Angata, M. Suzuki and M. Fukuda, J. Biol. Chem.. 1998, 273, 28524-28532. 83. P. L. Smith and J. B. Lowe, J. Biol. C%em.,1994, 269, 15162--15171. 84. H. Bakker, I. Friedmann, S. Oka, T. Kawasaki, N. Nifantev, M. Schachner and N. Mantei, J. Biol. Chenz., 1997, 272, 29942L29946. 85. W. M. Blanken, G. J. M. Hooghwinkel and D. H. Van den Eijnden, Eur. J. Biochem., 1982, 127, 547-552. 86. K. Sasaki, K. Kuratamiura, M. Ujita, K. Angata, S. Nakagawa, S. Sekine, T. Nishi and M. Fukuda, Proc. Nutl. Acud Sci. USA. 1997, 94, 14294-14299. 87. F. Piller, J. P. Cartron, A. Maranduba, A. Veyrieres, Y. Leroy and B. Fournet, J. Bid. Chrm., 1984,259. 13385-13390. 88. A. H. L. Koenderman, P. L. Koppen and D. H. Van den Eijnden, Eur. J. Biochem., 1987,166, 199--208. 89. Y. Sakamoto, T. Taguchi, Y. Tano, T. Ogawa, A. Leppanen, M. Kinnunen, 0. Aitio, P. Parmanne, 0. Renkonen and N. Taniguchi, J. Biol. Chem., 1998,273, 27625-27632. 90. P. Mattila, H. Salminen, L. Hirvas, J. Niittymaki, H. Salo, R. Niemela, M. Fukuda, 0. Renkonen and R. Renkonen, J. Bid. C/2ein., 1998, 273, 27633-27639. 91. J. C. Yeh, E. Ong and M. Fukuda, J. Bid. Chem., 1999,274, 3215-3221. 92. R. Niemela, J. Natunen, M. L. Majuri, H. Maaheimo, J. Helin, J. B. Lowe, 0. Renkonen and R. Renkonen, J. Biol. Chrm., 1998, 273, 4021-4026. 93. C. J. Britten, D. H. Van den Eijnden, W. McDowell, V. A. Kelly, S. J. Witham, M. R. Edbrooke, M. I. Bird, T. De Vries and N. Smithers, Glycobiology, 1998, 8, 321-327. 94. H. Salminen, K. Ahokas, R. Niemela, L. Penttila, H. Maaheimo, J. Helin, C. E. Costello and 0. Renkonen, FEBS Lett., 1997, 419, 220-226. 95. G. Pfeiffer, K. H. Strube, M. Schmidt and R. Geyer, Eur. J. Biochem., 1994,219, 331L348. 96. A. Dell, H. R. Morris, R. L. Easton, M. Panico, M. Patankar, S. Oehninger, R. Koistinen, H. Koistinen, M. Seppala and G. F. Clark, J. Biol. Chem., 1995,270, 24116-24126. 97. R. T. Camphausen, H.-A. Yu and D. A. Cumming, in Glycuproteins (J. Montreuil, H. Schachter and J. F. G. Vliegenthart, eds.), Elsevier, Amsterdam, 1995, pp. 391-414. 98. M. J. Treuheit, C. E. Costello and H. B. Halsall, Biochem. J., 1992, 283, 105-1 12. 99. D. A. Cumming, Glycobiology, 1991, I , 115-130.

622

23 On the Origin of Oligosaccharide Species-Glycosyltransferases in Action

100. S. Kornfeld, FASEB J., 1987, I , 462-468. 101. S. M. Manzella, L. V. Hooper and J. U. Baenziger, J. Biol. Chem., 1996, 271, 12117-12120. 102. E. D. Green, G. Adelt, J. U. Baenziger, S. Wilson and H. van Halbeek, J. Biol. Chem., 1988, 263, 18253-18268. 103. H. Yoshima, A. Matsumoto, T. Mizuochi, T. Kawasaki and A. Kobata, J. Biol. Chem., 1981, 256, 8476-8484. 104. A. Kobata, Methods Enzymol., 1972,28, 262- 27 1. 105. B. T. Sheares and D. M. Carlson, J. Biol. Chem., 1983,258, 9893-9898. 106. F. Kolbinger, M. B. Streiff and A. G. Katopodis, J. Biol. Chem., 1998,273, 433-440. 107. M. Amado, R. Almeida, F. Carneiro, S. B. Levery, E. H. Holmes, M. Nomoto, M. A. Hollingsworth, H. Hassan, T. Schwientek, P. A. Nielsen, E. P. Bennett and H. Clausen, J. Biol. Chem., 1998, 273, 12770--12778. 108. S. Isshiki, A. Togayachi, T. Kudo, S. Nishihara, M. Watanabe, T. Kubota, M. Kitajima, N. Shiraishi, K. Sasaki, T. Andoh and H. Narimatsu, J. Biol. Chem., 1999,274, 12499-12507. 109. D. Zhou, E. G. Berger and T. Hennet, Eur. J. Biochem., 1999,263, 571-576. 110. T. Hennet, A. Dinter, P. Kuhnert, T. S. Mattu, P. M. Rudd and E. G. Berger, J. Biol. Chem., 1998, 273, 58-65. 11I . J. C. Paulson, J. Weinstein and S. de Souza, Eur. J. Biochem., 1984, 140, 523-530. 112. H. T. de Heij, P. L. Koppen and D. H. Van den Eijnden, Carbohydr. Rex, 1986,149, 85-99. 113. A. P. Neeleman, W. P. W. Van der Knaap and D. H. Van den Eijnden, Glycobiology, 1994,4, 64 1-65 1. 114. J. Srivatsan, D. F. Smith and R. D. Cummings, J. Parasitol., 1994, 80, 884-890. 115. H. Mulder, B. A. Spronk, H. Schachter, A. P. Neeleman, D. H. Van den Eijnden, M. De Jong-Brink, J. P. Kamerling and J. F. G. Vliegenthart, Eur. J. Biochem., 1995,227, 175-185. 116. I. M. Van den Nieuwenhof, W. E. C. M. Schiphorst, I. van Die and D. H. Van den Eijnden, Glycobiology, 1999, 9, 115-123. 117. D. H. Van den Eijnden, A. P. Neeleman, W. P. W. Van der Knaap, H. Bakker, M. Agterberg and I. van Die, Biochem. Soc. Trans., 1995,23, 175-179. 118. T. P. Skelton, L. V. Hooper, V. Srivastava, 0. Hindsgaul and J. U. Baenziger, J. Biol. Chem., 1991,266, 17142- 17150. 119. H. Mulder, H. Schachter, M. de Jong Brink, J. G. Van der Ven, J. P. Kamerling and J. F. G. Vliegenthart, Eur. J. Biochem., 1991, 201, 459--465. 120. A. K. Datta and J. C. Paulson, J. Bid. Chem., 1995, 270, 1497-1500. 121. K. Drickamer, Glycobiology, 1993, 3. 2-3. 122. S. Tsuji, J. Biochem. Tokyo, 1996, 120, 1-13. 123. S. Tsuji, A. K. Datta and J. C. Paulson, Glycobiology, 1996, 6, R 5-R 7. 124. A. Ishii, M. Ohta, Y. Watanabe, K. Matsuda, K. Ishiyama, K. Sakoe, M. Nakamura, J. Inokuchi, Y. Sanai and M. Saito, J. Biol. Chem., 1998, 273, 31652-31655. 125. M. Kono, S. Takashima, H. Liu, M. Inoue, N. Kojima, Y . C. Lee, T. Hamamoto and S. Tsuji, Biochem. Biophys. Res. Commun., 1998,253, 170-175. 126. T. Okajima, S. Fukumoto, H. Miyazaki, H. Ishida, M. Kiso, K. Furukawa and T. Urano, J. Biol. Chem., 1999,274, 11479-11486. 127. M. L. E. Bergh and D. H. Van den Eijnden, Eur. J. Biochem., 1983,136, 113-118. 128. E. R. Sjoberg, H. Kitagawa, J. Glushka, H. van Halbeek and J. C. Paulson, J. Biol. Chem., 1996,271, 7450-7459. 129. Y. C. Lee, M. Kaufmannn, S. Kitazume-Kawaguchi, M. Kono, S. Takashima, N. Kurosawa, H. Liu, H. Pircher and S. Tsujii, J. Biol. Chem., 1999, 274, 11958-11967. 130. V. P. Rajan, R. D. Larsen, S. Ajmera, L. K. Ernst and J. B. Lowe, J. Biol. Chem., 1989,264, 11158-1 1167. 131. R. J. Kelly, S. Rouquier, D. Giorgi, G. G. Lennon and J. B. Lowe, J. Biol. Chem., 1995,270, 4640-4649. 132. T. Kumazaki and A. Yoshida, Proc. Nail. Acad. Sci. USA, 1984,81, 4193-4197. 133. A. Sarnesto, T. Kohlin, 0. Hindsgaul, J. Thurin and M. Blaszczyk-Thurin, J. Biol. Chem., 1992,267,2737-2744. 134. S. Hitoshi, S. Kusunoki, I. Kanazawa and S. Tsuji, J. Bid. Chem., 1996, 271, 16975- 16981. 135. B. W. Weston, P. L. Smith, R. J. Kelly and J. B. Lowe, J. Biol. Chern., 1992, 267, 24575.24584.

References

623

136. K. Sasaki, K. Kurata, K. Funayama, M. Nagata, E. Watanabe, S. Ohta. N. Hanai and T. Nishi, J. Bid. Chem., 1994,269, 14730-14737. 137. T. Kudo, Y. Ikehara, A. Togayachi, M. Kaneko, T. Hiraga, K. Sasaki and H. Narimatsu, J. Biol. Chem., 1998,273, 26729-26738. 138. M. Kaneko, T. Kudo, H. Iwasaki, Y. Ikehara, S. Nishihara, S. Nakagawa, K. Sasaki, T. Shiina, H. Inoko, N. Saitou and H. Narimatsu, FEBS Lett., 1999, 452, 237-242. 139. S. L. Martin, M. R. Edbrooke, T. C. Hodgman, D. H. Van den Eijnden and M. I. Bird, J. Biol. Chem.; 1997,272, 21349-21356. 140. Z. M. Ge, N. W. C. Chan, M. M. Palcic and D. E. Taylor, J. Biol. Chem., 1997,272, 2135721363. 141. F. Dupuy, J. M. Petit, R. Mollicone, R. Oriol, R. Julien and A. Maftah, J. Biol. Chem., 1999, 274, 12257-12262. 142. T. De Vries, M . P. Palcic, P. S. Schoenmakers, D. H. Van den Eijndeu and D. H. Joziasse, Glycobiology, 1997, 7, 921-927. 143. T. De Vries, T. Norberg, H. Lonn and D. H. Van den Eijnden, Eur. J. Biochem., 1993,216, 769-717. 144. T. De Vries and D. H. Van den Eijnden, Biochemistry, 1994, 33, 9937-9944. 145. F. Yamamoto, J. Marken, T. Tsuji, T. White, H. Clausen and S. Hakomori, J. Bid. Chem., 1990,265, 1146-1151. 146. D. H. Joziasse, J. H. Shaper, D. H. Van den Eijnden, A. J. Van Tunen and N. L. Shaper, J. Biol. Chem., 1989,264, 14290-14297. 147. R. D. Larsen, V. P. Rajan, M. M. Ruff, J. Kukowska Latallo, R. D. Cummings and J. B. Lowe, Proc. Natl. Acad. Sci. USA, 1989,86, 8227-8231. 148. D. B. Haslam and J. U. Baenziger, Proc. Nutl. Acad. Sci. USA, 1996, 93, 10697-10702. 149. A. D. Yates, P. Greenwell and W. M. Watkins, Biochem. Soc. Trans., 1983, 11, 300-301. 150. M. F. A. Bierhuizen and M. Fukuda, Proc. Nutl. Acud. Sci. USA, 1992, 89, 9326-9330. 151. T. Schwientek, M. Nomoto, S. B. Levery, G. Merkx, A. G. Van Kessel, E. P. Bennett, M. A. Hollingsworth and H. Clausen, J. Bid. Chem., 1999, 274,4504-4512. 152. M. F. A. Bierhuizen, M. G. Mattei and M. Fukuda, Genes & Dev., 1993, 7, 468-478. 153. M. Sekine, K. Nara and A. Suzuki, J. Biol. Chem., 1997,272,27246-27252. 154. F. K. Hagen, B. Van Wuyckhuyse and L. A. Tabak, J. Bid. Chem., 1993: 268, 18960-18965. 155. E. P. Bennett, H. Hassan and H. Clausen, J. Biol. Chern., 1996,271, 17006-17012. 156. F. K. Hagen, K. G. Tenhagen, T. M. Beres, M. M. Balys, B. C. Vanwuyckhuyse and L. A. Tabak, J. Biol. Chem., 1997,272, 13843-13848. 157. B. G. Tenhagen, F. K. Hagen, M. M. Balys, T. M. Beres, B. Vanwuyckhuyse and L. A. Tabak, J. Biol. Chem., 1998,273, 27749-27754. 158. E. P. Bennett, H. Hassan, U. Mandel, E. Mirgorodskaya, P. Roepstorff, J. Burchell, J. Taylorpapadimitriou, M. A. Hollingsworth, G . Merkx, A. G. Van Kessel, H. Eiberg, R. Steffensen and H. Clausen, J. Biol. Chem., 1998,273, 30472-30481. 159. S. Iida, H. Takeuchi, H. Hassan, H. Clausen and T. Irimura, FEBS Lett., 1999, 449, 230-234. 160. F. K. Hagen, B. Hazes, R. Raffo, D. deSa and L. A. Tabak, J. Biol. Chem., 1999,274, 67976803. 161. Y. Nagata, S. Yamashiro, 3. Yodoi, K. 0. Lloyd, H. Shiku and K. Furukawa, J. Bid. Chem., 1992,267, 12082-12089. 162. N. W. Lo, J. H. Shaper, J. Pevsner and N. L. Shaper, Glycobiology, 1998,8: 517-526. 163. N. L. Shaper, J. H. Shaper, J. L. Meuth, J. L. Fox, H. Chang, I. R. Kirsch and G. F. Hollis, Proc. Nutl. Acad. Sci. USA, 1986, 83, 1573-1577. 164. H. Narimatsu, S. Sinha, K. Brew, H. Okayama and P. K. Qasba, Proc. Natl. Acad. Sci. USA, 1986, 83,4720-4724. 165. H. E. Appert, T. J. Rutherford, G. E. Tarr, J. S. Wiest, N. R. Thomford and D. J. McCorquodale, Biochem. Biophys. Res. Commun., 1986, 139, 163-168. 166. R. Almeida, M. Amado, L. David, S. B. Levery, E. H. Holmes, G. Merkx, A. G. Van Kessel, E. Rygaard, H. Hassan, E. Bennett and H. Clausen, J. Biol. Chem., 1997,272, 31979-31991. 167. I. van Die, A. van Tetering, W. E. C. M. Schiphorst, T. Sato, K. Furukawa and D. H. Van den Eijndeu, FEBS Lett., 1999, 450, 52-56. 168. D. H. Van den Eijnden, W. E. C. M. Schiphorst and E. G. Berger, Biochim. Biophys. Acta, 1983, 755, 32-39.

624

23 On the Origin of’ Oligosaccharide Species- Glycosyltransjerases in Action

169. M. Ujita, J. McAuliffe, 0. Hindsgaul, K. Sasaki, M. N. Fukuda and M. Fukuda, J. Biol. Chem., 1999,274, 16717-16726. 170. U. Brodbeck, W. L. Denton, N. Tanahashi and K. E. Ebner, J. Biol. Chem., 1967,242, 13911397. 171. K. Brew, T. C. Vanaman and R. L. Hill, Proc. Nutl. Acad. Sci. USA, 1968, 59,491-497. 172. T. Schwientek, R. Almeida, S. B. Levery, E. H. Holmes, E. Bennett and H. Clausen, J. Biol. Chem., 1998,273, 29331-29340. 173. T. Sato, K. Furukawa, H. Bakker, D. H. Van den Eijnden and I. van Die, Proc. Natl. Acud, Sci. USA, 1998, 95, 472-477. 174. M. Ujita, J. McAuliffe, T. Schwientek, R. Almeida, 0. Hindsgad, H. Clausen and M. Fukuda, J. Biol. Chem., 1998, 273, 34843-34849. 175. T. Nomura, M. Takizawa, J. Aoki, H. Arai, K. Inoue, E. Wakisaka, N. Yoshizuka, G. Imokawa, N. Dohmae, K. Takio, M. Hattori and N. Matsuo, J. Biol. Chem., 1998, 273, 13570-13577. 176. T. Okajima, K. Yoshida, T. Kondo and K. Furukawa, J. Biol. Chem., 1999, 274, 2291522918. 177. H. Bakker, M. Agterberg, A. van Tetering, C. A. M. Koeleman, D. H. Van den Eijnden and I. van Die, J. Bid. Chem., 1994,269, 30326-30333. 178. H. Bakker, P. S. Schoenmakers, C. A. M. Koeleman, D. H. Joziasse, 1. van Die and D. H. Van den Eijnden, Glycobiology, 1997, 7, 539-548. 179. I. van Die, H. Bakker and D. H. Van den Eijnden, Glycobiology, 1997, 7, R 5-R 8. 180. A. P. Neeleman and D. H. Van den Eijnden, Proc. Nutl. Acud. Sci. USA, 1996, 93, 1011110116. 181. K. Terayama, S. Oka, T. Seiki, Y. Miki, A. Nakamura, Y. Kozutsumi, K. Takio and T. Kawasaki, Proc. Nutl. Acad. Sci. USA, 1997, 94, 6093-6098. 182. Y. Shimoda, Y. Tajima, T. Nagase, K. Harii, N. Osumi and Y. Sanai, J. Biol. Chem., 1999, 274, 17115-17122. 183. K. Terayama, T. Seiki, A. Nakamura, K. Matsumori, S. Ohta, S. Oka, M. Sugita and T. Kawasaki, J. Biol. Chem., 1998,273, 30295-30300. 184. H. Kitagawa, Y. Tone, J. Tamura, K. W. Neumann, T. Ogawa, S. Oka, T. Kawasaki and K. Sugahara, J. Biol. Chem., 1998,273, 6615-6618. 185. S. Yanagidani, N. Uozumi, Y. Ihara, E. Miyoshi, N. Yamaguchi and N. Taniguchi, J. Biochem. Tokyo, 1997, 121, 626-632. 186. D. P. Zhou, A. Dinter, R. G. Gallego, J. P. Kamerling, J. F. G. Vliegenthart, E. G. Berger and T. Hennet, Proc. Nutl. Acad. Sci. USA, 1999, 96, 406-41 I . 187. D. H. Van den Eijnden and D. H. Joziasse, Curr. Opin. Struct. B i d , 1993, 3, 71 1 -721. 188. A. Van Tetering, W. E. C. M. Schiphorst, D. H. Van den Eijnden and I. Van Die, FEBS Lett.. 1999, 461, 311-314. 189. Y. Ikehara, N. Shimizu, M. Kono, S. Nishihara, H. Nakanishi, T. Kitamura, H. Narimatsa, S. Tsuji and M. Tatematsu, FEBS Lett., 1999, 463, 92-96. 190. S. Nishihara, H. Iwasaki, M. Kaneko, A. Tawada, M. Ito and H. Narimatsu, FEBS Lett., 1999, 462, 289-294. 191. A. Yoshida, M. T. Minowa, S. Takamatsu, T. Hara, H. Ikenaga and M. Takeuchi, Glycoconjugate J., 1998, 15, l 115-1 123. 192. E. P. Bennet, H. Hassan, U. Mandel, M. A. Hollingsworth, N. Akisawa, Y. Ikematsu, G. Merkx, A. Geurts van Kessel, S. Olofsson and H. Clausen, J. Biol. Chem., 1999, 274, 2536225370. 193. K. G. Ten Hagen, D. Tetaert, F. K. Hagen, C. Richet, T. M. Beres, J. Gagnon, M. M. Balys, B. VanWuyckhuyse, G. S. Bedi, P. Degand and L. A. Tabak, J. Bid. Chem., 1999, 274, 27867-27874. 194. E. P. Bennet, H. Hassan, M. A. Hollingsworth and H. Clausen, FEBS Lett., 1999, 460, 226230. 195. R. Almeida, S. B. Levery, U. Mandel, H. Kresse, T. Schwientek, E. P. Bennett and H. Clausen, J. Biol. Chem., 1999,274, 26165-26171. 196. J. Tongzhong, K. W. Bremer, A. D’Souza, R. D. Cummings and W. M. Canfield, Glycobiology, 1999, 9, 1123.

Carbohydrates in Chemistry and Biology Editor Beat Ernst, Gerald W. Hart, Pierre Sinay copyright OWILEY-VCH Verlag GmbH, 2000

24 Synthesis of Sugar Nucleotides Reinhold Ohrlein

24.1 Introduction Recent progress in molecular glycobiology has revealed the important biological roles of numerous glycoconjugates [ 11. In particular cell-surface carbohydrates play key roles in cellular communication processes via selective adhesion phenomena [2, 31. They guide e.g. leukocyte extravasation [4], the homing of lymphocytes and adhesion of myeloid cells to activated platelets and the endothelium [5-7]. Besides these interactions carbohydrates are involved in fertilization events, cell growth, and parasitic infections of viruses and bacteria 181. A very recent finding is the involvement of certain cell surface a-galactosides in hyperacute rejection reactions; this is of pharmaceutical interest in organ transplantation 191. This makes sugars worthwhile targets for pharmaceutical applications [ 10, 1 11. Although chemical synthesis of oligosaccharides has reached a mature stage [ 12- 141 the overall synthetic sequences are still lengthy and cumbersome [ 151. Biologically active carbohydrates are, moreover, composed of various different monosaccharide units and are often only active if presented in a polyvalent or multiantennary array to their respective receptors [ 16, 171. These multiantennary heteropolymers, in particular, have resisted efficient chemical synthesis. Recently biocatalysts have been found to be valuable tools for complementing the classical synthetic approach [ 181. Nature has invented several ways of assembling complex oligosaccharides [ 11. The Leloir pathway is one of the preferred routes of mammalian systems. The Leloir enzymes-the glycosyl transferases-transfer a monosaccharide unit from a nucleotide-activated donor substrate to a growing oligosaccharide chain with rigorous regio- and stereoselectivity [19] (see Scheme 1). Exclusively one of the many OH groups of the acceptor is glycosylated, in either an a or p mode! Neither the donor nor the acceptor sugar needs protection from the highly homofunctional reactants-often OH groups only are present on the sugars. Thus, selective protecting group manipulations, a preponderant task in the classical, chemical oligosaccharide assemblage can be reduced to an absolute minimum.

626

24 Synthesis of Sugar Nucleotides

(Ho)n

DONOR

ACCEPTOR E.C.2.4....I

SIDE-PRODUCT

J. transferase

(HO),/

Scheme 1. Leloir glycosylation pathway.

The availability of the glycosyltransferases and the activated donor substrates is, however, a prerequisite for the application of the biocatalytic methodology. Initially the transferases were isolated from body fluids such as human and bovine milk, e. g. p( 1-4)galactosyl transferase [20, 211, or blood, e.g. a(1-3)fucosyl transferase [22],or a variety of animal tissues such as mouse kidney, e.g. the core-I1 N-acetylglucosaminyl transferase [23], and rat liver, e.g. a(2-3)- and a(2-6)sialyl transferases [24, 251, to mention just a few. Modern cloning and expression techniques make an ever growing number of glycosyl transferases available to the glycochemist [26-281. Astonishingly, only eight nucleotide-activated donors are used by mammalian cells (see Scheme 2) [29]. In principal all are commercially available, although on a small scale only. The high prices of the natural donor substrates and the unavailability of non-natural derivatives of those donors is the reason why several chemical and biomimetic pathways have been probed to increase access to these crucial compounds [29, 301. The ensuing sections will summarize the most recent attempts to synthesize these donors, with emphasis on non-natural derivatives.

24.2 Synthesis of Sugar Nucleotides 24.2.1 Chemical Synthesis

UDP-Activated Donors All the UDP-activated donors (Scheme 2) are a-pyrophosphate sugars and thus are inherently prone to degradation in acidic media, leading to monosaccharide units and uridine diphosphate [31].This lability must always be taken into account at the outset of an efficient synthetic strategy. In addition the separation of the highly polar target compounds from side-products in the reaction mixture necessitates the

24.2 Synthesis of Sugar Nucleotides HocOH

0

627

0

HocOH

. i !

[UDP-Galj

'OH

HO'

OH HO

O

H

@GziiiE]HO'

OH

q 'OH

0

0

II

0-P-0-P-0 AcH

[GzG]

HO

OH

OH

OH

Ho

Scheme 2. Nucleotide-activated donor sugars.

strict adherence to sophisticated purification protocols. Generally, overall yields are poor. General approaches towards UDP-activated sugars are outlined in Scheme 3. The key intermediates in the synthesis of nucleotide a-diphosphates are the asugar phosphates. The MacDonald procedure [ 321 proved to be a good protocol for generation of numerous a-glycosyl phosphates (see Scheme 3 ) . They are obtained in

628

24 Synthesis of Sugar Nucleotides /OH

OH Man

OH P' OH

e;p

H

OAc

N

RO

2) 1) Bn,P(O,)OH H,/Pd/C

OAc

3)NaOMe

OH QOj-OH

Ro

NHAc OH

R = Ac, perAc-GlcNAc

OH

R = H,GlcNAc

I'

OH

RrY

O

H 0 ManAc O \ l - O H OH

HC OH

W

ACHN~o O,l1 OH GalNAc P' OH

-. .

0

HO

OH

Scheme 3. Formation of nucleotide a-diphosphate sugars; Y: leaving group.

ca 30-40% yield by fusion of an alp mixture of the peracetylated hexopyranosides with phosphoric acid, then base-catalyzed deacetylation. The low yields give rise to the assumption of the concomitant formation of the undesired fi isomers. It has, however: been claimed that these hydrolyze more readily in the deacetylation step and can thus be removed completely [32, 331. u-Glucose-1-phosphate has also been prepared in 87% yield directly from glucose and phosphoric acid in the presence of propylene oxide [341. Alternative routes for the synthesis of u-sugar- 1 -phosphates started from protected sugars with a free anomeric OH group which is deprotonated in a first step and subsequently reacted with diphenyl chlorophosphate [35]. The authors found variable ratios of a / f i mixtures depending, e.g., on the base used, the reaction temperature, and the structure of the reacted hexoses. Hydrogenation over PtOl and treatment with base gave the unprotected sugar phosphates. Although the uphosphates of N-acetylglucosamine and also chitobiose could be obtained from the corresponding oxazoline precursors (Scheme 3), the a-sugar-I-dibenzylphosphate intermediates proved quite unstable [36]. Accordingly, the even more unstable a-

24.2 Synthesis of Sugar Nucleotides , OAc OAc *OAc A%’eA:O c OAc

OAc

629

1)NH31MeO!

OAc

2) phosphite

0-i-H 1

-

0

.OH

0

Scheme 4. Synthesis of a-cellobiosyl phosphate.

mannosyl-1-phosphate was obtained in 61% yield [37]. Yet another approach towards a-sugar-1-phosphates of mannose, galactose, glucose, and cellobiose was reported by Luu and coworkers; this is exemplified in Scheme 4 for a-cellobiosyl phosphate [38]. The peracetylated sugars were first treated with ammonia in methanol to remove the anomeric acetate and the resulting 1-OH sugars were subsequently phosphitylated. Only a-phosphites were obtained; in the cellobiose synthesis illustrated the final a-phosphate could be isolated in ca 50% yield after iodine oxidation and deacetylation. Besides the a-phosphates of the naturally occurring sugars (Scheme 2) a number of a-sugar-1-phosphates of non-natural derivatives have been prepared by the above phosphorylation procedures. These sugar phosphates are reacted with activated nucleotide phosphates as is exemplified in Scheme 3 to form the desired UDP sugars. Usually the commercial uridine 5’-monophosphomorpholidate has been used (Y = morpholine). We found the in situ activation of the uridine monophosphate by carbonyldiimidazole to be superior [39] (Y = imidazole). Some recent examples of the preparation of non-natural UDP- and dTDPactivated sugars by use of this general pathway (Scheme 3) are compiled in Table 1. The literature listed should be consulted because the individual experimental details can vary. The simultaneous replacement of uridine by deoxythymidine and galactose by a number of deoxy sugars to give non-natural donor substrates has also been reported [50].Partially protected a-sugar-1-phosphates were coupled to morpholidateactivated deoxythymidine-5-phosphateand subsequently deprotected in the presence of lithium hydroxide to give the corresponding dTDP sugars (see Table 1). CMP-Activated Sugars

The stereochemical preparation of CMP-sialic acid is quite a synthetic challenge. This is because of the absence of neighboring group participation to assist the

630

24 Synthesis of Sugar Nucleotides

Table 1. Yields and scales for the coupling of sugar phosphates to nucleotide phosphates Sugar

Couple %

Scale (mg)

17*

Couple Yo

Scale (ms)

14

60*

69

10*

15

82*

82

14*

12

10*

17

11*

17

22*

72

19*

85

13*

12

*

-

25"

18

32*

33

67*

90

71*

70

18*

-

66*

33

71*

35

80*

40

44**

42

34**

33

21**

36

37**

33

-

* UDP sugar ** dTDP sugar.

Ref.

Sugar

Ref.

24.2 Synthesis of Sugar Nucleotides pathway A OAc

I I

pathway B

ox

l

DAc

63 1

AcO

OMe

1 I

OAc

/

I

43 - 82%

couple

I

I

46.50%

I

47 - 61%

I

oxidize

0

OAc

I

I 1I

68%

4

couple

I

deprolect

I ‘NAO

~

HO

I I R = NHAc, OAc, NHC(O)CH,Ac, NHCBz R = NHAc. OH, NHC(O)CH,OH. NHCBz

I 1 I

OH

HO

R = NHAc, OAc, NHC(O)CH,Ac. NHCBz R

= NHAc, NHC(O)CH,OH

Scheme 5. Synthesis of CMP-sialic acid and derivatives.

stereoselective phosphorylation of the tertiary anomeric center. Two strategies have been followed recently to prepare the parent CMP-sialic acid and some derivatives thereof. Pathway A relies on the coupling of a protected cytidine phosphoramidite with protected sialic acid [51, 521. The resulting phosphite is subsequently oxidized with a t-butyl hydroperoxide solution and finally deprotected (see Scheme 5). Next to the natural CMP-sialic acid, CMP-ketodeoxyoctulonic acid and the CMP-Nglycolylsialic acid were obtained by this protocol, albeit in the low mg-range! Pathway B starts from dialkyl sialyl phosphites [53, 541 (see Scheme 5). In this case the sialyl phosphite group is directly replaced by the N/O-acetylated cytidine phosphate with retention of configuration. Application of strategy B also made accessible nucleotide monophosphosialic acid derivatives with the cytidine replaced by other pyrimidine and purine bases, although in modest overall yields only [55]. By following pathway B a CMP‘disialyl’ derivative could be synthesized. This compound has a second sialic acid unit a-linked to the 8-OH group of the CMP-activated sialic acid [56].

632

24 Synthesis of Sugar Nucleotides

GDP-Activated Donors D-Mannose and L-fucose are related biogenetically and are transferred from their corresponding guanosine diphosphate-activated donors [ 571 by the corresponding transferases. GDP-Man constitutes an a-phosphate whereas in GDP-Fuc the diphosphate bridge is P-linked to the fucose unit. The synthesis of GDP-Man was accomplished analogously as described for UDP-Gal (see Scheme 3). In this case an a-mannosyl phosphate [58] is coupled with a GMP-morpholidate in the presence of lH-tetrazole [59] to give GDP-Man. Coupling yields vary from 40-76% (scale: 13-147 mg). Also GDP-3-N-acetyl-3deoxy-a-D-mannose was obtained in 35% (46 mg) yield according the morpholidate procedure [601. Alternatively, the coupling of a phosphinothioate-activated GMP with amannosyl phosphates has also been successful [61] (see Scheme 6).

,o HOrP BnO

"1

0 0.11,OBn

P

HO

'OH

Ny

OBn

'r

i

5, GDP-Man

Scheme 6. Synthesis of GDP-man and congeners. 1) NaH, CCl,CN, 81%; 2) HOP(O)(OBn)z, 99%; 3) Na, liquid NH3 then BaC12, 60%; 4) BuZP(S)Br, B q N , THF, 61%; 5 ) AgOAc, pyr, 40%.

24.2 Synthesis of Sugar Nucleotides

633

2,3,4,6-Tetra-O-benzyl-1-hydroxymannose is first deprotonated and then treated with trichloroacetonitrile to give the trichloroacetamidate intermediate which is subsequently reacted with dibenzylphosphoric acid to give the a-perbenzylated phosphate ester. All benzyl groups could be removed by sodium in liquid ammonia. The resulting a-mannosyl phosphate is subsequently treated with the guanosine 5'monophosphate dibutyl phosphinothioic anhydride in the presence of silver acetate to form the desired GDP-Man. The necessary thiophosphoryl-activated anhydride is obtained from GMP and commercial dibutylthiophosphoryl bromide. The thiophosphoryl approach was also employed for the synthesis of GDP-Gal, GDPGlcNAc and GDP-Xyl (see Scheme 6). GDP-Fuc is a particularly delicate compound. This is reflected by the numerous papers describing various, albeit low-yielding synthetic approaches. The early investigations [62-641 were plagued by the inefficient and cumbersome access to the p-fucosyl- 1-phosphate (Scheme 7). The most reliable, fully chemical preparations have been reported by Hindsgaul [65] and Whitesides [66]. The full synthetic pathway is outlined in Scheme 7. Commercial L-fucose is first peracetylated and subsequently treated with conc. hydrobromic acid to give the a-fucosyl bromide in quantitative yield. The bromide can be stored at -20°C for several months as a solid. In the presence of silver carbonate the bromide is reacted with dibenzyl phosphate. This comphosphate to give the dibenzyl 2,3,4-tri-O-acetyl-a-~-fucosyl pound is completely deprotected to form the labile p-fucosyl phosphate which is stirred with morpholidate-activated guanosine-5'-monophosphateat room temperature for a prolonged time. The pure GDP-Fuc is obtained from this reaction mixture by repeated passages over Dowex and Sephadex columns. Thus gramamounts of GDP-Fuc were prepared. According the synthetic pathway depicted in Scheme 7, a series of non-natural GDP-deoxyfucoses have been prepared successfully (compare Table 2). Hindsgaul and coworkers [67, 681 also reported the synthesis of GDP-Fuc derivatives which bear alkyl chains, aminoalkyl residues capable of being linked to fluorescent labels, and oligosaccharide moieties on the fucose C-6. Surprisingly, those GDP-'fucose' derivatives are accepted by certain fucosyl transferases [67, 691.

Scheme 7. 1) AczO, pyr, 100%; 2) 30% HBr in AcOH, CH2C12, 100%; 3) HOP(O)(OBn)zxNEt,, Ag2C03, -78"C, CH2C12, 72%; 4) H2/Pd/C, MeOH, 100%; 5) NaOMe, MeOH, 92%; 6) GMPmorpholidate, pyr, 27-43%.

634

24 Synthesis of Sugar Nucleotides

Table 2. Successful preparations of non-natural GDP-deoxyfucoses. Position

Residue

Ref.

X = 0, unless otherwise stated. Always only one residue is altered, the remaining residues are either OH for R,, Rb, R, or CH3 for Rd. Spacer = fluorescent label added to spacer H2N(CH2)2S(CH2)20CH2-,CSHII-,C ~ H I S - .

The critical intermediate in the outlined synthesis is the p-fucosyl phosphate, which can hardly be separated cleanly from the concomitant a isomer [65, 701. The ensuing coupling to the morpholidate-activated GMP proceeds sluggishly and is always accompanied by epimerization and hydrolysis. This makes the synthesis a low-yielding process and impedes the isolation of the pure GDP-fucose from this reaction mixture. Comments

The first hurdle which must be overcome in the chemical synthesis of sugar nucleotides is the preparation of anomerically pure sugar phosphates. In sialic acid there is no neighboring group which can control the stereoselective phosphorylation of the C-2 center to form the desired p-phosphates selectively. Optimization of protecting groups and phosphorylation protocols led to modest improvements only of the alp ratios. This problem is not yet solved satisfactorily. In GDP-Fuc and derivatives the anomeric phosphate group is P-linked to the C 1 atom. Peracetate protection of the fucose enables the quantitative synthesis of the a-bromide intermediate which is converted to the desired p-phosphate almost exclusively (see Scheme 7). The UDP-activated sugars and GDP-Man are a-phosphates and their chemical synthesis poses greater problems. Although various protecting group combinations and phosphorylation protocols were investigated, none led to the exclusive formation of the desired a-phosphates (see Schemes 3 and 6). The purification of those compounds is further hampered by their notorious instability. The ensuing step is pyrophosphate formation, generally a very sluggish reaction, even if one of the two phosphate groups-usually that of the nucleotide-is activated by a good leaving group like morpholine. The diphosphate coupling reaction

24.2 Synthesis of Sugar Nucleotides

635

often takes up to 3-5 days at room temperature and is still not complete. An increase in temperature to promote the reaction is not advisable because of the lability of the reaction products. The introduction of in situ activation of one phosphate group by tetrazoles or imidazoles has only led to exceptional improvements in the coupling yields (see Scheme 7). The long reaction time is generally accompanied by decomposition of the sugar phosphates. Thus the reaction mixture often contains a number of polar compounds which are difficult to separate from the desired sugar nucleotides. This cuts down overall yields. A few sugars only could be obtained in a satisfactory yield (see Table 1). For these reasons the full chemical preparation of nucleotide donor sugars is restricted to the laboratory scale only. Large-scale production still needs much developmental effort.

24.2.2 Chemo-Enzymatic Synthesis The delicate nature of the activated donor sugars and the problems encountered with their chemical synthesis prompted glycochemists to investigate preparative alternatives mimicking the in viuo synthesis [30, 71, 721. The activated donors are obtained either by a combined chemo-enzymatic or a fully enzymatic approach. The individual approach depends on the availability of the required biocatalysts. Although some donor sugars could even be generated in situ via multi-enzyme systems before enzymatic transfer [73], this route seems limited to natural donors only; it will be discussed in more detail in Section 24.3. Uridine Diphosphate-Activated Donor Sugars

Whitesides and coworkers extensively studied various enzymatic routes offered by nature to prepare UDP-activated donors on a gram scale (see Scheme 8) [74, 751. It is apparent that the key intermediates are again the a-sugar- 1-phosphates. Galactose and 2-aminogalactose were phosphorylated directly at the anomeric centers in the presence of galactokinase and ATP [75]. The resulting a-sugar phosphates were only crudely purified and converted to the desired UDP-sugars in a multi-enzyme cycle in the presence of UTP, UDP-Glc, UDP-glucose pyrophosphorylase (E.C. 2.7.7.9), pyrophosphatase (E.C. 3.6.1.l), and galactose-l-phosphate uridyl transferase (E.C. 2.7.7.12). This last enzyme transfers a UMP moiety from UDP-Glc on to the phosphate group of galactose and 2-amino-2-deoxygalactose to produce the corresponding UDP-sugar (see Scheme 8) and glucose-1-phosphate, which is recycled by the other enzymes to form again the UDP-Glc cosubstrate with UTP as the phosphate source. With glucose or glucosamine [75] a phosphate group is first transferred from ATP on to the 6-OH group by a hexokinase. The 6-phosphate sugars are then rearranged with either a phosphoglucomutase or a N-acetylglucosamine phosphomutase, respectively, to give the a-phosphates of glucose and N-acetylglucosamine. Glucose-lphosphate is subsequently incubated with UTP and UDP-glucose pyrophosphorylase, and a pyrophosphatase to decompose the pyrophosphate side-product to give

636

24 Synthesis of Sugar Nucleotides

H?,

OH

no:

"OH

O:+ -OH

5)

HO HOfioi HO

0 OH

- P-

OH

3)0 H'0Oho$:Z$ HO

7)

NH*

HO

NHAc

AcHN 0

9 OH -7OH

,

I

,

,I

[ UDP-GlcNAc]

no.

on

Scheme 8. 1) galactokinase, ATP; 2) galactose-l-phosphate uridyltransferase, UDP-glucose, UTP, pyrophosphatase; 3) hexokinase, ATP, pyrophosphatase; 4) phosphoglucomutase; 5 ) UDP-glucose pyrophosphorylase, UTP, pyrophosphatase; 6) UDP-glucose dehydrogenase; 7) N-acetoxysuccinimide; 8) N-acetylglucosamine phosphomutase; 9) UDP-N-acetylglucosamine pyrophosphorylase, UTP, pyrophosphatase; 10) if R = OH then UDP-glucose-4-epimerase; 1 1) UDP-N-acetylglucose4-epimerase.

UDP-Glc. UDP-GlcNAc is obtained via an analogous incubation with UDP-Nacetylglucosamine pyrophosphorylase and UTP, and a pyrophosphatase. UDP-glucuronic acid is synthesized by oxidation of the 6-OH group of UDPglucose with UDP-glucose dehydrogenase and NAD+ as a hydrogen acceptor. The process works most efficiently if the generated NADHt is immediately re-oxidized e.g. with L-lactic acid dehydrogenase and pyruvic acid [74]. UDP-glucose and UDP-N-acetylglucosamine can also be converted in situ into UDP-galactose or UDP-N-acetylgalactosamine, and vice versa, by treatment with commercial UDP-glucose epimerase (E.C. 5.1.3.2) or UDP-N-acetylglucosaminyl epimerase (E.C. 5.1.3.7), respectively. These enzymes convert the equatorial OH groups of glucose and N-acetylglucosamine, into axial 4-OH groups by an oxidation-reduction sequence. The equilibria, however, lie far on the side of the starting gluco isomers. The application of the epimerization reaction is, therefore, useful only when the generated UDP-Gal and UDP-GalNAc are consumed immediately, e.g. by a successive transferase reaction. The enzymatic and semi-enzymatic routes towards UDP-sugars have been nicely compiled by Heidlas et al. [29]. Many other enzymes and enzyme sources are listed

24.2 Synthesis o j Sugar Nucleotides

637

by Elling [71]. Although a vast variety of mammalian and non-mammalian sugar nucleotides might be accessible in this manner, most of the enzymes or enzyme systems are not generally available and have not been probed on a preparative scale. There are, however, a limited number of reports concerning the enzymatic synthesis of non-natural UDP-sugars such as UDP-2F-galactose [76]. Also some uridine analogs, e.g. 5-fluorouridine 5’-monophosphate, have also been coupled to give the corresponding ‘UDP-galactose derivatives. This reaction was catalyzed by the microorganism C. sauitoana in 74% yield on a 250 mg scale [77]. The synthetic routes exactly follow the route outlined for the preparation of UDP-Gal (Scheme 8). A direct microbial conversion of glucose and UMP to UDP-Glc has also been explored [78] as has the production of a-galactosyl phosphate from bacteria [79]. Hollow fiber reactor technology has also been investigated for the production of UDP-glucosamine and UDP-GlcNAc [SO]. The principles of this preparation are similar to those outlined in Scheme 8. In a first step the sugars are phosphorylated at the 6-OH group by a hexokinase and UTP, which is regenerated in situ from phosphoenol pyruvate in the presence of a phosphokinase. The 6-phosphate is subsequently rearranged to the sugar-l-phosphate in the same reactor. In a second step the a-anomeric phosphate is treated with UTP and UDP-glucose pyrophosphorylase to give UDP-glucosamine. The amino group can be acetylated with acetic anhydride to give UDP-GlcNAc. CMP-Activated Sugars Following the biosynthetic pathway has proved to be the most efficient means of preparing CMP-sialic acid and some derivatives thereof [29, 30, 711. Glc-NAc is epimerized in a first step to give Man-NAc which is condensed with pyruvic acid in the presence of sialic acid aldolase to produce sialic acid (Scheme 9). The sialic acid is subsequently converted enzymatically to CMP-sialic acid by CMP-sialic acid synthetase. The required CTP can be produced in situ from CMP and phosphoenolpyruvate as the ultimate phosphate source via several phosphate-transfer reactions catalyzed by phosphokinases (see box in Scheme 9). During the coupling of sialic acid to CTP a pyrophosphate is again released. This side-product must be decomposed by a phosphatase to shift the equilibrium toward the product. The whole synthesis has been scaled up to a multi-gram production of pure CMP-sialic acid [81] by use of a synthetase cloned and overexpressed in E. coli. Recently, the sequence has been further improved by a Japanese group who employed a deacetylase to remove selectively any non-epimerized glcNAc which inhibits the ensuing aldolase-reaction [82]. The sialic acid synthetase could also be employed successfully to prepare a series of non-natural CMP-Sia analogs [83] (see Schemelo). The 9-amino derivative was used to attach a fluorescenyl photolabel [83]. All the donor substrates listed in the table have been probed on various a(2-6) and a(2-3) sialyl transferases [ 841. An important activated donor sugar of Gram-positive bacteria is CMP-ketodeoxyoctulosonic acid (3-deoxy-~-mannooctulosonicacid). This compound and its

638

24 Synthesis of Suyur Nucleotides

N"2

HO

H AcHN

O

HO

'

M

__ 2)

O

AcH

OH

CTP

HO

OH HO'

t

0

Scheme 9. 1) aldolase; 2) sialic acid synthetase; 3 ) pyrophosphatase; 4) phosphokinase; 5 ) deacylase.

Scheme 10. Synthesis of non-natural CMP-'Sia' derivatives with sialic acid synthetase. y

2

R

HO'

OH

R

R'

R

R'

OH

24.2 Synthesis of Sugar Nucleotides

.

R=OH,F

HO'

639

.

'OH

Scheme 11. 1) CMP-Kdo-synthetase, CTP.

V

R = OH, N, AcHN-

HO-

OH

Scheme 12. 1) Mannose phosphomutase; 2) GDP -mannose pyrophosphorylase, GTP, pyrophosphatase.

5-fluoro congener (see Scheme 11) have been prepared recently on a preparative scale by use of cloned and overexpressed microbial CMP-Kdo-synthetase (E.C. 2.7.7.38) [85]. GDP-Activated sugars The in vivo synthesis of GDP-mannose is equivalent to that of the UDP-sugars [29, 301 (see Scheme 12). Mannose- 1-phosphate, prepared from mannose-6-phosphate either by a phosphomutase reaction or chemically [61], is converted to GDPmannose by a GDP-mannose pyrophosphorylase in the presence of GTP [86]. Also in this case a pyrophosphate side-product is generated which must be decomposed by a pyrophosphatase to shift the equilibrium to the desired GDP-Man. GDP-3azido- and GDP-3-N-acetylmannose have also been obtained successfully by following this pathway [60]. Highly efficient access to GDP-fucose and several non-natural analogs is gained by the combined application of chemical and enzymological methods (Scheme 13). A series of peracetylated fucose analogs have been prepared by the chemical route described for the peracetylated p-fucosyl phosphate (Scheme 7) [87]. Two decisive improvements have been introduced with regard to the fully chemical synthesis of GDP-Fuc (Scheme 13). Firstly, the acetylated 'fucosyl' phosphates used for the coupling reaction are more stable than the unprotected variety and they are far more soluble in the coupling solvent than the unprotected ones. This speeds up the diphosphate formation without any detectable, concomitant anomerization of

640

24 Synthesis of Sugar Nucleotides 0

R R 2 F g f ’ ‘-OH ‘OH Ac OoAc

2 ’1 ) R 2

p 0 , ’

AcOoAC

NH2 HO

OH

R1 0

~

R2

__ OH NH, F OH OH OH

%

overall CH, CH, CH, H CH,OH CH,OH

89 50

80 77 70 46*

Scheme 13. 1) a: DMF, GMPxNBu3 and carbonyldiimidazole, b: DOWEX-H+; 2) H20, pH = 7, calf intestine alkaline phosphatase; 3) H20, pH = 6.8, acetylesterase, 0.1 M NaOH; *4-epi-OH.

the ‘fucosyl’ phosphates. In addition the resulting product mixture is subsequently treated with calf intestine alkaline phosphatase to remove any unreacted monophosphate selectively. This facilitates the isolation of the protected GDP-sugars. They are simply precipitated from this mixture by the addition of ethanol. In a subsequent step the desired unprotected donor-sugars are obtained under very mild conditions via incubation with commercial acetylesterase and final ethanol precipitation. The overall sequence benefits from the absence of any chromatographic purification step, high overall yields, and broad versatility. Thus the fucose moiety has been replaced by D-arabinose, L-glucose and L-galactose or 2-amino- and 2fluorofucose (Scheme 13). The procedure could be applied likewise to prepare adenosine diphosphate-, xanthosine diphosphate-, and inosine diphosphate-fucose 1871. Investigation of the human FX protein has recently revealed its remarkable involvement in the in uiuo synthesis of GDP-Fuc from GDP-Man. The purified enzyme had combined epimerase and NADPH-reductase activity, converting GDP4-keto-6-deoxymannose into GDP-Fuc [ 881. The very same reaction sequence is also achieved by use of a crude extract from A. aerogenes, and has been used to prepare 14C-labeledGDP-Fuc [89]. Another microbial production of GDP-Fuc has been reported by a Japanese group 1901. GDP-Man obtained by fermentation was converted to GDP-Fuc by A. rudiobacter. A crude extract from hog submaxillary glands was also found to convert fucose directly into GDP-Fuc by a two-step enzymatic process (see Scheme 14) [911. The protein extract probably contained both a fucokinase to generate fucose-1phosphate and GDP-fucose pyrophosphorylase. This same analytical protocol was later on applied by chemists to synthesize small amounts of GDP-Fuc [92].

641

24.3 In situ Generation of Sugar Nucleotides

Fuc

1) /7T-

ATP

Fuc-p-phosphate

*)

7-7

GDP-Fuc

GTP ADP

PYrophosphate

Scheme 14. 1) fucokinase; 2) GDP-fucose pyrophosphorylase; 3) pyrophosphatase.

Comments

In comparison with the fully chemical synthesis of nucleotide-activated donor sugars the chemo-enzymatic or enzymatic approaches are distinguished by fewer protecting group manipulations and high chemo- and stereoselectivity. This simplifies purification protocols significantly. These features reduce the number of synthetic steps and increase overall yields. The fully enzymatic preparation of sugar nucleotides, however, requires the availability of a series of enzymes. In addition, their mutual interactions presupposes exact knowledge of their physiological properties by the user; this, in particular, makes chemists hesitate to use biocatalysts. Not all the involved enzymes may tolerate altered, non-natural substrates. A combined chemo-enzymatic approach therefore offers the possibility of combining the advantages of the chemical synthesis and those of the enzymatic preparation, e. g. to prepare non-natural nucleotide activated substrates.

24.3 In situ Generation of Sugar Nucleotides Several papers describe enzymatic glycosylations with in situ generation of the required activated donor sugars; these are impressively compiled in Refs. [30] and [73]. Especially impressive is the synthesis of the trisaccharide Siau(2-6)Galp( 14)GlcNAcOH with the in situ generation of CMP-sialic acid from GlcNAc (compare Scheme 9) and UDP-galactose from glucose (compare Scheme 8) coupled with the enzymatic transfer of those sugars. In this one-pot reaction a total of nine enzymes has been applied successfully [93]! The synthesis of the linear B-trisaccharide Gala(1-3)Galu( 1-4)GlcNAcpOR, an important human xenoantigen [94], has also been tackled by a multi-enzyme onepot procedure, albeit only on a microscale (see Scheme15) [95]. Especially noteworthy is the use of sucrose, Glccx(l-2)Fru, as the ultimate source of UDP-Gal. In a first step sucrose synthase from rice cleaves the disaccharide and forms UDP-Glc in the presence of UDP. UDP-Glc is subsequently epimerized at the 4-position by UDP-galactose epimerase to give the required UDP-Gal. This donor substrate is first used by p( 1-4)galactosyl transferase to produce the Nacetyllactosamine intermediate and then by a(1-3)galactosyl transferase to gal-

642

24 Synthesis of Sugar Nucleotides

\

Glca(I-2)Fru Glca(I-2)Fru

A GlCNAcpI-OR

\

UDP-Gal UDP

A

I

I),

3)

UDP-Glc

Galp(1-4)GlcNAcpl-OR

Scheme 15. 1) sucrose synthase, UTP; 2) UDP-glucose-4’-epimerase; 3) ferase; 4) a( 1-3)galactosyl transferase.

p( 1-4)galactosyl

trans-

actosylate the intermediate disaccharide at the 3-OH group of the previously introduced galactose unit. The released UDP is recycled again. Neither the p(14)galactosyl transferase nor the a( 1-3)galactosyl transferase attacks the sucrose. The in situ generation of GDP-Fuc from p-fucose-1-phosphate coupled with the transfer of the fucose moiety on to N-acetyllactosamine has also been probed on an analytical scale [96]. In the presence of GTP P-fucose-1-phosphate is converted to GDP-Fuc by GDP-fucose pyrophosphorylase. GDP-Fuc is then used by an a(13)fucosyl transferase to transfer the fucose moiety on to N-acetyllactosamine. The released GDP is recycled and phosphorylated to give GTP by a phosphokinase with phosphoenol pyruvate as phosphate source. The pyrophosphate produced in the complete cycle is decomposed by pyrophosphatase (see Scheme 16). There is only one report of the transfer of a non-natural galactose derivative (see Scheme 17) [97]. 2-Deoxyglucose-6-phosphate, which is itself produced by a hexokinase reaction, is first equilibrated to a-2-deoxyglucose-1-phosphate by a phosphoglucomutase. In the presence of UTP and UDP-glucose pyrophosphatase this sugar is converted to UDP-2-deoxyglucose which is subsequently epimerized to UDP-2-deoxygalactose by galactose epimerase. The resulting UDP-2-deoxygalactose is then transferred by a p(l-4)galactosyl transferase on to the 4-OH group of N-acetylglucosamine to give the deoxylactosamine derivative. The released UDP is re-phosphorylated to UTP in the presence of phosphoenolpyruvate by pyruvate kinase to close the reaction cycle. Thus milligram amounts of 2’-deoxy-Nacetyllactosamine were synthesized in ca 40% yield. Comments

It goes without saying that the most elegant oligosaccharide synthesis is achieved by enzymatic assemblage of sugars starting from simple, unprotected monosaccharide building blocks and harvesting the desired oligosaccharide. This can be achieved by use of cyclic enzyme systems which activate and transfer the corresponding mono-

24.3 In situ Generation of Sugar Nucleotides

643

9 ~

H,C

H

HOoH 1) GTP

Scheme 16. 1) GDP-fucose pyrophosphorylase; 2) a( 1-3)fucosyl transferase; 3) phosphokinase; 4) pyrophosphatase. OH O,'P-OH P

H i o S O H

2)

I

0 O-prOH OH UTP,+

Scheme 17. 1) phosphoglucomutase; 2) UDP-glucose pyrophosphorylase; 3) UDP-galactose-4'epimerase; 4) j3( 1-4)galactosyl transferase; 5) phosphokinase.

644

24 Synthesis of Sugur Nucleotides

saccharides to form the desired product. No sugar phosphates and sugar nucleotides have to be isolated. In contrast, the released nucleotide diphosphate is re-used by the enzymes to start a new activation cycle (see Scheme 17). This task can also be achieved by use of microorganisms-the so called wholecell synthesis-provided the microorganisms do not metabolize the product oligosaccharide. The practicability of such cyclic multi-enzyme preparations has been confirmed, e. g. for the synthesis of N-acetyllactosamine or the tetrasaccharide, sialyl Lewis', containing four different monosaccharide units. There have also been a limited number of reports showing that slightly altered natural sugars can be introduced into such cyclic multi-enzyme systems to generate non-natural oligosaccharides. Carbohydrates with major non-natural elements will, however, still have to be prepared chemically in the near future.

24.4 Outlook The steady progress in cloning and expression techniques on the one hand and improvements in cell cultivation and fermentation procedures on the other hand will make an increasing number of enzymes available to the carbohydrate chemist. Tailor-made biocatalysts which will convert non-natural substrates also will be provided by these biotechnological methods. Additionally, screenings of microorganisms will increase access to rare and unknown sugars. Thus the preparation of natural and non-natural oligosaccharides will become cheaper. This will lead to the identification of a growing number of therapeutically active glycoconjugates and consequently an increased demand for sugars by the pharmaceutical industry. There are already suppliers who offer kilogram amounts of activated natural donor sugars for enzymatic synthesis on request (Boehringer, Wandrey, Yamasa Corp. Japan). Provided the required sugar transferases are available, combinatorial techniques will be probed in enzymatic oligosaccharide synthesis. This will give easy and rapid access to oligosaccharides and will promote studies of oligosaccharideprotein interactions. Nonetheless, chemistry will not be totally and abruptly banned from the carbohydrate field, although the era of lengthy and sophisticated assemblage of oligosaccharides by purely chemical means will wane.

References 1. 0. Hindsgaul and M. Fukuda, Molecular Glycobiology, Oxford University Press 1994. 2. A. Varki, Glycobiology 1993, 3, 97. 3 . R. A. Dwek, Chem. Rev. 1996, 96, 683. 4. P. Sears, C.-H. Wong, Chem. Commun. 1998, 1161. 5. F. R. Carbone, P. A. Gleesen, Glycobiology 1997, 7, 725. 6. C. R. Bertozzi, Chem. & Biol. 1995, 2, 703. 7. L. A. Lasky, Ann. Rev. Biochem. 1995, 64, 113.

References

645

8. A series of papers on parasites in Science 1994,264, 1857-1886. 9. K. Gustafsson, K. Strahan, A. Preece, Immunol. Rev. 1994, 141, 59. 10. J . M. McAuliffe, 0. Hindsgaul, Chem. & Ind. 1997: 3, 70. 11. Z. J. Witczak, Curr. Med. Chem. 1995, I , 392. 12. Compare preceding chapters of this volume. 13. S. Hanessian, Preparative Carbohydrate Chemistry, Marcel Dekker 1997. 14. F. Barresi, 0. Hindsgaul, Modern Synth. Methods 1995, 283. 15. F. Barresi, 0. Hindsgaul, Curbohydr. Chern. 1995, 14, 1043. 16. T. Feizi, Curr. Opin. Struct. Biol. 1993, 3. 701. 17. Y. Nagai, Pure & Appl. Chem. 1997, 69, 1893. 18. H. J. M. Gijsen, L. Qiao, W. Fitz, C.-H. Wong, Chem. Rev. 1996, 96, 443. 19. S. C. Crawley, M. M. Palcic, Modern Methods Carbohydr. Synth. 1996, 492. 20. P. H. Johnson, W. M. Watkins, Glycocojugate J. 1992, 9, 241. 21. H. E. Appert, T. J. Rutherford, G. E. Tarr, N. R. Thomford, J. McCorquindale, Biochem. Biophys. Res. Conzmun. 1986, 138, 224. 22. A. Sarnesto, T. Kohlin, 0. Hindsgaul, K. Vogele, M. Blaszczyk-Thurin, J. Thurin, J. Biol. Chem. 1992,267, 2745. 23. R. Ohrlein, 0. Hindsgaul, M. M. Palcic, Carbolzydr. Res. 1993, 244, 149. 24. L. J. Melkerson-Watson, L. C. Sweeley, J. Biol. Chern. 1991, 266, 4448. 25. J. C. Paulson, J. L. Rearick, R. L. Hill, J. Biol. Chem. 1977,252, 2363. 26. M. C. Field, L. J. Wainwright, Glycobiology 1995, 5, 463. 27. M. Fukuda, M. F. A. Bierhuizen, J. Nakayama, Gl.vcobiology 1996, 6, 683. 28. J. B. Lowe, Sem. Cell Biol. 1991, 2, 289. 29. J. E. Heidlas, K. W. Williams G. M. Whitesides, Acc. Chem. Res. 1992,25, 307. 30. C.-H. Wong, R. L. Halcomb, Y. Ichikawa, T. Kajimoto, Angew. Chem. Int. Ed. Engl. 1995,34, 412, 521. 31. N . Kochetkov, V. N. Shibaev, Adv. Carhohydr. Cliem. Biochem. 1973,28, 307. 32. D. L. MacDonald, Methods Enzymol. 1966, 8, 121. 33. C. D. Warren, R. W. Jeanloz, Biochemistry 1973, 12, 5031. 34. S. B. Tzokov, I. T. Devedjiev, E. K. Bratovanova, D. Petkov, Angew. Chem. Int. Ed. Engl. 1994, 33, 2302. 35. S. Sabesan, S. Neira, Carbohydr. Res. 1992,223, 169. 36. J. Lee, J. K. Coward, J. Org. Chem. 1992, 57, 4126. 37. T. Yamazaki, C. D. Warren, A. Herscovics. R. W. Jeanloz, Carbohydr. Res. 1980, 79, C9-CI2. 38. L. Knerr, X. Pannecoucke, B. Luu, Tetrahedron Lett. 1998, 39, 273. 39. F. Cramer, H. Neunhoeffer, K. H. Scheit, G. Schneider, J. Tennigkeit, Angew. Chem. 1962, 74, 387. 40. M.-C. Chapeau, P. A. Frey, J. Org. Chem. 1994, 59, 6994. 41. A. Burton, P. Wyatt, G. J. Boons, J. Chem. Soc. Perkin Trans. I1997, 16, 2375. 42. T. Endo, Y. Kajihara, H. Kodama, H. Hashimoto, Bioorg. Med. Chem. 1996, 4 , 1939. 43. Y. Kajihara, T. Endo, H. Ogasawara, H. Kodama, H. Hashimoto, Carbohydr. Res. 1995, 269, 273. 44. G. Srivastava, 0. Hindsgaul, M. M. Palcic, Carbohydr. Res. 1993, 245, 137. 45. H. Yuasa, 0. Hindsgaul, M. M. Palcic, J. Am. Cheni. Soc. 1992, 114, 5891. 46. H. Yuasa, 0. Hindsgaul, M. M. Palcic, Can. J. Chem. 1995, 73, 2190. 47. L. Thomas, S. A. Abbas, K. L. Abbas, Carbohydr. Res. 1988,184, 77. 48. 0. Tsuruta, G. Shinohara, H. Yuasa, H. Hashimoto, Bioorg. Med. Chem. Lett. 1997, 7, 2523. 49. T. Yamazaki, C. D. Warren, A. Herscovic, R. W. Jeanloz, Can. J. Chem. 1981, 59, 2247. 50. S. Wiemann, W. Klafie, Liebigs Ann. Chem. 1995, 1779. 51. M. D. Chappell, R. L. Halcomb, Terrahedron 1997, 53, 11109. 52. S. Makino, Y. Ueno, M. Ichikawa, Y. Hayakawa, T. Hata, Tetrahedron Lett. 1993, 34, 2775. 53. T. J. Martin, R. R. Schmidt, Tetrahedron Lett. 1992, 33, 6123. 54. T. Yoshino, R. R. Schmidt, Carhohydr. Lett. 1995, 1, 329. 55. T. J. Martin, H. Braun, R. R. Schmidt, Bioorg. Med. Chem. 1994,2, 1203. 56. Y. Kajihara, T. Ebata, K. Koseki, H. Kodama, H. Matsushita, H. Hashimoto, J. Org. Chem. 1995, 60, 5732.

646 57. 58. 59. 60. 61. 62. 63. 64.

24 Synthesis of Sugar Nucleotides

V. Ginsburg, J. Biol. Chem. 1961,236, 2389. com. available from SIGMA, CH-9741 Buchs, Switzerland, P.O. 260. V. Wittmann, C.-H. Wong, J. Org. Chem. 1997, 62, 2144. W. Klafie, Carbohydr. Res. 1995,266, 285. J. E. Pallanca, N. J. Turner, Chem. Soc. Perkin Trans. I 1993, 301 7. H. A. Nunez, J. O’Connor, P. L. Rosevar, R. Baker, Can. J. Chem. 1981,59, 2086. R. R. Schmidt, B. Wegmann, K.-H. Jung, Liehigs Ann. Chem. 1991, 121. B. M. Heskamp, H. J. G. Broxtermann, G. A. van der Marel, H. van Boom, J. Carbohydr. Chem. 1996, 15, 61 1. 65. U. B. Gokhale, 0. Hindsgaul, M. M. Palcic, Can. J. Chem. 1990, 68, 1063. 66. K. Adelhorst, G. M. Whitesides, Curbohydr. Rex 1993, 242, 69. 67. C. Hallgren, 0. Hindsgaul, J. Carbohydr. Chem. 1995, 14, 453. 68. C. Vogel, C. Bergmann, A.-J. Ott, T. K. Lindhorst, J. Thiem, W. V. Dahlhoff, C. Hallgren, M. M. Palcic, 0. Hindsgaul, Liebiqs Ann. Chem. 1997, 601. 69. G. Srivastava, K. J. Kaur, 0. Hindsgaul, M. M. Palcic, J. Biol. Chem. 1992, 267, 22356. 70. Y. Ichikawa, M. M. Sim, C.-H. Wong, J. Org. Chem. 1992, 57, 2943. 71. L. Elling, Adu. Biochem. Eng. & Biotrchnol. 1997, 58, 91. 72. S. David, C. Auge, C. Gautheron, Adv. Carbohydr. Chem. Biochem. 1991. 49, 175. 73. Y. Ichikawa, R. Wang, C.-H. Wong, Methods Enzymol. 1994, 247, 107. 74. E. J. Toone, E. S. Simon, G. M. Whitesides, J. Org. Chem. 1991, 56, 5603. 75. J. E. Heidlas, W. J. Lees, G. M. Whitesides, J. Org. Chem. 1992, 57, 146, 152. 76. T. Hayashi, B. W. Murray, R. Wang, C.-H. Wong, Bioorg. Med. Chem. 1997, 5, 497. 77. K.4. Fujita, A. Matsukdwa, K. Shibata, T. Tanaka, M. Taniguchi, S. Oi, Curbohydr. Res. 1994,265, 299. 78. J. H. KO, H . 3 . Shin, Y. S. Kim, D.-S. Lee, C.-H. Kim, Appl. Biochem. Biotechnol. 1996, 60, 41. 79. B. Nidetzky, A. Weinhausel, R. GrieBner, K. D. Kulbe, J. Curbohydr. Chem. 1995, 14, 1017. 80. P. A. Ropp, P.-W. Cheng, Anal. Biochem. 1990, 187, 104. 81. M. Kittelmann, T. Klein, U. Kragl, C. Wandrey, 0. Ghisalba, Appl. Microbiol. Biotechnol. 1995, 44, 59. 82. A. Kuboki, H. Okazaki, T. Sugai, H. Ohta, Tetrahedron 1997, 53, 2387. 83. R. Brossmer, H. J. Gross, Methods Enzymol. 1994, 247, 153. 84. H. J. Gross, R. Brossmer, Glycoconjugate J. 1995, 739. 85. T. Sugai, C.-H. Lin, G.-J. Shen, C.-H. Wong, Bioorg. Med Chem. 1995, 3, 313. 86. E. S. Simon, S.,,Grabowski, G. M. Whitesides, J. Ovg. Chem. 1990, 55, 1834. 87. G. Baisch, R. Ohrlein, Bioorg. Med. Chem. 1997, 5, 383. 88. M. Tonetti, L. Sturla, A. Bisso, U. Benatti, A. De Flora, J. Biol. Chem. 1996,271, 27274. 89. V. Ginsburg, Methods Enzymol. 1966, 8, 293. 90. K. Yamamoto, T. Maruyama, H. Kumagai, T. Tochkura, T. Seno, H. Yamaguchi, Agric. Biol. Chem. 1984, 48, 823. 91. R. Prohaska, H. Schenkel-Brunner, Anal. Biochem. 1975, 69, 536. 92. R. Stiller, J. Thiem, Liebigs Ann. Chem. 1992, 467. 93. Y. Ichikawa, J. L.-C. Liu, G.-J. Shen, C.-H. Wong, J. Am. Chem. Soc. 1991, 113, 6300. 94. U. Galili, K. Swanson, Proc. Natl. Acad. Sci. USA 1988, 7401. 95. C. H. Hokke, A. Zervose, L. Elling, D. H. Joziasse, D. H. van den Eijnden, Glycoconjugate J. 1996, 13, 687. 96. Y. Ichikawa, Y.-C. Lin, D. P. Dumas, G.-J. Shen, E. Garcia-Junceda, M. A. Williams, R. Bayer, C. Ketcham, L. E. Walker, J. E. Paulson, C.-H. Wong, J. Am. Chem. Soc. 1992, 114, 9283. 97. J. Thiem, T. Wiemann, Angew. Chem Znt. Ed. Engl. 1991, 30, 1163.

Carbohydrates in Chemistry and Biology Editor Beat Ernst, Gerald W. Hart, Pierre Sinay copyright OWILEY-VCH Verlag GmbH, 2000

25 Enzymatic Glycosylations with Glycosyltransferases Ossi Renkonen

25.1 Introduction Samples of pure molecular species of protein- and lipid-bound oligosaccharideg are needed for studies of their biological functions. These functions, in turn, are mediated by interactions between oligosaccharides and other biomolecules. Successful studies of these interactions require the use of naturally occurring compounds in the pure form and often also the use of non-natural analogs of protein- and lipid-bound oligosaccharides. The multitude of possible isomeric oligosaccharide structures has hampered the development of glycobiology [ 1) because isolation of pure samples from naturally occurring mixtures of oligosaccharide isomers has been a difficult task. To avoid this difficulty some researchers rely on enzyme-assisted in vitro synthesis of oligosaccharides. The major branch of this approach is based on Leloir’s description of biosynthesis of glycosidic bonds [2]. In Leloir-type of reactions, a monosaccharide unit is enzymatically transferred from a sugar nucleotide to a saccharide acceptor. The reactions proceed stereoselectively with unprotected acceptors, produce one glycosidic linkage in each step, and are applicable to a number of acceptors. A particular advantage of the Leloir type synthesis lies in the availability of several glycosyltransferases in recombinant form; the number of commercially available recombinant glycosyltransferases is in rapid growth and the number of enzymes cloned in research laboratories increases even more rapidly [3]. Another booster is the rapidly improving availability of sugar nucleotides required for the enzymatic synthesis of oligosaccharides (cf Chapter 24, by R. Ohrlein, in this book). Regeneration of sugar nucleotides during the transferase reactions shows promise in large scale enzymatic oligosaccharide synthesis (cf the chapter by C.-H. Wong in this book). This chapter will describe several Leloir-type transferase reactions that have been used for stepwise construction of a oligosaccharides comprising five different monosaccharides, GlcNAc, GalNAc, Gal, Fuc, and NeuSAc. This is a very limited approach among the myriads of possible oligosaccharides that can be potentially

648

25 Enzymatic Glycosylations with Glycosyltransferases

constructed from the few hundreds of different monosaccharides expressed in Nature. The limitations are remarkable even when considering the number of reported glycosyltransferase-catalyzed reactions. The examples actually selected for discussion represent steps to synthetic oligosaccharides with particularly interesting biomedical properties. Some of the final products are potent in vitro inhibitors of murine sperm-to-egg adhesion, and antagonists of mammalian leukocyte adhesion to activated endothelium. The examples discussed will show that relatively easily synthesized glycans can have considerable contraceptive and anti-inflammatory potential. To compensate for the limitations of this material, the reader is referred to the biosynthetic chapters of this book, and to recent reviews published elsewhere on the general theme, including those of Sears and Wong [3] and Van den Steen et al. [4].

25.2 In nitro Synthesis of the Core Region of 0-Glycans Many plasma membrane proteins contain domains of mucin type with clustered 0-linked glycans; the peptide cores within these clusters adopt stiff and extended conformations, representing rods from which the saccharides protrude [ 51. 25.2.1 Initialization of 0-Glycan Biosynthesis The addition of the first sugar in mucin-type 0-glycosylation is directed by the family of UDP-Ga1NAc:polypeptide N-acetylgalactosaminyltransferases (EC 2.4.1.41) (ppGaNTases). This activity has been partially purified from several sources [6]. More recently, several isoforms of the mammalian ppGaNTase family have been cloned and functionally expressed [7-91. The acceptor specificities of the individual enzymes are mostly broad and overlapping, yet distinct. The large number of the transferases is probably responsible for the lack of a general consensus sequence for mucin type 0-glycosylation. An algorithm for predicting 0-glycosylation sites in mammalian proteins is available, however [ 101. 25.2.2 Synthesis of Core 1 In vitro synthesis of 0-glycan Core 1 (Gal(D1-3)GalNAc) by UDP-Ga1:GalNAcR P3-Gal-transferase (EC 2.4.1.122) from pig submaxillary glands was described almost thirty years ago by Roseman and colleagues [ 111. Porcine submaxillary gland microsomes catalyzed the transfer of galactose from UDP-[ ''C]galactose to sialidase-treated ovine submaxillary mucin, which carries single a-linked GalNAc residues in clustered arrays on serine and threonine residues. As much as 3.4 pmol ['4C]galactose was transferred to the protein in an experiment involving 10 pmol multivalent mucin acceptor, 20 pmol UDP-[ 14C]Galand 23 mg microsomal pro-

25.2 In vitro Synthesis of the Core Region of 0-Glycans

649

tein. Treatment with P-galactosidase from C. perfringens, liberated [ ''C]galactose from the mucin product, and its treatment with 0.05 M KOH + 1 M NaBH4 released a reduced oligosaccharide. The latter gave [ 14C]galactoseand galactosaminitol in a molar ratio of 1.0:0.75 upon acid hydrolysis. Accordingly, it represented [ ''C]GalpGalNAc,,d. The product was chromatographically and electrophoretically identical with authentic Gal(p 1-3)GalNAc,d obtained from gangliosides by a known procedure; other positional isomers were not available for comparison and could not be excluded, however. This pioneering piece of work illuminates many features characteristic of enzymeassisted synthesis of oligosaccharides. Firstly, it is possible to obtain apparently homogeneous saccharide products by using unprotected acceptors, radiolabeled sugar nucleotides and even very crude enzymes. Secondly, product characterization is a difficult and a crucially important task. Later work by Medicino et al. [ 121 and Cheng and Bona 1131 on purified Core 1 galactosyltransferase from tracheal mucosa established the formation of the Gal( P 1-3)GalNAc linkage firmly; permethylation and periodate oxidation were used in the product analysis. A family of homologous P3-galactosyltransferase genes encoding enzymes that transfer to GlcNAc and to GalNAc has been identified recently 1141, but the Core 1 Gal transferase has not yet been cloned. 25.2.3 Synthesis of Core 2

In vitro synthesis of 0-glycan Core 2 [Gal(~l-3)[GlcNAc(~l-6)]GalNAc] by UDP-GlcNAc:Gal( p1-3)GalNAc-R (GlcNAc to GalNAc) p6-GlcNAc-transferase activity of dog submaxillary gland microsomes was reported by Williams and Schachter [ 151. The enzyme is conveniently abbreviated as C2GnT. Preliminary identification of the mucin product of this branch-generating reaction became possible when [ 14C]Gal(P 1-3)GalNAc-[ovine submaxillary mucin] was synthesized in vitro (see above) and was used as acceptor in the C2GnT reaction. The resulting mixture of acceptor- and product-oligosaccharides was released from the protein by reductive p-elimination, was methylated, and was finally hydrolyzed. Only 2,3,4,6tetra-@methyl-[ ''C]galactose was found in the hydrolysate, implying that the new GlcNAc unit had been transferred to the GalNAc residue of the mucin acceptor. Later, Williams et al. performed preparative experiments using Gal(p1-3)GalNAc-a-0-benzyl as the acceptor, generating a few hundred nanomoles of the trisaccharide product [ 161. With this amount of material they could apply methylation analysis and NMR spectroscopy for product characterization, establishing that the canine submaxillary microsomes had catalyzed the formation of the GlcNAc(pl6)GalNAc linkage. At least two enzyme isoforms are involved in the biosynthesis of Core 2; C2GnT of L-type is restricted to Core 2 synthesis and C2GnT of M-type synthesizes Core 2, Core 4 (see below) and GlcNAc( P1-3)[GlcNAc( p1-6)IGal-OR which is known as blood group I antigen (see below) [17]. A cDNA has been isolated that encodes the C2GnT of the L-type [18]. Also the C2GnT of the M-type has been cloned, and the recombinant forms of the two distinct C2GnTs have been used for in vitro synthesis of Core 2 0-benzyl glycosides

650

25 Enzymatic Glycosylations with Glycosyltransferases

[19]. In our laboratory the analogous transformation has been performed by use of hog gastric mucosal microsomes, which catalyzed the conversion of the free Core 1 disaccharide Gal@-3)GalNAc into Gal(~l-3)[GlcNAc(~1-6)]GalNAc [20]. Functionally active molecules of P-selectin glycoprotein ligand- 1 (PSGL- 1) are not expressed in wild type Chinese hamster ovary (CHO) cells, but are expressed in CHO cells transfected with the cDNA encoding L-type C2GnT [21]. This implies that the physiological counter-receptor of P-selectin has to carry 0-glycans of Core 2-type in order to be recognized by P-selectin.

25.2.4 Synthesis of Core 3 and Core 4 Brockhausen et al. [22] described in vitro synthesis of 0-glycan Core 3 (GlcNAc(P13)GalNAc) and Core 4 [GlcNAc(Pl-3)[GlcNAc( pl-6)]GalNAc] by UDP-GlcNAc: GalNAc-R P3-GlcNAc-transferase and UDP-GlcNAc:GlcNAc( Pl-3)GalNAc-R (GlcNAc to GalNAc) P6-GlcNAc-transferase, respectively. The enzymes are conveniently abbreviated as C3GnT and C4GnT. GalNAc-a-phenyl and GalNAc[ovine submaxillary mucin] served as acceptors in the reactions studied, and the mucin products (0.26-0.43 pmol) were characterized by liberating the oligosaccharides by reductive p-elimination, followed by NMR spectroscopy and methylation analysis of the liberated alditols. Later experiments have shown that C3GnT activity is high in human colonic and respiratory tissues and in salivary glands from several species. This enzyme has, however, proven difficult to solubilize by detergents and is unstable in the solubilized form, which has hampered its further study [23]. Recently, C4GnT activity was demonstrated in the recombinant form of C2GnT of M-type [ 191.

25.2.5 In vitvo Extension of Core 1 Glycans Mucin type 0-glycans carry often single or multiple N-acetyllactosamine (LacNAc) units. Brockhausen et al. [24] showed that pig gastric mucosal microsomes, upon incubation with UDP-[ ''C]GlcNAc and Gal@-3)GalNAcal-OR (where R is an appropriate mucin or antifreeze glycoprotein), are able to generate [ ''C]GlcNAc(Pl3) Gal@-3)GalNAcal-OR. The transfer of [ ''C]GlcNAc amounted to 1.65 pmol. The resulting [ ''C]glycans were released from the protein by reductive a-elimination, and HPLC was used to isolate altogether 100 nmol pure ['4C]GlcNAc((313)Gal(~I-3)GalNAcr,~. Methylation analysis and NMR spectroscopy were used to establish its structure. This P3-GlcNAc transferase is a distinct initiator of polylactosamine chain biosynthesis on Core 1, but it is unlikely to support the further growth of the chain. The data of Brockhausen et al. [25] showed that pig gastric mucosal microsomes fail to transfer GlcNAc to Gal@1-4)GlcNAc. The trisaccharide sequence of Core 1 type is most likely P4-galactosylated at the non-reducing end in vitro by a number of Gal-transferases and further extended in the polylactosamine mode (see below).

25.3 Enzymatic in vitro Synthesis o j Polylactosamine Backbones

65 1

25.2.6 In vitvo Extension of Core 2 Glycans

The growth of Core 2 structures resembles that of Core 1 saccharides, but can occur both at the Gal and the GlcNAc residues. Brockhausen et al. showed that pig gastric mucosa contains P3-GlcNAc transferase activity, which converts Core 2 into GlcNAc(~1-3)Gal(~1-3)[GlcNAc(~1-6)]GalNAcal-R, where R is the polypeptide backbone of either a mucin or antifreeze glycoprotein [24]. It is likely that this reaction is catalysed by the same enzyme that elongates Core 1 disaccharide. Core 2 trisaccharide in unconjugated form was quantitatively galactosylated by incubation with UDP-Gal and P4GalT I from bovine milk (EC 2.4.1.90), which is commercially available at reasonable prices [20]. The resulting tetrasaccharide Gal(~l-4)GlcNAc(~l-6)[Gal(~l-3)]GalNAc was isolated in pure form by gel filtration on Bio-Gel P-2 and characterized by NMR. P4Galactosylation of Core 2 is catalyzed even more effectively by P4GalT-IV than by P4GalT-I [26]. The P1,3-GlcNAc transferase activity (iGnT) of human serum is known to transfer to Gal(P1-4)GlcNAc, Gal(Pl-4)Glc, their simple glycosides as well as with Gal(~1-4)GlcNAc(~1-3)Gal(~l-4)Glc [27-291, but it works poorly with Gal(P13)GalNAc [28]. Put together, the acceptor-data on the P1,3-GlcNAc transferases of human serum and pig gastric mucosal microsomes seem to be distinct and complementary in their substrate specificities, such that the serum enzyme probably extends the 6-linked arm and the pig stomach mucosal enzyme extends the 3-linked arm of the tetrasaccharide Gal(~1-3)[Gal(~l-4)GlcNAc(~l-6)]GalNAc. 25.2.7 Extension of Core 3 and Core 4 Glycans Extension of Core 3 and Core 4 glycans is known to occur (4, 301, but in vitro syntheses of the extended species have not been described.

25.3 Enzymatic in uitvo Synthesis of Polylactosamine Backbones Polylactosamines consist of N-acetyllactosamine (LacNAc) chains linked to sphingoglycolipids of neolacto series and to proteins; both N- and 0-glycosidic forms of protein-bound polylactosamines are common. Polylactosamines have been known for a long time as the backbones of keratan sulfates and as undersulfated components in membranes of red blood cells and embryonic carcinoma cells. The polylactosamine backbones are expressed with and without distal decorations, which can consist of sulfate groups, single monosaccharides, e.g. a-NeuSAc, a-Fuc, a-Gal, a-GalNAc, P-GalNAc, a-GlcNAc, and sulfo-3-P-GlcA (HNK-1) and remarkably numerous oligosaccharide determinants. Fucose and sulfate decorations are expressed commonly also along the polylactosamine chain. The multitude of different capping decorations on lactosamineglycans are responsible for a large number of different saccharide-lectin interactions on mammalian cell surfaces.

652

25 Enzymatic Glycosylutions with Glycosyltvunsferuses

The polylactosamine backbones are bonded to: (i) the Man-cores of complex Nglycans, (ii) Cores 1-4 of O-glycans, and (iii) the lactose core of sphingoglycolipids. The same P3GlcNAc transferases are probably responsible for initiation and extension of some, but not all, polylactosamines. The present concept of polylactosamine biosynthesis after initialization involves: (i) the generation of primary backbones consisting of chains of LacNAc(P 1-3’)LacNAc type (LacNAc, Gal(P14)GlcNAc), (ii) generation of eventual backbone branches to yield arrays of the type LacNAc(pl-3’)[LacNAc( pl-6’)]LacNAc, and (iii) decoration reactions. 25.3.1 Enzymatic Synthesis of the Primary Chains of Blood Group i-Type The (33GlcNAc transferase responsible for extension of terminal N-acetyllactosamine residues of free oligosaccharides was reported independently by three groups, in Novikoff ascites tumor cells [31] and in human serum [27,28]. Yates and Watkins [27] succeeded in synthesizing 3.8 pmol purified [ ‘‘C]GlcNAc(~l-3)Gal(pl-4)GlcNAc from 80 pmol LacNAc and 12 pmol UDP-[ 14C]GlcNAc.They also synthesized 7.5 pmol pure [14C]GlcNAc(P1-3)Gal(P1-4)Glc from 200 pmol Gal@-4)Glc and 30 pmol UDP-[ 14C]GlcNAc. P-N-Acetylhexosaminidase treatment liberated the newly transferred [ 14C]GlcNAcresidues from both products, implying that Plinkages had been formed by the serum transferase. Methylation analysis revealed that the galactosyl groups of both products were 3-O-substituted with GlcNAc. Simple glycosides of LacNAc and lactose were also good acceptors, as was Gal(p 1-4)GlcNAc( P1-3)Gal( P 1-4)Glc. Piller and Cartron 1281 reported similar experiments. Their data show that Fuc(al-2)Gal(P1-4)GlcNAc and Gal(P1-4)[Fuc(al-3)]GlcNAc are not acceptors for the serum p3-GlcNAc-transferase, and the disaccharide Gal@-3)GlcNAc is a poor acceptor. By contrast, the polyand Gal(P1-4)lactosamines Gal(~l-4)GlcNAc(~1-3)Gal(~l-4)GlcNAc-O-Me GlcNAc(~1-3)Gal(~l-4)GlcNAc(~l-3)Gal(~1-4)GlcNAc-O-Me were as good acceptors as Gal(P 1-4)GlcNAc; the resulting products were not structurally characterized, however. Because linear polylactosamines are known as blood group i-determinants, the enzymes responsible for these reactions are called iGnT. The serum iGnT transferred GlcNAc effectively also to desialylated fetuin and desialylated a1 -acid glycoprotein and to the pentasaccharides Gal(P1-4)and Gal(Pl-4)GlcNAc(P1-2)GlcNAc(pl-2)Man(al-3)Man( P 1-4)GlcNAc Man(al-6)Man( Pl-4)GlcNAc, representing distinct branches of complex N-glycans [28] Accordingly, serum iGnT is able to transfer to terminal Gal(p1-4)GlcNAc residues of glycoprotein-linked complex N-glycans. By contrast, asialoglycophorin, which mostly contains Gal(Pl-3)GalNAc bound to serine or threonine residues, was not an acceptor. The serum iGnT converts also GalNac(P1-4)GlcNAc to GlcNAc(~1-3)GalNAc(~1-4)GlcNAc 1321. The serum iGnT extends also P6-linked branches of polylactosamines, converting the hexasaccharide LacNAc(P 13’)[LacNAc(~l-6’)]LacNAc into the octasaccharide GlcNAc(PI-3’)LacNAc(Pl3’)[GlcNAc(pl-3)’ LacNAc(p 1-6’)ILacNAc [ 331. iGnT has been purified from calf serum [34], and a human cDNA clone encoding a similar enzyme has been isolated by Sasaki et al. [35]. A transferase experiment

25.3 Enzymatic in vitvo Synthesis of Polylactosamine Backbones

653

with the crude recombinant enzyme and 500 nmol lacto-N-neo-tetraose gave 0.7 nmol [ 3H]GlcNAc(pl-3)Gal( p1-4)GlcNA( pl-3)Gal( pl-4)Glc.

25.3.2 Distal Branching of i-Type Polylactosamine Backbones

The first branching reaction of polylactosamine backbones was described by Piller et al. [36] who found an appropriate p1,6-GlcNAc transferase activity in hog gastric mucosal microsomes. This activity catalyzed the reaction shown in eq. (1).

GlcNAc(pl-3)Gal(pl-4)Glc + UDP-GlcNAc + GlcNAc(

p 1-3) [GlcNAc(pl-6)] Gal(pl-4)Glc

(1)

The product was identified by methylation, followed by methanolysis, O-acetylation and GLC-mass spectrometry of the derivatized monosaccharides. The GLC which was present data showed that 1,2,4,6-tetra-O-methyl-3-O-acetylgalactose, among the components obtained from the trisaccharide acceptor, had completely disappeared. It was replaced by a and p isomers of 1,2,4-tri-O-methy1-3,6-di-Oacetylgalactose. Compared with the acceptor, the amount of 1,3,4,6-tetra-O-methyI2-deoxy-N-methylacetamidoglucoseobtained from the tetrasaccharide product was twofold. Mass spectrometric analysis of the derivatized galactose obtained from the tetrasaccharide product confirmed that the new substituent was linked to carbon 6. The activity in hog gastric mucosal microsomes was unable to transfer to acceptors that terminated with a complete LacNAc unit at the non-reducing end. The microsomes transferred also to the glycolipid GlcNAc(p1-3)Gal( pl-4)Glcpl-Ceramide. The primary experiment of Piller et al. (loc. cit.) was repeated by Seppo et al. [37], using GlcNAc(pl-3)Gal(pl-4)GlcNAcas the acceptor. Here, the tetrasaccharide GlcNAc(~l-3)[GlcNAc(~l-6)]Gal(~l-4)GlcNAc was formed, as was shown by partial acid hydrolysis of the product and chromatographic identification of all diand trisaccharides of the hydrolysate. More recently, Helin et al. [38] incubated the doubly labeled pentasaccharide GlcNAc(pl-3)[ 3H]Gal([31-4)GlcNAc(p1-3)[ 14C]Gal(pl-4)GlcNAc with UDPGlcNAc and the gastric mucosal microsomes, obtaining the hexasaccharide GlcNAc(P 1-3)[GlcNAc(P 1-6)][3H]Gal(p 1-4)GlcNAc( p 1-3)[ 14C]Gal(p 1-4)GlcNAc. Here, product analysis was performed by partial acid hydrolysis. Paper chromatography of the hydrolysate gave [3H]-labeled, but not [ 14C]-labeled,glycans that contained the newly formed GlcNAc(p1-6)Gal bond. The diagnostic cleavage products included the disaccharide GlcNAc( pl-6)[ 3H]Gal, the trisaccharides GlcNAc(p1-6)[ 3H]Gal(pl -4)GlcNAc and GlcNAc( p1-3)[GlcNAc( pl-6)][ 3H]Gal, and the tetrasaccharide GlcNAc(p1-3)[GlcNAc(p1-6)][ 3H]Gal(P1-4)GlcNAc. In addition to the degradation data, positive-ion MALDI-TOF mass spectrometry of the original hexasaccharide product revealed a major peak of monoisotopic m/z 1177.1 that was assigned to (M + N)+ of GlcNAc4Galz (calculated m/z 1177.4). The 'H NMR spectrum of the hexasaccharide product also confirmed its structure. The spectrum revealed reporter group signals similar to those of the acceptor

654

25 Enzymatic Glycosylations with Glycosyltransferases

pentasaccharide, but also contained an additional, one-proton doublet at 4.592 ppm that was assigned H-1 of the pl,6-bonded GlcNAc. In summary, the data of Helin et al. show that only the penultimate galactose unit close to the non-reducing end of the pentasaccharide served as a primary acceptor site, despite the presence of an additional, internal galactose residue. We call the branching activity of the gastric distal site specificity and the formucosal microsomes as dIGnT to emphasize its mation of blood group 1-type structures. Interestingly, incubation of the hexasaccharide GlcNAc(pl-3)[GlcNAc( P1-6)]Gal(~1-4)GlcNAc(~l-3)Gal(~l-4)GlcNAc with UDP-GlcNAc and gastric mucosal microsomes gave a small amount of the doubly branched heptasaccharide product GlcNAc(pl-3)[ GlcNAc( 81-6)]Gal( Pl-4)GlcNAc( pl-3)[GlcNAc( pl-6)]Gal(P1-4)GlcNAc [38]. It was not clear whether an analogous, slow reaction at the mid-chain galactose had occurred also at the linear pentasaccharide GlcNAc(pl3)Gal(~lL4)GlcNAc(~1-3)Gal(~l-4)GlcNAc at a rate that was below the detection limit. Alternatively, dIGnT can catalyze reactions of different rates at the midchain galactoses of different polylactosamine acceptors. Quite recently, Yeh et al. [ 191 cloned the cDNA of C2GnT-M and expressed it in functionally active form. Besides the C2GnT activity, the recombinant enzyme had significant C4GnT activity and a small amount of dIGnT activity when tested using the acceptors GlcNAc(P1-3)GalNAcul-para-nitrophenol and GlcNAc(Pl-3)respectively. These data support Gal(pl-4)GlcNAc( ~1-6)Man(u1-6)Man~l-octyl, the notion that the dIGnT-activity of hog gastric mucosal microsomes used in the experiments of Piller et al. [36], Seppo et al. [37], and Helin et al. [38] might have represented C2GnT of M-type. Starting from GlcNAc, a polylactosamine consisting of seven N-acetyllactosamine units in a highly branched array was synthesized in our laboratory, by use of a multistep process involving the dIGnT branching reaction [ 391. Besides reactions catalyzed by hog gastric mucosal microsomes at two distinct stages, the process consisted also of several reactions catalyzed by the chain extension enzymes iGnT and P4GalT I (see below). The tetradecameric product was synthesized in milligram amounts and subsequently decorated by: (i) four distal al,3-bonded Gal residues, (ii) four distal pl,3-linked GlcNAc units [39], (iii) four distal u2,3-linked Neu5Ac residues, (iv) four peridistal a3-bonded Fuc units, and (v) a combination of four distal a2,3-linked NeuSAc and four peridistal a3-bonded Fuc units [40]. The products were characterized by NMR spectroscopy and MALDI-TOF mass spectrometry, and were studied as putative inhibitors of murine sperm-to-egg adhesion and binding of lymphocytes to capillary endothelium of inflamed tissue. A glycolipid from hog gastric mucosa contains a large, branched polylactosamine backbone [41] that is strikingly similar to the tetradecameric polylactosamine backbone of Seppo et al. [39]. 25.3.3 Central Branching of i-Type Polylactosamine Backbones

Another type of polylactosamine branching was reported by Anne Leppanen et al. [42]. In their experiments, the doubly labeled linear pentasaccharide GlcNAc(pl-3)[ 3H]Gal(p 1-4)GlcNAc( p1-3)[ I4C]Gal(p1-4)GlcNAc was converted

25.3 Enzymatic in vitvo Synthesis of Polylactosamine Backbones

655

into the branched hexasaccharide GlcNAc(P1-3)[ 3H]Gal(pl-4)GlcNAc(~1-3)[GlcNAc(P1-6)][ 14C]Gal(Pl-4)GlcNAc when incubated with UDP-GlcNAc and human serum. The product was identified by digestion with endo-P-galactosidase that cleaved the hexasaccharide completely as shown by paper chromatography. All of the 3H radioactivity of the digest was found in the disaccharide fraction, cochromatographing with the GlcNAc( pl-3)Gal marker. Nearly all of the [ 14C]label in the digest was found in a tetrasaccharide, which was identified as GlcNAc(P13)[GlcNAc(Pl-6)][ ‘‘C]Gal(~l-4)GlcNAcby partial acid hydrolysis and subsequent chromatography. These data imply that the original hexasaccharide product was cleaved by endoP-galactosidase only at the galactosidic bond of the [ 3H]Gal located close to the non-reducing end. This was compatible with the postulated structure of the hexadetersaccharide product, the GlcNAc(~l-3)[GlcNAc(~1-6)]Gal(~l-4)GlcNAc minant is known to resist the action of endo-P-galactosidase. Accordingly, the serum enzyme transferred solely to the [‘4C]-labeledmid-chain Gal of the pentasaccharide acceptor. We call this activity cIGnT to emphasize the central sitespecificity of its action. Gu et al. [43] have reported similar activity in rat serum. Later experiments showed that human serum cIGnT transfers to position 6 of the internal Gal also in the tetrasaccharide Gal(~1-4)GlcNAc(~1-3)Gal(~l-4)GlcNAc [44]. The cIGnT of rat serum gave a similar product. Accordingly, this enzyme is not sensitive to the distal galactosylation status of the chain. It works with chains capped by many different groups (Leppanen et al., unpublished experiments), but it transfers effectively only to acceptors having at least one complete Gal(p 1 4)GlcNAc unit bonded at position 3 to the acceptor Gal. The structure of the pentasaccharide product generated by rat serum cIGnT was established firmly by 2D NMR experiments, for which a relatively large sample of the pure pentasaccharide had to be prepared [45]. The tetrasaccharide Gal(814)GlcNAc(~1-3)Gal(Pl-4)GlcNAc was first synthesized in amounts of a few micromoles, starting from LacNAc that was extended in a reaction catalyzed by the iGnT activity of human serum; the resulting GlcNAc( Pl-3’)LacNAc was purified and then P1,4-galactosylated by bovine milk P4GalT I. Finally, the branching reaction of the tetrasaccharide using rat serum cIGnT gave 2.7 pmol of the pure pentasaccharide product. Enzymatic generation of multiple branches to a linear polylactosamine backbone was first described by Leppanen et al. [44]. In these experiments the linear hexasaccharide LacNAc( PI -3’)LacNAc( P 1-3’)LacNAc was converted into the doubly branched octasaccharide LacNAc(Pl-3’)[GlcNAc( p1-6’)]LacNAc( PI-3’)[GlcNAc(pl-6’)ILacNAc by incubation with UDP-GlcNAc and cIGnT activity of rat serum. Analogous experiments of Salminen et al. [46] with the linear octasaccharide LacNAc(~1-3’)LacNAc(~1-3’)LacNAc(~1-3’)LacNAc gave the triply branched undecasaccharide LacNAc( ~1-3’)[GlcNAc(~l-6’)]LacNAc(~1-3’)[GlcNAc(~l-6’)]LacNAc(~l-3’)[GlcNAc(~l-6’)]LacNAc, that was converted to the tetradecasaccharide LacNAc(~1-3’)[LacNAc(~l-6’)]LacNAc(~l-3’)[LacNAc(p1-6’)]LacNAc( p 1-3’)[LacNAc( p I -6’)lLacNAc by a treatment with UDP-Gal and P4-GalT I. Branched polylactosamine backbones of this type are expressed in human erythrocytes and embryonic carcinoma cells. Recent experiments have led to identification of cIGnT in human embryonic -

656

25 Enzymatic Glycosylations with Glycosyltransferases

carcinoma cells [47], cloning of the cDNA encoding the cIGnT of these cells [48], expression and purification of a truncated, functional form of this cIGnT [49], as well as purification of a cIGnT from hog intestinal mucosa [50].The recombinant cIGnT transfers efficiently also to Gal(~l-4)GlcNAc(~l-3)Gal(~l-4)GlcNAc(~l6)Man(al-6)Manpl-octyl, a mimic of the polylactosamine-bearing branch of Nglycans [ 191. In due time, these advances might enable transfer of multiple branches to linear polylactosamines on a large scale. 25.3.4 f34-Galactosylation in Polylactosamine Backbones

The P4GalT I from bovine and human milk was recognized as the only enzyme of its kind for a while, but novel members of this family (P4GalT-I1 to P4GalT-VI) have been isolated recently (reviewed by Ujita et al. [26]). p4GalT-IV initiates polylactosamine synthesis in the 6-linked arm of Core 2 0-glycan more effectively than P4GalT-I [26], but the bovine milk P4GalT-I elongates primary polylactosamine chains efficiently in vitro [44, 511. P4GalT-I initiates the growth of p1,6-linked GlcNAc branches of polylactosamines efficiently [52], and it also elongates P1,6bonded GlcNAc(~1-3)Gal(~1-4)GlcNAc branches [53].

25.4 a3-Sialylation of N-Acetyllactosaminoglycans at the Terminal Gal Distal u3-sialylation of branched polylactosamines was catalyzed successfully by the activity (ST3Gal IV?) present in human placental microsomes [40, 521; both 3and 6-linked short polylactosamine branches reacted well. More recently, large, tetravalent sialopolylactosamines have been synthesized in low micromolar amounts (L. Penttila et al., unpublished experiments). Even long polylactosamine chains of blood group i-type are effectively sialylated by placental microsomes [54]. Several a3-sialyltransferases have been cloned, and ST3Gal I11 is commercially available in recombinant form. A particularly interesting report of Gilbert et al. [55] describe a fusion protein consisting of CMP-Neu5Ac synthetase and a2,3-sialyltransferase from Neisseria meningitidis. This polypeptide was able to catalyze the reactions shown in eqs (2) and (3).

+ Neu5Ac + CMP-Neu5Ac + PP CMP-Neu5Ac + Gal-OR -+ Neu5Ac(a2-3)Gal-OR + CMP CTP

(2) (3)

In small-scale reactions the fusion protein worked well with glycosides of LacNAc and lactose, as well as the bi-antennary N-linked glycan, and used as donors N-acetylneuraminic acid, N-propionylneuraminic acid, and N-glycolylneuraminic acid. The two tethered enzymes were recovered by a simple procedure in functionally pure form, free from contaminating enzyme activity that can hydrolyze sugar nu-

25.5 ct3-Furosylationof Lactosamine Saccharides

651

cleotides or other components of the cofactor regeneration system. Accordingly, the fusion protein could be used in a sugar nucleotide cycle in which only catalytic amounts of expensive sugar nucleotides and transferase-inhibiting nucleoside phosphates are used; the former are continuously regenerated in situ from low cost precursors by appropriate enzyme reactions [56].In this way the fusion protein was used to produce a2,3-sialyllactose on a 100-g scale starting from lactose, sialic acid, phosphoenolpyruvate, and catalytic amounts of ATP and CMP. Taken together, the data of Gilbert et al. 15.51 show that production of the enzymes required for oligosaccharide synthesis might be simplified by fusing together carefully selected enzymes that catalyze neighboring steps in the synthetic processes.

25.5 a3-Fucosylation of Lactosamine Saccharides Gal(p1-4)[ Fuc(al-3)]GlcNAc- and NeuSAc(a2-3)Gal( pl-4)[Fuc(al-3)]GlcNAc sequences, known as Lewis’- and as sialyl Lewisx-determinants, respectively, are common distal elements in lactosaminoglycans. Lewis’ (Lex) is an adhesive sugar that can bind to other epitopes of its kind in the presence of Ca2+ [57],and sialyl Lewis’ (sLex) is the prototype ligand of the E-, P- and L-selectins, participating in leukocyte-endothelium adhesion, which is a process of considerable biomedical importance [58, 591. Also a3’-galactosylated and a2’-fucosylated LacNAc units at chain termini are acceptors for a3-fucosylation, and the products, i.e. Gal(a13)Gal(pl-4)[Fuc(a 1-3)IGlcNAc and Fuc(a 1-2)Gal( p I -4)[Fuc(ul-3)]GlcNAc, respectively, are also exciting cell adhesion molecules, participating in murine sperm-to-egg adhesion [60] and adhesion of Helicobacter pylorii to human stomach [61]. In man, at least five different transferases (FucT I11 to VII) catalyze a3-fucosylation. These reactions occur at distal LacNAc termini and also at mid-chain positions of polylactosamine extensions. Enzyme-assisted in vitro synthesis of a3-fucosylated LacNAc-glycans was initially performed with naturally occurring transferases, often with enzymes from human milk. For instance, de Vries et al. [62] showed that partially purified human milk a3/4 transferases (now assumed to be a mixture of Fuc TI11 and VI) reacted with all LacNAc units of LacNAc(~1-3’)LacNAc(~l-3’)LacNAcpl-OR converting the acceptor into the triply fucosylated product Lex(pl-3’)Lex(p1-3’)Lex~lOR [Lex, Lewis’, Gal(~1-4)[Fuc(ul-3)]GlcNAc]. More recently, Niemela et al. [63] studied partial u3-fucosylation of the free hexasaccharide LacNAc(pl-3’)LacNAc( pl -3’)LacNAc with a similar enzyme sample isolated from human milk. They found that among the three acceptor sites of the unconjugated hexasaccharide, the middle LacNAc reacted most rapidly and the non-reducing LacNAc most slowly. All individual isomers of the resulting monofucosyl- and difucosylderivatives were isolated in the pure form and characterized in these experiments. The ‘H NMR spectra of the monofucosyl hexasaccharides revealed distinct differences between the isomers.

658

25 Enzymatic Glycosylations with Glycosyltransferases

Conversion of the trisaccharide NeuSAc(a2-3’)LacNAc into the tetrasaccharide NeuSAc(a2-3’)Lex, was first performed using cloned human Fuc-TI11 [64]. In a similar experiment, de Vries et al. [65]incubated 6.6 pmol NeuSAc(a2-3’)LacNAc and 8.1 pmol GDP-[ ‘‘C]fucose for 96 h at 22 “C in 6 mL 50 mM sodium cacodylate buffer, pH 6.5 in the presence of 60 pmol MnC12 with 490 mU recombinant Fuc-TV immobilized uia protein A on 0.6 mL IgG-Sepharose. The resulting mixture was applied to a Bio-Gel P-2 column and eluted with 500 mM ammonium acetate, pH 6.8. The sialyl Lewis’ tetrasaccharide, eluting as a sharp peak at a position corresponding to 0.53 x Vtot, contained 81Y0 of the total [ 14C]fucosepresent in the reby action mixture. It was identified as Neu5Ac(a2-3)Gal(~l-4)[Fuc(al-3)]GlcNAc ID ‘ H NMR spectroscopy. Using NeuSAc(a2-3’)LacNAc( pl-3’)LacNAc( PI -3’)LacNAc as the acceptor Niemela et al. [54] showed that that the full length recombinant Fuc-TVII transfers primarily at the distal, sialylated LacNAc unit, converting the acceptor into a long chain [sialyl-Lexl-glycan. By contrast, recombinant Fuc-TIV transferred fastest to the two other LacNAc residues of this acceptor. Using Fuc-TIV and Fuc-TVII in a concerted manner, the authors synthesized in vitro the triply fucosylated determinant Neu5Ac(a2-3’)Lex(~l-3’)Lex(pl-3’)Lex,that is expressed on the P-selectin ligand PSGL-1 of HL-60 cells. Even polylactosamines carrying a3-fucosyl units at specific sites along the chains can be synthesized with enzyme preparations isolated from human milk, although these enzymes per se transfer in a rather random fashion. Fucosylation performed early, before the polylactosamine chains have been fully developed, leads to glycans which contain the Lewis’ epitope close to the reducing end [66, 671. Another possibility is to generate temporary GlcNAc branches to the linear i-type chains; the ensuing a3fucosylation reaction is prevented at the branch-bearing LacNAc units, but not elsewhere along the acceptor chain [68]. a3-Fucosyltransferase preparations isolated from human milk have also been used to convert two tetravalent [sialyl(a2-3’)LacNAc]-polylactosaminesinto the corresponding tetravalent [sialyl(a2-3’)Lex]-glycans [40,521. The fucosylation transfer was restricted to the distal, sialylated LacNAc units, because all other LacNAc residues carry branches. This restriction is probably general; all human a3fucosyltransferases seem to require the presence of a free hydroxyl group at position 6’ of the acceptor LacNAc units. As first described, these syntheses yielded ca 50-100 nmol of the final products, sufficient for Stamper-Woodruff binding experiments between lymphocytes and capillary-wall endothelium on tissue slices, MALDI-TOF mass spectrometry, and one dimensional H NMR spectroscopy. The tetravalent [sialyl-Lex] glycans proved to be very potent inhibitors of L-selectin mediated adhesion of lymphocytes to capillary endothelium of rejecting organ transplants [40, 52, 691. They also inhibited L-selectin-mediated adhesion of lymphocytes to peripheral lymph node high endothelial venules [70]. As L-selectin is expressed on lymphocyte surfaces in clustered arrays on tips of microvilli, we believe that the oligovalent [sialyl Lex] glycans might bind particularly tightly to lymphocyte L-selectin by crosslinking several monovalent L-selectin molecules in the cellular clusters. More recently, the syntheses of the two tetravalent [sialyl-Lex] glycans have been scaled up 20-50 fold, to

References

659

generate material for experiments involving in vivo inhibition of leukocyte rolling in mouse cremaster muscle capillaries (L. Penttila et al., unpublished work). Multifucosylated and terminally a3-sialylated polylactosamine chains of i-type, are also effective E-selectin ligands [71, 721. Using an enzyme sample purified from human milk, Toppila et al. [53] transferred four a3-fucose units to a bi-antennary array of five LacNAc units that was a3-sialylated at both long antennal. The resulting tetrafucosyl bisialosaccharide was a very potent inhibitor of L-selectinmediated lymphocyte-endothelium adhesion, both at capillaries of rejecting heart and at peripheral lymph nodes. Accordingly, the determinant Neu5Ac(a2-3’) Lex(pl-3’)Lex is recognized better than NeuSAc(a2-3’)Lex also by L-selectin. Of particular interest was that in the experiments of Toppila et al. (loc. cit.), the tetrafucosyl bisialosaccharide inhibitor was much more potent in its action at the rejecting heart than at the lymph node. This specificity of action suggests that by use of saccharide inhibitors of cell adhesion it might be possible to inhibit lymphocyte recruitment at inflamed tissues without endangering the normal lymphocyte patrolling that requires extravasation at lymph nodes. The molecular mechanisms responsible for this organ selectivity of the saccharide inhibitor remain unknown at present; one possibility is that endothelium of inflamed tissues carries less adhesionpromoting saccharides, or less active saccharides, than the endothelium of lymph nodes. References I. 2. 3. 4.

Laine, R. A (1994) Glycobiology, 4, 759-67. Leloir, L. F. (1971) Science 172, 1299-303. Sears, P. and Wong, C.-H. (1998) Cellular and Molecular Life Sciences 54, 223-252. Van den Steen, P., Rudd., P., Dwek. R . A., and Opdenakker, G. (1998) Crit. Rev. Biochem. Mol. Biol., 33, 151-208. 5. Jentoft, N (1990) Trends Biochem. Sci. 15, 291-94. 6. Elhammer, A. and Kornfeld, S. (1986) J. Biol. Chem. 261, 5249-55. 7. Wandall H. H., Hassan H., Mirgorodskaya E., Kristensen A. K., Roepstorff P., Bennett E. P., Nielsen P. A,, Hollingsworth M. A., Burchell J., Taylor-Papadimitriou J., Clausen H. (1997) J. Biol. Chem. 272, 23503-14. 8. Hagen, F. K. and Nehrke, K. (1998) J. Biol. Chem. 273, 8268-77. 9. Hagen, K. B., Hagen, F. K., Balys, M. M., Beres, T. M., Van Wuyckhuyse, B., Tabak, L. A. (1998) J. Biol. Chem. 273, 27749-54. 10. Hansen, J. E., Lund, O., and Brunak, S. (1997) Nucleic Acids Res. 25, 278-82. 11. Schachter H., McGuire, E. J. and Roseman, S. (1971) J. Biol. Chem. 246, 5321-28. 12. Mendicino, J., Sivakami, S., Davila, M. and Chandrasekaran, E. V. (1982) J. Biol. Chem. 257, 3987-94. 13. Cheng, P.-W and Bona, S. J. (1982) J. Biol. Chem. 257, 6251-58. 14. Amado M., Almeida R., Carneiro F., Levery S. B., Holmes E. H., Nomoto M., Hollingsworth M. A,, Hassan H., Schwientek T., Nielsen P. A., Bennett E. P., Clausen H. (1998) J. Biol. Chem. 273, 12770-08. 15. Williams D. and Schachter, H. (1980) J. Biol. Chem. 255, 11247-52. 16. Williams D., Longmore G. D.,~Matta, K. L. and Schachter, H. (1980) J. Biol. Chem. 255, 11253-61. 17. Kuhns, W., Ruth, V., Paulsen, H., Matta, K. L., Baker, M. A,, Barner, M., Granovsky, M., Brockhausen, I. (1993) Glycoconjugate J. 10, 381-394. 18. Bierhuizen, M. F. A. and Fukuda, M. (1992) Proc. Natl. Acad. Sci. USA, 89, 9326-30.

660

25 Enzymatic Glycosylations with Glycosyltransferases

19. Yeh, J-C., Ong, E. and Fukuda, M. (1999) J. Biol. Chem. 274, 3215-21. 20. Maaheimo, H., Penttila L. and Renkonen, 0. (1994) FEBS Lett. 349, 55-59. 21. Li, F., Wilkins, P. P., Crawley, S., Weinstein, J., Cummings, R. D. and McEver, R. P. (1996) J. Biol. Chem. 271, 3255-64. 22. Brockhausen I., Matta, K. L., Orr, J., and Schacter H. (1985) Biochemistry, 24, 1866-74. 23. Vavasseur, F., Yang, J.-M. Dole, K., Paulsen, H. and Brockhausen, I. (1995) Glycobiology, 5, 351-57. 24. Brockhausen, I., Orr, J. and Schachter, H. (1984) Can. J. Biochem. Cell Biol. 62, 1081-90. 25. Brockhausen, I., Williams, D., Matta, K. L., Orr, J. and Schachter, H. (1983) Can. J. Biochem. Cell Biol. 61, 1322-33. 26. Ujita, M., McAuliffe, J., Schwientek, T., Almeida, R., Hindsgaul, O., Clausen, H. and Fukuda, M. (1998) J. Biol. Chem. 273, 34843-49. 27. Yates, A. D. and Watkins, W. M. (1983) Carbohydrate Res. 120, 251-268. 28. Piller, F. and Cartron, J.-P. (1983) J. Biol. Chem. 258, 12993-99. 29. Hosomi, O., Takeya, A, and Kogure, T. (1984) J. Biochem. 95, 1655-59. 30. Podolsky D. K. (1985) J. Biol. Chem. 260, 8262-71. 31. Van den Eijnden, D. H. Winterwerp, H., Smeeman, P., and Schiphorst, W. E. C. M. (1983) J. Biol. Chem. 258, 3435-37. 32. Salo, H., Niemeli, R.. Ilves, K., Aitio, 0. and Renkonen, 0 (1998) XIXth International Carbohydrate Symposium, San Diego, Abstract BP 255. 33. Vilkman, A,, Niemela, R., Penttila, L., Helin, J., Leppanen, A,, Seppo, A,, Maaheimo, H., Lusa, S., and Renkonen, 0. (1992) Carbohydrate Res. 226, 155-74. 34. Kawashima, H., Yamamoto, K., Osawa, T. and Irimura, T. (1993) J. Biol. Chem. 268, 2711826. 35. Sasaki, K., Kurata-Miura K., Ujita, M., Angata, K., Nakagawa, S., Sekine, S., Nishi, T. and Fukuda, M. (1997) Puoc. Nutl. Acud. Sci. USA 94, 14294-99 36. Piller, F., Cartron, J-P., Maranduba, A,, Veyrieres, A,, Leroy, Y., Fournet, B. (1984) J. Biol. Chem. 259, 13385-90. 37. Seppo, A. Penttila, L., Makkonen, A,, Leppanen, A,, Niemela, R., Jantti, J., Helin, J., and Renkonen, 0. (1990) Biochem. Cell Biol. 68, 44-53. 38. Helin, J., Penttila, L., Leppanen, A., Maaheimo, H., Laauri, S. Costello, C. E., and Renkonen, 0. (1997) FEBS Lett. 412, 637-642. 39. Seppo, A,, Penttila, L., Niemela, R., Maaheimo, H., Renkonen, 0. and Keane, A. (1995) Biochemistry 34, 4655-61. 40. Seppo A,, Turunen, J.-P. Penttila, L., Keane, A. Renkonen, 0. and Renkonen, R. (1996) Glycobiology, 6, 65-71. 41. Slomiany, A. and Slomiany, B. L. (1980) Biochem. Biophys. Res. Commun. 93, 770-775 42. Leppanen A., Penttila, L., Niemela, R., Helin, J., Seppo, A,, Lusa, S., and Renkonen, 0. (1991) Biochemistry, 30, 9287-96. 43. Gu, J., Nishikawa, A,, Fujii, S., Gasa, S. and Taniguchi, N. (1992) J. Biol. Chem. 267, 2994-99. 44. Leppanen, A,, Salminen, H., Zhu, Y., Maaheimo, H., Helin, J., Costello, C. E. and Renkonen, 0. (1997) Biochemistry, 36, 7026-36. 45. Maaheimo, H., Rabina, J. and Renkonen, 0. (1997) Carbohydr. Res. 297, 145-51. 46. Salminen. H., Ahokas. K., Niemela, R., Penttila, L., Maaheimo, H, Helin, J., Costello, C. E. and Renkonen, 0.(1997) FEBS Lett. 419, 220-26. 47. Leppanen, A,, Zhu, Y., Maaheimo, H., Helin J., Lehtonen, E., Renkonen, 0. (1998) J. Biol. Chem. 273, 17399-405. 48. Bierhuizen, M. F. A. Mattei, M.-G. and Fukuda. M. (1993) Genes Dev. 7, 468-78. 49. Mattila, P., Salminen, H., Hirvas, L., Niittymaki, J., Salo, H., Niemela, R., Fukuda, M., Renkonen, 0. and Renkonen, R. (1998) J. Biol. Chem. 273.27633-39. 50. Sakamoto, Y. Taguchi, T. Tano Y., Ogawa, T., Leppanen, A., Kinnunen, M., Aitio, O., Parmanne, P., Renkonen, 0. and Taniguchi, N. (1998) J. Biol. Chem. 273, 27625-32. 51. Helin, J., Maaheimo, H., Seppo, A , Keane, A., Renkonen, 0. (1995) Carbohydrate Res. 266, 191-209. 52. Renkonen, 0, Toppila, S., Penttila, L., Salminen, H., Helin, J. Maaheimo, H., Costello, C. E., Turunen, J. P. and Renkonen, R. (1997) Glycobiology 7, 453-61.

References

661

53. Toppila, S., Renkonen, R. Penttila, L., Natunen, J., Salminen, H., Helin, J., Maaheimo, H. and Renkonen 0. (1999) Eur. J. Biochem. 261, 208-215. 54. Niemela R., Natunen, J., Majuri, M. L., Maaheimo, H., Helin, J., Lowe, J. B., Renkonen, O., and Renkonen, R (1998) J. Biol. Chem. 273, 4021--26. 55. Gilbert, M., Bayer R., Cunningham, A,-M., Defrees, S., Gao, Y., Watson, D. C., Young, N. M., and Wakarchuk, W. W. (1998) Nature Biotechnology, 16, 769-72. 56. Ichikawa, Y., Shen, G.-J., and Wong. C.-H. (1991) J. Am. Chem. SOC.113, 4698-700. 57. Eggens, I., Fenderson, B. A. Toyokuni, T., Dean, B., Stroud, M.; and Hakomori, S. (1989) J. Biol. Chem. 264, 9476-84. 58. Lasky, L. A. (1995) Annu. Rev. Biochem. 64, 113-39. 59. Springer, T. A. (1995) Annu. Rev. Physiol. 57, 827-72. 60. Johnston, D. S., Wright, W., Shaper, J. H., Hokke, C. H., van den Eijnden, D. H., Joziasse, D. H. (1998) J. Biol. Chem. 273, 1888-95. 61. Boren, T., Falk, P., Roth K. A,, Larson G., and Normark, S. (1993) Science, 262, 1892-95. 62. de Vries, T., Norberg, T., Lonn, H. and van den Eijnden, D. H. (1993) Eur. J. Biochem. 216, 769-77. 63. Niemela, R. Natunen, J. Penttila, L., Salminen, H., Helin, J., Maaheimo, H. Costello, C. E. and Renkonen 0, (1999) Glycobiology, 9: 517-526. 64. Dumas, D. P., Ichikawa, Y., Wong, C. H. Lowe, J. B. and Nair, R. P. (1991) Bioorg. Med. Chem. Lett. 8, 425-28. 65. de Vries, T. van den Eijnden, D. H., Schultz, J. and O’Neill, R. (1993) FEBS Lett., 330, 24348. 66. Kashem., M. A. Wlasichuk, K. B., Gregson, J. M. and Venot, A. P. (1993) Carbohydrate Res. 250. 129-44. 67. Rabina, J., Natunen, J., Niemela, R., Salminen, H., Ilves K., Aitio, O., Maaheimo, H., Helin, J. and Renkonen, 0. (1998) Carbohydrate Res. 305, 491-99. 68. Niemela, R., Rabina, J., Leppanen, A., Maaheimo, H Costello, C. E. and Renkonen, 0. (1995) Carbohydrate Res. 279, 331-38. 69. Turunen J. P., Majuri, M. L., Seppo, A., Seppo, A,, Tiisala, S., Paavonen, T., Miyasaka, M., Lemstrom, K., Penttila, L., Renkonen, 0. and Renkonen, R. (1995) J. Exp. Med. 182, 113342. 70. Toppila, S., Lauronen, J., Mattila, P., Turunen, J. P., Penttila, L., Paavonen, T., Renkonen, 0. and Renkonen, R. (1997) Eur. J. Immunol. 27, 1360-65. 71. Patel, T. P., Goelz, S. P., Lobb, R. R. and Parekh, R. B. (1994) Biochemistry, 33, 14815-24. 72. Stroud, M. R., Handa, K., Dalyan. M. E. K., Ito, K., Levery, S. B., Hakomori, S., Reinhold, B. B., and Reinhold V. N. (1996) Biochemistry, 35, 770-78.

Carbohydrates in Chemistry and Biology Editor Beat Ernst, Gerald W. Hart, Pierre Sinay copyright OWILEY-VCH Verlag GmbH, 2000

26 Recycling of Sugar Nucleotides in Enzymatic G1ycosylation Kathryn M. Koeller and Chi-Huey Wong

26.1 Introduction The sugar nucleotide-dependent glycosyltransferases are effective catalysts for the enzymatic synthesis of oligosaccharides, glycopeptides, and glycoproteins [ 11. These enzymes catalyze the formation of glycosidic linkages in a stereo- and regiospecific manner [2].The use of glycosyltransferases therefore largely eliminates the extensive protecting group manipulations required in the chemical synthesis of carbohydrates. Although in recent years, many of the glycosyltransferases and their sugar nucleotide substrates have become commercially available, two major drawbacks remain to utilizing glycosyltransferases for synthetic purposes. First and foremost, nucleoside diphosphates (NDPs) generated as products during transferase reactions inhibit the enzyme [3]. For micro-scale reactions this problem can be solved by addition of a phosphatase to break down the NDP product [4]. For preparative scale synthesis, however, this solution does not overcome the expense of stoichiometric quantities of sugar nucleotides. To circumvent these barriers, methods for recycling sugar nucleotides in enzymatic glycosylation have been developed [5].These multi-enzyme systems require that sugar nucleotides be present in only catalytic quantities, thereby eliminating product inhibition by NDPs and reducing expense. Regeneration strategies have been employed in multi-gram syntheses of complex carbohydrates in high yields. This chapter reviews the sugar nucleotide recycling systems utilized to date for carbohydrate synthesis and highlights related advances in the field.

26.2 Glycosyltransferasesof the Leloir Pathway and their Sugar Nucleotide Substrates A considerable number of eukaryotic glycosyltransferases have been cloned to date, each with characteristic linkage specificity and substrate preference [ 1b]. Leloir

26 Recycling of Sugar Nucleotides in Enzymatic Glycosylution

664

NH2

0.

0.

0.

no a-UDP-Glucose (UDP-Glc)

0. Ho

OH

OH

a-GDP-Mannose (GDP-Man)

“&

HO

HO

’ ” ! ! z p HOoH

Hof!)UDP

a-UDP-N-Acetylgluwsamine (UDP-GlcNAc)

a-UDP-Galactose (UDP-Gal)

0

P-GDP-L-Fucose (GDP-FUC)

CAo HOHO o& HoOUDP

PCMP-N-Acetylneuraminic acid (CMP-NeuAc)

Figure 1. Sugar nucleotide substrates for glycosyltransferases of the Leloir pathway.

pathway glycosyltransferases utilize sugar nucleotides as donor substrates, and are responsible for the synthesis of most cell-surface glycoforms in mammalian systems [6]. It is remarkable that such a broad range of enzymes has converged on eight general sugar nucleotides for use as glycosyl donor substrates (Figure 1). Glucosyl- and galactosyltransferases employ substrates activated with uridine diphosphate as the anomeric leaving group (a-UDP-Glc, a-UDP-GlcNAc, a-UDPGlcUA, a-UDP-Gal, a-UDP-GalNAc), whereas fucosyl- and mannosyltransferases utilize guanosine diphosphate (P-GDP-Fuc, a-GDP-Man). Sialyltransferases are unique in that the glycosyl donor is activated by cytidine monophosphate (PCMP-NeuAc). Preparative-scale syntheses of relevant sugar nucleotides have been developed previously [ 71. Sugar nucleotide recycling systems attempt to take advantage of the efficiency of biosynthetic pathways in enzymatic synthesis. In biological systems, simple monosaccharides are the precursors to sugar nucleotides (Figure 2) [6b]. In NDP-sugar biosynthesis an initial glycosyl phosphate is formed through kinase-mediated phosphorylation of a monosaccharide. Reaction of the glycosyl phosphate with a nucleoside triphosphate (NTP) is then catalyzed by the corre-

Gal

Glc

-

26.3 Design of Regeneration Systems

665

I--

Gal-i-P

-

e l I- , ---

Glc-6-P -Glc-1-P

GlcN-6-P

Man

J l

Glc Ac-i-P

4DP-Xyl

ManNAc

- -1NeuAc

1 UDP-GalNAc

Figure 2. Biosynthetic pathways to sugar nucleotides from simple monosaccharides.

sponding NDP-sugar pyrophosphorylase (eq. (1)). This results in the production of the activated NDP-sugar. In contrast, the nucleoside monophosphate (NMP)sugar CMP-NeuAc is formed by direct condensation of NeuAc with CTP, catalyzed by the enzyme CMP-NeuAc synthetase (eq. (2)). Sugar-1-phosphate NeuAc

+ CTP

+ NTP

+ NDP-sugar

+ CMP-NeuAc

+ PPi

+ PPi

(1)

(2)

26.3 Design of Regeneration Systems The similarities between glycosyltransferase reactions that employ NDP-sugars as substrates enable the construction of a nearly universal NDP-sugar recycling scheme (Scheme 1). In theory, the NDP that is released by the transferase can be

666

26 Recycling of Sugar Nucleotides in Enzymatic Glycosylation

NDP-sugar

pyrophosphorylase NDP-s4iar

NDP

NTpA Kinase

I!

Sugar-1-P

A;

Scheme 1. A general theoretical recycling system for NDP-sugars.

converted into the corresponding NDP-sugar in a two-step enzymatic process. Such a method would involve kinase-mediated phosphorylation of the NDP to convert it to an NTP. Condensation of the NTP with a glycosyl phosphate would then afford the activated sugar nucleotide. The regeneration of CMP-NeuAc is slightly more complicated, in that the product released by the transferase is an NMP rather than an NDP. Thus, an additional phosphorylation step and, likewise, an additional kinase are required before condensation of NeuAc with CTP can occur.

26.4 Practical Regeneration Systems In practice, the availability of the biosynthetic enzymes and substrates is often a limiting factor in the design of a sugar nucleotide recycling system. This generally makes the overall regeneration strategy more complex and lengthier than the twoenzyme theoretical scheme. The following sections describe recycling systems for sugar nucleotides that have found application in carbohydrate synthesis. 26.4.1 UDP-Galactose

In a system where UDP-Gal regeneration is the goal, the inclusion of a one-step conversion of UTP to UDP-Gal using the enzyme UDP-Gal pyrophosphorylase would be ideal. Unfortunately, this enzyme is not commercially available, nor is it easy to prepare from biological sources [8]. Luckily, an alternative biosynthetic pathway to UDP-Gal exists. In nature, UDP-Glc is a central intermediate for the biosynthesis of UDP-Gal and UDP-GlcUA, as well as other sugar nucleotides. The enzyme UDP-Glc pyrophosphorylase (UDPGP; EC 2.7.7.9) required for the conversion of UTP to UDP-Glc can be purchased. The UDP-Gal 4-epimerase

26.4 Practical Regeneration Systems

OH Ho

& O &:H

.

661

NHAcOH

A

/

NHAc

HO

HO

PPase

-

HoHO &$,.,OH

opo; OH

2 pi Scheme 2. Recycling of UDP-Gal employing UDP-Gal4-epimerase.

(UDPGE; EC 5.1.3.2) needed for the subsequent conversion of UDP-Glc to UDPGal is also readily available. A system employing these concepts was the first sugar nucleotide recycling method described in the literature. Wong et al. used this scheme to synthesize multigram quantities of N-acetyllactosamine (LacNAc) from N-acetylglucosamine (Scheme 2) [9]. This method has several salient features worth discussion that are not altogether obvious. For example, in the UDPGE-catalyzed reaction the equilibrium lies in the direction of the undesired product UDP-Glc. Continuous removal of UDP-Gal by pl,4-galactosyltransferase(P1,4-GalT; EC 2.4.1.22) drives the reaction in the direction of UDP-Gal production, however. The UDPGP substrate glucose-1-phosphate (Glc-1-P) is provided by the action of phosphoglucomutase (PGM; EC 5.4.2.2) on Glc-6-P, which is more stable and more readily available than Glc-1-P. Occasionally hexokinase (EC 2.7.1.1) can be employed to provide Glc-6-P from glucose. Finally, pyrophosphatase (PPase; EC 3.6.1.1) functions to remove the inorganic pyrophosphate product generated in the UDPGP reaction, thereby driving it forward. Other groups have used routes based on this initial concept for the synthesis of oligosaccharides [lo]. A similar recycling system was reported by Auge et al. for

668

26 Recycling of Sugar Nucleotides in Enzymatic Glycosylution

OH

HO

OR UDP-Gal

UDP

f

Scheme 3. Recycling of UDP-Gal employing sucrose synthetase for the one-pot synthesis of the a-gal epitope.

complex branched penta- and heptasaccharide synthesis [ 1I]. Thiem and Wiemann have also applied this strategy to the recycling of UDP-2-deoxy-Gal for the synthesis of 2’-deoxy-LacNAc [ 121. The UDP-Gal epimerase system was also one of the early methods to employ pyruvate kinase (PK; EC 2.7.1.40) for the conversion of UDP to UTP, with the simultaneous formation of pyruvate from phospho(eno1)pyruvate (PEP). Previous kinase systems employed were either more expensive (nucleoside diphosphate kinase(EC 2.7.4.6)/ATP) or were thermodynamically less favorable (acetate kinase(EC 2.7.2.l)/acetyl phosphate). Another potentially useful kinase system has recently been described. Noguchi and Shiba have utilized polyphosphate kinase (PPK) for the formation of UTP from UDP in a UDP-Gal recycling system [13]. Previously, PPK was shown to catalyze the reversible transfer of phosphate from ATP to ADP. Recently, it has been discovered that PPK will also accept other NDP and NTP substrates. For application to recycling systems, poly(phosphate) is more affordable than PEP as a phosphoryl donor. Synthetically, replacement of PK with PPK in the UDPGE-based recycling system efficiently provided LacNAc from GlcNAc on a multi-gram scale. The epimerase-based recycling system for UDP-Gal has been modified to contain the enzyme sucrose synthetase (EC 2.4.2.13) (Scheme 3) [14]. This enzyme catalyzes the formation of sucrose (Glc-al,2-Fru) from fructose with simultaneous conversion of UDP-Glc to UDP. The readily reversible nature of the synthetase reaction enabled the authors to employ it in the production of UDP-Glc. As previously, removal of UDP-Glc by UDPGE provided the desired forward driving force in this enzymatic cascade. Initially, this strategy was utilized for the production of LacNAc from GlcNAc. Hokke et al. then expanded this methodology to synthesis of the a-gal trisaccharide epitope, employing both a1,3-GalT and p1,4GalT in a one-pot reaction [15]. The success of this strategy relied on the fact that only the p1,4-GalT accepts the non-reducing terminal GlcNAc residue as a

26.4 Practical Regeneration Systems

HP
O

q HoOUDP

669

EzL IUTP

DVq #

“ I

I

Scheme 4. Recycling of UDP-Gal employing Gal-I-P uridyltransferase for the synthesis of LacNAc.

substrate. Subsequently, the Wang group also made use of a one-pot, two-GalT system for the synthesis of the a-gal pentasaccharide epitope [16]. In an alternative strategy for the regeneration of UDP-Gal, Wong et al. employed Gal- 1-P-uridyltransferase (Gal-1-P UT; EC 2.7.7.12) for the conversion of Gal-1-P to UDP-Gal (Scheme 4) [17]. This conversion is coupled to the formation of Glc-1-P from UDP-Glc by use of UDPGP. An attractive feature of this method is the utilization of galactose as starting material. This economical approach is possible through use of galactokinase (EC 2.7.1.6), an ATP-dependent enzyme that produces the galactosyl phosphate. LacNAc and several LacNAc derivatives have been synthesized using this recycling scheme. The Gal-1-P UT regeneration system has also been used in syntheses of glycopeptides and homogeneous glycoproteins. This UDP-Gal recycling system was combined with a subtilisin-based peptide synthesis to afford several disaccharidelinked glycopeptides [ 181. Furthermore, in a study of glycoprotein remodeling, the Gal-1-P UT method was employed to append galactose on to an N-linked GlcNAc on the surface of RNase B (Scheme 5) [19]. Further elaboration to the sialyl LewisX (sLe”) tetrasaccharide was then performed with u2,3-sialyltransferase (a2,3-SiaT) and ul,3-fucosyltransferase (al,3-FucT). Thus, recycling systems have shown their utility in simple saccharide synthesis and in the construction of complex glycopeptides and glycoproteins. 26.4.2 Other UDP-Sugars Although UDP-Glc has great utility as an intermediate in the regeneration of other sugar nucleotides, it also occasionally functions as the desired glycosyl donor.

610

26 Recycling of Sugar Nucleotides in Enzymatic Glycosylation

I -0 0

I5

26.4 Pructicul Regeneration Systems

67 1

Haynie and Whitesides have developed a regeneration scheme for UDP-Glc in the syntheses of sucrose and trehalose [20]. As previously, the sucrose synthetase conversion of UDP to UDP-Glc is coupled to the transfer of Glc from sucrose. The authors obtained the synthetase in partially purified form from wheat germ. UDPGlc is, furthermore, an intermediate in the biosynthesis of UDP-GlcUA, a donor substrate for P-glucuronosyltransferases (EC 2.4.1.17) [21]. For the recycling of UDP-GlcUA, the enzyme UDP-Glc dehydrogenase (EC 1.1.1.22) is required for the oxidation of the UDP-Glc 6-hydroxyl to the 6-carboxylic acid. This conversion is coupled with the reduction of 2 NAD+ to 2 NADH. Through this recycling system it was shown that GlcUA could be transferred to various phenolic acceptors. Instead of individually immobilized enzymes, Gygax et al. utilized a crude liver homogenate containing all the enzymes necessary for the regeneration scheme. Regenerations of UDP-GlcNAc and UDP-GlcUA have been successfully combined in the synthesis of a hyaluronic acid polymer of -1500 sugar residues (Scheme 6) [22]. In the presence of Glc-1-P and GlcNAc-1-P, both UDP-GlcUA and UDP-GlcNAc can be recycled. The presence of UDP-Glc dehydrogenase and lactate dehydrogenase are also required for the oxidation of UDP-Glc to UDPGlcUA. For this study, the hyaluronic acid synthase and most other enzymes were commercially available, although the UDP--GlcNAc pyrophosphorylase was obtained by overexpression in E. coli. A recycling system for UDP-GlcNAc has been reported for use with P1,6GlcNAc-transferase (EC 2.4.1.148) [23]. In this study, the authors attempted to use crude yeast cell extracts as the source of UDP-GlcNAc pyrophosphorylase (EC 2.7.7.23). Because of problems with contaminating phosphatases, however, this method was unsuccessful. Stepwise synthesis with whole yeast cells eventually provided the desired trisaccharide product. 26.4.3 CMP-NeuAc

The regeneration system for CMP-NeuAc is more complicated than that for NDPsugars (Scheme 7) [24]. An additional phosphorylation step must be incorporated, because CMP, a nucleoside monophosphate, is released after reaction with the sialyltransferase. For recycling purposes, nucleoside monophosphate kinase (NM K; EC 2.7.4.4) or myokinase (MK; EC 2.7.4.3) is added for the conversion of CMP to CDP. In this reaction, the phosphoryl donor is ATP. Subsequently, both CDP and ADP must be re-phosphorylated to CTP and ATP, respectively. Thus, for regeneration of CMP-NeuAc, an additional kinase and two equivalents of PEP are required. The condensation of NeuAc with CTP is catalyzed by CMP-NeuAc synthetase (EC 2.7.7.43). This system was used for the large-scale synthesis of 6’sialyl-LacNAc(6’-SLN) from LacNAc catalyzed by a2,6-SiaT (EC 2.7.7.43) in 97% yield. A recent modification of this recycling system was in the expression of an a2,3SiaT/CMP-NeuAc synthetase fusion protein [25]. This construct was more soluble than a2,3-SiaT alone, which is a poorly soluble transmembrane protein. The fusion

E7

B-co,-

PPi

-

r

J

l

UDP-GlcNAc

V

UDP-GlCUA

NADH

UTP

E6 = Lactate Dehydrogenase E7 = Inorganic Pyrophosphatase

Scheme 6 . Recycling of UDP-GlcUA and UDP-GlcNAc for the synthesis of a hyaluronic acid polymer.

HO& HO

Pi

NAD

x

= UDP-GlcNAcPyrophosphorylase

E5 = Pyruvate Kinuse

E4

E3 = UDP-Glc Pyrophosphorylase

El = Hyaluronic Acid Synthase E2 = UDP-Gk Dehydrogenase

0

R

m

ru

613

26.4 Practical Regeneration Systems

OH

H O W ; o 2 H

2Pi

€5

PPi

NeuAc

AcHN HO OH

Y CTP C

H

o OH w o

&

o

&

AcHN HO OH

OH

Ho

CMP

Scheme 7. Recycling of CMP-NeuAc for the synthesis of 3’-sialyl-LacNAc. El = a2,3-sialyltransferase; E2 = nucleoside monophosphate kinase; E.i = pyruvate kinase; Eq = CMP-NeuAc synthetase; Es = pyrophosphatase.

protein functioned efficiently at pH 7.5, even though the optimum pH for a2,3-SiaT (pH 6.0) and CMP-NeuAc synthetase (pH 8.5) differ significantly. This heterodimer was utilized in a 100-gram synthesis of 3/-sialyllactose. In another study, the CMP-NeuAc recycling system was coupled to UDP-Gal regeneration for a one-pot synthesis of 6’-SLN (Scheme 8) [26]. In the strategy reported by Ichikawa et al., NeuAc is generated from the reaction of ManNAc with pyruvate, catalyzed by NeuAc aldolase (EC 4.1.3.3). In turn, the pyruvate generated by the UDP-Gal and CMP-NeuAc recycling systems serves as a substrate for NeuAc aldolase. Remarkably, the multi-enzyme systems operated efficiently in tandem, without problems of product inhibition. Thus only the monosaccharide substrates GlcNAc, ManNAc, and Glc-1-P were required for the formation of the 6/-SLN trisaccharide. In principle, this one-pot multi-system strategy could be extended to the production of other complex oligosaccharides, such as the branched tetrasaccharide sLeX. The monosaccharide sialic acid remains relatively costly for preparative-scale synthesis. Although it has been demonstrated that NeuAc can be generated in situ by reaction of ManNAc with pyruvate, catalyzed by NeuAc aldolase [27], ManNAc is also relatively expensive and difficult to prepare. Therefore, a method for the generation of NeuAc from the inexpensive monosaccharide GlcNAc in a twoenzyme system might also prove useful in regeneration schemes. It has been shown that GlcNAc can be converted to ManNAc chemically [28], or enzymatically by the

OH NHAc

614

26 Recycling of Sugar Nucleotides in Enzymatic Glycosylution

PPi

6

noHO

on

NHAc

Scheme 8. Recycling of UDP-Gal and CMP-NeuAc for the one-pot synthesis of 6’-sialylLacNAc.

action of GlcNAc-2-epimerase [29]. Recently, an overexpression system for the epimerase was developed for industrial-scale synthesis [ 301. The resulting ManNAc product provided by the epimerase could then be condensed with pyruvate to produce NeuAc. CMP-NeuAc regeneration has also been coupled to p-galactosidase (EC 3.2.1.23) activity (Scheme 9). [31] The biological role of glycosidases is the hydro-

t

OH

ManNAc

CMP-NeuAc

Scheme 9. Combination of P-galactosidase and a2,3-SiaT for the synthesis of 3'-sialyl-LacNAc.

,OH

AcHN

Y

PEP + NeuAc

NHAc

676

26 Recycling of Sugar Nucleotides in Enzymutic Glycosylution

lysis of glycosidic linkages, but these enzymes can be made to catalyze the reverse reaction under appropriate conditions. As such, removal of product by a transferase can shift the equilibrium to the bond-forming direction. p-Galactosidases can utilize either the lactose disaccharide or a p-nitrophenyl galactoside as the galactosyl donor substrate for synthetic purposes [32]. In this respect, 0-galactosidase has been shown to be an effective catalyst in the synthesis of LacNAc and various derivatives thereof. In one system, LacNAc then served as a substrate for a2,6-SiaT in the production of 6’-SLN. Gambert and Thiem have, moreover, used this strategy in the synthesis of the sialylated Thomsen-Friedenreich antigen [33]. Compared with glycosyltransferases the use of glycosidases is advantageous in terms of stability, cost, and availability. As a drawback, glycosidases are also more promiscuous with regard to linkage specificity. Another system that couples glycohydrolase activity to glycosyltransferase activity has been reported. Ito and Paulson have developed a CMP-NeuAc recycling scheme that operates in conjunction with a trans-sialidase activity from Trypanosoma cruzi (Scheme 10) [34].This trans-sialidase catalyzes the reversible transfer of NeuAc from NeuAc(a2,3)Gal-OR to virtually any p-linked galactoside substrate. The CMP-NeuAc recycling system is employed to circumvent the expense of the sialylated pentasaccharide donor substrate. Synthetically, NeuAc is first transferred under catalysis by a2,3-SiaT. This pentato Gal(~l,3)-GlcNAc(~l,3)Gal(~1,4)Glc saccharide product is then a substrate for the trans-sialidase. Several novel sialylated molecules were produced by this technique.

26.4.4 GDP-Sugars The regeneration of GDP-Man has been employed in the synthesis of mannosebased oligosaccharides (Scheme 11) [35]. GDP-Man can be produced from Man-lP and GTP by the action of GDP-Man pyrophosphorylase (GDPMP; EC 2.7.7.22) in a manner analogous to previously described for NDP-sugars. The system described employed an a l ,2-mannosyltransferase (al,2-ManT) that was overexpressed in E. coli. The a1,ZManT accepted mannose, mannobiose, and 0-mannosyl glycopeptides as substrates. In the same fashion, GTP can be converted to GDP-Man as an intermediate in the recycling of GDP-Fuc (Scheme 12) [36]. This conversion can be accomplished by utilizing GDPMP from dried yeast cells and ‘GDP-Fuc-generating enzymes’ partially purified from the bacterium Klebsiella pneumonia. This system must be coupled to an alcohol dehydrogenase (EC 1.1.1.2), which catalyzes the oxidation of 2-propanol to acetone, along with the reduction of NADP+ to NADPH. Alternatively, Fuc-1-P can be biosynthesized from fucose by the action of fucokinase (EC 2.7.1.52) from porcine liver in the presence of ATP (Scheme 13). Reaction of Fuc-1-P with GTP catalyzed by GDP-Fuc pyrophosphorylase (EC 2.7.7.30) then affords GDP-Fuc [37]. Both of these methods have been utilized in the synthesis of sLeXfrom 3’-SLN.

26.4 Practical Regeneration Systems

QY

a

0

611

678

26 Recycling of Sugar Nucleotides in Enzymatic Glycosylation

zE

HO

HO HO

1

9

HO HO

NHCbz

f

0

OH

/

nu

GDP

HO

GTP

HO HO

OPO3'

Scheme 11. Recycling of GDP-Man for the synthesis of mannosyl glycopeptides. OH

HO OH

Sialyl Lex a1 ,&FucT

pyrophosphorylase PPiGDP-Fuc

ActlN

GW-Fuc

GDp-Man

* SynmeSIllng

OH

NHAc

NADPH

p:

dehydroge-

NADP

A

Scheme 12. Recycling of GDP-Fuc through GDP-Man for the synthesis of sLe"

PPase

2P

26.4 Practical Regeneration S y s t e m

679

680

26 Recycling of Sugar Nucleotides in Enzymatic Glycosylation

3-phosphoglycerate phosphatase

I

phosphogfucose isomerase

phosphate 4-=

Scheme 14. Multi-enzyme synthesis of D-fructose from starch.

In situ recycling methods for GDP-Fuc are not yet optimized for preparativescale synthesis. This is mainly because of difficulty obtaining adequate quantities of the key biosynthetic enzymes. GDP-Man pyrophosphorylase, ‘GDP-synthesizing enzymes’, fucokinase, and GDP-Fuc pyrophosphorylase are currently not available commercially.

26.4.5 Other Carbohydrate-Based Regeneration Systems

Other recycling systems that do not involve the synthesis of sugar nucleotides are also relevant in carbohydrate synthesis. For example, the production of fructose from starch has been accomplished in a five-enzyme system (Scheme 14) [38]. Conversion of starch to Glc-1-P is catalyzed by phosphorylase a. Isomerization to Glc6-P, and further conversion to Fru-6-P is then achieved by phosphoglucomutase and phosphoglucose isomerase, respectively. A trans-aldolase then converts Fru-6-P to fructose, with the simultaneous conversion of glycerol to 3-phosphoglycerate. The action of 3-phosphoglycerate phosphatase then irreversibly hydrolyses the phosphate, providing the forward driving force, and completing the cycle by recycling inorganic phosphate. Sulfation of carbohydrates is a topic of growing interest. The first recycling system used for the sulfation of carbohydrates was reported by Lin et al. The regeneration of 3’-phosphoadenosine 5’-phosphosulfate (PAPS), the universal biological sulfuryl donor, was accomplished by a multi-enzyme cascade (Scheme 15) [39]. In conjunction with a Nod factor sulfotransferase, N,N’-diacetylchitobiose was sulfated at the reducing terminal GlcNAc 6-hydroxyl. This sulfation system was also used to generate 6-sulfo-LacNAc, a key intermediate in the preparation of 6-sulfo-sLe' . A newly developed sulfation system accomplishes a similar goal, while also reducing the number of coupling enzymes necessary for the regeneration of PAPS (Scheme 16) [40]. In this protocol, PAPS is recycled from PAP through catalysis

J

0

n

- '\ -

n

.

HO

.\I OH

.

JCOi

0

ope,*-

E1+mco,

ATP

ir,

Scheme 15. Recycling of PAPS for the synthesis of sulfated carbohydrates.

M A O Y

O

ecoi\/*Dp

6- (PAPS)

~~

GDP-FUC al,3-FucT

~

CMP-NeuAc a2.3-SiaT

E3: Pyrophosphatase

E2: Myokinase

El: Pyruvate Kinase

SLex 6-6ulfate

Q

P

hJ

682

26 Recycling of Sugar Nudeotides in Enzymatic Glycosylation

HO

HO

0

OH

PAPS

0

OH

PAP

I 0

Scheme 16. Two-enzyme system for the recycling of PAPS.

by a single enzyme, aryl sulfotransferase. Because the only cofactor in this scheme is p-nitrophenol sulfate, the regeneration system has been greatly simplified, and is much more cost-effective. Chitotriose was found to be the best acceptor for the sulfotransferase. Notably, although LacNAc was a poor substrate, LacNAc(bl,3)LacNAc was a good acceptor in this system.

26.5 Conclusion Methods for the regeneration of sugar nucleotides in enzymatic glycosylation have proven extremely useful for synthetic purposes. Enzymatic synthesis can be both efficient and cost-effective in an appropriate system. Although recycling schemes for sugar nucleotides such as UDP-Gal and CMP-NeuAc are well-stocked by a commercial arsenal, others have not been as readily developed. For example, GDP-Fuc and UDP-GlcNAc regenerating schemes have yet to prove effective on a preparative-scale, because of lack of availability of the relevant biosynthetic enzymes. This field is, however, actively maturing with the continuous development of novel en-

References

683

zymes for use in recycling schemes. Complex substrates, such as glycopeptides, glycoproteins, and sulfated oligosaccharides, are currently being pursued to further the utility of recycling systems in carbohydrate synthesis.

References I. a) Y. Ichikawa in Glycopeptides und Related Compounds (Eds. D. G. Large, C. D. Warren), Marcel Dekker, New York, 1997, p. 79; b) P. Sears, C.-H. Wong, Cell Mol. Life Sci. 1998. 54, 223; c) C.-H. Wong, R. L. Halcomb, Y. Ichikawa, T. Kajimoto, Angew. Chem. Znt. Ed. Engl. 1995, 34, 412 and 521. 2. A. T. Beyer, J. E. Sadler, J. I. Rearick, J. C. Paulson, R. L. Hill, Adv. Enzymol. 1981, 52, 24. 3. B. W. Weston, R. P. Nair, R. D. Larsen, J. B. Lowe, J. Biol. Chem. 1992, 267, 4152. 4. C. Unverzagt, H. Kunz, J. C. Paulson, J. Am. Chen7. Soc. 1990, 112, 9308. 5. Y. Ichikawa, R. Wang, C.-H. Wong, Methods Enzymol. 1994, 247, 107. 6. a) L. F. Leloir, Science 1971, 172, 1299; b) R. Kornfeld, S. Kornfeld, Ann. Rev. Biochem. 1985, 54, 63 1. 7. C.-H. Wong, G. M. Whitesides, Enzymes in Synthetic Organic Chemistry, Pergamon, Oxford, 1994. D. 252. 8. S. E.'Heidlas, W. J. Lees, G. M. Whitesides, J. Org. Chem. 1992, 57, 152. 9. C.-H. Wong, S. L. Haynie, G. M. Whitesides, J. Org. Chem. 1982, 47, 5418. 10. a) M. M. Palcic, 0. P. Srivastava, 0. Hindsgaul, Curbohydr. Res. 1987, 159, 315; b) C. Auge, S. David, C. Mathieu, C. Gautheron, Tetrahedron Lett. 1984, 25, 1467. 11. C. Auge, C. Mathieu, C. Merienne, Curbohydr. Res. 1986, 151, 147. 12. J. Thiem, T. Wiemann, Angew. Chenz. Int. Ed. Engl. 1991, 30, 1163. 13. T. Noguchi, T. Shiba, Biosci. Biotechnol. Biochem. 1998, 62, 1594. 14. L. Elling, M. Grothus, M.-R. Kula, Glycohiology 1993, 3, 349. 15. C. H. Hokke, A. Zervosen, L. Elling, D. H. Joziasse, D. H. van den Eijnden, Glycoconjugate J. 1996, 13, 687. 16. J. Fang, J. Li, X. Chen, Y. Zhang, J. Wang, Z. Guo, W. Zhang, L. Yu, K. Brew, P. G. Wang, J. Am. Chem. Soc. 1998, 120, 6635. 17. C.-H. Wong, R. Wang, Y. Ichikawa, J. Org. Chem. 1992, 57, 4343. 18. C.-H. Wong, M. Schuster, P. Wang, P. Sears, J. Am. Chem. Soc. 1993, 115, 5893. 19. K. Witte, P. Sears, R. Martin, C.-H. Wong, J. Am. Chem. Soc. 1997, 119, 2114. 20. S. L. Haynie, G. M. Whitesides, Appl. Biochem. Biotech. 1990,23, 155. 21. D. Gygax, P. Spies, T. Winkler, U. Pfaar, Tetrahedron 1991, 47, 5119. 22. C. DeLuca, M. Lansing, 1. Martini, F. Crescenzi, G.-J. Shen, M. O'Regan, C.-H. Wong, J. Am. Chem. Soc. 1995, 117, 5869. 23. G. C. Look, Y. Ichikawa, G.-J. Shen, P.-W. Cheng, C.-H. Wong, J. Org. Chem. 1993, 58, 4326. 24. Y. Ichikawa, G.-J. Shen, C.-H. Wong, J. Am. Chem. SOC.1991,113, 4698. 25. M. Gilbert, R. Bayer, A.-M. Cunningham, S. DeFrees, Y. Gao, D. C. Watson, N. M. Young, W. W. Wakarchuk, Nat. Biotechnol. 1998, 16, 769. 26. a) Y. Ichikawa, J. L.-C. Liu, G.-J. Shen, C.-H. Wong, J. Am. Chem. Soc. 1991, 113, 6300; b) P. Stangier, W. Treder, J. Thiem, GlycoconjuguteJ. 1993, 10, 26. 27. a) C. Auge, S. David, C. Gautheron, Tetruhedron 1984,25, 4663; b) M.-J. Kim, W. J. Hennan, H. M. Sweers, C.-H. Wong, J. Am. Chem. SOC.1992,114, 10138. 28. E. S. Simon, M. D. Bednarski, G. M. Whitesides, J. Am. Chem. Soc. 1988, 110, 7159. 29. U. Kragl, D. Gygax, 0. Ghisalba, C. Wandrey, Angew. Chem. Int. Ed. Engl. 1991, 30, 82. 30. I. Maru, J. Ohnishi, Y . Ohta, Y. Tsukada, Carbohydr. Res. 1998, 306, 575. 31. G. F. Herrmann, Y. Ichikawa, C. Wandrey, F. C. A. Gaeta, J. C. Paulson, C.-H. Wong, Tetrahedron Lett. 1993, 34, 3091. 32. S. Takayama, M. Shimazaki, L. Qiao, C.-H. Wong, Bioorg. Med. Chem. Lett. 1996, 6 , 1123. 33. U. Gambert, J. Thiem, Eur. J. Org. Chem. 1999, 1, 107.

684

26 Recycling of Suyur Nucleotides in Enzymatic Glycosylation

34. Y. Ito, J. C. Paulson, J. Am. Chem. SOC.1993, 115, 7862. 35. P. Wang, G.-J. Shen, Y.-F. Wang, Y. Ichikawa, C.-H. Wong, J. Ory. Chem. 1993, 58, 3985. 36. Y. Ichikawa, Y.-C. Lin, D. P. Dumas, G.-J. Shen, E. Garcia-Junceda, M. A. Williams, R. Bayer, C. Ketcham, L. E. Walker, J. C. Paulson, C.-H. Wong, J. Am. Chem. SOC.1992, 114, 9283. 37. See also R. Stiller, J. Thiem, Liebigs Ann. Chew. 1992, 5, 467. 1992, 114, 6980. 38. A. Moradian, S. A. Benner, J. An?. Chem. SIC. 39. C.-H. Lin, G.-J. Shen, E. Garcia-Junceda, C.-H. Wong, J. Am. Chem. Soc. 1995, 117, 8031. 40. M. D. Burkart, M. Izumi, C.-H. Wong, Anyew. Clzrm. Int. Ed. Enyl. 1999, 38, 2147.

Carbohydrates in Chemistry and Biology Editor Beat Ernst, Gerald W. Hart, Pierre Sinay copyright OWILEY-VCH Verlag GmbH, 2000

27 Enzymatic Glycosylations with Non-Natural Donors and Acceptors Xiangping Qian, Keiko Sujino, and Monica M. Palcic

27.1 Introduction The demonstration that oligosaccharides can serve as recognition markers in diverse biological events [ I , 21 has stimulated interest in the synthesis of oligosaccharides and their analogs [3-71. The availability of such molecules can provide further insights into their biological functions and might lead to the discovery of novel carbohydrate-based therapeutics [ 81. Despite many advances in the chemical synthesis of oligosaccharides, it still remains a challenge, especially for the preparation of analogs where additional steps must be employed in the synthetic route to achieve a modification at a specified position [4]. As well, a replication system for amplifying minute amounts of carbohydrates is not currently available, nor is instrumentation commercially available for the solid-phase synthesis of oligosaccharides. Oligosaccharides are biosynthesized in vivo by glycosyltransferases that sequentially transfer a single pyranosyl residue from a sugar nucleotide donor to a growing carbohydrate chain [9- 121. Enzymatic synthesis using glycosyltransferase has several advantages over the chemical synthesis of oligosaccharides-high regio- and stereoselectivity, no requirement for chemical protection and deprotection, and very mild reaction conditions [ 13-1 51. Although there are more than 100 different mammalian glycosyltransferases, each biosynthesizing a unique glycosidic linkage, only nine main nucleotide donors are used as building blocks by these glycosyltransferases to construct the diverse and complex oligosaccharides found in mammals. Recognition of the donor substrates by glycosyltransferases is primarily based on the nucleotide portion. Nonnatural oligosaccharide analogs can be prepared by using glycosyltransferases to transfer a modified monosaccharide from donors. Studies probing the specificity of glycosyltransferases using acceptor analogs where the hydroxyl group is replaced with H, OCH3, NH2 or other substituents, have helped to establish structural requirements for acceptor recognition [16]. It has been found that only a few of the

686

27 Enzymatic Glycosylations with Non-Natural Donors and Acceptors

hydroxyl groups of the acceptor, termed ‘key polar groups’ [ 17- 191, are necessary for binding to glycosyltransferases. Chemical modifications can therefore be introduced at positions other than those bearing key polar groups to produce nonnatural acceptor precursors for enzymatic synthesis. With advances in molecular biology and biotechnology more than 30 glycosyltransferases have been cloned and many glycosyltransferases are now readily available in quantities sufficient for use in in vitro synthesis [20]. The increasing availability of glycosyltransferases along with their flexibility in the recognition of both donor and acceptor substrates makes enzymatic glycosylation with non-natural donors and/or non-natural acceptors an increasingly practical alternative for the preparation of non-natural oligosaccharides [ 16, 2 1-24]. The term ‘non-natural’, in the current context, is used to indicate that the acceptor or donor is chemically modified and different from the natural ones that are used by glycosyltransferases. Biosynthetically glycosyltransferases act on very large and complex glycoproteins and glycolipids, therefore small molecule acceptors might be viewed as non-natural. In this contribution, however, we do not consider a change of aglycone or adding/ removing a sugar unit(s) at the reducing end as creating a non-natural acceptor.

27.2 Enzymatic Glycosylations 27.2.1 Galactosylations

p1,4-Galactosyltransferase ~1,4-Galactosyltransferase(Pl,4-GalT, E.C. 2.4.1.22/38/90) has been commercially available for many years in unit quantities. It is the most widely studied glycosyltransferase with regard to substrate specificity and use in preparative synthesis. It catalyzes the transfer of Gal with inversion of configuration from UDP-Gal to OH4 of terminal P-linked GlcNAc to form N-acetyllactosamine. In the presence of alactalbumin, the enzyme prefers to use glucose as an acceptor to produce lactose. Non-natural donors ~l,4-Galactosyltransferasehas been shown to tolerate modifications on any OH group of the Gal residue of UDP-Gal donors. It transfers 2-deoxy-Gal at a rate similar to Gal [25-281. 3-Deoxy [29], 4-deoxy [30], 6-deoxy [31, 321, and 6-deoxy-6fluoro Gal [31, 321 are also transferred. The enzyme can also utilize UDP-Ara [30, 331, UDP-GalNAc [34], UDP-GalNH2 [27], UDP-Glc [30, 35, 361, and UDPGlcNH2 [34] as donors (Scheme 1). Interestingly, UDP-5’-thio-Gal, with the ring oxygen of Gal replaced by a sulfur atom, was active as a donor substrate for 01,4-GalT [37]. UDP-5’-thio-GalNAc was also found to be a substrate in the presence of lactalbumin [38].

27.2 Enzymatic Glycosylations

687

HO

H

I

H, NHz, NHAc

NHAc

I Non-Natural Donors I OMe, 0-Fucose, 0-Sialic acid methyl ester, H, F, SH

r-

A

H, 0x0, 0-Ally1

HO

OR

HO

NHAc

1

N33 NHAcyl

OH

HO

NHAc

OH I-Deoxy-nojirimycin

aHoaO 4 HO&

AcHN OH (-) Conduritol B

Glucal

a

" 1

" 1

H0-L

OH

AcHN

AcHN

OH

OH

-.

nu,

OH

OH

OH

I Non-Natural Acceptors Scheme 1. Non-natural donors and acceptors for P1,4-GalT.

Non-natural acceptors

As shown in Scheme 1, the acceptor specificity of B174-GalTis extremely relaxed because numerous modified GlcNAc analogs are active as acceptors. Basically, Pl,4-GalT tolerates modifications everywhere on the sugar ring, including the ring oxygen, as long as the 4-OH remains available for glycosylation.

688

27 Enzymatic Glycosylations with Non-Natural Donors and Acceptors

The 2-NHAc group can be replaced with azido [27, 391, N-propanoyl [40, 411, N-butanoyl [40], allylcarbamate [42], and many amide derivatives [43] including charged groups, highly bulky heterocycles, and glycuronamides. 2-Ethylamino-2deoxy, 2-N-methylacetamido-2-deoxy, and 2-O-acetyl-P-~-glucosidewere, however, inactive as acceptors and inhibitors for P1,4-GalT [44]. Analogs with the 3-OH group deoxygenated [44, 451, alkylated with a methyl or ally1 group, or oxidized to the ketone are active as acceptors although the relative rates of transfer are much lower than to N-acetylglucosamine [45]. The 6-OH group of GlcNAc can be methylated [41, 451, fucosylated [41], deoxygenated [44], or substituted with F or SH [44]. Although addition of a-linked sialic acid to 6-OH of GlcNAc is not tolerated [41], when the carboxylic acid of NeuAc is derivatized to the methyl ester, the resulting compound proved to be a weak acceptor for p1,4-GalT. Although the relative rate of transfer is only 4% that for GlcNAc, this is sufficient for preparative synthesis and generation of product [411. The great tolerance of pl,4-GalT for acceptor modifications is further exemplified by its transfer of Gal from UDP-Gal to 5’-thio-Glc and l-deoxynojirimycin, which have the ring oxygen modified, and to glucal, which has a flattened ring conformation [45]. The enzyme can even resolve racemic (+) conduritol B to give a single galactosylated product of (-)-conduritol B [46]. More interestingly, the enzyme transfers galactose to the p-anomeric position of 3-acetamido-3-deoxy-~glucose acceptors resulting in the formation of an unusual pl-1 (trehalose type) linkage [47]. This ‘frame-shifted’ galactosylation [48] was also observed with Nacetylgentosamine [49], N-acetyl-5’-thiogentosamine[ 501, and xylose [ 5 11 acceptors. A large variety of immobilized acceptors has also been employed in preparative reactions with Pl,4-GalT [52-571. al,3-Galactosyltransferase

a1,3-Galactosyltransferase (al,3-GalT, E.C. 2.4.1.151) catalyzes the transfer of Gal with retention of configuration from UDP-Gal to 3-OH of the Gal residue in Gal(P1-4)GlcNAc-R to form Gal(al-3)Gal(Pl-4)GlcNAc epitopes [58-611, the major xenoactive antigens responsible for hyperacute rejection in xenotransplantation [62].An enzyme for preparative synthesis has been isolated from porcine and bovine tissues and recombinant porcine al,3-galactosyltransferase is now commercially available in unit quantities. Non-natural donors Unlike pl,4-GalT, al,3-galactosyltransferase has a rather restricted specificity for the nucleotide donor substrate as revealed by studies with recombinant murine al,3-GalT [63]. Although UDP-Glc, UDP-GalNAc, UDP-GlcNAc, and UDPglucuronic acid are not substrates for the enzyme [63], UDP-2-deoxy-Gal is a better substrate than UDP-Gal. The relative rates of transfer of 3-deoxy, 4-deoxy, or 6deoxy donors are 0.2, 0.6 and 2% that of UDP-Gal [64].

27.2 Enzymatic Glycosylutions

689

I Non-Natural DonorsI

H, OCHzR

OH, N3! Succinimido, NHCHO, NHAcyl

1

6 -

HO

OH

OR

H, OCH2R

N HAc

v

H -

Non-Natural Acceptors Scheme 2. Non-natural donors and acceptors for ctl,3-GalT.

Non-natural acceptors

Besides using Gal(p1-4)GlcNAc as an acceptor, the enzyme can transfer Gal to Gal(plL3)GlcNAcand Gal(al-4)Glc [65]. Substitutions of 2-NHAc of the GlcNAc residue with azido and succinimido groups are tolerated [63]. The N-acetyl group can be replaced with many acyl groups of various sizes, hydrophilicities, or lipophilicities [66], although replacement of the 2-NHAc with an amino group abolishes activity [63]. Deoxygenation of the 3-OH of the GlcNAc residue is tolerated whereas any substitution or derivatization at this position is not. As shown in Scheme 2, analogs with modifications (deoxygenation and 0-alkylation) on 6-OH of GlcNAc, or 2-OH, and 6-OH of the terminal Gal residue are substrates [67]. Modification on 4-OH of Gal is not tolerated by the enzyme, suggesting that this group is a key polar group essential for binding to a1,3-GalT [67]. The 4-deoxy analog of Gal(p1-3)GlcNAcp-OR was found to be as active as Gal(p1-3) GlcNAcp-OR [67].

OR

690

27 Enzymatic Glycosylations with Non-Nutural Donors and Acceptors

27.2.2 Fucosylations The fucosylated oligosaccharides on cell surfaces are involved in numerous intercellular recognition events [ 11. al,3/4-Fucosyltransferasescatalyze the transfer of a fucose residue with retention of configuration from GDP-fucose to a variety of acceptors to form the blood-group related antigenic determinants Lea and LeX[68]. al,3/4-Fucosyltransferasesare a multigene family of enzymes with different substrate specificities toward Gal(p1-4)GlcNAc, Gal(pl-3)GlcNAc, and NeuAc(a23)Gal(p1-4)GlcNAc acceptors, inhibitor sensitivity, pH optima, and tissue distributions [12]. The most common source for isolation of a preparatively useful fucosyltransferase is human milk [ 101. There are two different human milk fucosyltransferases, al,3/4-fucosyltransferase and al,3-fucosyltransferase.Five different human al,3/4-fucosyltransferaseshave been cloned, FucT 111 to FucT VII, and both FucT V and VI are commercially available. Human Milk a1,3/4-Fucosyltransferase Non-natural donors

GDP-3-deoxyfucose, GDP-arabinose and GDP-L-galactose are active as donor substrates for human milk a1,3/4-FucT (E.C. 2.4.1.65) when Gal(plL3)GlcNAcpO(CH2)gCOOMe is used as an acceptor [69]. GDP-3-deoxy-Fuc and GDP-L-Gal were shown to be preparatively useful, using Gal( pl-4)GlcNAcp-O(CH2)&OOMe as an acceptor for human milk a1,3/4-FucT [70]. The enzyme was also shown to tolerate the addition of a propyl group at 6-OH of L-Gal in GDP-L-Gal. Large substituents, e.g. tethered blood group B trisaccharide, can be introduced at C-6 of the Fuc residue in GDP-Fuc [71].Essentially, any group can be attached to the C-6 position of FUC,as was demonstrated by the transfer of modified fucose residues containing biotin and blood group A trisaccharide biolabels (Scheme 3 ) [72]. Non-natural acceptors

Human milk a1,3/4-FucT uses both Gal(p1-3)GlcNAc (type I) and Gal(p14)GlcNAc (type 11) acceptors. Chemical mapping studies with a series of monodeoxygenated and modified acceptor substrates showed that modifications are tolerated at every hydroxyl group in the sugar rings except 6-OH of the Gal and 3- or 4-OH of the GlcNAc residue to which fucose is transferred [73].The 2-NHAc group of the GlcNAc residue in NeuAc(a2-3)Gal(~l-4)GlcNAc~-O(CH~)~COOMe or NeuAc(a2-3)Gal(~1-3)GlcNAc~-O(CH~)~COOMe can be replaced with azido, amino, or propionamido groups [74].Thio-linked N-acetyllactosamine, in which the inter-glycosidic oxygen is replaced by sulfur, is also a good acceptor for the enzyme [75]. Ether- and imino-linked octyl N-acetyl-5a’-carba-~-lactosamides were also found to be acceptors for human milk al,3/4-FucT [76]. Surprisingly, human milk al,3/4-FucT can even tolerate the introduction of a large methyl group directly at the site of fucosylation; it transfers fucose to the hindered tertiary alcohol acceptor 3-C-methyl-N-acetyllactosamine[ 771. This unexpected finding demonstrated that glycosyltransferases can overcome inherent steric limitations and can be used to

27.2 Enzymatic Glycosylations

CHPOH, H &OH

691

0-GDP

I I

CH20CH2CH2CH3

0 aGal( 1-13),

aFuc(l+2)

0

II

L N , C 6 - F u c PGal-0 H

,

I Non-Natural Donors I

NHAc OR

OR

NH

S

HO

H AcHN O W

i

@0 O

o

%

OH

NHAc

HO

I Non-Natural Acceptors1 Scheme 3. Non-natural donors and acceptors for human milk a1,3/4-FucT.

OR

692

27 Enzymutic Glycosylations with Non-Natural Donors and Acceptors

produce oligosaccharide analogs that cannot easily be prepared by chemical methods. FucT I11 and VI FucT I11 transfers fucose on to the 4-OH group of GlcNAc residues of Gal(b1-3)GlcNAc or NeuAc(a2-3)Gal( PI-3)GlcNAc to form Lea or sialyl Lea respectively, whereas FucT VI uses Gal(PI-4)GlcNAc or NeuAc(a2-3)Gal(~l-4)GlcNActo form LeXor sialyl Le" products. Despite their close sequence homologies, they behave differently in their recognition of non-natural donors [78]. FucT 111 transfers LGal, arabinose, L-Glc, 2-amino-2-deoxyfucose, and 2-fluoro-2-deoxyfucosewhereas FucT VI cannot tolerate modifications on the 2-OH group of the fucose. FucT VI still transfers L-Gal and arabinose, however. Both enzymes tolerate diverse replacements of the N-acetyl group of the GlcNAc unit, as seen with Pl,4-GalT [78SO]. FucT VI has been shown to tolerate replacement of the GlcNAc unit with glucal and even with cyclohexane diol[24]. The utility of FucT I11 and FucT VI has further been expanded to enable construction of libraries of sialyl Lea or sialyl Le" derivatives using both non-natural donors and non-natural acceptors (Schemes 4 and 5) [Sl-831. FucT I11 and FucT VI have been probed with non-natural donors where the purine base was modified. Both enzymes can tolerate exchange of the guanine by other purine bases such as adenine [84]. These non-natural donors proved to be preparatively efficient in the enzymatic synthesis of Lea or Le", suggesting that costly sugar nucleotide donors can be replaced with inexpensive ones to reduce the cost of enzymatic synthesis. FucT V Similar to human milk a1,3/4-FucT, FucT V requires the 6-OH group of Gal and OH-3 or OH-4 of GlcNAc for substrate binding. Gal(P1-4)Glucal [39], Gal(P1-4) (-5-S)-Glc [39], 3-sulfo-Gal(Pl-4)GlcNAc [85] and Gal(~l-4)-6-sulfo-GlcNAc[85] are acceptors for FucT V. 27.2.3 Sialylations

Cell-surface sialic acid residues play important roles in diverse biological processes [86]. Sialic acids are usually found in terminal positions linked through an a-glycosidic linkage. The stereoselective synthesis of a-sialosides remains a challenge because the glycosides have the thermodynamically unfavorable equatorial orientation and the anomeric carbon is a very hindered quaternary center [4]. Enzymatic sialylation is, therefore, an attractive alternative means of preparation of a-sialosides in an efficient and stereocontrolled manner. a2,3-Sialyltransferase and a2,6-Sialyltransferase

a2,3-Sialyltransferase and a2,6-sialyltransferase from rat liver (a2,3-SialT, E.C. 2.4.99.6; a2,6-SialT, E.C. 2.4.99.1) have been cloned and expressed and are now

27.2 Enzymatic Glycosylations

693

{f: N

L-Glc

I

I CHPOH,H

OH OH

1 Non-Natural Donors I

HOWi&\EJ HO

AcHN

OH

HO

NHAc

NHAcyl

OR

I

I Non-Natural Acceptors) Scheme 4. Non-natural donors and acceptors for FucT 111.

commercially available in quantities sufficient for use in preparative synthesis. a2,3SialT from rat liver transfers a sialic acid unit from CMP-sialic acid to OH-3 of the terminal Gal residue in Gal(P1-4)GlcNAc or Gal(pl-3)GlcNAc sequences whereas rat liver a2,6-SialT transfers the sialic acid to OH-6 of the terminal Gal residue in Gal@-4)GlcNAc [87]. Non-natural donors

The finding that CMP-sialic acid synthetase can accept many sialic acid analogs has facilitated the synthesis of CMP-sialic acid analogs for studies of the donorspecificity of these enzymes [88-901. Both a2,3-SialT and a2,6-SialT tolerate substitutions at C-9 of CMP-sialic acid (Schemes 6 and 7) [89, 91, 921. The 9-OH group can be replaced with fluoro, azido, amino, acetamido, hexanoylamido, benzamido, and even fluorescent labels. 9-0-Acetyl sialic acid can also be transferred. Donors in which the 5-NHAc group of sialic acid is replaced with OH, NHC(O)CHzOH, or NHCbz (Cbz = benzyloxycarbonyl) are also accepted by a2,3-SialT [93]. CMP-4-deoxysialic acid was found to be a donor substrate for a2,6-SialT [94].

694

27 Enzymatic Glycosylations with Non-Natural Donors and Acceptors

OH OH

I Non-Natural Donors)

NHAcyl

Ho

A

C

H HO

N

~

HO

O ,H

OH


I Non-Natural Acceptors] Scheme 5. Non-natural donors and acceptors for FucT VI.

Non-natural acceptors

Chemical mapping studies have indicated that OH-6 of Gal and 2-NHAc are required for binding to rat liver a2,6-SialT whereas rat liver a2,3-SialT requires an intact 3,4,6-triol system on the Gal residue [95].Analogs of Gal(pl-4)GlcNAc(pl2)Mana-O-Oct, where the 3- or 4-OH group of the Gal residue is deoxygenated or substituted with a fluoro atom, were found to be active as substrates for a2,6-SialT (Scheme 7) [96]. The 4/’-O-methyl derivative of the trisaccharide is an acceptor for a2,3-SialT [96]. A wide range of substitutions on the N-acetyl group of both type I or type I1 acceptors were accepted by a2,3-SialT [97, 981. a2,3-SialT also transfers sialic acid to lactal, lactose, and 2-0-pivaloyllactose [39].

27.2 Enzymatic Glycosylutions F, N3, NH2, NHAcyl, NH-FIuows, OAC

695

0

cop

AcHN

HO

I

OH

Ho OH, NHCOCHzOH, NHCbz

I Non-Natural DonorsI N3, NH2, NHCOCH2CH3, NHAcyl, OH, OCOC(CH3)3 H, Br H H, OCH3

HO

OR

HO

I H

N3, NH2, NHAcyl

OMe

-@

I

H

,OH

OOct

1 Non-Natural Acceptor3 Scheme 6. Non-natural donors and acceptors for a2,3-SialT

696

27 Enzymatic Glycosylutions with Non-Natural Donors and Acceptors F, N3, NH2, NHAcyl, NH-FIuo~~s, OAC

0

AcHN

I Non-Natural Donors I

I

ooci

I Non-Natural Acceptors) Scheme 7. Non-natural donors and acceptors for a2,6-SialT.

27.2.4 N-Acetylglucosaminylation The branching pattern of complex N-glycans is controlled by a series of N-acetylglucosaminyltransferases numbered GnT I-VI [ 12, 991. These enzymes transfer an N-acetylglucosamine residue from UDP-GlcNAc to different sites on the tri-mannose core structure Man(a1-6 [Man(al-3)]ManfJ as shown in Scheme 8. Biosynthetically, N-acetylglucosaminyltransferases catalyze the transfer of a GlcNAc residue from UDP-GlcNAc to oligosaccharide acceptors having the trimannose core structure attached to GlcNAc(81-4)GlcNAcp-Asn. The finding that GlcNAc(~1-4)GlcNAc~-Asn can be replaced with a hydrophobic group made it feasible to study the donor and acceptor specificities of these enzymes using smaller oligosaccharide acceptors which are readily prepared [ 100, 1011. N-AcetylglucosaminyltransferaseI, 11, and I11

The C-3, C-4, and C-6 deoxy analogs of UDP-GlcNAc were found to act as donor substrates for GnT I (E.C. 2.4.1.101) from human milk using Man(a1-6 [Man(al-

27.2 Enzymutic Glycosylutions

GnT VI

Hm

HO HO

GnT 111

GnT IV -HO

697

HoJ

H

AH-

O

=

o

~

e

~

&

HO Asn-R

\OH

NHAc

GnTl

Scheme 8. Glycosylation sites for GnT ILVI.

OMe OR

H

HO

m

OOct

HNHAc 0-UDP

Non-Natural Donors

HO

IGnTl

Non-Natural Acceptors

Scheme 9. Non-natural donors and acceptors for GnT I

3)]Manp-O(CH2)sCOOMeas the acceptor [ 1021. The tetrasaccharide products of the above reactions have been evaluated with GnT I1 [102]. Deoxygenation at C-3 of GlcNAc unit was not tolerated by GnT I1 and removal of the 4-OH or 6-OH of GlcNAc caused only modest decreases in activity [ 1021. As shown in Scheme 9, GnT I can use non-natural acceptors that are deoxygenated or 0-alkylated on the aMan(1-6) residue [103-1051. Alkylations at OH-4 of clMan1-3, the mannose residue to which transfer occurs, are tolerated. Analogs where OH-3, 4, or 6 of the aManl-6 residue was deoxygenated or methylated were found to be substrates of GnT I1 (Scheme 10) [103].

698

27 Enzymatic Glycosylations with Non-Natural Donors und Acceptors

H,OMe H,OMe

-

0 NHAc

H,OMe

I

-

HO

ooct H,OMe

-

H

#" HoHO

Scheme 10. Non-natural acceptors for GnT I1 and 111.

Deoxygenation at C-4 [ 1061, C-6 [ 1071 or methylation at C-4 [99, 1031 of aMan13 are also tolerated by GnT XI. A dideoxygenated pentasaccharide, which was chemo-enzymatically synthesized by use of GnT I1 [ 1061, and a di-0-methylated pentasaccharide were active as substrates for GnT 111 (E.C. 2.4.1.144) producing the expected hexasaccharide analogs [108, 1091. N-AcetylglucosaminyltransferaseV

There has been particular interest in N-acetylglucosaminyltransferase V (GnT V, E.C. 2.4.1.155) because the activity of this enzyme has been shown to correlate with the metastatic potential of tumor cell lines [ 110-1 141. Whereas a heptasaccharide is the smallest physiological substrate, trisaccharides GlcNAc(~l-2)Man(alL6)ManP-OR[115, 1161 and GlcNAc(P1-2)Man(al-6)GlcP-OR [ 1171 were found to be excellent acceptors for GnT V. The 0-GlcNAc residue can also be replaced by a P-Glc residue [ 1181. Studies using analogs where OH-3, 4, or 6 of the P-GlcNAc residue of GlcNAc(~l-2)Man(al-6)Glc~-OR were modified indicated that an intact 3,4,6-triol system is required for recognition by GnT V [ 119, 1201. Modifications on the a-Man residue and P-Glc residues are tolerated by this enzyme (Scheme 11) [117, 121-1231. Kinetic studies using restricted trisaccharide analogs revealed that GnT V preferentially recognizes the gg rotamer [124]. Analogs where the oxygen atoms in the glycosidic linkages were replaced with sulfur are also good acceptors for GnT V producing the corresponding tetrasaccharides [ 1251.

699

27.2 Enzymatic Glycosylutions

\ o

8'

c

3

d

X 8

I 0

Z

c

o x I

-0

8

0

s

700

27 Enzymatic Glycosylations with Non-Natural Donors and Acceptors

27.3 Summary Glycosyltransferases are remarkably flexible in their donor and acceptor specificities. The use of glycosyltransferases greatly simplifies the synthetic procedure for the preparation of non-natural oligosaccharides starting from non-natural donors and/or non-natural acceptors. With progress in molecular biology, even more glycosyltransferases will become available in sufficient quantities for use in chemical synthesis. As the structures of glycosyltransferases become well defined, they can be tailored using site-directed mutagenesis to have enhanced stabilities, even broader specificities and high affinities towards non-natural substrates while maintaining regio- and stereospecificity. Large-scale synthesis of oligosaccharides using a CMPfusion has been reported [ 1261. CombinNeuSAc synthetase/a2,3-~ialyltransferase ing two or more enzymes into a fusion protein will simplify the production and purification of enzymes, reduce costs and facilitate synthesis. As improved and alternate methods for the regeneration [ 1I] and large-scale production [ 1271 of sugar nucleotide donors are developed, enzymatic glycosylation will become increasingly routine for the construction of large quantities of many non-natural oligosaccharides. Acknowledgments Research support from the Natural Sciences and Engineering Research Council of Canada (to M.M.P.) and the Alberta Research Council for a graduate scholarship in Carbohydrate Chemistry (to X.Q.) are gratefully acknowledged.

References 1. 2. 3. 4.

Varki, A. Glycobiology 1993, 3, 97. Dwek, R. A. Chem. Rev. 1996, 96, 683. Toshima, K.; Tatsuta, K. Chem. Rev. 1993, 93, 1603. Khan, S. H.; Hindsgaul, 0. In Moleculur Clycobiology; Fukuda, M., Hindsgaul, 0. Eds.; Oxford: New York, 1994; pp 206. 5. Barresi, F.; Hindsgaul, 0. J. Carbohydr. Chem. 1995, 14, 1043. 6. Boons, G.-J. Tetrahedron 1996, 52, 1095. 7. Schmidt, R. R. In Glycosciences;Gabius, H.-J., Gabius, S. Eds.; Chapman & Hall: Weinheim, 1997; pp 31-53. 8. McAuliffe, J.; Hindsgaul, 0. Chem. Ind. 1997, 170. 9. Wong, C.-H.; Whitesides, G. M. Enzymes in Synthetic Organic Chemistry; Elsevier: Oxford, 1994; pp. 252-311. 10. Palcic, M. M. Methods Enzjwzol. 1994, 230, 300. 11. Wong, C.-H.; Halcomb, R. L.; Ichikawa, Y.; Kajimoto, T. Anyew. Chem., Int. Ed. Engl. 1995, 34, 521. 12. Brockhausen, I.; Schachter, H. In Glycosciences;Gabius, H.-J., Gabius, S. Eds.; Chapman & Hall: Weinheim, 1997; pp 79-113. 13. Gijsen, H. J. M.; Qiao, L.; Fitz, W.; Wong, C.-H. Chem. Rev. 1996, 96, 443. 14. Takayama, S.; McGarvey, G. J.; Wong, C.-H. Chem. Soc. Rev. 1997,26,407.

References

701

15. Ichikawa, Y. In Glycopeptides and Related Compounds: Synthesis, Analysis and Application, Large, D. G., Warren, C. D. Eds.; Dekker: New York, 1997; pp 79-205. 16. Palcic, M. M.; Hindsgaul, 0. Trends Glycosci. Glycotechnol. 1996, 8, 37. 17. Lemieux, R. U. In Proc. VIIIth Int. Symp. Med. Chem.; Swedish Pharmaceutical Press: Stockholm, 1985; pp. 329-351. 18. Lemieux, R. U. Chem. SOC.Rev. 1989, 18, 347. 19. Lemieux, R. U. Acc. Chem. Res. 1996,29, 373. 20. Sears, P.; Wong, C.-H. Cell. Mol. Life Sci. 1998, 54, 223. 21. Matta, K. L. In Modern Methods in Carbohydrate Synthesis; Khan, S . H.; O’Neil, R. A. Eds.; Harwood Academic: Amsterdam, 1995; pp 437-466. 22. Crawley, S. C.; Palcic, M. M. In Modern Methods in Carbohydrate Synthesis; Khan, S. H.; O’Neil, R. A. Eds.; Hanvood Academic: Amsterdam, 1995; pp 492-517. 23. Elhalabi, J. M.; Rice, K. G. Curr. Med. Chem. 1999, 6, 93. 24. Ohrlein. R. Top. Curr. Chem. 1999, 200, 227. 25. Thiem, J.; Wiemann, T. Angew. Chem., Intl. Ed. Engl. 1991, 30, 1163. 26. Thiem, J.; Wiemann, T. Synthesis 1992, 141. 27. Wong, C.-H.; Wang, R.; Ichikawa, Y. J. Org. Chem. 1992, 57, 4343. 28. Srivastava, G.; Hindsgaul, 0.;Palcic, M. M. Carbohydr. Res. 1993, 245, 137. 29. Hindsgaul, 0.;Kaur, K. J.; Gokhale, U. B.; Srivastava, G.; Alton, G.; and Palcic, M. M. ACS Symp. Ser. 1991, 466, 38. 30. Berliner, L. J.; Robinson, R. D. Biochemistry 1982, 21, 6340. 31. Kodama, H.; Kajihara, Y.; Endo, T.; Hashimoto, H. Tetrahedron Lett. 1993, 34, 6419. 32. Kajihara, Y., Endo, T., Ogasawara, H., Kodama, H.; Hashimoto, H. Carbohydr. Rex 1995, 269, 273. 33. Arlt, M.; Hindsgaul, 0. J. Org. Chem. 1995, 60, 14. 34. Palcic, M. M.; Hindsgaul, 0. Glycobiology 1991, I , 205-209. 35. Do, K.-Y.; Do, S.-I.; Cummings, R. D. J. Biol. Chem. 1995, 270, 18447. 36. Andree, P. J.; Berliner, L. J. Biochim. Biophys. Acta 1978, 544, 489. 37. Yuasa, H.; Hindsgaul, 0.;Palcic, M. M. J. Am. Chem. Soc. 1992, 114, 5891. 38. Tsuruta, 0.;Shinohara, G.; Yuasa, H.; Hashimoto, H. Bioorg. Med. Chem. Lett. 1997, 7,2523. 39. Ichikawa, Y.; Lin, Y.-C.; Dumas, D. P.; Shen, G.-J.; Garcia-Junceda, E.; Williams, M. A,; Bayer, R.; Ketcham, K.; Walker, L. E.; Paulson, J. C.; Wong, C.-H. J. Am. Chem. Soc. 1992, 114, 9283. 40. Berliner, L. J.; Davies, M. E.; Ebner, K . E.; Beyer, T. A.; Bell, J. E. Mol. Cell. Biochem. 1984. 62, 37. 41. Palcic, M. M.; Srivastava, 0. P.; Hindsgaul, 0. Carbohydr. Res. 1987, 159, 315. 42. Ohrlein, R.; Ernst, B.; Berger, E. G. Carbohydr. Res. 1992,236, 335. 43. Baisch, G.; Ohrlein, R.; Ernst, B. Bioory. Med. Chem. Lett. 1996, 6, 749. 44. Kajihara, Y.; Kodama, H.; Endo, T.; Hashimoto, H. Carbohydr. Res. 1998, 306, 361. 45. Wong, C.-H.; Ichikawa, Y.; Krach, T.; Gautheron-Le Narvor, C.; Dumas, D. P.; Look, G. C. J. Am. Chem. SOC.1991, 113, 8137. 46. Yu, L.; Cabrera, R.; Ramirez, J.; Malinovskii, V. A,; Brew, K.; Wang, P. G. Tetruhedron Lett. 1995, 36, 2897. 47. Nishida, Y.; Wiemann, T.; Sinnwell, V.; Thiem, J. J. Am. Chem. Soc. 1993, 115, 2536. 48. Gambert, U.; Thiem, J. Top. Curr. Chem. 1997, 186, 21. 49. Nishida, Y.; Wiemann, T.; Thiem, J. Tetrahedron Lett. 1992, 33, 8043. 50. Nishida, Y.; Wiemann, T.; Thiem, J. Tetrahedron Lett. 1993, 34, 2905. 51. Wiemann, T.; Nishida, Y.; Sinnwell, V.; Thiem, J. J. Org. Chem. 1994, 59, 6744. 52. Nishimura, S.-I.; Lee, K. B.; Matsuoka, K.; Lee, Y. C. Biochem. Biophys. Res. Commun. 1994, 199, 249. 53. Matsuoka, K.; Nishimura, S.-I. Macromolecules 1995, 28, 2961. 54. Wiemann, T.; Taubken, N.; Zehavi, U.; Thiem, J. Carbohydr. Res. 1994,257, C1. 55. Kopper, S. Carbohydr. Res. 1994,265, 161. 56. Schuster, M.; Wang, P.; Paulson, J. C.; Wong, C.-H. J. Am. Chem. Soc. 1994, 116, 1135. 57. Meldal, M.; Auzanneau, F.-I.; Hindsgaul, 0.;Palcic, M. M. J. Chem. Soc., Chem Commun. 1994, 1849.

102

27 Enzymatic Glycosylations with Non-Natural Donors and Acceptors

58. Joziasse, D. H.; Shaper, N. L.; Salyer, L. S.; van den Eijnden, D. H.; van der Spoel, A. C.; Shaper, J. H. Eur. J. Biochem. 1990, 191, 75. 59. Fang, J.; Li, J.; Chen, X.; Zhang, Y.; Wang, J.; Guo Z.; Zhang, W.; Yu, L.; Brew, K.; Wang, P. G. J. Am. Chem. Soc. 1998,120,6635. 60. Hokke, C. H.; Zervosen, A.: Elling, L.; Joziasse, D. H.; van den Eijinden, D. H. Glycoconjugate Journal 1996, 13, 687. 61. Galili, U. Immunol. Today 1993, 14, 480. 62. Butler, D. Nature 1998, 391, 320. 63. Stults, C. L. M.; Macher, B. A.; Bhatti, R.; Srivastava, 0. P.; Hindsgaul, 0. Glycobiology 1999, 9, 661. 64. Sujino, K.; Uchiyama, T.; Hindsgaul, 0.;Seto, N. 0. L.: Wakarchuk, W. W.; Palcic, M. M. J. Am. Chem. SOC.2000, 122, 1261. 65. Blanken, W. M.; Van den Eijnden, D. H. J. Biol. Chem. 1985,260, 12927. 66. Baisch, G.; Ohrlein, R.; Kolbinger, F.; Streiff, M. Bioorg. Med. Chem. Lett. 1998, 8, 1575. 67. Sujino, K.; Malet, C.; Hindsgaul, 0.;Palcic, M. M. Carbohydr. Rex 1998, 305, 483. 68. Erika, S. Trends Glycosci. Glycotechnol. 1996, 8, 391. 69. Gokhale, U. B.; Hindsgaul, 0.;Palcic, M. M. Can. J. Chem. 1990, 68, 1063. 70. Stangier, K.; Palcic, M. M.; Bundle, D. R.; Hindsgaul, 0. Thiem, J. Curbohydr. Res. 1998, 305, 511. 71. Srivastava, G.; Kaur, K. J.; Hindsgaul, 0.;Palcic, M. M. J. Biol. Chem. 1992, 267, 22356. 72. Hallgren, C.; Hindsgaul, 0. J. Curbohydr. Chem. 1995, 14, 453-464. 73. Gosselin, S.; Palcic, M. M. Bioorg. Med. Chem. 1996, 4, 2023. 74. Nikrad, P. V.; Kashem, M. A,; Wlasichuk, K. B.; Alton, G.; Venot, A. P. Carbohydr. Res. 1993,250, 145. 75. Ding, Y . ;Hindsgaul, 0.; Li, H.; Palcic, M. M. Bioorg. Med. Chem. Lett. 1998, 8, 3199. 76. Ogawa, S.; Matsunaga, N.; Li, H.; Palcic, M. M. Eur. J. Org. Chem. 1999, 631. 77. Qian, X.; Hindsgaul, 0.;Li, H.; Palcic, M. M. J. Am. Chem. Soc. 1998, 120, 2184. 78. Baisch, G.; Ohrlein, R.; Katopodis, A,; Streiff, M.; Kolbinger, F. Bioorg. Med. Chem. Lett. 1997, 7, 2441. 79. Baisch, G.; Ohrlein, R.; Katopodis, A.; Ernst, B. Bioorg. Med. Chem. Lett. 1996, 6, 759. 80. Baisch, G.; Ohrlein, R.; Streiff, M. Bioorg. Med. Chem. Lett. 1998, 8, 161. 81. Baisch, G.; Ohrlein, R.; Katopodis, A. Bioorg. Med. Chem. Lett. 1997, 7, 2431. 82. Baisch, G.; Ohrlein, R.; Streiff, M.; Kolbinger, F. Bioorg. Med. Chem. Lett. 1998, 8, 755. 83. Baisch, G.; Ohrlein, R. Bioorg. Med. Chem. 1998, 6, 1673. 84. Baisch, G.; Ohrlein, R.; Katopodis, A. Bioorg. Med. Chem. Lett. 1996, 6, 2953. 85. Chandrasekaran, E. V.; Jain, R. K.; Larsen, R. D.; Wlasichuk. K.; DiCioccio, R. A.; Matta, K. L. Biochemistry 1996, 35, 8925. 86. Reuter, W.; Stache, R.; Stehling, P., Baum, 0. In Glycosciences; Gabius, H.-J., Gabius, S. Eds.; Chapman & Hall: Weinheim, 1997; pp 245-259. 87. Harduin-Lepers, A,; Recchi, M.-A.; Delannoy, P.; Glycobiology 1995, 5 , 741. 88. Higa, H. H.; Paulson, J. C. J. Biol. Chem. 1985, 260, 8838. 89. Gross, H. J.; Bunsch, A.; Paulson, J. C.; Brossmer, R. Eur. J. Biochem. 1987, 168, 595. 90. Brossmer, R.; Gross, H. J. Methods Enzymol. 1994, 247, 153. 91. Gross, H. J.; Rose, U.; Krause, J. M.; Paulson, J. C.; Schmid, K.; Feeney, R. E.; Brossmer, R. Biochemistry 1989, 28, 7386. 92. Gross, H. J.; Brossmer, R. Eur. J. Biochem. 1988, 177, 583. 93. Chappell, M. D.; Halcomb, R. L.; J. Am. Chem. Soc. 1997, 119, 3393. 94. Gross, H. J.; Brossmer, R. Glycoconjugate J. 1987, 4 , 145-156. 95. Wlasichuk, K. B.; Kashem, M. A,; Nikrad, P. V.; Bird, P.; Jiang, C.; Venot, A. P. J. Biol. Chem. 1993,268, 13971. 96. Van Dorst, J. A. L. M.; Tikkanen, J. M.; Krezdorn, C. H.; Streiff, M. B.; Berger, E. G.; Van Kuik, J. A.;,Kamerling, J. P.; Vliegenthart, J. F. G. Eur. J. Biochem. 1996,242, 674. 97. Baisch, G.; Ohrlein, R.; Streiff, M.; Ernst, B. Bioorg. Med. Chem. Lett. 1996, 6, 755. 98. Baisch, G.; Ohrlein, R.; Streiff, M. Bioorg. Med. Chem. Lett. 1998, 8, 157. 99. Taniguchi, N.; Ihara, Y. Glycoconjugate J. 1995, 12, 733. 100. Vella, G. J.; Paulsen, H.; Schachter, H. Can. J. Biochem. Cell. Biol. 1984, 62, 409.

References

703

101. Palcic, M. M.; Heerze, L. D.; Pierce, M.; Hindsgaul, 0. Clycoconjugate J. 1988, 5, 49. 102. Srivastava, G.; Alton, G.; Hindsgaul, 0. Carbohydr. Res. 1990, 207, 259. 103. Reck, F.; Springer, M.; Paulsen, H.; Brockhausen, I.; Sarkar, M.; Schachter, H. Carbohydr. Res. 1994, 259, 93. 104. Moller, G.; Reck, F.; Paulsen, H.; Kaur, K. J.; Sakar, M.; Schachter, H.; Brockhausen, I. Glycoconjugate J. 1992, 9, 180. 105. Reck, F.; Springer M.; Meinjohanns, E.; Paulsen, H.; Brockhausen, I.; Schachter, H. Clycoconjugate J. 1995, 12, 747. 106. Kaur, K. J.; Hindsgaul, 0. Carbohydr. Res. 1992, 226, 219. 107. Reck, F.; Meinjohanns, E.; Springer M.; Wilkens, R.; Van Drost J. A. L. M.; Paulsen, H.; Moller, G.; Brockhausen, I.; Schachter, H. Glycoconjugate J. 1994, 11, 210. 108. Alton, G.; Kanie, Y.; Hindsgaul, 0. Carhohydr. Rex 1993,238, 339. 109. Khan, S. H.; Compston, C. A,; Palcic, M. M.; Hindsgaul, 0. Carbohydr. Res. 1994,262, 283. 110. Yamashita, K.; Tachibana, Y.; Ohkura, T.; Kobata, A. J. Biol. Chem. 1985, 260, 3963. 1 1 1. Arango, J.; Pierce, M. J. Cell. Biochem. 1988, 37, 225. 112. Dennis, J. W.; Kosh, E.; Bryce, D.-M.; Breitman, M. L. Oncogene 1989, 4, 853. 113. Dennis, J. W.; Laferte, S.; Waghorne, C.; Breitman, M. L.; Kerbel, R. S. Science 1987, 236, 582. 114. Dennis, J. W. Cancer Surveys 1988, 7, 573. 115. Tahir, S. H.; Hindsgaul, 0. Can. J. Chem. 1986, 64, 1771. 116. Hindsgaul, 0.;Tahir, S. H.; Srivastava, 0. P.; Pierce, M. Carbohydr. Res. 1988, 173, 263. 117. Srivastava, 0. P., Hindsgaul, 0.;Shoreibah, M.; Pierce, M. Carbohydr. Res. 1988, 179, 137. 118. Kanie, 0.;Palcic, M. M.; Hindsgaul, 0. RZKEN Rev. 1995, 853. 119. Kanie, 0.;Crawley, S. C.; Palcic, M. M.; Hindsgaul, 0. Carbohydr. Rex 1993, 243, 139. 120. Kanie, 0.;Crawley, S. C.; Palcic, M. M.; Hindsgaul, 0. Bioorg. Med. Chem. 1994, 2, 1231. 121. Linker, T.; Crawley, S. C.; Hindsgaul, 0. Carbohydr. Res. 1993,245, 323. 122. Khan, S. H.; Duus, J. D.; Crawley, S. C.; Palcic, M. M.; Hindsgaul, 0. Tetrahedron Asym. 1994, 5, 2415. 123. Ogawa, S.; Furuya, T.; Tsunoda, H.; Hindsgaul, 0.;Stangier, K.; Palcic, M. M. Carhohydr. Res. 1995,271, 197. 124. Lindh, I.; Hindsgaul, 0. J. Am. Chern. Soc. 1991, 113, 216. 125. Lu, P.-P.; Hindsgaul, 0.;Li, H.; Palcic, M. M. Cun. J. Chem. 1997, 75, 790. 126. Gilbert, M.; Bayer, R.; Cunningham, A.-M.; DeFrees, S.; Gao, Y.; Watson, D. C.; Young, N. M.; Wakarchuk, W. W. Nature Biotech. 1998, 16, 769. 127. Koizumi, S.; Endo, T.; Tabata, K.; Ozaki, A. Nature Biotech. 1998, 16, 847.

Carbohydrates in Chemistry and Biology Editor Beat Ernst, Gerald W. Hart, Pierre Sinay copyright OWILEY-VCH Verlag GmbH, 2000

28 Solid-Phase Synthesis with Glycosyltransferases Claudine A @ , Christine Le Nurvou, and And& Lubineau

28.1 Introduction In recent years chemical solid-phase synthesis has gained growing interest in parallel to the development of the concept of combinatorial chemistry, a promising approach to the discovery of new biologically active compounds. In peptide and oligonucleotide chemistry this technology has been established for a long time, but not in oligosaccharide chemistry. Pioneering studies in this field suffered from the lack of effective glycosyl donors and high stereoselective coupling methods [ 1-31, Recent advances in this area, and development of highly reactive sugar donors and protective group chemistry have stimulated a renewed interest in solid-phase synthesis of oligosaccharides [4-61 and glycopeptides [7, 81.

28.2 General Aspects Solid-phase synthesis has several advantages over traditional solution-based methods. Firstly, large excesses of reagents can be used so that the coupling steps can be repeated to drive reactions to completion. Secondly, products are isolated at each step by simple filtration, an obvious simplification of the work-up procedures, which might enable future automation. Development of polymer-supported synthesis of oligosaccharides using purely chemical glycosylation methods, a chemo-enzymatic approach based on the use of glycosyltransferases has also started to develop. Glycosyltransferases have become widely used in the past ten years as efficient tools for glycosylation; as opposed to glycosidases, another class of enzymes of interest in oligosaccharide solution phase synthesis, these enzymes catalyze sugar unit transfer from a sugar-nucleotide donor on to an unprotected sugar acceptor, with complete regio and stereoselectivity.

106

28 Solid-Phase Synthesis with Glycosyltransferases

Thus these also seem appropriate reagents for use in solid-phase synthesis. These enzymes have, moreover, become increasingly available as a result of genetic engineering, and soluble forms of glycosyltransferases, originally membrane-bound proteins, have been generated. Solid-phase enzymatic synthesis raises extra problems, however, such as accessibility of the interior of the solid matrix to the enzyme or, for chemo-enzymatic synthesis, the biocompatibility of the support with both aqueous and organic solvents. Indeed this approach requires that the resins swell in aqueous buffers which, for example, precludes the use of the hydrophobic Merrifield type resin. Roughly, the amounts of swelling in different solvents give information on the overall polar or hydrophobic character of the support [9]. But even the microporous polyoxyethylene-polystyrene, sold under the trade name of Tentagel, which has good swelling properties in polar solvents, does not allow macromolecules such as enzymes of 20-50 kDa and more, to penetrate into the interior of the beads. A recent study on chymotrypsin-catalyzed proteolysis of peptide substrates bound to Tentage1 beads has shown that only the surface of the beads, i.e. 15% of the total functional sites at most, are available to the enzyme [lo]. The general scheme for solid-phase synthesis using glycosyltransferases is depicted in Scheme 1. Solid-phase synthesis requires a covalent linker group to attach the small molecule, a monosaccharide or an amino acid in the case of a glycopeptide, on to the polymeric resin. Such a linker must be stable to the reaction consupport

monosaccharide cleavable linker

SugarGlycosyltransferase 1

Y c

Sugarnucleotide 2

Glycosyltransferase 2

Oligosaccharide release from matrix

Scheme 1 . General schemc for the solid-phase synthesis of oligosaccharides using glycosyltransferases.

28.3 Enzymatic Synthesis on Insoluble Supports

701

ditions used during the synthesis but it needs to be cleaved selectively at the end of the synthesis, releasing the oligosaccharide or the glycopeptide from the resin into solution; light-sensitive or enzyme-susceptible linkers are particularly suitable for this purpose. A spacer-arm providing distance between the support and the acceptor substrate to eliminate steric problems can be optionally intercalated between the matrix and the linker. Sugar-chain elongation is the result of the sequential action of glycosyltransferases in the presence of the corresponding sugar-nucleotide donor. We shall describe enzymatic glycosylations that have been achieved on both insoluble and soluble supports, because, irrespective of the nature of the support, polymer-supported synthesis is based on the same concept, and good results with soluble supports have been reported in the literature.

28.3 Enzymatic Synthesis on Insoluble Supports 28.3.1 Enzymatic Synthesis of Oligosaccharides Use of an Amino-Functionalized Water-Compatible Polyacrylamide Gel Pioneering work in the solid-phase enzymatic synthesis of oligosaccharides was accomplished by Zehavi and co-workers. The authors used an amino-functionalized water-compatible polyacrylamide gel; it is well-known that polyacrylamide gels, the preferred polymer gels for electrophoresis, allow proteins to diffuse inside. Zehavi et al. first attached one glucose unit to a photo-removable linker, methyl 4-hydroxymethyl-3-nitrobenzoate. In the second step the unprotected glycoside 1 was linked to aminoethyl-substituted polyacrylamide gel beads via an amide linkage (Scheme 2). The glucopolymer 2 although obtained in a low yield, served as an acceptor in p( 1-4)galactosyltransferase-catalyzed reaction, in the presence of a-lactalbumin, with I4C radiolabeled UDP-Gal. The lactose disaccharide 4 was subsequently released from the lactosyl polymer 3 by irradiation at 320 nm, in quantitative yield, but nevertheless the overall yield of lactose, estimated from radiolabelling was disappointingly very low (< YO), probably because of inaccessibility of the substrate to the enzyme [ 111. Such light-sensitive polymers with maltose and maltotriose side chains were also prepared for use as acceptors in the glycogen synthase reaction, but only slightly better transfers (7%) were observed [12, 131. To improve the accessibility of saccharide acceptors in glycosylation reactions, polymers with longer spacers were prepared [14]. Thus a glucopolymer 5 with an aminooctyl spacer between the polyacrylamide support and the photolabile linker became an effective support for enzymatic glycosidation catalyzed by p( 1-4)galactosyltransferase. The reaction was conducted on 0.5 mmol scale with in situ generation of UDP-Gal from UDP-Glc by adding UDP-Glc epimerase. Lactose 4 released from lactopolymer 6 by photolysis was isolated in an overall yield of 51% (Scheme 2) [15].

708

28 Solid-Phase Synthesis with Glycosyltransferases

HO

OH

HO

OH

OH

3,n=2

6,n=8

I

iii

4

Scheme 2. Synthesis of lactose uia enzymatic glycosylation using a photolabile solid support. Reagents: i) NH2-(CH2),-NH-CO-P, water-DMF, EDCD, 10% for 2, 37% for 5; ii) a-lactalbumin, GT, UDP-Gal (3 eq.), cacodylate buffer pH 7, 3 mM MnC12, for 3; a-lactalbumin, GT, UDP-Glc (1.1 eq), UDPGE, cacodylate buffer pH 7, 0.3 mM MnC12, 0.02% NaN3, for 6; iii) hv, 1% for 4 from 2. 51% for 4 from 5.

Use of a Sepharose Matrix

In 1980 Barker and Nunez [16] reported the synthesis of Gal~l-4GlcNAcp-hexanolamine by use of a solid-phase system in which the GlcNAcp-hexanolamine glycoside was covalently linked to CNBr-activated Sepharose 4B, with 6 to 10 pmol ligand mL-' packed wet gel; it is noteworthy that the same coupling method has been widely applied by David et al. to enzyme immobilization [17]. With partially purified p( 1-4)galactosyltransferase from bovine milk, and a twofold excess of UDP-Gal, the Sepharose-bound N-acetylglucosamine 7 was converted into 8 in

28.3 Enzymutic Synthesis on Insoluble Supports

709

iii HO

NHAc

HO

bH

'OH

9 Scheme 3. Enzymatic synthesis of LacNAcD-hexanolamine on Sepharose. Reagents: i) GT, UDPGal (2 eq.), cacodylate buffer pH 7, 5 mM MnC12, 0.1'%1mercaptoethanol, 18 h, 35"C, 85%; ii) 2 M NaOH, 40°C, 3 h, 80%.

85% yield; then the disaccharide glycoside 9 could be cleaved from the matrix, in 80% yield, by alkaline treatment (Scheme 3). Very recently, in a related approach, Norberg et al. presented the solid-phase enzymatic fucosylation of a disaccharide acceptor linked to Sepharose via a disulfide linkage and a linker arm of 12 atoms with a loading of 5 to 10 pmol disaccharide ligand mL-' wet gel [ 181. The sepharose-bound disaccharide 10 incubated with human milk a( 1-3/4)fucosyltransferase and GDP-Fuc in fivefold excess was converted into the Sepharose-bound trisaccharide 11 in 68% glycosylation yield (Scheme 4). The Lewisa trisaccharide derivative 12 could be released from the matrix, together with the starting disaccharide 13, by treatment with 2-mercaptoethanol, in 91% yield. Because the difficulties encountered in this enzymatic fucosylation step could not be improved, even after extensive washing of the gel and recycling with fresh enzyme and sugar-nucleotide, it was assumed that low reactivity was a result of steric factors. The same matrix was, therefore, further substituted with longer linkers of the PEG-type. Sepharose gels with linker lengths of 35, 47, 59 and 71 atoms between the matrix and the N-acetylglucosamine acceptor were built and subjected to enzymatic galactosylation with /3( 1-4)galactosyltransferase and excess UDP-Gal [19]. Aliquots were removed and treated with DTT; the 'H NMR spectrum of the

7 10

28 Solid-Phase Synthesis with Glycosyltransferases

10

ii

Scheme 4. Solid-phase enzymatic synthesis of a Lewisa trisaccharide by use of a disaccharide acceptor bound to Sepharose. Reagents: i) cacodylate buffer pH 6.8, 5 mM MnC12, 0.05% NaN3, FT, GDP-Fuc ( 5 eq.) 68%; ii) 2-mercaptoethanol, 1 h, 60"C, 91%.

filtrate enabled estimation of the yield. The longest linker turned out to give the best yield (98%). The GlcNAc Sepharose with the longest linker was then successfully converted into the sialyl Lewis' tetrasaccharide 16, by successive incubation of the gel 14 with p( 1-4)galactosyltransferase, recombinant a(2-3)sialyltransferase and human milk a(1-3/4)fucosyltransferase, in the presence of the corresponding sugar-nucleotides, used in three to fourfold excess, and alkaline phosphatase,

28.3 Enzymatic Synthesis on Insoluble Supports

71 1

14

15, R =

15

I1

16, R = H

Scheme 5. Solid-phase enzymatic synthesis of a Lewis" tetrasaccharide using a linker of 71 atoms between the monosaccharide acceptor and the Sepharose matrix. Reagents: i) 1. cacodylate buffer pH 7.5, 5 mM MnC12, x-lactalbumin, GT, UDP-Gal (3 eq.), CIAP, 3 7 T , 5 days; 2. cacodylate buffer pH 7.35, 0.1% triton X-100, ST3, CMP-NeuAc (2.5 eq.), 3 mM MnCll, 0.02% NaN3, 35 "C, 5 days; 3. cacodylate buffer pH 6.5, 5 mM MnC12, FT, GDP-Fuc (4 eq.), 37"C, 5 days; ii) 1. DTT, 0.1 M phosphate buffer; 2. BioGel P-2, 57% from 14.

followed by final treatment of the gel 15 with dithiothreitol (57% overall yield, Scheme 5 ) . Use of Controlled-Pore Glass

Wong and coworkers promoted the use of aminopropyl silica for solid-phase enzymatic synthesis [20]. This rigid, non-swelling support turned out to be compatible with biomolecules, but it required a spacer for rendering the substrate more accessible to the enzyme. The authors prepared a disaccharide target carrying at the reducing end a C-6 spacer-arm with a carboxylic function that was condensed via its cesium salt to commercially available N-iodoacetylaminopropyl controlled-pore glass, affording the solid-supported disaccharide acceptor 17 (Scheme 6). This ac-

712

28 Solid-Phase Synthesis with Glycosyltransferuses

17

OH AcHN ~

Ji

OH

~

o

H

HO

~ OH

o

+

o

OH

~ OH

o

~

H o 0

18

J ii

HO

OH

OH

OH

19 Scheme 6. Solid-phase enzymatic synthesis of NeuAca2-3Gal/ll-4GlcNAc~1-3Gal on controlledpore glass. Reagents: i) 1. HEPES buffer pH 7.2, 5 m M MnC12, GT, UDP-Gal (3 eq.), 0.5% DTT, 48 h; 2. cacodylate buffer pH 7.5, CMP-NeuAc (2 eq.), CIAP, 48 h, 9.y.; ii) NHz-NHz, room temp., 24 h.

ceptor could be enzymatically galactosylated with /I1( -4)galactosyltransferase and UDP-Gal, and then sialylated upon treatment with CMP-NeuAc, a(2-3)sialyltransferase and alkaline phosphatase, commonly used to hydrolyze the inhibitor CMP, released in the reaction. Both glycosylation steps seemed quantitative, because after cleavage of the ester linkage on the conjugate 18 by treatment with hydrazine, the tetrasaccharide hydrazide 19 was the only observed product.

28.3.2 Enzymatic Synthesis of Glycopeptides Use of Controlled-Pore Glass

The same functionalized silica support was used for glycosyltransferase-catalyzed sugar-chain elongation on a glycopeptide [21]. A N-linked glycodipeptide was attached to the support via a spacer made of six glycine units and an a-chymotrypsinsensitive phenylalanine ester bond, giving the silica-supported N-Boc-Asn (G1cNAcp)-Gly-Phe 20 with a loading of ca 0.2 mmol glycopeptide gg’ dry silica (Scheme 7). This supported glycopeptide was first galactosylated with p( 1-4)

~

28.3 Enzymatic Synthesis on Insoluble Supports

O 'H

I1 3

Ji

Scheme 7. Solid-phase synthesis of a sialyl LewisXglycopeptide on aminopropyl silica. Reagents: i) 1. HEPES buffer pH 7, 10 mM MnC12, GT, UDP-Gal (1.5 eq.), 55%; 2. HEPES buffer pH 7, 5 mM MnC12, ST3, CMP-NeuAc (1.5 eq.), 65%; ii) 1. a-chymotrypsin; 2. ultrafiltration; 3. FT, GDP-Fuc (2.5 eq.), HEPES buffer pH 7, 95%).

7 14

28 Solid-Phase Synthesis with Glycosyltransferases

galactosyltransferase and UDP-Gal in 55% yield, and subsequently sialylated in the presence of CMP-NeuAc and a(2-3)sialyltransferase in 65% yield affording the silica-bound sialylated glycopeptide 21. Yields were evaluated by cleavage of the glycopeptide from an aliquot of the functionalized silica with a-chymotrypsin. Because galactosylation and sialylation reactions were incomplete, the sialylated triwas released from peptide Boc-Asn-(NeuAca2-3Gal~l-4GlcNAc~)-Gly-Phe-OH the solid support, together with partially glycosylated products, by cleavage with achymotrypsin. Then fucosylation in a traditional solution-phase reaction with a(13/4)fucosyltransferase and GDP-Fuc, afforded a mixture of the sialyl Lewis' tripeptide 22 (35%), the fucosylated tripeptide 23 (20%) and unreacted glycopeptide 24 (45%). The solid-phase synthesis of an 0-glycopeptide by a chemo-enzymatic approach was reported by Wong and Seitz [22]. By use of the same support, aminopropylCPG, the starting amino acid, alanine, was attached through the flexible, acid- and base-stable Hycron anchor, first introduced by Kunz [23], which can be cleaved as an ally1 ester with palladium(0) under mild conditions. The glycooctapeptide 26 incorporating the preformed glycosyl amino acid, Thr(GlcNAcp), was first assembled according to classical solid-phase peptide synthesis in an overall yield of 20% starting from 25 (Scheme 8). According to the authors, CPG turned out not to be the

Ac-Lys-Pro-Pro-Asn-Tr-Tr-Ser-Ala-HY CRON-NHGlcNAc

GalPl -4GlcNAc

+

b

CPG

GalP 1-4GlcNAc

Ac-Lys-Pro-Pro-Asn-Thr-Thr-Ser-Ala-OH

B

NeuAccr2-3Gal~l-4GlcNAc

29

Scheme 8. Chemo-enzymatic synthesis of a 0-sialyl-LacNAc octapeptide. Reagents: i) HEPES buffer pH 7, 5 mM MnC12, UDP-Gal (2 eq.), GT, CIAP, 3 7 T , 3 days; ii) HEPES buffer pH 7, ST3, CIAP, CMP-NeuAc (1.5 eq.), 3 7 T , 4 days, 76%.

28.4 Enzymatic Synthesis of Oligosaccharides and Glycoconjugates

7 15

optimum support for chemical peptide synthesis, but the most suitable for enzymatic reactions [24]. The loading of the support used in this preparation was also very low (60 pmol g-'). The glycopeptide 26 was then submitted to enzymatic galactosylation in the presence of p( l -4)galactosyltransferase and a twofold excess of UDP-Gal under the usual conditions. Unexpectedly, during incubation at neutral pH and 37"C, hydrolysis of the ally1 ester linkage was observed in 53% yield, affording the O-LacNAc octapeptide 27. The remaining supported product could be released from the silica beads by treatment with palladium(0) and morpholine. Altogether the O-LacNAc peptide 27 was synthesized in 15% yield on the basis of the loading of the initial support 25, giving a 75% yield for the galactosylation step. Enzymatic sialylation was subsequently performed on the supported glycopeptide 28 but again, peptide cleavage from the support occurred during incubation and the major part (76%) of the sialylated glycopeptide 29 was finally recovered from the filtrate. Use of Polyethylene Glycol Polyacrylamide (PEGA)

An alternative to the use of silica gel described above is a PEGA resin originally developed by Meldal for peptide solid-phase synthesis [25]. PEGA is a poly(ethy1ene glycol)-polyacrylamide copolymer with high swelling capacity both in organic media and in water, and has free amino groups for functionalization. The acid-labile Rink's linker (4-(a-amino-2',4'-dimethoxybenzyl)phenoxyaceticacid) was first attached to the PEGA resin and this supported linker was then reacted with FmocGly-O-pentafluorophenyl ester and, after removal of the Fmoc group, the resinbound amino acid was coupled to protected glycopeptide Asn(G1cNAcp) by conventional peptide chemistry, to give, after deprotection, the resin-bound glycopeptide 29 with a loading capacity of 0.16 mmol gg' resin (Scheme 9). Compound 29 was then galactosylated in the presence of p( 1-4)galactosyltransferase and a large excess of UDP-Gal, in quantitative yield [26]. It is worth mentioning that pre-incubation of the resin with enzyme and buffer for three days at 4 "C before adding UDP-Gal was beneficial, affording conversion of 29 into 30 in a much shorter time. The final lactosamine glycopeptide 31 was recovered by cleavage of 30 with 95% aqueous trifluoroacetic acid. Diverse PEGA resins with different amounts of cross-linking and loading capacity could be prepared and studied for their compatibility with biomolecules; this study led to the conclusion that a more cross-linked resin gave a lower yield in enzymatic galactosylation [27].

28.4 Enzymatic Synthesis of Oligosaccharides and Glycoconjugates on Soluble Supports 28.4.1 Enzymatic Synthesis of Oligosaccharides Use of Water-Soluble Amino-Substituted Poly(viny1 alcohol)

In view of the low yields observed for enzymatic transfer of galactose on insoluble polyacrylamide polymers [ 111, Zehavi et al. developed the use of a water-soluble

7 16

28 Solid-Phase Synthesis with Glycosyltransferuses

,OJ . ti

'N H

..J H

0

PEG$

0 Me0

Po

" H

HO

29

OH OH

OMe Me

,4 H

0

PEGL

HO

30

OMe

N

HO

OH

OH

H

31

Scheme 9. Chemo-enzymatic synthesis of a N-glycopeptide on a polyethylene glycol polyacrylamide (PEGA) resin. Reagents: i) cacodylate buffer pH 7.4, 5 mM MnC12, UDP-Gal (3 eq.), GT, 23 "C, 48 h, 95%; ii) 95% aqueous TFA, 23 "C, 2 h.

polymer that afforded improved accessibility of substrate to enzyme. This linear polymer consisted of poly(viny1 alcohol) substituted with amino groups. The lightsensitive gluco polymer 32 carrying 0.15 mmol D-glucose g-', was glycosylated as previously described using 14C-labeledUDP-Gal and p( 1-4)galactosyltransferase, affording, after purification by ultrafiltration, the lactopolymer 33 in 34% yield (Scheme 10). Upon UV irradiation of 33, radiolabeled lactose 4 was subsequently recovered in 88% yield [28].

28.4 Enzymatic Synthesis of Oligosaccharides and Glycoconjugates

'OH

7 17

bH

32

HO

-!&&:=-$7-+fi 'OH

O 'H 4

Scheme 10. Enzymatic synthesis of lactose by use of a photolabile water-soluble amino-substituted poly(viny1 alcohol). Reagents: i) cacodylate buffer pH 7, 3 mM MnC12, UDP-Gal (1 eq.), GT, alactalbumin, 0.1% mercaptoethanol, 37 "C, 18 h, 34%; ii) 1. hv, 20 h, 88%);2. ultrafiltration.

Use of Water-Soluble Glycopolymer Synthesized by Polymerization

Nishimura et al. introduced the use of water-soluble glycopolymers. They synthesized the glycopolymer 34 by radical copolymerization of sugar monomer having an n-pentenyl group as a polymerizable aglycone, with acrylamide in an aqueous medium in the presence of ammonium persulfate and N,N,N',N'-tetramethylethylenediamine as initiators, according to Scheme 11 [29, 301. Polymers of more than 40 kDa were obtained with a composition varying from 1 : 5 to 1 :23 for the carbohydrate monomer-to-acrylamide ratio, depending on copolymerization conditions. Galactosylation of the GlcNAc polymer 34, performed by means of UDP-Gal and /I( 1 -4)galactosyltransferase, led to the LacNAc polymer 35 in quantitative yield, the n-pentenyl group providing the acceptor sugar with an adequately flexible spacer-arm. As opposed to insoluble polymer, such a water-soluble polymer enabled direct reaction monitoring by ' H NMR measurement in D20 [31]. As an extension of this work, the authors applied the same approach to the preparation of free oligosaccharides, by introducing suitable, removable linker arms between the saccharide and the polymer backbone. Thus, radical copolymerization with acrylamide of monomers 36 and 37 having spacer-arms of different lengths, under the previous conditions, yielded water-soluble GlcNAc acceptor polymers 38 and 39, both having a linker selectively cleavable by hydrogenolysis [32, 331. Enzy-

7 18

28 Solid-Phase Synthesis with Glycosyltransferases NHAc

NHAc 1

OH

x:y=1:8

%

0

Scheme 11. Enzymatic galactosylation of a water-soluble GlcNAc-bearing polyacrylamide. Reagents: i) CH2=CH-CO-NH2, TEMED, APS, H20, room temp., 2 h; ii) 1. HEPES buffer pH 6, 10 mM MnC12, UDP-Gal (1.5 eq.), GT, 37"C, 24 h; 2. gel filtration on Sephadex (3-50, q.y.

matic sugar chain-elongation was performed with p( 1-4)galactosyltransferase and UDP-Gal. Quantitative galactosylation could be achieved with polymer 39 bearing the long spacer-arm, whereas only 30% galactosylation was observed with polymer 38 bearing the short spacer-arm. N-acetyllactosamine 40 was finally released in high yield from polymer 41 by hydrogenolysis (Scheme 12). Other types of GlcNAc polymer, 43 and 44, containing an L-phenylalanine residue in the spacer-arm moiety as an a-chymotrypsin-sensitive structure, were also prepared in a similar manner [33, 341. Galactosylation with /3( 1-4)galactosyltransferase and subsequent sialylation with rat liver a(2-6)sialyltransferase in the presence of the sugar-nucleotides were conducted as shown in Scheme 13. Sugar transfer reactions were achieved with high efficiency using only a small excess of UDP-Gal and CMP-NeuAc, affording, after filtration and lyophilization, the polymer 45 bearing sialyl a(2-6)N-acetyllactosamine branches, in almost quantitative yield. Finally by hydrolytic action of a-chymotrypsin, sialyl a(2-6)N-acetyllactosamine derivative 46 was recovered in 72% overall yield from 44.It is noteworthy that whereas quantitative galactosylation was also observed with polymer 43, it was not possible to release lactosamine from the polymer by the action of achymotrypsin, because of the lack of the flexible spacer-arm between phenylalanine and the polymer backbone [33].

28.4.2 Enzymatic Synthesis of Glycolipids on Water-Soluble PolyacrylamidePoly(N-acryloxysuccinimide) (PAN) The water-soluble polyacrylamide-poly(N-acryloxysuccinimide) (PAN) first described by Whitesides for enzyme immobilization [35],was used by Zehavi et al. in

28.4 Enzymatic Synthesis of Oligosaccharides and Glycoconjugates

36 R = NHCOCH=CH2 37 R = NHCO(CH~)SNHCOCH=CH~

7 19

38 R = S 39 R = (CH2)sNHCO-S

HO OH

42 R = S 41 R = (CH2)sNHCO-S

HO HO OH

OH

40

Scheme 12. Enzymatic synthesis of N-acetyllactosamine using a water-soluble glycopolymer with a linker cleavable by hydrogenolysis. Reagents: i) CHz=CH-CO-NH2, TEMED, APS, DMSOH20, room temp., 24 h, 61% for 37, 83% for 38; ii) 1. HEPES buffer pH 6, 10 mM MnC12, UDPGal (1 eq.), GT, 37 "C, 24 h; 2. gel filtration on Sephadex G-50, q.y. for 40, 30% for 41; iii) H2, Pd/ C, H20-MeOH, 95%.

enzymatic supported synthesis of glycolipids. The hydrazide function in compound 47 was reacted with PAN yielding the glucopolymer 48 which served as an acceptor in the p( 1-4)galactosyltransferase-catalyzed reaction in the presence of a-lactalbumin and 14C labeled UDP-Gal [36]. The lactosyl sphingosine polymer 49 was obtained in 36% yield, after purification by extensive dialysis; further photolysis of the 2-nitrobenzylurethane group followed by acylation with stearoyl chloride, provided lactosylceramide 50 in 54% yield (Scheme 14). Alternatively the polymer 51 carrying 0.24 meq lactosyl/3(l-1)sphingosine g-', chemically prepared from PAN, was sialylated in 55% yield by use of recombinant rat liver a(2-3)sialyltransferase, 14C labeled CMP-NeuAc, and calf intestinal alkaline phosphatase [37]. Again photolysis of the 2-nitrobenzylurethane group in the sialylated polymer 52 followed by acylation with stearoyl chloride afforded the GM3 ganglioside, NeuAccc(2-3)Galp( 1-4)Glcp( l-1)Cer 53.

720

28 Solid-Phase Synthesis with Glycosyltransferases

NHAc

OH

43 n=O 44 n = 6

x:y= 1 :4 Ii

AcHN HO

AcHN

HO

N H 2O 46

Scheme 13. Enzymatic synthesis of a sialylated trisaccharide derivative, by stepwise sugar-chain elongation, using a water-soluble chymotrypsin-sensitive glycopolymer. Reagents: i) 1. HEPES buffer pH 6, UDP-Gal (1.5 eq.), GT, 37"C, 24 h; 2. cacodylate buffer pH 7.4, ST6, CIAP, CMPNeuAc (1.5 eq.), 37 "C, 48 h, gel filtration on Sephadex G-25, 9.y.; ii) 1. a-chymotrypsin, Tris HCl buffer pH 7.8, 4 0 T , 24 h; 2. gel filtration on Sephadex G-15, 87%.

In conclusion, compared with solution-phase synthesis, solid-phase enzymatic synthesis, generally, needs higher concentrations of enzyme, longer reaction times, and a large excess of donors. Although there is still room for improvement in terms of yields and scale-up of syntheses, in solid-phase synthesis purification procedures are straightforward; so it is expected that this approach will open the route to automated oligosaccharide synthesis in the near future.

28.4 Enzymatic Synthesis of Oligosucchurides

72 1

OH

OH 49 R = NHNHCO-P 51 R = NH(CH*)*NHCO-P

OH OH

OAR

p AcHN +o& oHO” & :

o HO

OH

OH

y

(

H

N

C

H

d 12CH3

y (CH2) 16CH3

53 0 Scheme 14. Enzymatic synthesis of lactosyl ceramide and GM3 using a water-soluble polyacrylamide gel. Reagents: i) 1. PAN, EtjN; 2. Sephadex (2-25; ii) cacodylate buffer pH 7, 3 mM MnC12, UDP-Gal (0.3 eq.), GT, a-lactalbumin, 0.1% mercaptoethanol, 37 “C,24 h, 36%; iii) 1. hv, THF-water, room temp., 9.5 h, 2. 50% CH3COONa-THF, CH3(CH2)16COCl, 54% for 50; iv) cacodylate buffer pH 6.5, ST3, CIAP, CMP-NeuAc (2 eq.), 55%).

722

28 Solid-Phase Synthesis with Glycosyltransferases

References 1. J. M. Frechet, C. Schuerch, Carbohydr. Res., 1972,22, 399-412. 2. G. Excoffier, D. Gagnaire, J.-P. Utille, M. Vignon, Tetrahedron, 1975, 31, 549-553. 3. R. Eby, C. Schuerch, Carbohydr. Res., 1975,39, 151-155. 4. S. J. Danishefsky, K. F. McClure, J. T. Randolph, R. B. Ruggeri, Science, 1993, 260, 13071309. 5. R. Liang, L. Yan, J. Loebach, M. Ge, Y. Uozumi, K. Sekanina, N. Horan, J. Gildersleeve, C. Thompson, A. Smith, K. Biswas, W. C. Still, D. Kahne, Science, 1996,274, 1520-1522. 6. J. Rademann, R. R. Schmidt, Tetrahedron Lett., 1996,23, 3989-3990. 7. Z.-W. Guo, Y. Nakahara, Y. Nakahara, T. Ogawa, Angew. Chem. Int. Ed. Engl., 1997, 36, 146441466, 8. W. Kosch, J. Marz, H. Kunz, Reactive Polymers, 1994, 22, 181-187. 9. M. Meldal, Methods Enzymol., 1997, 289, 83-104. 10. J. Vagner, G. Barany, K. S. Lam, V. Krchnak, N. F. Sepetov, J. A. Ostrem, P. Strop, M. Lebl, Proc. Null. Acad. Sci, USA, 1996, 93, 8194-8199. 11. U. Zehavi, S. Sadeh, M. Herchman, Carbohydr. Rex, 1983, 124, 23-34. 12. U. Zehavi, M. Herchman, Carbohydr. Res., 1986,151, 371-378. 13. U. Zehavi, M. Herchman, S. Kopper, Carbohydr. Rex, 1992, 228, 255-263. 14. S. Kopper, U. Zehavi, Reactive Polymers, 1994,22, 171-180 15. S. Kopper, Carbohydr. Rex, 1994,265, 161-166. 16. H. A. Nunez, R. Barker, Biochemistry, 1980, 19, 489-495. 17. S. David, C. AugC, C. Gautheron, Ado. in Carbohydr. Chem. Biochem., 1991, 49, 175-237. 18. 0. Blixt, T. Norberg, J. Carbohydr. Chem., 1997, 16, 143-154. 19. 0. Blixt, T. Norberg, J. Org. Chem., 1998,63, 2705-2710. 20. R. L. Halcomb, H. M. Huang, C.-H. Wong, J. Am. Chem. Soc., 1994, 116, 11315-11322. 21. M. Schuster, P. Wang, J. C. Paulson, C.-H. Wong, J. Am. Chem. SOC.,1994, 116, 1135-1136. 22. 0. Seitz, C.-H. Wong, J. Am. Chem. SOC.,1997,119, 8766-8776. 23. 0. Seitz, H. Kunz, J. Org. Chem., 1997, 62, 813-826. 24. U. Slomczynska, F. Albericio, F. Cardenas, E. Giralt, Biomed. Biochim. Acta, 1991, 50, 67-73. 25. M. Meldal, Tetrahedron Lett., 1992,33, 3077-3080. 26. M. Meldal, F.-I. Auzanneau, 0. Hindsgaul, M. Palcic, J. Chem. SOC.Chem. Commun., 1994, 1849-1 850. 27. F.-I. Auzanneau, M. Meldal, K. Bock, J. Peptide Sci., 1995, I , 31-34. 28. U. Zehavi, M. Herchman, Carbohydr. Res., 1984, 128, 160-164. 29. S.-I. Nishimura, K. Matsuoka, K. Kuritd, Macromolecules, 1990, 23, 41 82-4184. 30. K. Matsuoka, S.-I. Nishimura, Macromolecules, 1995, 28, 2961-2968. 31. S.-I. Nishimura, K. Matsuoka, Y. C. Lee, Tetrahedron Lett., 1994, 35, 5657-5660. 32. %-I. Nishimura, K. B. Lee, K. Matsuoka, Y. C. Lee, Biochem. Biophys. Rex Conimun., 1994, 199, 249--254. 33. K. Yamada, E. Fujita, S.-I. Nishimura, Carbohydr. Res., 1997, 305, 443-461. 34. K. Yamada, S.-I. Nishimura, Tetrahedron Lett., 1995, 36, 9493-9496. 35. A. Pollak, A. Blumenfeld, M. Wax, R. L. Baughn, G. M. Whitesides, J. Am. Chem. Soc., 1980, 102, 6324-6336. 36. U. Zehavi, M. Herchman, R. R. Schmidt, T. Bar, GZycoconjugateJ., 1990, 7, 229-234. 37. U. Zehavi, A. Tuchinsky, Glyconconjugate J., 1998, 15, 657-662.

Carbohydrates in Chemistry and Biology Editor Beat Ernst, Gerald W. Hart, Pierre Sinay copyright OWILEY-VCH Verlag GmbH, 2000

29 Glycosidase-Catalysed Oligosaccharide Synthesis David J. Vocadlo and Stephen G. Withers

29.1 Introduction The use of glycosidases for the synthesis of oligosaccharides has a long history, dating back over 85 years [ 11. The most difficult problem with their use as synthetic catalysts relates to their normal role being in hydrolysis. Given the 55 M concentration of water, hydrolysis is generally the preferred process in aqueous solution. A second problem can be that of control of the regiochemistry of bond formation. Several approaches have been followed in attempts to solve these problems, and these are discussed briefly within this article. More detailed accounts can be found in some useful recent reviews [2-41 and in some earlier overviews [5-7]. In this chapter we have chosen to present a brief review including some background on the enzymes themselves, the basic approaches currently used to solve the hydrolysis and regiochemistry problems, and some future perspective. The major content of this chapter is, however, a tabulation of a large number of examples of the use of glycosidases in the synthesis of specific linkages. Hopefully this will prove a useful resource for those interested in the application of such technology.

29.2 Background on Glycosidases Glycosidic bonds occur within a fantastically broad range of contexts in natural systems, as described in Volume 2 of this series. Given the large number of glycosidic bonds and the frequent need for relatively specific cleavage thereof it is no surprise that very large numbers of glycosidases exist to carry out this function. Amino acid sequences of well over 1000 different glycoside hydrolases are now available, primarily deduced from their gene sequences, and these have been arranged into some 70 different families on the basis of sequence similarities [8]. At the time this article is being written an excellent web site is available that gives up-

724

29 Glycosiduse-Cutulysed Oligosucchuride Synthesis

dated listings of these family members, plus some other information on the identities of important amino acid residues (http://afmb.cnrs-mrs.fr/-pedro/CAZY/ db.html). The site also provides links to the original papers and three-dimensional structures of enzymes, where available. This site therefore provides an excellent starting point in the search for a specific enzyme activity because each family contains enzymes of closely related activity and the full spectrum of specificities is represented.

29.3 Basic Mechanisms Hydrolysis occurs with one of two possible stereochemical outcomes, inversion or retention of anomeric configuration. All enzymes within a family hydrolyze their substrate with the same stereochemical outcome. These different outcomes demand different mechanisms, proposals for which were made by Koshland in 1953 [9]. Subsequent mechanistic and structural studies have largely substantiated these proposals and have identified the key active site residues as almost always a pair of carboxylic acids [ 10- 131. Glycosidases hydrolysing with inversion of anomeric configuration employ a mechanism involving direct displacement of the leaving group in an acid- or base-catalyzed process proceeding via an oxocarbenium ionlike transition state as shown in Figure 1. The two carboxylic acids are appropriately spaced (10-1 1 A)to perform their role as general acidic or basic catalysts. The vast majority of glycosidases that hydrolyze with net retention of anomeric configuration also have an active site containing a pair of carboxylic acids, but now only approximately 5 A apart. This smaller separation of the two groups reflects their different roles in the two-step double-displacement mechanism. One residue functions as a general acid catalyst, protonating the glycosidic oxygen, whereas the other functions as a nucleophile, attacking at the anomeric center to form a covalent glycosyl-enzyme intermediate as shown in Figure 2. In a second step, water attacks this intermediate in a general base-catalyzed process to yield the product of retained anomeric configuration. Both steps again proceed via oxocarbenium ion-like transition states (not shown). A variation on this mechanism is seen for some N-acetylhexosaminidases (at this stage apparently restricted to those from family 20 and chitinases from family 18). Such enzymes do not seem to contain a carboxylic acid that functions as a nucleophile. Rather, the Nacetyl group functions in that role, the reaction proceeding through an oxazoline intermediate as shown in Figure 3.

29.4 Synthesis by the ‘Thermodynamic’ Approach One approach that has been employed to favor synthesis over hydrolysis involves running the reaction in the presence of very high concentrations (typically molar) of

29.4 Synthesis by the ‘Thermodynamic’ Approach 0

3

14

0

k

P

0

3:

0

I

a:

tt

I

I

I

0

I

I

I 0

0

d oI $ z

t

0

Y

P

725

726

29 Glycosidase-Catalysed Oligosaccharide Synthesis

29.4 Synthesis by the ‘Thermodynamic’ Approach

727

OH

A. phoenicis a-Mannosidase

HO

HO OH

HO ‘OH

22% Yield Figure 4. Enzymatic synthesis of a-D-Man-(1 -2)-~-Man using a-mannosidase from Aspergillus phoenicis.

sugar under so-called ‘thermodynamic’ conditions such that the condensation reaction becomes favorable. Organic solvents can help in displacing this equilibrium, as long as the enzyme is compatible with, and the sugars soluble in, this medium. Similarly, high salt concentrations can be used to reduce the activity of the water and reduce product oligosaccharide solubility, thereby displacing the equilibrium [ 141. The advantage of this approach is its simplicity, hence its cost-effectiveness. Yields are, however, typically poor, rarely exceeding 15%, and control of regiochemistry is essentially non-existent. The approach is therefore of limited use for the coupling of two sugars because it frequently produces intractable mixtures. There are, however, occasional exceptions involving the coupling of two units of the same sugar; useful (20%) yields of a disaccharide have occasionally been obtained [15, 161, Figure 4. This approach can, however, provide an attractive route to the synthesis of glycosides of simple hydrophilic (ideally liquid) alcohols; reactions are performed in the presence of a very high concentration of the alcohol as shown in the example in Figure 5.

HO

A. oryzae P-Galactosidase *

HO

OH

HO-(CH,),-OH

”&

(cH2I6oH

I

HO

OH

48% Yield

Figure 5. Enzymatic synthesis of 6-hydroxyhexyl P-D-galactoside using P-galactosidase from Aspergillus oryzae.

728

29 Glycosidase-Catalysed Oligosaccharide Synthesis

Crout has recently shown that this approach can provide good (up to 61%) yields if the reaction is performed at an elevated temperature (50 "C) and time is spent in determining the optimum water activity to be used for the enzyme under study [ 171. This can be achieved either by directly adjusting the alcohol/water ratio or by varying the concentration of a co-solvent. This approach can therefore provide useful syntheses of simple glycosides as starting materials for chemical or enzymatic synthesis.

29.5 The Kinetic Approach The alternative approach, which can be employed only with retaining glycosidases, involves intercepting the glycosyl-enzyme intermediate that is formed at the active site with an acceptor moiety which reacts in place of water as shown in Figure 6. This approach relies upon the presence of an aglycone binding site that will bind a sugar in the correct orientation for one of the hydroxyl groups to react with the glycosyl-enzyme. Interactions between this second sugar and the protein will serve to stabilize the transition state for glycosyl transfer, reducing the activation barrier relative to that for hydrolysis in exactly the same way that such interactions serve to stabilize the transition state for disaccharide hydrolysis. The process therefore involves reaction of an activated donor sugar with a suitable acceptor sugar in the presence of the appropriate retaining glycosidase as shown in the example in Figure 7. The success of this approach depends upon two factors. Firstly, the transglycosylation process must occur in preference to hydrolysis of the glycosyl-enzyme. This can be controlled to some extent through the concentration of the acceptor sugar used, because occupancy of the acceptor site requires sugar concentrations higher than the dissociation constant for that site. The second important point is that the product disaccharide must function as a significantly poorer glycosyl donor than the activated donor substrate employed. This is best controlled by use of the most reactive (highest kCat/Km)donor possible, because the ratio of the rates of consumption of donor and product depends upon the relative kcat/Kmvalues for the two species and on the concentration of each species present at that time according to the expression:

Nitrophenyl glycosides and glycosyl fluorides are particularly attractive donors in this regard, not only because of their generally high k,,,/K,,, values, but also because the departed aglycone is itself a very poor acceptor and thus will not compete as an acceptor. It is also very easily separated from the products. Although some disaccharides, e.9. lactose, can serve as very inexpensive donors, complications can arise if the glucose liberated itself acts as an acceptor. A more unusual donor that has been used is the sialyl lactoside of a water-soluble polymer,

729

29.5 The Kinetic Approach

9 v)

VJ \

8

0 10

cd

~

t

I

0

4: s x 0

0 x

v)

c-

0

d8 +

d

3: 0

3: 0

730

29 Glycosiduse-Cutulysed Oligosacchuride Synthesis

which was used in conjunction with a leech ceramide glycanase for the transfer of sialyl lactose to ceramide, thereby yielding ganglioside GM3 in high (61%) yield [ 181. Sugar oxazolines have been used as donors for N-acetylhexosaminidases that react through the oxazoline mechanism. For example, incubation of Bacillus sp. chitinase with the oxazoline derivative of chitobiose resulted in rapid production of chito-oligosaccharides as shown in Figure 8 [ 191. If these factors are controlled well, respectable yields in the range of 10-60% can be obtained. This, however, generally requires exquisite timing of product harvesting or the use of some trick to ‘pull’ the equilibrium of the reaction in the synthetic direction. Because no universal approach exists, these ‘tricks’ these must be developed separately for each enzyme. One way of achieving this end again involves the use of an organic co-solvent to reduce the activity of the water, thereby minimizing the hydrolytic process. For example, when p-nitrophenyl p-galactoside was used as donor and 3-0-methyl glucose as acceptor in reactions catalyzed by the Kluyveromycesfragilis P-galactosidase a 14% yield of the 6-linked disaccharide product was obtained if reaction was performed in 67%)triethyl phosphate whereas only hydrolysis was observed when the reaction was conducted in aqueous buffer [20]. Another approach has been to remove one product continuously in some way. In a very few fortuitous cases this has occurred via crystallization of the product in situ; this is obviously not a general approach. Removal of disaccharide products by preferential absorption on to activated charcoal chromatography columns has also been reported [21]. A more generally useful approach has been to use a second ‘coupling’ enzyme, which will react with the product of interest only and, ideally, convert it directly to the desired product. This approach is particularly useful if the second enzyme is a glycosyl transferase, directly yielding the trisaccharide desired as shown below in Wong’s synthesis of sialyl LacNAc (Figure 9) [22]. Other examples include Paulson’s synthesis of sialyl 2,3-Gal P-OR [23] and Thiem’s preparation of the sialyl T-antigen [24]. Control of the regiochemistry of bond formation is more difficult to achieve. Probably the best method at present is to screen the available glycosidases, hopefully guided by the literature, for the enzyme that gives the product desired. We hope our compilation of enzymes will be useful in this process. A logical approach to such screening, if the product is available, is to screen the enzymes for their ability to cleave the product of interest. Although the principle of microscopic reversibility would indicate that an enzyme capable of hydrolysing the desired product can form the product of interest, a positive result would not necessarily prove that it would form the desired product preferentially. A somewhat improved approach, which does not require access to the product, is described in Section 29.6. All such screening approaches, however, are currently limited by the availability of enzymes. A second approach that has met with some success has involved manipulation of solvent conditions to change the regiochemistry of bond formation. For example, in studies on the Bacillus circulans pgalactosidase using p-nitrophenyl galactoside as donor and p-nitrophenyl GlcNAc as acceptor the Gal+- 1,4-GlcNAc-pNP product was obtained as the exclusive product (21% yield) when reaction was performed in the presence of 50% acetoni-

29.5 The Kinetic Approach

3: 0

8$ 03: 3:

d

0

X 0

z

0

731

732

29 Glycosidase-Catalysed Oliyosaccharide Synthesis

trile. When reaction was performed at lower (20%) acetonitrile concentrations, however, Gal-p-1,6-GlcNAc-pNP was instead obtained in 25.5% yield [25]. Finally, there have been several examples of success in manipulating regiochemical outcomes by changing the anomeric stereochemistry of the acceptor sugar. Thus, for example, Nilsson has shown that when methyl a-galactoside is used as the acceptor for the transfer of Gal from p-nitrophenyl a-galactoside by the coffee bean a-galactosidase, the predominant product (27%) is the 1-3 linked isomer whereas when methyl p-galactoside is the acceptor the 1-6 linked product is formed preferentially (18% compared with 9% of the 1-3 isomer). Similarly with E. coli pgalactosidase, using p-nitrophenyl p-galactoside as donor, the 1-6 isomer was dominant (14% yield) for transfer to methyl a-galactoside whereas when transfer was to methyl p-galactoside the 1-3 linked product predominated over the 1--6 (22% compared with 3%) [26]. In a further example Crout demonstrated a major difference in product distribution when the N-acetylgalactosaminidase from Aspergillus ovyzae transfers GalNAc to methyl aGalNAc or to methyl PGalNAc. Transfer to methyl a-galactoside resulted in 81% yield of the 1-4 linked product and only 9% of the 1-6 isomer whereas transfer to methyl P-galactoside resulted in almost equal amounts of the two isomers, but at lower yields (20Y0of the 1-4 isomer and 23% of the 1-6 isomer) ~71. In evaluating the synthetic potential of glycosidases using the ‘kinetic’ method it is important to recognize that the yields and even the regiochemical outcomes will vary significantlywith enzyme concentration and the reaction time employed. These differences in regiochemical outcome can arise because the transglycosylation products obtained are, themselves, substrates for the enzyme and can, therefore, be processed further. The net consequence will therefore tend to be a progression from an initial ‘kinetically controlled’ product mix towards a ‘thermodynamically controlled’ transglycosylation product mixture and ultimately to hydrolysis products. Consequently it is frequently observed that the product ratios obtained do not reflect expectations on the basis of cleavage specificities because the product that forms fastest will probably also be rearranged fastest. Different regiochemical outcomes arising from variation of solvent composition or acceptor structure could, therefore, be a consequence of the different relative rates of these processes rather than specific effects on enzyme structure or binding mode.

29.6 Recent Developments and New Directions A completely new approach to the problem of controlling the hydrolytic reaction has been developed in the Withers group through the use of glycosynthases, specific mutants of retaining glycosidases that are used in conjunction with glycosyl fluoride donors of the opposite anomeric configuration to that hydrolyzed by the wild-type enzyme [28]. These mutants are those in which the amino acid functioning as the catalytic nucleophile has been replaced by a residue with a non-nucleophilic sidechain, alanine being the substitution of choice at the time of writing. Such mutants

29.6 Recent Developments and New Directions

733

are completely hydrolytically inactive towards disaccharide substrates because they are unable to form the reactive glycosyl-enzyme intermediate. If, however, the mutant is presented with a glycosyl fluoride substrate of the same anomeric configuration as the intermediate that would ordinarily have formed, it is accepted by the enzyme as a surrogate for this intermediate and is transferred to a suitable acceptor as shown in Figure 10. Because the products so formed are completely hydrolytically stable, excellent yields (up to 92% isolated yields to date) can be obtained. This approach was published first with the P-glucosidase from Agrobacterium sp. [28]. A second example has recently appeared [29]and several others are being developed at the time of writing. A more direct large-scale screen for enzymes of interest has very recently been developed (Blanchard & Withers, unpublished). This involves first inactivating the enzyme under investigation via the trapping of a 2-deoxy-2-fluoroglycosyl-enzyme intermediate by treatment of the enzyme with the appropriate 2,4-dinitrophenyl 2deoxy-2-fluoro glycoside. Once freed of excess inactivator aliquots of this inactive, yet catalytically competent, intermediate species are incubated in the presence of a series of potential acceptors for a defined time period during which transglycosylation, releasing free enzyme, can occur, Figure 11. Each aliquot is then assayed with a standard substrate, positive results indicating that the ligand in question is capable of acting as an acceptor. The approach has been adapted to a 96-well format, enabling the facile screening of large numbers of acceptors, although the approach does not provide any insights into the regiochemistry of bond formation. Most published work to date has been on the use of exo-glycosidases, which usually cleave a single sugar residue at a time from the non-reducing terminus. Approaches involving the transfer of large blocks of sugars will, however, require the use of endo-glycosidases. Some very impressive work in this area has been pioneered by Takagawa and coworkers, who demonstrated that an endo-N-acetyl glucosaminidase from Arthrobacter protophormiae (endoA) has transglycosylation activity and could be used to synthesize neoglycoproteins. They used (Man)6(GlcNAc)2 as donor and transferred to GlcNAc(Asn)peptide acceptors, thereby producing (Man)6(GlcNAc)Z(Asn)-peptide products. They also showed that the approach could be used with partially deglycosylated ribonuclease B as acceptor thereby producing a modified glyco form, as shown below (Figure 12) [30-341. This approach was further developed by Y.-C. Lee for the production of (Man)g(GlcNAc)z-peptide analogs by use of pre-synthesized GlcNAc-peptide analogs as acceptors [35]. Lee also showed that organic solvents such as acetone could significantly improve (up to 89%) yields of products in such block transfer reactions [33, 361. This approach has enormous potential in the remodeling of glycoproteins. The role of oligosaccharides in biological processes has become a field of intensive research and accordingly facile methods for their preparation are highly desirable. The following tables should hopefully provide a useful resource for the researcher interested in synthesizing such oligosaccharides via an enzymatic or chemoenzymatic route. Table 1 lists linkages formed by exo-glycosidases, Table 2 includes all products formed by endo-glycosidases, and Table 3 includes alkyl glycosides synthesized by exo-glycosidases. References included in each table were selected primarily on the basis of interest and adequate product characterization.

HO CH3

F

Figure 11. Reactivation of a 2-fluoro-a-glucosyl-enzyme intermediate by transglycosylation to an acceptor sugar.

Figure 10. Glucosynthase-catalyzed synthesis of a cellobioside.

HO

0

-0A

%

s

F

v3

ru

P

W

4

29.6 Recent Developments and New Directions

0

(GlcNAc)2(Man),

Ribonuclease

EndoA

a

___)

GlcNAc

Ribonuclease

735

0

(GlcNAc)2(Man)6

End?

Ribonuclease

Figure 12. Remodeling of RNase B glycoforms through the use of endo A.

In each table the linkages formed are listed alphabetically with a-linkages listed first followed by P-linkages. In the linkage column, where the configuration of the anomeric carbon in the second saccharide unit is indicated, the acceptor molecule is either an oligosaccharide or a glycoside. Glycoside acceptors are differentiated by the inclusion of the hetero-atom involved in the glycosidic linkage after the second saccharide unit. Yields listed are calculated based on donor, acceptor, or in a few cases, transfer ratios. Identical products synthesized by different groups are listed as separate entries. When the same principal investigator has referred to the synthesis in different publications there is only one entry listing each reference.

12

11

10

9

8

I

6

5

4

3

2

1

Enzyme Source

Trehdlase ~ - ~ - 2 d G1-])-alc( Lobosphaera sp. D-2dGIC Trehalase u - D - ~ ~ Gl-I)-UIc( Lohospkaera sp. D-G~ ~ - ~ - 2 d G1-3)-Dlc( Glucoamylase Aspergillus niger 2dGlc a-~-2dGlc( 1-3)-~- a-Glucosidase Aspergillus niger 2dGlc Glucoamylase ~ - ~ - 2 d G1l-4)-Dc( Aspergillus niger 2dGlc u-D-2dGIc(1-4)-D- a-Glucosidase Asperg illus n ig er 2dGlc a-D-2dGk(I - 6 ) - ~ - Glucoamylase 2dGlc Aspergillus niger ~ - ~ - 2 d G1-6)-Dk( a-Glucosidase Aspergillus niger 2dGlc a-D-Gal(l-l)-a-D- a-Galactosidase Gal Candida guilliermondii H-404 P-Galactosidase a-D-Gal(1-1)-DEscherichia coli Fruj P-Galactosidase a-D-Gal(l-I)-DAsperg illus Fwf oryzae u-D-Gal(1-2)-U-D- a-Galactosidase Coffeebean Gal-0

Entry Linkage

16

~ - ~ - 2 d G Il -cI)-a-D-2dGk ( u-D-2dGk(1-l)-a-~-Glc ~ - ~ - 2 d G 1l -3)-D-2dGk c( a-D-ZdGIc(1-3)-~-2dGlc 1 -4)-~-2dGlc u-~-2dGlc( u-D-~~GIc( 1-4)-~-2dGk

2dGlc Glc 2dGlc 2dGlc 2dGlc 2dGlc

2dGlc 2dGlc 2dGlc 2dGlc 2dGlc 2dGlc

NA

1-6)-~-2dGlc ~~-~-2dG lc(

a-D-Gal(I-l)-a-D-Gal

U - D - G d ( 1-I)-D-Fruf

a-D-Gal(1-1)-u-Fruf

a-u-Gal( 1 -2)-a-~-Gal-OpNP

2dGlc Lactose

Fruj

Fruf

apNPGal

2dGlc Lactose

Gal Gdl

upNPGal

2

52

34

NA

NA

~ - ~ - 2 d G Il -6)-D-2dGk c(

2dGlc

2dGlc

NA

NA

NA

NA

6

Yield ("/')

Product

Accept01

Donor

~

Table 1. Enzymatic synthesis of oligosaccharides using em-glycosidases.

26

40

40

39

38

38

38

38

38

38

31

31

Reference

26

25

24

23

22

21

20

19

18

17

16

15

14

13

a-Galactosidase Coffee bean a-Galactosidase Coffee bean a-Galactosidase Coffee bean a-Galactosidase Coffee bean a-Galactosidase Coffee bean a-Galactosidase Cundidu guilliermondii H-404 a-D-Gal(1-3)-a-D- a-Galactosidase Coffee bean (2,6-OAll)-Gal0 a-u-Gal( 1-3)-a-~- a-Galactosidase (2-OAlI)-Gal-O Coffee bean a-u-Gal( 1-3)-a-u- a-Galactosidase (2-0Bn)-Gal-0 Coffee bean a-D-Gal(1 -3)-a-u- a-Galactosidase Gal-0 Coffee bean a-D-Gdl(1-3)a-D- a-Galactosidase GalNAc-0 Coffee bean a-D-Gal(1-3)-a-~- P-Galactosidase Bacillus GdlNAc-O circulans a-~-Gal( 1-3)-a-u- p-Galactosidase GalNAc-0 Bacillus circuluns a-~-Gal( 1-3)-a-D- a-Galactosidase Gal-0 Coffee bean

a-u-Gal( I -2)-a-DGal-0 a-D-Gal( 1-2)-a-DGlc a-D-Gal(I-z)-a-DGlc a-D-Gal(1-2)-a-DGlc a-~-Gal( 1-2)-a-~Glc a-D-Gal(1 -2)-DGal

a-D-GalNAc-OEt a-D-GalNAc-OBn

apNPGal PpNPGal

apNPGal

PpNPGal

a-u-Gal-OMe

a-u-GalNAc-OMe

a-~-Gal( 1-3)-a-~-GalNAcOMe

a-D-Gal-OAll

Raffinose

46 26

53 27

46

a-D-Gal(1-3)-a-~-GalNAcOEt a-D-Gal(1-3)-a-~-GalNAcOBn

a-u-(2-OBn)-Gal-OMe

apNPGal

62

a-u-Gal( 1-3)-a-~-(2-OAll)Gal-OMe a-~-Gal( 1-3)-a-u-(2-OBn)Gal-OMe a-D-Gal(1-3)-a-u-Gal-oAll

a-u-(2-OAll)-Gal-OMe

apNPGal

45

43

a-D-Gal(1-3)-a-D-(2,6-OAll)- 14 Gal-OMe

apNPGal

21

39

NA

a-D-Gal(1-2)-~-Gal

Lactose

Lactose

44

42

5

a-u-Gal( 1-2)-CIs

CIS

Melibiose

4.9

41

1.8

a-~-Gal( I-2)-y-CD

Y-CD

Melibiose

43

41

2.3

a-u-Gal(1-2)-P-CD

p-CD

Melibiose

12

41

2.2

a - ~ - C a l1(-2)-a-CD

a-CD

Melibiose

26

6

a-u-Gal(l-2)-a-~-Gal-OoNP

apNPGal

apNPGal

ii.

F'

0

(0

?

-

b

35

&

a 2 a

:

g

b m

C

3\

Fu

u,

31

36

35

34

33

32

31

30

29

28

21

Enzyme Source

a-D-Gal(1-3)-a-~- a-Galactosidase Coffee bean Gal-0 a-~-Gal( 1-3)-a-~- a-Galactosidase Coffee bean Gal-0 a-D-GaI(1-3)-a-~- a-Galactosidase Coffee bean Gal-0 a-D-GaI(1-3)-a-D- a-Galactosidase Coffee bean Glc a-D-Gal(1-3)-D-D- a-Galactosidase Coffee bean (2,6-OAll)-Gal0 a-D-Gal(1-3)-P-D- a-Galactosidase (6-OAIl)-Gal-O Coffee bean a-D-Gal(I -3)-P-D- a-Galactosidase Asperg illus Gal oryzae a-D-Gal(1-3)-8-D- a-Galactosidase Gal Coffee bean a-D-Gal(I -3)-P-D- a-Galactosidase Gal Coffee bean a-D-Gal(1-3)-P-D- P-Galactosidase Asperg illus Gal oryzae a-D-Gal(1-3)-p-D- P-Galactosidase Asperg illus Gal oryzae

Entry Linkage

Table 1 (continued)

P-~-(2,6-OAll)-Gal-OMe

P-~-(6-0All)-Gal-OMe P-D-Gal(1-4)-p-~-GlcNAcSEt

P-D-Gal(1-4)-P-~-GlcNTeoc- a-D-Gal(I-~)-P-D-G~I( I-4)-P- 20 49 D-GlcNTeoc-SEt SEt P-Lactose-SEt a-D-GaI(1-3)-P-~-Gal( 1 4 - P - NA 50 D-GIC-SEt ~-D-G~I(~-~)-P-D-G~I(~-~)-P15 0-Lactose-SEt 50 D-Glc-SEt P-Lactose-SBu

apNPGal apNPGal

apNPGal

PpNPGal

upNPGal

apNPGal

a-D-Gal(1-3)-P-~-Gal( 1-4)-PD-Glc-SBU

a-D-Gal(1-3)-P-~-(6-OAll)Gal-OMe a-~-Gal(l-3)-P-~-Gal( 14-PD-G~NAC-SE t

a-D-Gal(1-3)-P-~-(2,6-OAll)Gal-OMe

18

16

3

3

5

apNPGal

a-D-Gal(I -3)-CI8

CIS

50

48

43

43

42

41

Melibiose

apNPGal

a-D-GaI(1-3)-a-~-Gal-OpNP NA

26

26

Reference

apNPGal

1

Yield ("%)

apNPGal

Product

a-D-Gal(1 -3)-a-~-Gal-OoNP

apNPGal

Acceptor

aoNPGal

apNPGal

Donor

50

49

48

41

46

45

44

43

42

41

40

39

38

P-Galactosidase Asperg illus oryzae a-D-Gal(1-3)-P-D- a-Galactosidase Coffee bean Gal-0 a-~-Gal( 1-3)-8-~- a-Galactosidase Coffee bean Gal-0 a-Galactosidase a-D-Gal(1-3-DCundida guillierGal rnondii H-404 a-D-Gal(1-4)-P-D- a-Galactosidase Morttierella Glc vinacea a-~-Gal( 1-4)-p-D- a-Galactosidase Absidiu reflexa Glc a-D-Gal(1-4)-P-D- a-Galactosidase Morttierella Glc vinacea a-D-Gal(1-4)-P-D- a-Galactosidase Absidia refexa Glc a-D-Gal(1-4)-P-~- P-Galactosidase Bacillus GlcNAc-S circulans P-Galactosidase a-D-Gal(1-4)-DEscherichia coli Fruf a-D-Gal(1-4)-~- 0-Galactosidase Asperg illus Fruf oryzue P-Galactosidase a-D-Gal(1-5)-DEscherichia coli Fruf P-Galactosidase a-D-Gal(1-5)-DAsperg illus FWf oryzae

a-D-Gal(1 - 3 ) - P - D Gal

52

13-P-~-Glc-19-a-~-Ga1( 1-6)- 11 !3-D-Glc-O-steviol 13-u-~-Gal(l-6)-a-~-Gal(l- 3.3 6)-P-D-Gk-19-P-D-GIC-Osteviol 13-a-~-Gal(l-6)-P-~-Glc-1911 P-D-Glc-0-steviol a-D-Gal(1-4)-P-~-GlcNAc- 50 SEt

Rubusoside

Rubusoside

Raffinosel Melibiose Raffinosel Melibiose Raffinosel Melibiose

10

10

6.6 9.2

a-D-Gal(1-4)-~-Fruf a-D-Gal(1-4)-~-Fruf

a-D-Gal(l-S)-D-Fruf a-D-Gal(1-5)-~-Fruf

0-D-GlcNAc-SEt

Fruf Fruf

Fruf Fruf

Gal Gal

Gal Gal

Rubusoside

Raffinosel Melibiose apNPGal

40

40

40

40

48

52

52

52

13-a-~-Gal( 1-6)-P-~-Glc-19- 13 P-D-Glc-0-steviol

Lactose

Lactose

Rubusoside

39

NA

a-D-Gal(1-3)-~-Gal

P-D-Gal-OMe

26

apNPGal

9

0-D-Gal-OMe

apNPGal

a-D-Gal(1-3)-fi-~-Gal-OMe

P-Lactose-SPh

PpNPGal

~

62

61

60

59

58

51

56

55

54

53

52

51

~~~

~

~

~~

~

Enzyme Source

a-D-Gal(1-6)-a-~- a-Galactosidase Glc Coffee bean a-D-Gal(1-6)-a-~- a-Galactosidase Gal-0 Coffee bean a-D-Gal(1-6)-a-~- a-Galactosidase Coffee bean Glc a-D-Gal(1-6)-a-~- a-Galactosidase Coffee bean Glc a-D-Gal(I -6)-a-~- a-Galactosidase Glc Coffee bean a-D-Gal(1-6)-a-~- a-Galactosidase Glc Coffee bean a-D-GaI(1-6)-a-~- a-Galactosidase Glc Coffee bean a-~-Gal( 1-6)-a-~- a-Galactosidase Morttierella Glc vinacea a-D-Gal(1-6)-p-~- a-Galactosidase (2-OAlI)-Gal-O Coffee bean a-D-Gal(1-6)-p-~- a-Galactosidase (2-0Bn)-Gal-0 Coffee bean a-D-Gal(1-6)-P-~- a-Galactosidase Asperg illus Gal oryzae a-D-Gal(1-6)-P-~- a-Galactosidase Coffee bean Gal

Entry Linkage

~~

Table 1 (continued)

24

a-D-GaI(ld)-y-CD a-~-Gal( 1-6)-a-CD a-D-Gal(1-6)-a-CD a-D-Gal(1-6)-P-CD a-D-Gal(1-6)-y-CD a-D-Gal(1-6)-a-~-GIc( 1-2)-0-

7-CD a-CD a-CD 0-CD y-CD Sucrose

Melibiose Melibiose Melibiose Melibiose Melibiose

@-D-Gal( 1-4)-b-~-GlcNAcSEt

p-D-Gal(1-4)-P-~-GlcNTeoc- a-D-GaI(1-6)-P-~-Gal( 14-0SEt D-GlcNTeoc-SEt

apNPGal

a-D-Gal(1-6)-P-~-(2-OAll)Gal-OMe a - ~ - G a l1-6)-P-~-(2-OBn)( Gal-OMe a-D-Gal(1-6)-P-~-Gal( 1-4)-PD-GlcNAc-SEt

apNPGal

apNPGal

apNPGal

Gal

D-FrUf

NA

a-D-Gal(1-6)-a-~-Gal-OMe

a-D-Gal-OMe

apNPGal

3.5

16

14

12

18

22

26

38

2

49

48

43

43

55

41

41

41

54

53

26

53

(?h)

NA

a-D-Gal(I-6)-P-CD

Melibiose

~

Reference

0-CD

~~~~

Yield

Product

~~

Acceptor

~~

Donor

74

73

72

71

70

69

68

67

66

65

64

63

a-D-Gal(1-6)-P-D- a-Galactosidase Aspergillus niger Gal a-~-Gal( 1-6)-P-D- a-Galactosidase Coffee bean Gal a-D-Gal(1-6)-P-~- P-Galactosidase -Gal Asperg illus oryzae a-D-Gal(1-~)-P-D- 0-Galactosidase Gal Asperg illus oryzae a-D-Gal(1-6)-p-~- P-Gdlactosidase Asperg illus Gal oryzue a-D-Gal(1-6)-P-D- a-Galactosidase Gal-0 Coffee bean a-D-Gal(1-~)-O-D- a-Galactosidase Gal-0 Coffee bean a-D-Gal(1-6)-P-D- P-Galactosidase Escherichia coli Glc a-D-Gal(1-6)-~- b-Gdlactosidase Escherichia coli Fmf a-~-Gal( I-~)-D- 8-Galactosidase Asperg illus Fmf oryzae a-D-Gal(I-~)-D- a-Galactosidase Candidu yuillierGal mondii H-404 a-D-GalNAc(1a-N-Acetyl3)-a-~-Gal-0 galactosaminidase Chamelea gullinu 5.2

NA

w-D-Gal(1 -6)-~-Fruf

a - ~ - G a l1-6)-~-Gal (

P-D-Gal-OMe Sucrose Fruj Fruf

apNPGal Gal Gal

Gal

a-D-Gal-OMe

40

4.4

a-D-Gal(1-6)-a-~-Glc( 1-2)-PD-Fruf a - ~ - G a l1-6)-~-Fruf (

P-u-Gal-OMe

apNPGal

aoNPGalNAc

40

a-D-GaI(1-6)-P-~-Gal-OMe NA

P-Lactose-SPh

PpNPGal

Lactose

51

a-~-Gal(l-6)-P-~-Gal-OMe 18

P-Lactose-SBu

PpNPGal

Lactose

26

a-~-Gal(l-h)-fi-~-Gal(I-4)-P17 D-Glc-SPh

P-Lactose-SEt

PpNPGal

D-Glc-SBU

11

39

55

50

b

z

3

4

n a

$

2

,$

c:

2

i

co r: co

;tr

o\

9

Eu

~-D-G~I(~-~)-P-D-G~I(~-~)-P13 50

50

50

P-Lactose-SEt

50

apNPGal

a-D-Gal(l-6)-P-~-Gal( I-4)-P- 20 D-GIC-SEt a-D-Gal(1-6)-P-~-Gal( l-4)-@- NA D-GIC-SEt a-~-Gal( 1-6)-P-~-Gal( 14-P9 D-Gk-SEt

P-Lactose-SEt

clpNPGal

142

29 Glycosidase-Catalysed Oligosaccharide Synthesis

W

d

z

0 W

0

r-

2

W

N

0

5 4

-3 v 3

&

5 b

&

&

m

0

V

5

5 0,

m

r-

w r -

r - r -

W

r-

rch

g

z

63

64

11

14

a-D-GIc(1-4)-P-D-Glc-OpNP

w-D-GIc(1-4)-~-Fruf

PpNPGlc

PpN PGlc

Fruf

PpNPGlc

PpNPGlc

Glc

96

95

94

93

92

66 66

6.4 3.5 4.5

2 6

u-D-GIc(1-4)-~-Glc a-D-Glc(1-4)-~-Glc ~-D-GIC( 1-5)-~-Fruf

a-D-GlC(1-6)-a-D-GIC(1-4)-DGlc u-D-GIc(1-6)-a-~-Gk( 1 -6)-DGlc

Glc Glc Fruf

Glc Glc

Glc Glc Glc

Glc Glc

59

66

65

60

2.2

u-~-Glc-( 1-4)-~-Glc

Glc

Glc

SP.

Glucoamylase Rhizopus sp. Glucoamylase Rhizopus oryzue Glucoamylase Rhizopus oryzae a-Glucosidase Sacchuromyces SP. U-D-GlC( 1-6)-U-D- Glucoamylase Rhizopus oryzue Glc u-D-G~c( 1-6)-a-~- Glucoamylase Rhizopus oryzue Glc

91

a-D-GIC-(1-4)-DGlc u-D-GIc(1-4)-~Glc u-D-GIc(1-4)-~Glc u-D-GIc(1-5)-DFruf

60

1.8

Glc

a-~-Glc-( 1-4)-D-Gk

Glc

a-Glucosidase Succhuromy ces

u-D-G~c-( 1-4)-DGlc

SP.

59

62

a-D-Glc(1-4)-a-D-Glc(l-l)-a2 D-GIC a-D-Glc(l-4)-fi-~-Glc-OpNP 1 1

Trehalose

Glc

SP.

~-D-CIC( 1-4)-U-D- Glucoamylase Rhizopus nivrus Glc 1-4)-P-D- P-Galactosidase u-D-G~c( Asperg illus Glc-0 oryzae (Impurity?) u-D-G~c( 1-~)-P-D- P-Galactosidase Glc-0 Asperg illus oryzue (Impurity?) a-Glucosidase Succharomyces

62

Trehalose

U - D - G ~ ( ~ - ~ ) - ~ -l-l)-uD - G ~ C (2.1 D-Glc

Glc

u-D-GIc(1-4)-a-~- a-Glucosidase Succhuromyces Glc

90

89

88

87

86

85

t

s

107

106

105

104

I03

102

U-D-Gk-(I-6)-DGlc U-D-GIC(I-6)-DGlc a-D-GIc(1-6)-DGlc

SP.

Glucoamylase Rhizopus sp. Glucoamylase Rhizopus orvzae Glucoamy lase Rhiz0pu.r oryzae Glc Glc G Ic

Glc Glc Glc

a-D-Glc(1-6)-~-Glc

60

26

26

24

24

a-D-Glc-(1-6)-~-Glc

Glc

Fruf'

Glc

Glc

Glc

Glc

Glc

1 -6)-~-a- Glucoamylase u-~-Glc( Rhizopus oryzae Glc a-D-Glc(I-6)-D-a- Glucoamylase RhizopuJ oryzae Glc a-D-Gk(1 -6)-Da-Glucosidase Saccharonzy ces Fmf SP. u-~-Glc-( I-~)-D- a-Glucosidase Sacchuromycex Glc

101

66

65

60

65

65

61

61

59

Stevioside

Maltose

a-D-GIc(1-6)-P-D- P-amylase Aspergillus sp. Glc Glc

Stevioside

Maltose

62

1- 1)-a- 13 a-D-Glc(1-6)-a-~-Gk( D-GlC 13-P-D-GIC(I-2)-P-D-GIC-I9- NA a-D-Gk(1-6)-b-~-Glc-Osteviol NA 13-a-D-Gk(I -6)-P-D-Gk(12)-P-D-Gk-19-P-D-GIC-0steviol a-D-GIc(l-6)-a-u-Glc(1-4)-~- 3.5 Glc a-D-Glc(1-6)-a-~-Glc( 1-6)-~- 6 Clc a-~-Glc( I-6)-~-Fruf 5

Trehalose

Glc

SP.

a - ~ - G k1-6)-a-~( Glucoamylase Rhizopus niveus Glc a-D-GlC(1-6)-p-O- P-amylase Aspergillus sp. Glc

62

6.4

a-D-Glc(1-6)-a-~-Glc( 1-I)-aD-GlC

Trehalose

Reference

Yield ("/.)

Product

Acceptor

Glc

Donor

a-D-Gk(1-6)-a-o- a-Glucosidase Saccharomy ces Glc

Enzyme Source

100

99

98

97

Entry Linkage

Table 1 (continued)

a-D-Man(1-2)-aD-Man-0 a-D-Man( 1-2)-aD-Man-0 a-D-Man( 1-2)-DMan a-D-Man( 1 - 2 ) - ~ Man a-D-Man( 1 - 2 ) - ~ Man a-D-Man(1-2)-DMan

a-D-Man(1-2)-DMan a-D-Man( 1 4 - D Man a-D-Man(1-3)aMan a-D-Man( 1- 3 ) - ~ Man a-D-Man( 1 - 3 ) - ~ Man a-D-Man( 1 - 4 ) - ~ Man a-D-Man( ]-4)-uMan

110

116

122

121

120

119

118

111

115

114

113

112

111

109

a-D-Man( 1 -1)-aD-Man a-D-Man( 1 -2)-aD-Man

108

a-Mannosidase Jack bean a-Mannosidase Aspergillus phoenicis a-Mannosidase Jack bean a-Mannosidase Jack bean a-Mannosidase Jack bean a-Mannosidase Jack bean a-Mannosidase Jack bean a-Mannosidase Aspergillus phoenicis a-Mannosidase Aspergillus niger a-Mannosidase Jack bean a-Mannosidase Jack bean a-Mannosidase Jack bean a-Mannosidase Aspergillus niger a-Mannosiddse Jack bean a-Mannosidase Jack bean 26 26 68

15 68

a-D-Man( 1-2)-a-~-Man-OMe 18 8 16 1.2 NA 22

2

3.6 1.3

0.2 4.2 NA

a-D-Man(1-2)-a-~-ManOpNP a-D-Man( l - 2 ) - ~ - M a n a-D-Man(1 -2)-~-Man a-u-Man( 1-2)-~-Man u-D-Man( I-z)-~-Man

a-D-Man( 1-2)-~-Man a-D-Man(1 -3)-~-Man a-D-Man( 1-3)-~-Man

a-D-Man( 1-3)-~-Man a-D-Man(1- 4 ) - ~ - M a n a-D-Man( 1-4)-~-Man

a-D-Man-OMe apNPMan Man Man Man Man

Man Man Man Man Man Man Man

apNPMan apNPMan Man Man Man Man

Man Man Man Man Man Man Man

69

61

15

67

16

69

67

16

8

a-D-Man(1 -2)-a-~-Man( 1-2)D-Man

Man

Man

61

3.3

a-D-Man( l-l)-a-D-Man

Man

Man

2

8'

2

2

2

i

$ B

3

$-

C

m

b

2

m

r5

2

m

\o

ru

135

134

133

132

131

130

129

128

127

126

125

124

a-D-Man( 1-6)-a-CD a-D-Man( Id)-P-CD a-D-Man( 1-6)-y-CD a - ~ - M a nl-6)-a-u-Glc( ( 1-6)p-CD a-D-Man( I -6)-a-~-Glc( 1-4)a - ~ - G l c1(-6)-a-CD a-D-Man( I-6)-a-~-Glc( 1-4)a - ~ - G l c1-6)-p-CD ( a-D-Man( 1-6)-a-u-Man-OMe

a-CD p-CD ?I-CD a-D-Glc(1-6)-P-CD

I -6)-aa - ~ - G l c1(-4)-a-~-Glc( CD 1-6)-pa-D-Glc( 1-4)-a-~-Glc( CD a-D-Man-OMe

Man Man

a-D-Man-OMe a-u-Man-OMe

a-D-Man( 1-6)-o-Man a-D-Man(1-6)-Man

Man Man

p-D-Gal(1-4)-p-~-Xyl-OpNP a-~-NeuAc(2-3)-P-~-Gal( 14)-P-D-XyI-OpNP

Man aMuNeuAc

Man

33

7.6

9.8

14

15

67

68

14

a - ~ - M a n1-6)-u-Man (

Man

Man

PpNPGal

73

clpNPMan

a-D-Man(l-6)-@-~-Gal-OpNP 5

12

12

72

71

71

71

70

Reference

26

6.1

6.4

5.3

11

10

14

0.3

Yield (%)

4

apNPMan

a-D-Man-OMe

Man

a-D-Man( I-4)-o-Man

Man

Man

a-Mannosidase Rhizopus niveus a-Mannosidase Jack bean a-Mannosidase Jack bean a-Mannosidase Jack bean a-Mannosidase Jack bean a-Mannosidase Jack bean a-Mannosidase Jack bean a-Mannosidase Jack bean a-Mannosidase Jack bean a-Mannosidase Jack bean a-Mannosidase Jack bean a-Mannosidase Aspergillus niyer Sialidase Trypanosoma cruzi

a-D-Man(I -4)-DMan a-D-Man(1-6)-aD-Glc a - ~ - M a nl-6)-a( D-G~c a-D-Man(1-6)-aD-Glc a-D-Man( 1-6)-aD-G~c a - ~ - M a n1-6)-a( D-Glc a-D-Man(1-6)-aD-Glc u-D-Man(l-6)-aD-Man-0 a - ~ - M a n1-6)-8( D-Ga i-0 a-D-Man( 1 - 6 ) - ~ Man a-D-Man(1-6)-uMan a-D-Man(1-6)-0Man a-~-NeuAc(2-3)a-D-Gal

123

Product

Acceptor

Donor

Enzyme Source

Linkage

Entry

Table 1 (continued)

29.6 Recent Developments and New Directions

v,

r-

v,

r-

m 3

x

m

1;3

.;!

ar

v,

W

W

W

v, e

m

v

l

r-

r

-

0

0

r - r - r - r - r -

e

o

W

?

m

w w r - r -

W

r-

W

ar

m

&

m

h

m

747

157

156

155

154

153

152

151

150

149

148

Enzyme Source

a-~-NeuAc(2-3)P-D-GIC-O a-~-NeuAc(2-6)a-D-Gal-0 a-~-NeuAc(2-6)P-D-Gal a-~-NeuAc( 2-6)P-D-Gal a-~-NeuAc(2-6)(3-D-Gal

Sialidase Vibrio cholerae Sialidase Vibrio cholerae Sialidase Vibrio cholerue Sialidase Vibrio cholerae Sialidase Clostridium perfringens

a-~-NeuAc(2-3)- Sialidase Newcastle P-D-G~~ disease virus a-~-NeuAc(2-3)- Sialidase Newcastle P-D-Gal disease virus a-~-NeuAc(2-3)- Sialidase Newcastle P-D-Gal disease virus a-~-NeuAc(2-3)- Sialidase Vibrio cholerae P-D-G~I-O a-~-NeuAc(2-3)- Sialidase Trypanosoma cruzi P-D-Gal-0

Entry Linkage

Table 1 (continued)

P-D-Gal(1-3)-~-GlcNAc

apNPNeu Ac

a-~-NeuAc(2-3)-P-~-Gal( 13)-r,-GlcNAc

3)-D-Gk

a-~-NeuAc(2-3)-P-~-Gal( 1-

3)-D-Gk

a-~-NeuAc(2-3)-P-~-Gal( 1-

Product

12

7.7

2.7

MU)

Yield

apNPNeuAc

P-D-Gal-OMe

a-~-NeuAc(2-3)-P-~-Gal7 OMe NA CH~)~ a-~-NeuAc(2-3)- P - D - G ~ ~ - O - ( C H ~ ) ~ S ~ ( a-~-NeuAc(2-3)-P-D-Gal-Op-D-GaI(1-3)-p(CH2)2Si(CH3)3 D-GalNAc-0(CH2)2 Si(CH313 a-~-NeuAc(2-3)-P-~-Glc4 apNPNeuAc P-D-Glc-OMe OMe apNPNeuAc a-D-Gal-OMe a-~-NeuAc(2-6)-a-~-Gal- 2 1 OMe a-~-NeuAc(2-6)-P-~-Gal( 111 apNPNeuAc P-Lactose-OMe 4)-P-~-Glc-oMe 1a-~-NeuAc(2-6)-P-~-Gal( apNPNeuAc p-D-Gal(1-4)-~-GlcNAc 10 ~)-D-G~cNAc 18.5 a-~-NeuAc(2-6)-P-~-Gal( a-~-NeuAc(2-8)- p-D-Gal(1-3)-~-Gk 3)-D-GIC D-NeuAc

P-D-Gal(1-3)-~-Glc

Acceptor

apNPNeuAc

Donor

78

76

76

76

76

2 23

0,

2.

Tc

9E

g

e

R 5

0 g 7

B

g-

a,

(?

?

0

g

x9

76

78

78

78

Reference

00

P

4

169

168

167

166

165

164

163

162

161

160

159

158

a-~-NeuAc(2-6)- Sialidase Clostridium B-D-Gal perfringens a-~-NeuAc(2-6)- Sialidase P-D-Gal Clostridiuni peyfringens a-~-NeuAc(2-6)- Sialidase P-wGal Clostridium perfringens Sialidase Arthrobucter ureufuciens a-~-NeuAc(2-6)- Sialidase Arthrohurter P-D-Gal ureufuciens a-~-NeuAc( 2-6)- Sialidase Vihrio p-D-Ga1 cholerue a-~-NeuAc(2-6)- Sialidase Vibrio P-D-Gal cholerae a-~-NeuAc(2-6)- Sialidase P-D-Gal Newcastle disease virus a-~-NeuAc(2-6)- Sialidase B-D-Cal Newcastle disease virus a-~-NeuAc(2-6)- Sialidase Vibrio cholerae P-D-Gal-0 a-~-NeuAc(2-6)- Sialidase Vibrio cholerue P-D-Gk-0 a-L-Fuc(1-2)-P- a-Fucosidase Porcine liver D-Gal-0 78

4.1 9.2

a-~-NeuAc(2-6)-P-~-Gal( 13)-~-Glc a-~-NeuAc(2-6)-P-~-Gal( 13)-D-GIc a-~-NeuAc(2-6)-P-~-Gal( 13)-D-Gk

a-~-NeuAc(2-8)- p-D-Gal(1-3)-~-Glc D-NeuAc apNPNeuAc P-D-Gal(1-3)-D-GlC

p-D-Gal(1-3)-D-GlC

P-D-Gal-OMe P-D-Glc-OMe P-D-Gal-OMe

apNPNeuAc

apNPNeuAc apNPNeuAc a-L-Fuc F

2.5

a-~-NeuAc(2-6)-P-~-Gal- 15 OMe a-~-NeuAc(2-6)-P-~-GIc- 12 OMe a-L-Fuc(1-2)-P-~-Gal-OMe 6.5

3)-D-GIC

a-~-NeuAc(2-6)-P-~-Gal( 1-

78

7.4

a-~-NeuAc(2-6)-P-~-Gal( 13)-D-GIC

p-D-Gal(1-3)-~-Glc

apNPNeuAc

0.9

78

5.6

a-~-NeuAc(2-6)-P-~-Gal( 13)-D-GlC

a-~-NeuAc(2-8)- p-D-cal( 1 -3)-~-Glc D-NeuAc

79

76

76

78

78

78

78

18.7 u-~-Ne~uk(2-6)-P-~-Gal( 3)-D-Gk

b-~-Gal( 1-3)-D-Glc

upNPNeuAc

78

2.4

P-D-Gal(1 -3)-~-Glc

{a-~-NeuAc(28)),-~-NeuAc

1a-~-NeuAc(2-6)-P-~-Gal( 3)-D-Gk

78

5.5

a-~-NeuAc(2-6)-P-~-Gal( 13)-~-GlcNAc

a-~-NeuAc(2-8)- p-~-Gal( 1-3)-~-GlcNAc o-NeuAc

2

5g.

b -.

e

2 n

2Q

5

2

Y 3

6

2 n

c:

2

B

5

o\

!c

t u

179

178

177

176

175

174

173

172

a-L-Fuc( 1-2)-~- a-Fucosidase apNPFuc Gdl Corynebacterium

171

a-L-Fuc( 1- 3 ) - ~ - a-L-Fucosidase Glc Penicillium multicolor a-L-Fuc(l-3)-D- a-Fucosidase Aspergillus niger Glc a-L-Fucosidase a-L-FuC(1-3)-DPenicillium GlcNAc multicolor ~-L-FuC(I-~)-D- a-Fucosidase Aspergillus niger GlcNAc a-L-Fuc(l-S)-~- a-Fucosidase Aspergillus niyer GlcNAc a-Fucosidase a-L-Fuc(l-4)-PBovine testes ~-(6-0Bn)GlcN-S a-L-Fuc(l-4)-(3- P-Fucosidase Bovine kidney ~-(6-0Bn)GlcN -S a-Fucosidase a-L-Fuc( 1 -6)-PPorcine liver D-Gal-0

SP.

SP.

49

58 24

a-L-Fuc(1-3)-D-GlcNAc

a-L-Fuc( I -3)-D-GlcNAc a-L-Fuc(1-3)-~-GlcNAc a-L-Fuc(1-4)-P-~-(6-OBn)GlcN-SEt a-L-Fuc( 1-4)-P-~-(6-OBn)GlcN-SEt a-L-Fuc(1-6)-P-~-Gal-0Me

GlcNAc

GlcNAc GlcNAc f3-~-(6-OBn)-GlcN-SEt

p-~-(6-0Bn)-GlcN-SEt

P-D-Gal-OMe

apNPFuc

apNPFuc apNPFuc apNPFuc

apNPFuc

a-L-Fuc F

10

50

33

61

a-L-Fuc(1-3)-~-Glc

Glc

apNPFuc

18

u,-L-Fuc(1-3)-D-Gk

a-L-Fuc(1-2)-~-Gal

79

84

83

82

80

81

80

80

80

Refer(YO) ence

Yield

a-L-Fuc(l-2)-P-~-Gal-OMe 25

Product

Glc

Gal

P-D-Gal-OMe

Acceptor

apNPFuc

apNPFuc

a-Fucosidase Corynehacterium

a-L-Fuc( 1-2)-PD-Gal-0

170

Donor

Enzyme Source

Entry Linkage

Table 1 (continued)

25;.



%$@

g

9

3

r5

0

=: 2

‘p n

tr,

g

tsg CI

v3

ly

3

4 v,

190

189

188

187

186

185

184

183

182

181

I80

SP.

P-~-Fruf(2-l)-a- P-Fructofuranosidase ArthroD-Glc bacter sp. K-1 P-D-Fmf(2- I)-a- P-FructofuranoD-G~c sidase Arthrohacter sp. K-l P-Fructofuranosidase Aspergillus sydowi P-~-Fruf(2-I)+ P-FructofuranoD-Fruf sidase Aspevgillus sydowi P-D-Fruf( 2- 1)-PP-FructofuranoD-FWf sidase Aspergillus oryzae P-Fructofuranosidase Aspergillus oryzae

SP.

P-Glucosidase Agrobacteriuni

Sucrose

Sucrose

Sucrose

Sucrose

Sucrose

Sucrose

Glucal

Glucal

SP.

P-Glucosidase Agrobacterium

Sucrose

Sucrose

Trehalose

Trehalose

a-D-Gk(1-6)-D-GIc

Maltose

P-D-Xyl-SBn

P-D-Glc-SPh

P-D-Xyl-SBn

87

NA

P-~-Fruf’(2-1 )-P-~-Fruf(z-1 )P-~-Fruf(2-l)-a-D-Gk

Glc(l-l)-a-~-Glc P-D-Fruf( 2- 1)- P-D-Fruf( 2- 1)a-D-Glc

17

21

88

88

87

86

86

85

85

85

NA

51

55

~(2-1)-b-~-Fruf(2-6)-a-~-

P-~-Fruf’(2l)-P-D-Fru$P1-I)-a-D~(2-6)-a-~-Glc( Glc P-~-Fruf(2-l)-P-D-Fruf-P-

P-~-Fruf(2-l)-a-D-Glc( 1-6)D-Glc

P-~-Fruf(2-l)-a-~-Glc( 1-4)D-GlC

16

12

2-deoxy-P-~-Glc(1-~)-P-DGlc-SPh 2-deoxy-P-~-Glc(1-~)-P-DXyl-SBn

16

2-deoxy-P-~-Glc(1-3)-P-DXyl-SBn

85

15

1- 3 ) - f i - ~ 2-deoxy-P-~-Gk( Glc-SPh

Glucal

Glucal

80

14

a-L-Fuc(1 -6)-b-~-Gal-OMe

apNPFuc

P-Glucosidase Agrobacterium

SP.

CL-L-FUC(I-~)-P-a-Fucosidase Ampullaria D-Gal-0 P-D-2dGIC(1 4 4 - P-Glucosidase Agrobacterium D-GlC-S

199

198

197

196

195

194

193

192

191

55

7.5

P-~-Fruf(2-I)-D-xyl

P-~-Fruf(2-1)-L-Ara

@-~-Fruf(21)-L-Fuc

P-~-Fruf(2-2)-L-Sor

P-~-FruJ’(2-3)-~-Gal

Gal

XYI L-Ara

L-FUC

L-Sor

Gal

L-Ara

Sucrose

Sucrose

Sucrose

Sucrose

Sucrose

Sucrose

Sucrose

P-~-Fruf(2-4)-L-Ara

27

86

70

86

86

86

86

86

86

86

86

Reference

33

62

52

40

P-~-Fruf(2-l)-D-Ara

D-Ara

Sucrose

P-~-Fruf(2-l)-D-Gal

44

P-~-Fruf(2-1)-P-~-Gal( 1-4)D-G~c

Lactose

Sucrose

P-~-Fruf(2-1)-PD-Gal

P-Fructofuranosidase Arthrobacter sp. K-1 P-D-Fruf( 2- l)-D- P-Fructofuranosidase ArthroAra bacter sp. K-1 P-D-Fruf( 2- 1)-D- P-Fructofuranosidase ArthroGal hacter sp. K-1 P-~-Fruf(2-l)-D- 0-Fructofuranosidase ArthroXYl bacter sp. K-1 P-~-Fruf(2-1)-L- P-Fructofuranosidase ArthroAra bacter sp. K-1 P-~-Fruf(2-l)-L- P-Fructofurdnosidase ArthroFuc hacter sp. K-1 P-D-Fmf(2-2)-L- 0-Fructofuranosidase ArthroSor bacter sp. K-1 P-o-Fruf(2-3)-~- P-Fructofuranosidase ArthroGal bacfer sp. K-1 Pm-FrUf(2-4)-L- P-Fructofuranosidase ArthroAra bacfer sp. K-l

Yield (“%)

Product

Acceptor

Donor

~

Enzyme Source

Entry Linkage

~~

Table 1 (continued)

29.6 Recent Developments and New Directions

t-

oo

00 m

0

m

3

m

W

00

oo Q\

N

m

m

m

m

m

N

a

+

3

A

v

3

3

0 N 0

u 9

-9

ci

ci

w

N N 0

N 0

4 v

6A ci

u

r-

00

0 m

N

N

N

4

-x h

9

u m N 0

d 0 N

u

ci In

0

N

W N 0

0

0

753

219

218

217

216

215

214

213

212

21 1

210

Enzyme Source

P-D-Gal(1-2)-a-D- P-Galactosidase Penicillium Glc multicolor p-D-Gal(1-2)-a-D- P-Galactosidase Bacillus Glc circulans P-Galactosidase p-D-Gal(1-2)-DEscherichiu coli Glc P-Galactosidase P-D-Gal(1-2)-DBacillus Glc circulans 0-Galactosidase P-D-Gal(1-2)-DBacillus Glc circulans p-~-Gal( 1-2)-~- 0-Galactosidase Bovine testes XYl P-Galactosidase P-D-Gal(1-2)-DIntestinal lactase XYl Lamb P-Galactosidase P-D-Gal(1-2)-DAspery illus XYl oryzue 0-Galactosidase P-D-Gal(1-2)-DSuccharomyces XYl fruyilis P-Galactosidase p-D-Gal(1-2)-DEscherichiu coli XYl

Entry Linkage

Table 1 (continued)

40

0.5

1

3

PoNPGal

PoNPGal

95

9

PoNPGal

95

95

95

95

4

PoNPGal

18

94

NA

Lactose

PoNPGal

94

NA

Lactose

Gal

42

42

Reference

17

(“h)

Yield

Lactose

Product

6.7

Acceptor

Lactose

Donor

P-D-GaI(1-2)-LXYl P-D-Gal(1-3)-(60Ac)-Glucal P-,-Gal( 1-3)-a-DGalNAc-0 B-D-Gal(1-3)-a-DGalNAc-0

225

233

232

23 I

230

229

228

221

P-D-Gal(1-3)a-DGalNAc-0 P-D-Gal(1-3)-a-DGalNAc-0 P-D-Gal(1-3)-a-DGalNAc-0 P-D-Gal(1-3)-a-DGalNAc-0 P-D-Gal(1-3)-U-DGalNAc-0

P-D-Gal(1-2)-LXYI

224

226

P-D-Gal( I -2)-LXYI

XYI P-D-Gal(1 -2)-LXYl P-D-Gal(1-2)-LXYl

P-D-Gdl( 1-2)-D-

223

222

22 1

220

P-Galactosidase Escherichia coli P-Galactosidase Bovine testes P-Galactosidase Intestinal lactase Lamb P-Galactosidase Aspery illus oryzae P-Galactosidase Saccharomyces fragilis P-Galactosidase Escherichia coli P-Galactosidase Escherichia coli P-Galactosidase Bovine testes P-Galactosidase Bacillus circulans P-Galactosidase Bovine testes P-Galactosidase Bovine testes P-Galactosidase Porcine testes P-Galactosidase Bovine Testes P-Galactosidase Bovine Testes PpNPGal

PpNPGal

Lactose

PpNPGal

PpNPGal

PpNPGal

PpNPGal

PpNPGal

PoNPGal

apNPGalN Ac

a-D-GalNAc-0-Thr

a-D-GalNAc-0-Ser-(N-Z)OAll

a-D-GalNAc-OEt

L-Xyl

L-Xyl

PoNPCa1 PoNPGal

L-Xyl

L-Xyl

[JoNPGal

PoNPGal

PpNPGal

P-D-Gal(1-2)-L-Xyl

P-D-Gal(1-2)-L-Xyl

P-D-Gal(1-2)-L-Xyl

P-D-Gal(1-2)-L-Xyl

11

3

5

2

95

95

95

95

244

243

242

24 1

240

239

238

237

236

235

p-D-Gal(1-3)-a-D- P-Galactosidase Penicillium Glc multicolor

oryzae

p-D-Gal(1-3)-a-~- P-Galactosidase Bovine Testes GalNAc-0 P-D-Gal(1-3)-a-~- P-Galactosidase GalNAc-0 Bovine Testes P-D-Gal(1-3)-a-D- P-Galactosidase Bovine Testes GalNAc-0 P-D-Gal(1-q-a-D- P-Galactosidase GalNAc-0 Bovine Testes P-D-Gal(1-3)-a-D- P-Galactosidase GalNAc-0 Bovine Testes P-D-Gal(I-3)-a-D- 0-Galactosidase Bovine testes GalNAc-0 p-~-Gal( 1-3)-a-D- P-Galactosidase Aspergillus Gal-0 oryzue p-D-Gal(1-3)-a-D- P-Galactosidase Gal-0 Aspergillus oryzae p-D-Gal(1-3)-a-D- P-Galactosidase Bovine Testes Gal-0 P-D-Gal(1-3)-a-~- P-Galactosidase Aspergillus Glc

234

Enzyme Source

Linkage

Entry

Table 1 (continued)

Lactose

Gal

PpNPGal

PVGal

PoNPGal

Lactose

PpNPGal

PpNPGal

PpNPGal

PpNPGal

PpNPGal

Donor

Trehalose

9.7

1.8

15

a-~-Gal-OAll

p-~-Gal( 1-3)-a-~-Gal-OAll

93

42

62

101

102

101

101

101

101

6

101

Reference

a-~-Gal-OMe

Yield (%)

93

Product

P-D-Gal(1-3)-a-~-GalNAc-O- 35 (CHz)zO(CHz)2N3 P-D-Gal(1-3)-a-~-GalNAc-O- 22 CH2CH(NHz)COOH P-D-Gal(1-3)-a-~-GalNAc- 21 OpNP P-D-Gal(1-3)-u-~-GalNAc- 22 OAll a-D-GalNAc-0p-D-Gal(1-3)-a-~-GalNAc-o- 28 CH(CH3)CH(NH2)COOH CH(CH3)CH(NH2)COOH a-D-GalNAc-0-Ser-(Np-~-Gal( 1-3)-a-~-GalNAc-O- 20 Aloc)-OMe Ser-(N-Aloc)-OMe a-D-Gal-OMe p-D-Gal(1-3)-a-~-Gal-OMe 6 a-D-GalNAc-0(CHz)20(CHz)zN3 a-D-GalNAc-0CH2CH(NH2)COOH apNPGalNAc

Acceptor

258

257

256

255

254

253

252

25 1

250

249

248

247

246

245

p-D-Gal(1-3)-a-D- 0-Galactosidase Bovine Testes GlcNAc-0 P-u-Gal(1-3)-a-D- P-Galactosidase Bovine Testes GlcNAc-0 P-D-Gal(1-3)-a-L- P-Galactosidase Barley FUC-0 p-D-Gal(1-3)-P-D- P-Galactosidase (6-O-Ac)-Gal Escherichia coli p - D - G d ( 1-3)-P-D- P-Galactosidase Barley (6-0Ac)-Gal-0 P-D-Gal(1-3)-0-D- P-Galactosidase (6-OBn)-GlcNBovine testes S P-u-Gal(1-3)-P-D- P-Galactosidase Bijidobacterium Gal bzjidum P-D-Gal(1-3)-P-D- P-Galactosidase Bacillus Gal circulans p-D-Gal(1-3)-P-D- P-Galactosidase Bovine testes GalNAc-0 P-D-Gal(1-3)-P-D- P-Galactosidase GalNAc-0 Porcine testes D-D-Gal(1-3)-8-D- P-Galactosidase Gal-0 Asperg illus oryzae P-u-Gal(1-3)-8-D- P-Galactosidase Escherichia coli Gal-0 P-D-Gal(1-~)-P-D- 0-Galactosidase Escherichia coli Gal-0 P-D-Gal(1-3)-P-D- P-Galactosidase Escherichia coli Gal-0 1-~)-P-DP-~-(6-0-Ac)-Gal( Xyl-OMu P-~-(6-0Ac)-Gal-OMe

PpNPGal

PoNPGal

PoNPGal P-D-Gal-OMe P-D-Gal-OAII

PoNPGal PoNPGal Lactose

a-D-GalNAcOpNP

P-D-GalNAc-0-EtBr

NeuAc

PoNPGal

Lactose

PoNPGal

Lactose

Lactose

PpNPGal

51

28

2.5

30

4

P-D-Gal(I -3)-P-~-Gal-ooNP 14

63

100

45

104

84

51

103

101

32

16

101

I3

P-D-GaI(1-3)-P-o-GalNAc-O- 2.5 EtBr p-D-GaI(1-3)-P-~-GalNAc- 13 OpNP P-D-Gal(l-3)-P-~-Gal-ooNP 11

P-D-Gal(1-3)-P-~-Gal( 1-8)-DNeuAc

a-L-Fuc-OMe

PpNPGal

Lactose

p-D-GaI(1-3)-!3-~-(6-0-Ac)Gal( 1-4)-P-~-Xyl-OMu p-~-Gal( 1-3)-P-~-(6-OAc)Gal-OMe P-D-Gal(1-3)-P-~-(6-OBn)GlcN-SEt

a-D-GlcNAc-OBn

PpNPGal

PpNPGal

P-D-Gal(1-3)-a-~-GlcNAcOAll p-D-Gal(1-3)-a-~-GlcNAcOBn p-D-Gal(1-3)-a-L-Fuc-OMe

a-D-GlcNAc-OAll

PpNPGal

3f

=L

2 a s

3

2

C

g

B

o\

u,

Fu

p-D-Gal(1-3)-P-D- P-Galactosidase Bacillus Gal-0 circulans P-D-GaI(1-3)-p-D- P-Glucosidase A grobacterium Gal-S SP. P-D-Gal(1-3)-P-D- P-Glucosidase A yrobacterium Gal-S SP. p-~-Gal( 1-3)-p-D- P-Galactosidase GIcNAc-0 Bovine testes p-D-Gal(14-P-D- 0-Galactosidase Bovine testes GIcNAc-0 P-D-Gal(1-3)-P-D- P-Galactosidase Bovine Testes GIcNAc-S P-D-Gal(1-3)-P-D- P-Galactosidase Bacillus GIG-0 circulans

269

268

267

266

265

264

P-D-GlcNAc-0-EtSiMe3 P-D-GlcNAc-OMe P-D-GlcNAc-SEt 0-D-Glc-OMe

PpNPGal Lactose Lactose

P-D-Gal(1-3)-P-~-GlcNAc-0- 7 EtSiMe3 p-D-Gal(1-3)-P-~-GlcNAc- about OMe 10 p-D-Gal(1-3)-P-u-GlcNAc17 SEt p-D-Gal(1-3)-P-~-Glc-OMe 19

8.8

P-D-Gal(1-3)-P-~-Gal-SBn

92

108

45

45

85

85

9.5

P-D-Gal-SPh

P-D-Gal(1-3)-D-~-Gal-SPh

2

2 5.

9

n &

s-

0 r5

3 u

0

Q

z is

k

k?

%.

%-

92

PoNPGal

PpNPGal

PpNPGal

Lactose

106

44

Q

h, \n

00

a

5.8

44

Reference

107

Lactose PoNPGal

p-~-Gal( 1-3)-P-~-Gal-oEtSi(Me)3

P-D-Gal-O-EtSi(Me)3

Lactose

263

262

261

260

10

Yield (“A)

wl

4

P-D-Gal-O-Ser-(N-Aloc)P-D-Gal(l-3)-P-D-Gal-O-Ser- 13 (N-Aloc)-OMe OMe monoallyl ether (p-~-Glc-O)- monoallyl ether (P-~-Gal(l- 2 3)-P-~-Glc-0)-1,3,5I , 3,Sbenzene-trirnethanol benzene-trimethanol p-D-Gal-OMe 5.5 P-D-Gal(l-3)-P-~-Gal-OMe

P-D-Gal(l-3)-P-~-Gal-OBn

P-D-Gal-OBn

Lactose

P-Galactosidase Escherichia coli P-Galactosidase Escherichia coli P-Galactosidase Escherichia coli 0-Galactosidase Bovine liver

p-o-Gal(1-3)-P-DGal-0 P-D-Gal(1-3)-P-DGal-0 P-D-GaI(1-3)-P-DGal-0 p-D-Gal(1-3)-P-DGal-0

Product

Acceptor

Donor

Enzyme Source

259

Entry Linkage

Table 1 (continued)

280

279

278

277

276

215

274

273

272

27 1

270

aGalF

PpNPGal

P-D-Gal(1 -~)-P-D- P-Glucosynthase Agrobacterium Xyl-0

P-D-Gal(1-3)-P-D- P-Galactosidase Escherichia coli Xyl-0 p-~-Gal( 1-3)-P-D- 0-Glucosidase A yrobacterium Xyl-s

GalNAc GalNAc

PpNPGal

P-D-Xyl-SBn

Lactose

PpNPGal

P-D-Gal(1-3)-P-D- P-Galactosidase Bacillus Xyl-s circulans P-Galactosidase p-~-Gal( 1-q-DBovine testes GalNAc P-Galactosidase P-D-Gal(1-3)-DBijidobacterium GalNAc blfidurn

SP.

PpNPGdl

P-D-Gal(1-3)-P-D- P-Glucosidase A yrobacterium Xyl-s

SP.

SP.

SP.

PpNPGal

PpNPGal

SP.

p-D-Gal(1-3)-P-D- P-Glucosidase Agrobacterium Xyl-0

38

P-D-Gal(l-3)-P-~-Glc-SPh

P-D-GIG-SPh

PpNPGal

PpNPGal

3.7

b-~-Gal( 1-3)-(3-Glc-OpNP

PpNPGlc

Lactose

p-D-Gal(1-3)-8-D- (3-Glucosidase A yrobacterium Glc-S

SP.

P-Galactosidase Bacillus circulans 1-3)-(3-D- P-Glucosidase (3-~-Gal( Agrobacterium Glc-S

p-~-Gal( 1-3)-(3-DGlc-0 85

109

P-D-GaI(1-3-DGalNAc

P-D-Gal(1-3)-DGalNAc

fl-~-Gal( 1-q-DGlc P-D-Gal(1 -3)-DGlc

282

283

284

b-~-Gal( 1-q-DGlc

p-~-Gal( 1-q-DGlcA

P-D-GaI(1-q-DGlcNAc

b-~-Gal( 1-3)-DGlcNAc

286

287

288

289

285

p-D-Gal(1-3)-DGalNAc

28 1

Entry Linkage

Table 1 (continued)

113

113

104

94

105

113

108

trace

3.7

1.4

0.5 NA

NA

23

4.4

12

P-D-Gal(1-3)-~-GalNAc

p-D-Gal(1-3)-~-GalNAc

b-~-Gal( 1-3)-~-GalNAc

b-D-Gal(1- 3 ) - ~ - G k p-D-Gal( 1- 3 ) - ~ - G k

b-~-Gal( I-~)-D-G~

P-D-Gal(1-3)-~-GlcA

p-D-Gal(1-3)-~-GlcNAc

b-~-Gal( 1-3)-~-GlcNAc

GalNAc

GalNAc

GalNAc

Glc Lactose

Lactose

GlcA

GIcNAc

GlcNAc

PpNPGal

PpNPGal

PpNPGal

Gal Lactose

Lactose

Lactose

PpNPGal

Lactose

P-Galactosidase Penicillium multicolor P-Galactosidase Streptococcus 6646K P-Galactosidase Asperg illus oryzae P-Galactosidase Escherichia coli P-Galactosidase Bijidobacterium bijidum P-Galactosidase Bacillus circuluns P-Galactosidase Bacillus circuluns P-Galactosidase Streptococcus 6646K P-Galactosidase Bovine Testes

40

113

Reference

Yield ("%)

Product

Acceptor

Donor

Enzyme Source

29.6 Recent Developments and New Directions

761

311

310

309

308

307

306

305

304

303

302

Enzyme Source

P-Galactosidase Escherichia coli P-Galactosidase Barley P-Galactosidase Snail P-Galactosidase Bacillus circulans P-D-Gal(1-4)-a-D- P-Galactosidase Bacillus Glc circulans P-D-Gal(1 -4)-a-D- P-Galactosidase Glc Bacillus circulans p-D-Gal(1-4)-a-~- P-Galactosidase Glc Bacillus circulans P-D-Gal(1-4)-a-D- 0-Galactosidase Bacillus Glc circulans P-D-Gal(1-4)-a-D- P-Galactosidase Bacillus Glc circulans P-D-Gal(1-4)-a-D- P-Galactosidase Bacillus Glc circulans

P-D-Gal(1-3)-LXYl P-D-Gal(1-4)-U-DGal-0 p-D-Gal(1-4)-a-~Gal-0 P-D-Gal(1-4)-a-DGlc

Entry Linkage

Table 1 (continued)

22

20

P-D-Gal(1-4)-a-D-Glc(1-4)-a~ - G l c1-6)-a-CD ( 1-6)-Pp-~-Gal( ]-4)-a-~-Glc( CD p-D-Gal(I-4)-a-~-Glc( 1-4)-a~ - G l c1-6)-P-CD (

P-~-Gal(l-4)-a-~-Glc(l-6)-y17 CD P-D-Gal(l-4)-a-D-GIc(1-4)-aD-G~c( 1-6)-y-CD

a-~-Glc( I-4)-a-~-Glc( 1 -6)-aCD u-D-GIc(1-6)-P-CD

a-D-Glc(1-4)-a-~-Glc( 1-6)-8CD a-D-Glc(1-6)-y-CD

1-6)-ya-~-Glc( l-4)-a-~-Glc( CD

Lactose

Lactose

Lactose

Lactose

Lactose

21

22

20

P-~-Gal(1-4)-a-D-Gk( 1-6)-aCD

a-D-Glc(1-6)-a-CD

Lactose

. I

12

Lactose

NA

P-D-Gal(l-4)-{a-~-Glc( 1 4)}4-a-D-Glc-OpNP

a-D-Gal-OMe

PpNPGal

P-D-GaI(l-4)-a-~-Gal-OMe

15

{u-D-GIc(1-4)}4-a-~-GlcOpNP

a-D-Gal-OMe

PpNPGal

P-D-Gal(1-3)-L-Xyl

(oh)

Yield

NA

L-Xyl

PoNPGal

Product

P-D-Gal(1-4)-a-~-Gal-OMe

Acceptor

Donor

115

115

1 15

115

115

115

114

51

51

95

Reference

G.

2

PB

&

e

;s

ff0

0 =: 9 h u

B

g-

y

A

E 3

E

0

s-g.

a N

4

32 1

320

319

318

317

316

315

314

313

312

P-D-Gal(1-4)-a-D- P-Galactosidase Asperg illus Glc oryzae p-D-Gal(1-4)-a-D- P-Galactosidase Bacillus GlcNAc-0 circulans p-~-Gal( 1-4)-a-D- P-Galactosidase Bacillus Glc-S circulans P-wGal(1-~)-P-D- P-Glucosynthase Agrobacterium 2-deoxy-2-ClGlc-0 SP. P-D-GaI(1-4)-P-D- P-Galactosidase Bacillus Gal circulans P-D-Gal(1-4)-P-D- P-Galactosidase Bacillus Gal circulans P-D-Gal(1-4)-P-D- 0-Galactosidase Bacillus Gal circulans p-~-Gal( 1-4)-P-D- 0-Galactosidase Bacillus Gal circulans P-u-Gal(1-~)-P-D- P-Galactosidase Bacillus Gal circulans p-~-Gal( 1-4)-P-D- 0-Galactosidase Bacillus Gal circulans 111

28

105

30

P-~-2-deoxy-2-C1-Glc-O-2,4- p-~-Gal( 1-4)-P-~-2-deoxy-2- 64 CI-Glc-O-2.4-DNP DNP 5

p-~-Gal( 1-4)-u-~-Glc-SEt

a-D-Glc-SEt

aGalF

a-D-Glc(1-6)-y-CD

Lactose

5x 115

a-~-Glc( 1-4)-a-o-Glc(l-6)-PCD

C

115

2

0

3

p-~-Gal( 1-4)-P-~-Ga1(1-4)-a- 5.3 115 D-GIc(1-4)-a-D-GIC(I-6)-PF b CD ? P-D-G~~(~-~)-P-D-G~I(I-~)-~3.2 115 rc ~ - G l c1-6)-y-CD ( 2.

8

z -s

rc

b

1

a-~-Glc( 1-4)-a-~-Glc(l-6)-a- p-~-Gal( 1-4)-P-~-Gal(l-4)-a- 5.3 CD ~ - G l c1-4)-a-~-Glc( ( 1-6)-aCD fl-~-Gal(l-4)-P-~-Gal(l-4)-a-4.2 a-~-Glc( ld)-P-CD ~ - G l cId)-P-CD (

Lactose

Lactose

Lactose

"J

e rc

115

p-~-Gal( 1-4)-p-~-Gal( 1-4)-a~ - G l c1(-6)-a-CD

a - ~ - G l cI-6)-a-CD (

Lactose

b 0

b,

5.7

p-D-Gal(1-4)-P-~-Gal( 1-q-DGlcA

GlcA

Lactose

116

PpNPGal

14

p-~-Gal( I-4)-a-~-GlcNAcOBn

62

a-D-GlcNAc-OBn

D-GIC PpNPGal

P-~-Gal(l-4)-u-~-Glc(l-l)-a1.2

Trehalose

Gal

330

329

328

327

326

325

324

323

322

Enzyme Source

P-Galactosidase Bacillus circulans P-D-Gal(1-4)-P-D- 0-Galactosidase Bacillus Gal circulans P-D-Gal(1-4)-P-D- P-Galactosidase Bacillus Gal circulans p-D-Gal(1-4)-P-D- P-Galactosidase Bacillus GalNAc circulans P-D-Gal(1-4)-P-D- P-Galactosidase Barley Gal-0 P-D-GaI(1-4)-0-D- P-Galactosidase Bacillus Gal-0 circulans P-D-Gal(1-4)-P-D- 0-Galactosidase Gal-S Bacillus circulans P-D-Gal(1-4)-P-~- exo-P-galactanase Aspergillus niyer Gal P-D-Gal(1-4)-P-D- P-Galactosidase Bacillus Glc circulans

Entry Linkage

Table 1 (continued)

Lactose

Galactan

P-D-Gal-{P-D-Gal(I-~)),-DGal Rubusoside

7.3

31

60

52

118

111

92 32

Lactose

P-D-Gal(1-4)-b-~-Gal-SEt

51 3

PpNPGal

PpNPGal

25

PpNPGalNAc

Lactose

17

Lactose

Lactose

117

115

Reference

p-D-Gal( 1-4)-P-~-GalNAcOpNP

Lactose

PpNPGal

p-D-Gal(1-4)-P-~-Gal( 1-4)-a- 8.2 ~ - G l c1-4)-a-~-Glc( ( 1-6)-yCD P-D-Gal(1-4)-P-~-Gal( 1-4)-D- 52 Glu

Yield ("%)

94

a-D-Gk(1-4)-a-D-GIc(1-6)-yCD

Lactose

Product

1- 4 ) - ~ - NA P-D-Gal(1-4)-P-~-Gal( Glc

Acceptor

Donor

P-D-Gal(1-4)-P-DGlcN3-S

P-D-Gal(1-4)-P-DGlcNAc

P-D-Gal(1-4)-P-DGlcNAc

p-D-Gal(1-4)-P-DGlcNAc

P-D-Gal(1-4)-P-DGlcNAc-0

P-D-Gal(1-4)-P-DGlcNAc-0

336

337

338

339

340

P-Galactosidase Bullera singularis P-Galactosidase Bacillus circulans P-Galactosidase Bacillus circulans P-Galactosidase Bacillus circulans P-Galactosidase Diplococcus pneumoniae P-Galactosidase Bullera singularis P-Galactosidase Bacillus circulans

SP.

B-D-Gal(1-4)-P-D- P-Glucosynthase Agrobacterium Glc

SP.

P-D-Gal(1-4)-P-D- P-Galactosidase Bacillus Glc circulans P-D-Gal(1-4)-P-D- P-Glucosynthase Agrobacterium Glc

335

334

333

332

33 1

Lactose

Lactose

PpNPGal

PpNPGal

PpNPGal

PpNPGal

Lactose

13

PpNPGlcNAc

P-D-GlcNAc-0-Ser-(N-Z)OEt

1-6)-~-Man P-D-G~cNAc(

P-D-G~cNAc( 1-2)-~-Man

119

1 17

p-D-Gal(1-4)-P-~-GlcNAcOpNP

Ser-(N-Z)-OEt 21

P-D-Gal(1-4)-P-~-GlcNAc-o- 20

6)-~-Man

25

102

P-~-Gal(l-4)-P-~-GlcNAc(l17 119

2)-~-Man

P-~-Gal(l-4)-P-~-GlcNAc(l4.2

1-~)-P-D-G~cNAc( 1- 48 1-6)-~-GalNAc p-~-Gal( P-D-G~cNAc( 6)-~-GalNAc

z

2

F'

2

2'

b

5P

rL

n 3:

$

(D

B

5

0

(D

P-D-G~~(~-~)-P-D-G~cN~-SE~ 4 116 ru F m

p-D-Gal(1-4)-P-~-GlcN3OMe

83

28

P-D-GlC(1-4)-P-D-GlCOp(0Me)Ph

aGalF

88

1-4)-P- 92 P-D-G~c( 1-4)-P-~-Glc-opNP P-~-Gal(l-4)-P-~-Glc( D-GIc-O~NP

aGalF

P-~-Gal(l-4)-(3-~-Glc( 14-PD-Glc-OpOMePh

28

Rubusoside

Lactose

349

348

347

346

345

344

343

342

341

Enzyme Source

p-D-Gal(1-4)-P-D- 0-Galactosidase Bacillus GlcNAc-0 circulans P-D-Gal(1-4)-B-D- P-Galactosidase GlcNAc-0 Bacillus circulans P-Galactosidase Bacillus circulans p-D-Gal(1-4)-P-D- P-Galactosidase GIcNAc-0 Bovine testes P-D-Gal(1-4)-P-D- P-Galactosidase Bacillus GlcNAc-0 circulans P-D-Gal(1-4)P-D- P-Galactosidase GlcNAc-S Bacillus circulans P-D-Gal(1-4)-P-~- P-Galactosidase Bullera GIcNAc-S singularis P-D-Gal(1-~)-P-D- P-Galactosidase Bacillus GIcNAc-S circulans P-D-Gal(1-4)-P-D- 0-Galactosidase Bullera GlcNPht-S singularis

Entry Linkage

Table 1 (continued)

P-~-GlcNAc-O-6-hydroxymethyl-hexanoate P-D-GlcNAc-OBn

P-D-G~CNAC-O-(CH&CH~

P-D-GlcNAc-OMe

PpNPGal

PpNPGal

PpNPGal

PpNPGal

P-D-GlcNAc-SEt

P-D-GlcNAc-SEt

P-D-GlcNAc-SEt

P-D-GlcNPht-SEt

PpNPGal

Lactose

PpNPGal

Lactose

6'-oxo-P-~-Gal-O- 0-D-GlcNAc-OMe PNP

Acceptor

Donor ("h)

Yield

P-D-Gal(1-4)-P-~-GlcNPhtSEt

P-D-Gal(1-4)-p-~-GlcNAcSEt

p-D-Gal(1-4)-P-~-GlcNAcSEt

P-D-Gal(1-4)-p-~-GlcNAcSEt

B-D-Gal(1-4)-p-~-GlcNAcOMe P-D-Gal(1-4)-b-~-GlcNAcOMe

20

49

11

14

60

10

about

10

P-D-Gal(1-4)-P-~-GlcNAc-0- 57 6-hydroxy-methylhexanoate p-D-Gal(1-4)-p-~-GlcNAc- 14 OBn

Product

84

111

83

116

120

45

116

116

116

Reference

29.6 Recent Developments and New Directions m 00

w m

m

00

00

N

E:

2

2

d

00

N

cx

m

CI

cx

vi

vi

00

00

I-

0

2

E:

3

N

-

w

u3

u3 iD

3 3

m

N

b

9

h

P

t

u

b ci

ci

+ b

4

b

9

@?

t

t

ci

Ci

h

Tt

ci

767

~

369

368

361

P-D-Gal(1-4)-DGalNAc

p-D-Gal(1-4)-DGalNAc

P-Galactosidase Bacillus circulans P-Galactosidase Bifdobacterium bzj5durn P-Galactosidase Penicillium multicolor

SP.

SP.

P-D-Gal(1 -4)-P-D- P-Glucosidase Agrobacterium Xyl-s

365

364

363

366

SP.

P-Galactosidase Escherichia coli P-Galactosidase Escherichiu coli P-Glucosidase Guinea pig liver 0-Glucosidase Agrobacterium

b-~-Gat( 1-4)-P-DXyl-0 p-D-Gal(1-4)-P-DXyl-0 p-~-Gal( 1-4)-P-DXyl-0 P-D-Gal(1-4)-P-DXyl-s

362

85

14 NA

9

11

16

33

1

Lactose PpNPGal PpNPGal

PpNPGal

Lactose

PpNPGal

PpNPGal

113

113

92

85

121

17

PpNPGal

103, 110 74

85

11

PpNPGal

P-D-Gal(I-4)-P-D- P-Glucosidase Xyl-0 A grobacterium

361

SP.

85

Reference

24

Yield (%)

PpNPGal

Product

P-Glucosidase Agrobacterium

Acceptor

P-D-Gd(1-4)P-DMan-S

Donor

~~

Enzyme Source

360

Entry Linkage

~

Table 1 (continued)

29.6 Recent Developments and New Directions

769

p-D-Gal(1-4)-DGlcNAc

p-D-Gal(1-4)-DGlcNAc

p-D-Gal(1-4)-DXYl p-D-Gal(1-4)-DXYI

fi-~-Gal( 1-4)-~XYl

p-D-Gd(1-4)-DXYl

P-D-Gal(1-4)-DXYI

384

385

386

388

389

390

387

P-D-Gd(1-4)-DGlcNAc

P-D-Gal(1-4)-DGlcNAc P-D-Gd(1 -4)-DGlcNAc

383

382

381

Entry Linkage

Table 1 (continued)

P-Galactosidase Escherichiu coli P-Galactosidase Aspergillus oryzae p-Galactosidase Clonezyme libraryTM 0-Galactosidase Bacillus circulans P-Galactosidase Bacillus circuluns P-Galactosidase Bovine testes P-Galactosidase Intestinal lactase Lamb P-Galactosidase Asperg illus oryzue 0-Galactosidase Saccharomy ces fragilis P-Galactosidase Escherichia coli

Enzyme Source

GlcNAc

GlcNAc

XYI XYl

XYl

XYl

XYl

Lactose

PoNPGal PoNPGal

PoNPGal

PoNPGal

SoNPGal

GlcNAc

PoNPGal

Lactose

GlcNAc

Gal

p-~-Gal( 1-4)-D-xyl

p-D-Gal(1-4)-~-GlcNAc

43

42

61

0.9

P-D-Gal(I -4)-~-GlcNAc

GlcNAc

Gal

p-D-Gal(1-4)-~-GlcNAc

Yield ("%)

Product

Acceptor

Donor

95

124, 125

123

21

Reference

4

i?

0 u -

0

4

2 s

0

4.

$

+&

403

402

40 1

400

399

398

397

396

395

394

393

392

39 1

0-Galactosidase Escherichia coli P-Galactosidase Bovine testes P-Galactosidase Intestinal lactase Lamb P-Galactosidase Asperg illus oryzae P-Galactosidase Saccharomyces fragilis p-D-Gal(1-4)-LP-Galactosidase Escherichia coli XYl P-D-Gal(1-5)-a-L- P-Glucosidase Guinea pig liver Araf-0 P-D-Gal(1-6)-a-~- P-Galactosidase GalNAc-0 Porcine testes p-~-Gal( 1-6)-a-D- P-Galactosidase Escherichia coli Gal-0 P-D-Gal(1-6)-a-~- P-Galactosidase Barley Gal-0 1-6)-a-~- 0-Galactosidase p-~-Gal( Snail Gal-0 p-~-Gal( 1-6)-a-~- P-Galactosidase Asperg illus Gal-0 oryzae p-~-Gal( 1-6)-a-~- P-Galactosidase Gal-0 Aspergillus ovyzae 14 NA NA 57 70

P-D-Gal(1-6)-a-~-Gal-OMe P-D-Gal(l-6)-a-~-Gal-OMe P-D-GaI(l-6)-a-~-Gal-OMe p-~-Gal( l-6)-a-~-Gal-OMe

a-~-Gal-OMe a-~-Gal-OMe a-~-Gal-OMe a-D-Gal-OMe

PoNPGa1 PpNPGal PpNPGal PoNPGal

SVGal

I00

P-D-Gal(]-6)-a-~-GalNAcOpNP P-D-GaI(l-6)-a-~-Gal-OMe

apNPGalNAc

Lactose

93

93

51

51

26

121

p-~-Gal( l-5)-a-L-Araf-OpNP NA

a-L-Araf OpNP

PpNPGal

5.6

95

20

P-D-Gal(1-4)-L-xyl

L-Xyl

L-Xyl

PoNPGal

96

PoNPGal

L-Xyl

(3oNPGal

39

L-XyI

L-Xyl

PoNPGal

b-~-Gal( 1-4)-D-xyl

PoNPGal

XYl

PpNPGal

2

6'

2 2

5

35

9%

2 5

C

5

?

5

rp

2J cs

o\

P

b

~

~~

412

41 1

410

409

408

407

406

405

404

Enzyme Source

P-D-Gal(I-~)-cL-D-0-Galactosidase Bacillus Glc circulans P-D-Gal(1-6)-a-~- P-Galactosidase Bacillus Glc circulans P-D-Gal(1-6)-a-~- P-Galactosidase Bacillus Glc circulans P-D-Gal(1-6)-a-~- P-Galactosidase Bacillus Glc circulans P-D-Gd(1-6)-a-~- B-Galactosidase Bacillus Glc circulans P-D-Gal(1-6)-a-~- 0-Galactosidase Penicillium Glc multicolor P-D-Gal(1-6)-a-~- P-Galactosidase Penicillium Glc multicolor P-D-Gal(1-6)-a-~- 0-Galactosidase Penicillium Glc multicolor (I-D-Gal(1-6)-a-~- P-Galactosidase Penicillium Glc multicolor

Entry Linkage

~~

Table 1 (continued)

115

115

1.2

2.3

1.6

6.8

10

14

P-D-Gal(1-6)-a-~-Glc( I -6)-PCD p-D-Gal(1-6)-a-~-GIc( 1-4)-a~ - G l c1 (-6)-P-CD

1-6)-aP-D-GaI(I -6)-a-~-Glc( CD

B-D-Gal(l-6)-a-~-Glc( 1-4)-aD-GIc(1-6)-a-CD P-D-GaI(I-6)-a-D-GIC(1-6)-PCD

P-D-GaI(1-~)-cL-D-GIc( 1-4)-~- 13 D-Gk(1-6)-P-CD

a-D-Glc(ld)-P-CD

a-D-GIC(I-.l)-a-D-GlC(1 -6)-PCD a-~-Glc( 1-4)-a-~-Glc( 1-6)-yCD a-~-Glc( I -6)-a-CD

a-D-Glc(1-4)-a-~-Glc( 1-6)-aCD

Lactose

Lactose

Lactose

Lactose

Lactose

Lactose

Lactose

1-4)-a-D-GlC(1-6)-Pa-~-Glc( CD

115

2.6

P-D-GaI(1-6)-a-D-GIC(I - ~ ) - U D-GIc(1-6)-a-CD

a-D-Gk(1-4)-a-D-Gk(1-6)-aCD

Lactose

115

115

1I5

115

115

115

p-D-Gal(1-6)-a-~-Glc( 1-6)-1~- 1.1 CD

a-~-Glc( 1-6)-a-CD

Reference

Lactose

Yield (%)

Product

Acceptor

Donor

424

423

422

42 1

420

419

418

417

416

415

414

413

P-D-Gal(1-6)-a-~- S-Galactosidase Penicillium Glc multicolor P-D-Gal(1-6)-a-~- P-Galactosidase Penicillium Glc multicolor P-D-Gal(1-6)-U-D- P-Galactosidase Escherichia coli Glc p-D-Gal(1-6)-a-~- P-Galactosidase Asperg illus Glc oryzae S-D-Gal(1-~)-P-D- P-Galactosidase GalNAc-0 Porcine testes P-D-Gal(1-6)-P-D- P-Galactosidase Asperg illus Gal-0 oryzae p-D-Gal(1-~)-P-D- P-Galactosidase Escherichia coli Gal-0 P-D-GaI(1-6)-P-~- 0-Galactosidase Escherichia coli Gal-0 P-D-Gal(1-~)-P-D- P-Galactosidase Escherichia coli Gal-0 P-D-GaI(1-6)-P-D- b-Galactosidase Escherichia coli Gal-0 p-D-Gal(1-6)-p-~- P-Galactosidase Escherichia coli Gal-0 P-D-Gal(1-6)-P-D- P-Galactosidase Bovine liver Gal-0 PoNPGal

Lactose

Lactose

Lactose

PoNPGal

PoNPGal

DoNPGal

Lactose

Gal

ru 9

2

P

44

106 107

3

P-D-GaI(1-6)-P-~-Gal-O-Ser- 6 P-u-Gal-O-Ser-(N-Aloc)(N-A1oc)-OMe OMe 10 monoallyl ether ( P-D-G~c-O)- monoallyl ether (P-D-Gal(16)-P-~-Gal-o)-( P-D-G~c1,3,5-benzene-trimethanol 0)-1,3,5-benzenetrimethanol

S-u-Gal-OBn

W

4 4

2

F'

2

2

s

m

4

3

n

$

2

% 44

2.4

P-D-Gal-OAl1

P-D-Gal(1-6)-P-~-Gal-OA11

C

26

2

o\

3

63

p-D-Gal(1-6)-p-~-Gal-OMe

DoNPGal

P-D-Gal(1-6)-p-~-Gal-OoNP

62

13

floN PGa1

127

NA

P-D-Gal(1-6)-a-~-Glc( 1-4-PD-Fruf 1- 1)-uP-D-Gal(1-6)-a-~-Glc( D-Glc

100

115

8

1-4)-aP-D-Gal(1-6)-a-~-Glc( ~ - G l c1-6)-y-CD (

2.6 p-D-Gal(1-6)-(3-~-GalNAcOpNP P-D-Gal(1-6)-P-~-Gal-ooNP 21

115

9.5

P-D-Gal(1-6)-a-~-Glc( 1-6)-yCD

PpNPGalNAc

Trehalose

Sucrose

a-u-Glc( 1-4)-~-D-Glc( 1-6)-yCD

Lactose

PoNPGal

a-~-Glc( 1-6)-y-CD

Lactose

434

433

432

431

430

429

428

421

426

425

Enzyme Source

P-Galactosidase Bacillus circuluns p-~-Gal( 1-6)-8-~- P-Galactosidase Escherichiu coli GIcNAc-S P-D-GaI(I -6)-P-D- 0-Galactosidase Bovine testes GlcNPth-S P-D-Gal(1-6)-P-D- P-Galactosidase Glc-0 Bacillus circulans p-~-Gal( 1-6)-P-~- P-Galactosidase Bacillus Glc-0 circulans P-D-GaI(I -6)-P-D- 0-Galactosidase Escherichiu coli Glc-0

SP.

1-~)-P-D- P-Glucosidase p-~-Gal( Agrobucterium Gal-S

SP.

p-D-Gal(1-~)-P-D- P-Galactosidase Gal-0 Bucillus circuluns P-D-Gal(1-~)-P-D- 0-Glucosidase Gal-0 Guinea pig liver P-D-Gal(]-~)-P-D- P-Glucosidase Agrobucteriurn Gal-S

Entry Linkage

Table 1 (continued)

Rubusoside

Rubusoside

Rubusoside

Lactose

Lactose

B-D-GlcNPth-SEt

Lactose

Lactose

13-P-~-Gal( 1-6)-P-~-Glc-19P-D-Glc-0-steviol

I~-P-D-GIC-I 9-P-D-Gal(1-6)P-D-Glc-0-steviol

0.5

12

p-D-Gal(1-6)-p-~-GlcNAc- 28 SEt P-D-Gal(1-6)-P-o-GlcNPth10 SEt I~-P-D-G~I( 1-6)-P-~-Glc-l9- 2.6 P-D-Glc-0-steviol

P-D-GlcNAc-SEt

Lactose

26

5.2

p-~-Gal( 1-6)-P-~-GlcNAcOpNP

P-D-Gal(1-6)-p-~-Gal-SBn

PpNPGlcNAc

p-D-Gal-SBn

Lactose

PpNPGal

52

52

52

83

108

25

85

85 8.8

P-D-Gal(1-h)-b-~-Gal-SPh

p-~-Gal-SPh

PpNPGal

121

P-D-Gal(1-6)-P-~-Gal-opNP NA

Reference

PpNPGal

PpNPGal

Yield (“A)

92

Product

1.1

P-D-Gal-OMe

Acceptor

Lactose

Donor

445

444

443

442

441

440

439

438

431

436

435

P-Galactosidase Bacillus circulans p-tA+aI(1-6)-P-D- P-Galactosidase Bacillus Glc-0 circulans p-D-Gal(1-6)-P-D- P-Galactosidase Glc-S Bacillus circulans P-D-Gal(1-6)-~- 0-Galactosidase Escherichia coli (3-OMe)-Glc Pa-Gal( 1-6)-~- P-Galactosidase Kluyueromyces (3-OMe)-Glc fray iris

P-D-Gal(1-6)-P-D- P-Galactosidase Escherichia coli Glc-0 p-D-Gal(1-6)-P-D- P-Galactosidase Aspergillus Glc-0 oryzae P-D-Gal(1-6)-P-D- P-Galactosidase Glc-0 Aspergillus oryzae P-D-Gal(1-6)-P-D- P-Galactosidase Glc-0 Penicillium multicolor p-D-Gal(1-6)-P-~- P-Galactosidase Glc-0 Penicillium multicolor p-~-Gal( 1-6)-P-D- P-Galactosidase Glc-0 Bovine liver Rubusoside

Rubusoside

monoally1 ether ( p-~-Glc-O)- monoally1 ether (p-D-Gd(16)-P-~-Glc-O)-1,3,51,3.5-benzene-trimethanol benzene-trimethanol P-D-Gal(1-6)-P-~-Glc-oMe P-D-Glc-OMe

Lactose

Lactose

PoNPGal

SoNPGal

p-D-Gal(1-6)-~-(3-OMe)-Glc

P-D-Gd(1-6)-P-D-Gk-SEt

P-D-Glc-SEt

PpNPGal

PpNPGal

p-D-Gal(1-6)-P-Glc-OpNP

PpNPGlc

P-D-Glc-0-steviol 3.8

2.3 13-P-~-Gal(l-6)-P-~-Glc-19-

Lactose

Lactose

52

1 ~ - P - D - G ~ c - ~ ~ - P1-6)- D - G ~ ~3.2 (

Rubusoside

Lactose

52

52

52

2.2

Rubusoside

Lactose

P-D-Glc-0-steviol

52

12

Rubusoside

Lactose

116

29 Glycosidase-Catalysed Oliyosaccharide Synthesis

8 0

c 0

3

d

4

W

* d

ci

ci

ci

u

ci

r-

M

o\

0 m

-

d

d

d

d

d

e

d

m

ci N

m

? l

d

d

d

m

b-D-Gal(1- 6 ) - ~ Glc

p-D-Gal(1- 6 ) - ~ Glc

P-D-Gal(1-6)-~Glc

p-D-Gal(1 - 6 ) - ~ Glc3NAc

p-D-Gal(1 - 6 ) - ~ GlcN

P-D-Gal(1-6)-~GlcNAc p-D-Gal(1 - 6 ) - ~ GlcNAc

451

458

459

460

46 1

462

p-D-Gal(1 - 6 ) - ~ GlcNAc

p-D-Gal(1-6)-~GlcNAc

464

465

463

456

P-D-Gal(I - ~ ) - D Glc P-mGal(1 -6)-DGlc

455

P-Calactosidase Eschrvichiu coli j3-Galactosidase BiJidobacterium hiJidum j3-Galactosidase Bijidobacterium bijidum P-Galactosidase Bacillus circulans P-Galactosidase Bacillus circuluns P-Galactosidase Bacillus circulans P-Galactosidase Clonezyme libraryTM P-Galactosidase Eschrrichiu coli P-Galactosidase Kluyveromy ces fragilis P-Galactosidase Bzjidobacterium bijidum P-Galactosidase Eschericliia coli

3.2 NA

NA

NA

NA

9.3

10

30

2.5

9.1

P-u-Gal(1-6)-~-Glc P-D-Gal(1-6)-D-Glc

P-D-Gal(1-6)-[p-~-Gal( 1-4)]D-G~ p-D-Gal(I-6)-u-Glc

p - ~ - G d1-6)-[ ( p-~-Gal( 141~-Glc p-~-Gal( 1-6)-D-Gk3NAC

P-D-Gdl( I-6)-D-GlcN

P - D - G ~1-6)-~-GlcNAc (

p-D-Gal(1-6)-~-GlcNAc

P-D-Gal(1-6)-~-GlcNAc

Glc Lactose

Lactose

Lactose

Lactose

Glc3NAc

GlcN

GlcNAc GlcNAc

GlcNAc

GlcNAc

Gal Lactose

Lactose

Lactose

Lactose

Lactose

Lactose

PoNPGal PoNPGal

PpNPGal

Gal

21

113

20

122

92

94

94

104

104

40

4 4 4

z

2 6'

rh

$.

3F

2

C

B $-

5

rh

? n

ca

P

t u

Table 1 (continued)

b-~-GalNAc( 1-3)-P-o-Glc(1 ~)-P-D-GIc-SE~

1-6)-p-~-Glc-SEt P-D-G~c(

PpNPGalNAc

B-u-GalNAc(13)-P-~-Glc

474

17

6

P-D-GalNAc( 1-3)-P-~-GalOMe

P-o-Gal-OMe

PpNPGalNAc

P-D-GalNAc(13)-p-~-Gal-O

473

8.5

P-D-Gal(1-9)-~-NeuAc

NeuAc

Lactose

P-D-Gal(1-9)aNeuAc

8.5

p-D-Gal(1-8)-~-NeuAc

NeuAc

15

1.9

18

D-D-Gal(1-6)-~-Glucal

Lactose

412

47 1

D-GlUCd

PpNPGal

P-D-Gal(1-6)GlUcdl P-D-Gal(1-8)-~NeuAc

470

S-D-Gal(1-6)-~-GlcNAc

GlcNAc

Lactose

p-~-Gal( 1-6)-~GlcNAc

469

GlcNAc

Lactose

P-D-Gal(l-6)-DGlcNAc

468

129

56

105

105

97

92

126

126

2.3

p-D-Gal( 1-6)-~-GlcNAc

GlcNAc

Lactose

P-D-Gal(1 - 6 ) - ~ GlcNAc

467

40

12

P-D-Gal(1-6)-~-GlcNAc

GlcNAc

Gal

P-Galactosidase Asperg illus oryzae P-Galactosidase Bacillus circulans P-Galactosidase Kluyveromyces lactis P-Galactosidase Bucillus circuluns P-Galactosidase Escherichia coli P-Galactosidase Bacillus circuluns P-Galactosidase Bacillus circulans P-Hexosaminidase Chamelea gallinu P-Hexosaminidase Asperg illus oryzue

p-D-Gal(1-6)-DGlcNAc

466

Reference

("%)

Yield

Product

Acceptor

Donor

Enzyme Source

~~

Linkage

Entry

~

P-D-GalNAc(13)-P-D-GIC-O

P-D-GalNAc(14)-a-D-GIC

P-D-GalNAc(14)-a-DGIcNAc-0 P-D-GalNAc(14)-a-~-GIc-o

P-D-GalNAc(14)-P-~-Glc

P-D-GalNAc(14)-P-D-GIC

P-D-GalNAc(14)-P-DGIcNAc-0 P-D-GalNAc(14)-P-~-Glc-o

P-D-GalNAc(14)-~-GlcNAc

476

477

478

480

48 1

482

484

483

479

P-D-GalNAc(13)-P-D-GIC-O

475

0-Hexosaminidase Aspergillus oryzae P-Hexosaminidase Aspergillus oryzae P-Hexosaminidase Aspergillus oryzae 0-Hexosaminidase Aspergillus oryzae P-Hexosaminidase Aspergillus oryzae P-Hexosaminidase Aspergillus oryzae P-Hexosaminidase Aspergillus oryzae P-Hexosaminidase Aspergillus oryzae P-Hexosaminidase Aspergillus oryzae P-Hexosaminidase Aspergillus oryzae PpNPGalNAc

PpNPGalNAc

PpNPGalNAc

PpNPGalNAc

PpNPGalNAc

PpNPGalNAc

PpNPGalNAc

PpNPGalNAc

PpNPGalNAc

PpNPGalNAc

30

31

P-D-GalNAc(]-4)-a-~-GlcOMe 1P-D-GalNAc(I -4)-P-~-Glc( 4)-p-~-Glc-SEt

a-D-Glc-OMe

GlcNAc

P-D-Glc-OMe

P-D-GlcNAc-OMe

P-D-GIc(1-6)-p-~-Glc-SEt

P-D-GIc(1-4)-p-~-Glc-SEt

10

P-D-GalNAc(1-4)-P-~-GlcOMe

P-D-GalNAc(1-4)-~-GlcNAc 72

20 P-D-GalNAc(1-~)-P-DGlcNAc-OMe

27

129

27

81

P-D-GalNAc(1-4)-a-~GlcNAc-OMe

a-D-GlcNAc-OMe

20

129

29

P-D-GalNAc(1-4)-a-~-Glc( 14)-P-D-Gk-SEt

1-4)-p-~-Glc-SEt a-~-Glc(

P-D-GalNAc(1-4)-P-~-Glc( 16)-P-D-Gk-SEt

130

29

P-D-GalNAc(1-3)-P-~-GlcOMe

P-D-Glc-OMe

129

1P-D-GalNAc(1-3)-P-~-Glc( 6)-p-~-Glc-SEt

41

P-D-GIc(1-4)-P-mGlc-SEt

h

?E

C

$-

B

494

493

492

49 1

490

489

488

487

486

485

Enzyme Source

P-D-GalNAc(16)-a-~GIcNAc-0 P-D-GalNAc(16)-a-~-Glc-O

SP.

P-Hexosaminidase Asperg illus oryzae P-Hexosaminidase Asperg illus oryzae P-Hexosaminidase P-D-GalNAc(1 Asperg illus 6)-~-GalNAc oryzae P-Hexosaminidase P-D-GalNAc(1~)-D-G~cNAc Asperg illus oryzae P-D-Glc(1-2)-P-D- P-Galactosidase Asperg illus Glc-0 oryzae P-D-GIc(1-2)-P-D- 0-Galactosidase Asperg illus Glc-0 oryzae 0-Glucosidase P-D-GlC-(1 -2)-DAlmond Glc P-D-Glc-(1-2)-D- P-Glucosidase Penicillium Glc funiculosum P-Glucosidase P-D-GlC(1 -2)-DAlmond Glc P-D-Glc(14-P-D- P-Glucosynthase Agrobacterium Xyl-0

Entry Linkage

Table 1 (continued)

60, 132 60

133 28

I9

3.1 3.3

5.7 I2

P-D-Gk(1-2)-P-D-GlC-OpNP

P-D-GIc-(I -2)-D-GIC P-D-GIC-(1-2)-D-Gk

P-D-Gk( 1-2)-D-Gk P-D-GlC(I-3)-P-D-Xyl-OpNP

GalNAc

GlcNAc

PpNPGlc

PpNPGlc

Glc Glc

Glc PPNPXYI

PpNPGalNAc

PpNPGalNAc

PpNPGlc

PpNPGlc

Glc Glc

Glc aGlcF

131

P-D-GalNAc(1-6)-~-GlcNAc 33

64

117

130

P-D-GalNAc(1-6)-~-GalNAc 38

OMe

6

P-D-GalNAc(I-6)-a-u-Glc-

a-D-Gk-OMe

PpNPGalNAc

27

8

P-D-GalNAc(1-6)-a-~GlcNAc-OMe

a-D-GlcNAc-OMe

Reference

PpNPGalNAc

Yield (YO)

Product

Acceptor

Donor

2

b rp

3-

h

2

u2

0 =r

il

P

83

%.

%-

0

Q

b

0

00

4

506

505

504

503

502

50 1

500

499

498

491

496

495

SP.

SP.

P-D-Gk(1-4)-P-D- P-Glucosynthase Agrobacterium Glc aGlcF

aClcF

Glc

P-D-GlC(1-4)-P-D- P-Clucosidase Glc Almond P-D-GlC(1-4)-P-D- P-Glucosynthase Agrobacteriurn Glc

SP.

aClcF

aGlcF

aClcF

aGlcF

aClcF

Glc

3

PpNPGlc

PPhGlc

P-D-GlC(1-4)-P-D-Gk(I-4)-PD-GlC-OPh

24

34

P-D-GlC(I-4)-P-D-Gk(1 4 4 - 42 P-~-2-deoxy-2-F-Glc-O-2,4~-2-deoxy-2-F-Glc-O-2,4DNP DNP 4 P-~-2-deoxy-2-F-Glc-O-2,4- P-D-GlC(1-4)-P-D-GlC(1-4)-P~ - G l c1(-4)-P-~-2-deoxy-2DNP F-Glc-O-2,4-DNP NA Cellobiose Glc3

D-D-G~C( 1-4)-P-~-Glc( 1-4)-PD-GlC(1-3)-P-D-Xyl-OpNP

51

28

62

Trehalose

1-4)-a-~-Glc( 1-1)-aP-D-G~c( 0.9 D-GlC P-D-GIC(I -4)-P-~-2-deoxy-2- 38 P-o-2-deoxy-2-F-Glc-0-2,4DNP F-Glc-0-2.4-DNP

Glc

Glc

3.1

P-D-GlC-(1-3)-D-GlC

60, 132 60

133

Glc

Glc

3.4

P-D-GlC-(1-3)-D-GlC

4.2

Glc

Glc

P-D-GlC(1-4)-P-D- P-Glucosynthase Agrobacterium Glc

SP.

P-D-GlC(1-4)-P-D- P-Glucosynthase Agrobacterium Glc

SP.

P-D-GIc(1-4)-P-~- P-Glucosynthase A grobacterium Glc

SP.

P-Glucosidase Almond P-Glucosidase Penicillium funiculosum P-Glucosidase P-D-GlC(1-3)-DAlmond Glc P-D-GlC(1-4)-a-D- P-Glucosidase Almond Glc P-D-GlC(1-4)-P-D- P-Glucosynthase Agrobacterium 2-deoxy-2-FGk-0 SP. P-D-GlC(1-4)-P-D- 0-Glucosynthase Agrobacterium Glc

P-D-GlC-(1-3-DGlc P-D-GlC-(1-3)-DGlc

515

514

513

512

51 1

510

509

508

507

SP.

P-D-GIc(1-4)-P-D- P-Glucosynthase Agrobacterium Glc SP. P-D-GlC(1-4)-P-D- P-Glucosynthase Agrobacteriurn Glc

SP.

p-~-Glc( I -4)-P-D- P-Glucosynthase Glc Agrobacteriurn

SP.

SP.

P-D-GIC(I-4)-P-D-Gk(I-4)-P8

PpOMePhGlc

aGlcF

aGlcF

D-GIc(1-4)-P-~-Glc-opNP

P-D-GIc(1 -4)-P-~-Glc-opNP P-D-GIc(1-4)-P-D-Gk(1-4)-P-

64

79

66

P-D-Glc(1-4)-P-D-GIc(I -4)-PD-Glc-OpOMePh

PpOMePhGlc

aGlcF

~ - G l c1(-4)-P-D-GkOpOMePh I-~)-P-D-GIc( 1-4)-PP-D-GIc(1-4)-P-~-Glc-opNP j3-~-Glc( D-GIc-O~NP

54

P-D-GIc(1-4)-P-~-Glc( 1-4)-P~ - G l c1-4)-P-~-Glc-oMu (

PMuGlc

aGlcF

aGlcF

75

P-D-Glc(I-'I)-P-D-GIc(1-4)-8D-Glc-OMU

PMuGlc

6

aGlcF

P-D-Gk(I -4)-P-D- P-Glucosynthase Glc A yrobacterium SP. P-D-G~c( I -4)-P-D- P-Glucosynthase Glc Agrobacterium SP. P-D-GIc(I-4)-P-D- P-Glucosynthase Agrobacterium Glc

SP.

P-D-GIc(I -4)-P-D-Gk(I -4)-PD-Glc-OoNP

PoN PG1c

P-D-Glc(1-4)-P-D-Gk(1 4 - 0 D-Glc(1-4)-P-D-Glc-OoNP

aGlcF

P-D-Glc(1-4)-P-D- P-Glucosynthase Glc Agrobacterium

10

P-D-Glc(I-'I)-P-D-Gk(I-4)-8D-GlC(I -4)-P-D-Gk-OpNP

PpNPGlc

29

Yield (%)

Product

Acceptor

PoNPGlc

aGlcF

SP.

P-D-Glc(I-4)-P-D- P-Glucosynthase Glc Agrobacterium

Donor

aGlcF

Enzyme Source

P-D-GIc(1-4)-P-D- P-Glucosynthase Glc Agrobacterium

Entry Linkage

Table 1 (continued)

28

28

28

28

28

28

28

28

28

Reference

P-D-G~c-( 1-4)-~- P-Glucosidase Almond Glc

SP.

526

SP.

0-Glucosynthase Agrobacterium

1-4)b-DP-D-G~C( Man-0 Glc

aGlcF

Glc

PpNPMan

P-D-G~c-( 1-4)-D-Gk

5.9

P-~-Glc(l-4)-P-~-Man-opNP 31

OpOMePh

60, 132

28

28

44

P-D-G~c( I-4)-p-~-Glc-

PpOMePhGlc

aGlcF

P-D-G~c( 1-4)-P-u- P-Glucosynthase Agrohacterium Glc-0

SP.

28

P-D-G~c( 1-4)-P-~-Glc-ooNP 41

PoNPGlc

28

aGlcF

38

28

83

83

P-D-GlC(1-4)-P-D- 0-Glucosynthase Agrobacterium Glc-0

SP.

P-D-GlC(1-4)-p-D- 0-Glucosynthase Agrobacterium Glc-0

SP.

b-~-Glc( 1-4)-P-D-Gk-OpNP

48

P-D-G~c( 1-4)-P-D-Glc-OPh

PPhGlc

aGlcF

PpNPGlc

6

1-4)-P-~-GlcNPth P-D-G~c(

P-D-GlcNPth

Ccllobiose

SP.

aGlcF

8

P-D-G~c( 1-4)-P-~-GlcNPthSEt

P-D-GlcNPth-SEt

Cellobiose

83

8

P-D-GlcN3-OMe

Cellobiose

P-D-Glc(I-4)-P-U- 8-Galactosidase Bullera GkN3 singularis 1-4)-P-D- P-Galactosidase P-D-G~c( Bullera GlcNPth singularis P-D-GIc(1-~)-P-D- P-Galactosidase Bullera GlcNPth singularis p-~-Glc( 1-4)-P-D- P-Glucosynthase Agrobacterium Glc-0

28

6

p-D-GlC(I-'I)-P-D-GlC(1-4)-PD-GIc(1-4)-P-~-ManOpNP P-D-GIc(1-4)-P-D-GIcNjOMe

PpNPMan

aGlcF

SP.

b-~-Glc( 1-4)-P-D- P-Glucosynthase Agrohacterium Glc

28

42

p-~-Glc( 1-4)-P-~-Glc( 1-4)-8D-Man-OpNP

PpNPMan

aGlcF

P-D-G~c( 1-4)-P-~- 0-Glucosynthase Agrobarterium Glc

525

524

523

522

521

520

519

518

517

516

s ro

g.

2

Ti

35 9

2.

ro

22

%

C (5 1

B

2

z2

Eu

p-~-GlcNAc( 13)-b-~-Gal

531

0-Hexosaminidase Bucillus circuluns P-Hexosaminidase Nocardia orientalis

P-D-GIcNAc(12)-~-Man

536

535

534

533

532

531

530

529

528

P-Glucosidase Penicillium funiculosum P-D-GIc(1-4)-DP-Glucosidase Almond Glc P-D-GIc(1-6)a-D0-Glucosidase Almond Glc P-D-GlC(1-6)-P-D- P-Glucosidase Sesame Glc P-D-GIc-(I-6)-DP-Glucosidase Almond Glc P-D-GIc-(1-6)-~- P-Glucosidase Penicillium Glc funiculosum 1 - 6 ) - ~ - P-Glucosidase P-D-G~c( Almond Glc b-~-GlcA( 1-3)-P- P-Glucuronidase Bovine liver D-Gal p-~-GlcA( 1-3)-P- P-Glucuronidase Bovine liver D-Gal

Enzyme Source

521

P-D-G~c-( 1-4)-DGlc

Entry Linkage

Table 1 (continued)

GICNAC~

GlcNAc

PpNPGlcA

12

26

p-~-Gal( l-4)-P-~-GlcNAcOpNP

Man

D-XYI-OMU

P-D-GIcNAc(1-3)-P-~-Gal( 14)-fl-D-GlcNAc-OpNP

p-D-Gal(1-4)-P-D-xylOMu 1- 2 ) - ~ - M a n P-D-G~cNAc(

1.5

0.2

P-D-G~~(I-~)-P-D-G~~(~-~)-PP-~-GlcA(1-3)-P-~-Gal(l-4)20

p-~-GlcA( 1-3)-P-~-Gal( 1-4)P-D-XYI-OMU

D-D-Gal( 1-4)-P-~-Xyi-oMu

PpNPGlcA

P-D-GIc(1-6)-~-Glc

Glc

Glc

136

1I9

I03

103

133

17

P-D-GIc-(I - 6 ) - ~ - G k

Glc

Glc

60, 132 60

19

Glc

Glc

62 135

Cellobiose

Cellobiose

5

133

60

Reference

<25

Treh a1ose

Glc

5.1

6.7

Yield ("/)

P-D-Glc(1-6)-a-D-Glc(I -1)-aD-Glc p-~-Glc( 1-6)-fi-~-Glc( 14-PD-Glc P-D-GIc-(1-6)-~-Glc

P-D-G~c( 1-4)-D-Gk

Glc

Glc

~-D-GIc-( 1-4)-~-Glc

Product

Glc

Acceptor

Glc

Donor

548

547

546

545

544

543

542

54 1

540

539

538

P-Hexosaminidase Nocurdiu orientalis P-Hexosaminidase P-D-GIcNAc(1Nocardia 3)-(3-~-Gal orientalis P-Hexosaminidase P-D-GIcNAc(13)-P-~-Gal Nocardiu orientalis P-D-GIcNAc(1P-Hexosaminidase 3)-P-~-Gal Nocurdiu orientalis P-D-GIcNAc(1P-Hexosaminidase 3)-P-~-Gal-o Chamelea gallina P-D-GIcNAc(1P-Hexosaminidase 3)-P-~-Glc Aspergillus oryzae P-D-G~cNAc( 1P-Hexosarninidase 4)-a-DAspergillus oryzae GIcNAc-0 P-D-GIcNAc(10-Hexosaminidase Aspergillus 4)-a-~-Glc-O oryzae P-D-GIcNAc(1P-Hexosaminidase 4)-P-~-Glc Asperg illus oryzae P-D-G~cNAc( 1P-Hexosaminidase ~)-P-D-G~cNAc Aspergillus oryzae P-D-GIcNAc(1P-Hexosaminidase 4)-P-D-GlcNAc Aspergillus oryae

P-D-GlcNAc(13)-P-D-Gal

138

138

56

139

3.3

1.2

4

13

55

b-~-GlcNAc( 1-3)-P-D-Gal(13)-a-~-GalNAc-opNP 1p-~-GlcNAc( ]-3)-P-~-Gal( ~)-P-D-G~INAc-O~NP

P-D-GIcNAc(1-3)-P-~-GalOMe P-D-GlcNAc(1-3)-P-~-GlcOMe P-D-G~cNAc( I -4)-a-~GlcNAc-OMe

P-D-Gal(1-3)-a-~-GalNAcOpNP b-~-Gal( 1-3)-p-~-GalNAcOpNP P-D-Gal-OMe

P-D-Glc-OMe

a-~-GlcNAc-oMe

GIcNAc~

PpNPGlcNAc

PpNPGlcNAc

130

130

140

141

12

10

55

10

P-D-GICNAC( 1-4)-a-~-GlcOMe P-D-GIcNAc(1-4)-P-D-GlCOMe P-D-GIcNAc(1-4)-P-DGlcNAc G 1cN Ac5

a-D-Glc-OMe

P-D-Glc-OMe

GlcNAc

GICNAC~

PpNPGlcNAc

PpNPGlcNAc

PpNPGlcNAc

G1cN Ac3

PpNPGlcNAc

GICNAC~

130

137

1.9

1-4)-b-~-Glc-OpNP P-D-GkNAc(1-3)-P-D-GaI(1p-~-Gal( ~)-P-D-GIC-OPNP

GlcNAcz

137

3.4 ~-~-GkNAc(l-3)-fi-~-Gal( 14)-P-~-Glc-oMe

P-D-Gal(1-4)-fi-~-Glc-OMe

GlcNAc2

558

551

556

555

554

553

552

551

550

P-Hexosaminidase 1P-D-GICNAC( ~)-P-D-G~cNAc Asperg illus oryzae P-Hexosaminidase P-D-GlcNAc(1~)-P-D-G~cNAc Asperg illus oryzae 0-Hexosaminidase b-~-GlcNAc( 1~)-P-D-G~cNAc Asperg illus oryzae 1P-D-G~cNAc( Chitinase Nocardia ~)-P-D-G~cNAc orientalis P-D-GIcNAc(1P-Hexosaminidase ~)-P-D-GIcNAc Nocardia orientalis 1b-~-GlcNAc( P-Hexosaminidase Asperyi1lu.c 4)-P-Doryzae GlcNAc-0 Chitobiase Bacillus P-D-GIcNAc(14)-D-GkNAc SP. P-D-G~cNAc( IP-Hexosaminidase 4)-D-GlcNAc Asperg illus oryzae P-D-GIcNAc(10-Hexosaminidase Asperg illus 4)-D-GkNAc oryzae P-Hexosaminidase P-D-GICNAC( 1Nocardia 6)-a-~orientalis GalNAc-0

549

Enzyme Source

Linkage

Entry

Table 1 (continued)

23

1.5

GlcNAc3

G1cN Ac4

p-~-GlcNAc( 1-6)-[P-o-Gal(13)1-a-~-GalNAc-OpNP

G 1cN Ac2

GlcNAcz

P-D-Gal(1 -3)-a-~-GalNAcOpNP

GlcNAc?

GlcNAc?

6.2

43

P-D-GlcNAc(1-4)-D-GlcNAc

GlcNAc

GlcNAcoxazoline GlcNAc?

I38

145

145

144

139

24

P-D-G~cNAc( 1-~)-P-DGlcNAc-OMe

0-D-GlcNAc-OMe

PpNPGlcNAc

143

13

GIcNAc~

GlcNAcz

GkNAc:!

142

G1cN Ac6

G1cN Ac4

GlCNAc4

141

141

141

Reference

34

1

GlcNAch

GlcNAc4 GlcNAc4

GlcNAc4

GICNAC~

Yield (%)

20

GlCNAcs

GICNAC~

Product

GlcNAC5

Acceptor

Donor

29.6 Recent Developments and New Directions

m W

0

m

4

W

m

m

w

* r = I

cici

r-

r-

00

3

3

3

m

m

m

M

m 3

00

m 3

v, \D

787

579

578

577

576

575

574

573

572

57 1

570

Enzyme Source

Donor

P-D-GIcNAc(16)-P-~-Gal-o P-D-GIcNAc(16)-P-DGlcNAc-0 P-D-GIcNAc(1 6)-P-DGlcNAc-0

P-Hexosaminidase PpNPGlcNAc Jack bean P-Hexosaminidase GlcNAcz Nocardia orientalis P-Galactosidase &NPGlcNAc Asperg illus oryzae (Impurity?) P-Hexosaminidase GICNAC~ P-D-GIcNAc(16)-P-D-GlC-O Nocardia orientalis P-D-GIcNAc(10-Hexosaminidase GIcNAcz Nocardia 6)-b-D-Gk-O orientalis P-Hexosaminidase PpNPGlcNAc 1b-~-GlcNAc( 6)-~-GalNAc Asperg illus oryzae P-Hexosaminidase PpNPGlcNAc P-D-GIcNAc(1Aspergillus 6)-D-GlcNAc oryzae P-Hexosaminidase GlcNAc P-D-GIcNAc(1~)-D-GICNAC Asperg illus oryzae P-Hexosaminidase GlcNAcz P-D-GlcNAc(1Nocardia 6)-D-GlcNAC orientalis P-Hexosaminidase GlcNAc P-D-GlcNAc(1Bacillus 6)-~-Man circulans

Entry Linkage

Table 1 (continued)

143

119

P-D-GlcNAc(~ - ~ ) - D - G ~ c N A15c

P-D-GlcNAc(1 -~)-D-GIcNAc 25

P-D-GIcNAc(1-6)-~-Man

GlcNAc

GICNAC~

Man

1.7

146

140

P-D-GkNAc(1-6)-P-DGlcNAc

GlcNAc

22

117

1-6)-~-GalNAc 26 p-~-GlcNAc(

137

137

GalNAc

2.1

10

1P-u-GlcNAc(1-6)-[p-~-Gal( 4)]-P-u-Gk-OMe

p-~-Gal( 1-4)-P-~-Glc-OMe

I -6)-[b-~-Gal( 1P-D-Gal(1-4)-P-~-Glc-opNP p-~-GlcNAc( 4)I-P-D-GIC-OpNP

4

j3-~-GlcNAc( 1-6)-P-DGICNAc-OpNP

PpNPGlcNAc

64

136

0.8

P-D-Gal(1-4)-p-~-GlcNAcOpNP

45

14

P-D-GIcNAc(I -6)-P-~-GalOMe p-~-GlcNAc( 1-6)-[P-~-Gal( 1~)]-P-D-GICNAC-OPNP

P-D-Gal-OMe

Reference

Yield ("/')

Product

Acceptor

593

592

59 1

590

589

588

587

586

585

584

583

582

581

580

SP.

P-Glucosidase Agrobacterium

SP.

P-Mannosidase Guinea pig liver P-Mannosidase Guinea pig liver 0-Galactosidase Aspergillus oryzae P-Glucosidase A yrobacterium

0-Mannanase Aspergillus niger P-Mannosidase Aspergillus oryzae D-~-Man(l4)-P- P-Galactosidase Aspergillus D-Man-0 oryzae P-D-Man(1-6)-P- P-Mannosidase D-Man-0 Guinea pig liver P-D-Man(1-6)-P- P-Mannosidase D-Man-0 Guinea pig liver p - ~ - X y l ( l - l ) - P - ~ -P-Xylosidase Aspergillus niger XYl P-D-XYl(1-4)-DP-Xylosidase Aspergillus niyer Man p-~-Xyl( 1-6)-a-~- P-Xylosidase Glc Aspergillus awamori K4 P-Xylosidase P-D-xyl(1-6)-DGlcNAc Aspergillus niger

P-D-Man(1-4)-PD-GlcNAc P-D-Man(1-4)-PD-GIcNAc

P-D-Man(1-2)-PD-Man-0 P-D-Man(1-2)-PD-Man-0 P-D-Man(1-3)-aD-Man-0 a/PpNPMan (mixture)

P-D-GIG-SPh

a/PpNPMan (mixture) PManF

64

85

7

7

147 150 150 151

152

3.7 NA NA 10

36

P-D-XYI(I -4)-~-Man

PpNPMan

Man

PpNPMan

GlcNAc

Trehalose

1-6)-~-GlcNAc p-~-Xyl(

147 0.7 P-D-Man(1-6)-P-~-ManOpClP P-D-Man(1-6)-P-~-ManOpNP p-~-Xyl( l-l)-P-~-Xyl

PpClPMan

PpCIPMan

XYl

64

13

P-D-Man(1-4)-P-D-ManOpNP

149

148 26

3.7

85

147

1.3

9

147

0.9

a/PpNPMan (mixture)

P-D-Man(1-4)-P-DGlcNAc( 1-4)-~-GlcNAc P-D-Man(1-4)-P-DGlcNAc( 1-4)-~-GlcNAc

P-D-Man(1-3)-P-~-Glc-sPh

P-D-Man(1-2)-p-D-ManOpClP P-D-Man(1-2)-P-~-ManOpNP P-D-Man(]-3)-a-~-ManOpNP

a/ppNPMan (mixture)

PpNPMan

Man3

P-D-Xyl-SPh

PpNPMan

PpNPMan

SManF

PpClPMan

PpCIPMan

p-~-Xyl( 1-6)-~Man p-~-Xyl( 1-6)Mannitol

p-~-Xyl( 1-6)Sorbitol

594

596

595

Linkage

Entry

Table 1 (continued)

Asperg illus awamori K4 P-Xylosidase Asperg illus awamori K4

P-Xylosidase Aspergillus niger P-Xylosidase

Enzyme Source

XYl2

XYl2

XYl2

Donor

Sorbitol

Mannitol

Man

Acceptor

P-D-XYI(1-6)Sorbitol

P-D-xyl(l-6)-Mannitol

fb~-Xyl( 1-6)-~-Man

Product

<10

<10

NA

(%)

Yield

151

151

150

~~

Reference

4.

g

P s

3

zg

F

Q

\o

0

v3

4

7

6

5

4

3

2

1

EnzymeJSource

(l-?)-D-

a-D-GalNAc-

Endo-a-N-acetylgalactosaminidase Diplococcus (6-0Me)pneumoniae Gal Endo-a-N-acetyla-D-GalNAc(1-?)-D-Fuc galactosaminidase Diplococcus pneumoniae Endo-a-N-acetyla-D-GalNAcgalactosaminidase ( 1-?)-D-Gal Diplococcus pneumoniae Endo-a-N-acetyla-D-GalNAc(1-?)-D-GIC galactosaminidase Diplococcus pneumoniae Endo-a-N-acetyla-D-GalNAcgalactosaminidase (1 -1)-0 Diplococcus pneumoniae a-D-GalNAc-0 Endo-a-N-acetylgalactosaminidase Diplococcus pneumoniae a-D-GalNAc-0 Endo-a-N-acetylgalactosaminidase Diplococcus pneumoniue

Entry Linkage

~~

D-FUC

Gal

Gk

Glycerol

Ser

Thr

p-D-Gal(1-3)-C(-DGalNAcasialoglycoprotein p-D-Gal(1-3)-a-DGalNAcasialoglycoprotein p-u-Gal(1-3)-a-DGalNAcasialoglycoprotein p-D-Gal(1-3)-a-~GalNAcasialoglycoprotein P-D-Gal(1-3)-a-DGalNAcasialoglycoprotein

(6-OMe)-~-Gal

Acceptor

p-D-Gal(1 -3)-a-DGalNAcasialoglycoprotein

p-~-Gal( 1-3)-a-DGa1N Acasialoglycoprotein

Donor

Table 2. Enzymatically synthesized oligosaccharides using endo-glycosidases. (9%)

153

153

153

153

153

153

20

26

22

69

17

12

p-D-Gal(1-3)-a-DGalNAc(1-?)-D-Fuc

p-~-Gal( 1-3)-a-DGalNAc(1-?)+Gal

p-D-Gal(1-3)-a-~GalNAc( l-?)-~-Glc

D-D-Gal(1-3)-a-~-GalNAc(1-1)-Glyceryl

p-~-Gal( 1-3)-a-~-GalNAc0-Ser

p-D-Gal(1-3)-a-~-GalNAc0-Thr

153

Yield Reference

17 p-D-Gal(1-3)-a-~GalNAc(1-?)-~-(6-0Me)Gal

Product

4

9

a

OpNP a-D-GlC(I-4)-a-D-Gk(1-4)a-~-Glc( 1-3)-P-D-GlCOClNP { a-~-Glc( 1-4))B-D-GIc

18

17

16

15

14

13

12

a-~-Glc( 1-4)-a- Cyclomaltodextrinase y-CD Glc 20 D-G~c Bacillus sphaericus E-244 a-~-Glc( 1-4)-a- Amylase Klebsiella {a-D-Glc(M)),~-D- apNPGlc { a-~-Glc( 1-4)) ,j-a-~-Glc13 D-G~c-O pneumoniue Glc OpNP a-~-Glc( 1-4)-a- a-Amylase Aspera-Maltosyl-F {~-D-G~C(~-~))~-P-D-G~C{a-~-Gk(l-4))2-{a-~-Glc(l59 D-G~c gillus oryzae 0-indoyl ethanol 3))2 - P-~-Glc-O-indoyl ethanol a-D-Glc(1-4)-a- Cyclomaltodextrinase p-CD Glc { ~-D-GIC( 1-4))7 - ~ - G l ~ 39 D-G~c Bacillus sphuericus E-244 u-D-G~c( 1-4)-a- Pullulanase Klebsiella Pullulan Stevioside 13-{a-D-GlC(1-4))l-p-DNA D-G~c SP. Glc(1-2)-P-D-Glc-19-P-~Glc-0-steviol u-D-G~c( 1-4)-a- Pullulanase Klebsiella Pullulan Stevioside 13-a-D-GlC(l-4)-P-D-Gl~(lNA D-G~c SP. 2)-P-D-GlC-19-P-D-GlC-0steviol a-~-Glc( 1-4)-a- Pullulanase Klebsiella Pullulan Stevioside 13-P-D-GlC(1-2)-P-D-Glc-19- NA D-G~c SP. {a-~-Glc( 1-4))2-P-~-Glc0-steviol

PClNPGlc

a-D-Glc(1-3)-P-~-Glc-

61

61

61

155

157

156

155

154

154

1.5 154

11

Glc

{ a-D-Glc(1-4))3-~-

I./“(

Yield Reference

U-D-GIC(l-4)-a-D-GlC(1-4)- 16

1Y.-D-Gk(l-4)-a-D-Gk(1-4)a-~-Glc( I -3)-U-D-GlCOpNP

Product

u-~-Glc( 1-3)-P- a-Amylase StreptoD-Glc-0 myces griseus

Glc

PpNPGlc

apNPGlc

(U-D-GIC( 1-4))3-~Glc { a-~-Glc( 1-4))3 - ~ -

Acceptor

Donor

10

myces griseus

a-~-Glc( 1-3)-P- a-Amylase Strepto-

9

D-Glc-0

a-~-Glc( 1-3)-a- a-Amylase StreptoD-Glc-0 myces griseus

Enzyme/Source

8

Entry Linkage

Table 2 (continued)

2F’

9 s

ff2

803-

2

2 u 2

0

B

6 E 8

4

s E?

%0

m

0

Eu

4 \o E3

a-Maltosyl-F

a-~-Glc( 1-6)-a- Isoamylase D-G~ Pseudomonas umyloderamosa a-~-Glc( 1-6)-a- Pullulanase Klebsiella D-G~c aerogenes a-~-Glc( 1-6)-a- Pullulanase Aerobacter aerogenes D-GIc a-D-Glc(1-6)-a- Pullulanase AeroD-GIc bacter aerogenes a-D-Glc(1-6)-a- Pullulanase Bacillus D-G~c acidopullulyticus a-D-Glc(1-6)-a- Pullulanase Bacillus D-G~c acidopullulyticus a-~-Glc( 1-6)-a- Isoamylase D-G~c Pseudomonas arnyloderamosa Pullulanase Bacillus acidopullulyticus Isoamylase Pseudornonas amyloderamosa

24

32

31

30

29

28

21

26

25

a-Maltosyl-F

a-Maltosyl-F

a-Maltosyl-F

a-Maltosyl-F

a-Maltosyl-F

a-Maltosyl-F

a-Maltosyl-F

Maltose

((r-D-Gk(1-4)}3-~- PpNPGlc Glc

a-CD

y-CD

y-CD

a-CD

b-CD

p-CD

1-4)-a-~-Glc( 1 a-~-Glc( 6)-y-CD a-CD

p-CD

PClNPGlc

apNPGlc

apNPGlc

apNPGlc

apNPGlc

1 -4)-p- a-Amylase Streptoa-~-Glc( D-Glc-0 myces griseus

{a-D-GlC(l-4)}4-~Glc {a-D-GlC(1-4)}4-DGlc { a-~-Glc( 1-4))3 - ~ Glc {a-D-Gk(I-4)}s-DGlc {a-D-GlC(l-4)}3-DGlc

23

22

21

20

19

Amylase Pseudomonas stutzeri Amylase Pseudomonas stutzeri a-Amylase Streptomyces griseus Amylase Bacillus licheniformis a-Amylase Streptomyces griseus

a-~-Glc( 1-4)-aD-Glc-0 a-~-Glc( 1-4)-aD-G~c-O a-D-Glc(1-4)-aD-Glc-0 a-D-Glc(1+aD-G 1c-0 a-~-Glc( 1-4)-pD-Glc-0

18

60

25

NA

NA

62

NA

60

11

13

13

15

54

12

a-D-Glc(1-4)-a-~-Glc( 1-6)-y- 43 CD a-D-Glc(1-4)-a-~-Gk( 1-6)60 a-CD

6))2-y-CD a-D-Glc(]-4)-a-~-Glc( 1-6)a-CD a-~-Glc( 1-4)-a-~-GIc( 1-6)p-CD a-~-Glc( 1-4)-a-~-Glc( 1-6)p-CD a-~-Glc( 1-4)-a-~-Glc( 1-6)a-CD a-~-Glc( 1-4)-a-~-Glc( 1-6)-yCD

{ a-o-Glc( 1-4)-a-~-Glc( 1-

{a-D-Gk(l-4)}~-a-D-GkOpNP (a-~-Gk(l-4)}~-a-~-GlcOpNP { a-D-Glc(1-4)}3-a-D-GlcOpNP {a-D-Glc(1-4)}s-a-~-GlcOpNP a-D-GlC(1-4)-a-D-GIC(1-4)a-D-Glc(1-4)-P-~-GlcOClNP 1-4)a-D-Glc(1-4)-a-~-Glc( a-D-Glc(1-4)-b-D-GkOpNP a-D-Glc(1-4)-a-~-Glc( 1-6)p-CD

161

161

161

161

161

161

161

162

161

154

154

160

154

159

158

w

W

4

46

45

44

43

42

41

40

39

38

37

36

35

34

33

Pullulanase Aerobarter aerogenes Pullulanase Klebsiellu uerogenes Pullulanase Klebsiellu uerogenes Pullulanase Klebsiellu uerogenes Pullulanase Klebsiella aerogenes Pullulanase Klebsiellu uerogenes Isoamylase Pseudomonus

Emy me/Source

~-D-GIC( 1-6)-aD-Gk a-D-Glc(1-6)-aD-Glc 1-6)-aa-~-Glc( D-Glc a-D-GIc(1-6)-aD-Glc a-~-Glc( 1-6)-aD-Glc

Isoamylase Pseudomonas Isoamylase Pseudomonus Isoamylase Pseudomonus Isoamylase Pseudomonas Isoamylase Pseudomonas

a-~-Glc( 1-6)-a- Isoamylase Pseudomonas D-Glc a-~-Glc( 1-6)-a- Isoamylase D-Glc Pseudomonus

a-D-Glc(1-6)-aD-Glc a-D-Glc(1-6)-aD-Glc a-~-Glc( 1-6)-aD-Glc a-~-Glc( 1-6)-aD-Glc a-~-Glc( 1-6)-aD-G~c 1-6)-a~-D-GIC( D-Glc a-u-Glc(l-6)-aD-Glc

Entry Linkage

Table 2 (continued)

y-CD a-~-Glc( 1-4)-a-~-Glc( 16)-a-CD p-CD

Maltose Maltose

a-CD

{a-D-Glc(1-4))~-DGlc {a-~-Glc(I-4))2-DGlc

p-CD

y-CD

p-CD

Maltose Maltose

a-CD

{a-~-Glc( 1-4)}*-D-Glc

Maltose

{a-D-GIC(I-4))2-DGlc { a-D-Glc(1-4))2-DGlc

{a-D-Glc(1-4))~-DGlc

Maltose {a-D-Gk(1-4))2-~-Glc

a-D-Gk(1-4)-a-~-Glc( 16)-P-CD a-CD

Maltose Maltose

y-CD

Acceptor

a-Maltosyl-F

Donor

("w

NA

163

163

163

163

163

163

163

163

6. 162 NA

C'

52

R @

2.

;s

0

R

2

9

z

P g-

?

162

162

162

0

$

2

%0

NA

39

NA

GIG(1 -6)I-D-Gk {a-~-Glc(l-4)}2-a-~-Glc(lNA 6)-{a-D-Ck(1-4))z-D-Gk a-D-Glc(1-4)-a-~-Gk(l-4)- NA a - ~ - G k1(-6)-[a-~-Gk( 1.I)]-a-D-GIC(1-4)-D-Gk 1-4)-a-~-Glc( 1-6)- NA a-~-Glc( a-CD a-D-Glc( 1-4)-a-~-Glc( 1-6)- NA p-CD 1-6)-y- NA a-D-Glc(1-4)-a-~-Gk( CD {a-~-Gk(l-4))~-a-~-Glc(lNA 6)-a-CD {a-D-GIC(I-4)}2-a-D-GIC(I-NA 6)-P-CD

a-D-Glc(1-4)-a-D-Glc(1-6)a-CD a-D-Glc(1-4)-a-~-Gk( 1-6)-yCD { ~-D-GIC( 1-4)-a-~-Glc( 16))z-a-CD a-D-GIc(1-4)-a-~-Gk( 1-6)p-CD a-D-GlC(1-4)-a-D-Gk(1-4)[ {a-~-Glc( 1-4)}px-~-

162

161

Yield Reference

a-D-Glc( 1-4)-a-~-GIc( 1-6)-y- N A CD { a-D-Gk( I-'I)-a-D-GIC(1NA 6)Iz-P-CD

Product

iD P

4

59

58

57

56

55

54

53

52

51

50

49

48

47

a-D-Glc(1-6)-a- Isoamylase Pseudomonas D-Glc a-~-Glc( 1- 6 ) - ~ - Neopullulanase Bacillus subtilus Glc ~-D-GIC( I - 6 ) - ~ -Neopullulanase Bacillus subtilus Glc 1- 6 ) - ~ - Neopullulanase ~-D-GIC( Bacillus subtilus Glc a-D-Glc(1- 6 ) - ~ - Neopullulanase Bacillus subtilus Glc Rice debranching a-~-Glc-O enzyme P-~-3-deoxy-3- Cellulase Trichoderma F-Glc viridae P-Galactanase Penicillium citrinum P-Galactanase Penicillium citrinum P-Galactanase Penicillium citrinum P-Galactanase Penicillium citrinum P-Galactanase Penicillium citrinum P-Galactanase Penicillium citrinum

{a-~-Glc( 1-4)}*-a-~-Glc( 16)-y-CD a-D-GlC(1-6)-D-GlC 164

10

17

18

1-3)p-D-Gal(1-3)-~-GalNAc P-D-Gal(I-~)-P-D-G~I( D-GalNAc 1-6)P-D-Gal(1-h)-~-GlcNAc p-D-Gal(1-4)-P-~-Gal( D-GIcNAc

Arabinogalactan

Arabinogalactan

19

P-D-Gal(1-4)-a-~-Gal-OMe

16

3.5

p-D-GaI(1-4)-b-~-Gal( 1-4)a-D-Gal-OMe

p-D-Gal(1-4)-~-GlcNAc P-~-Gal(l-4)-P-o-Gal( 1-4)D-GIcNAc

a-D-Gal-OMe

166, 167

165

164

164

164

163

NA

5.7 1-6)1-4)-a-~-Gk( { a - ~ - G l c ( l - 4 ) } ~ - ~ - ~-D-GIC( ~1~ a-D-Gk(1-4)-~-Gk a-~-Glc( 1-4)-a-~-Glc( 1-6)- NA {a-~-Glc( 1-4)}2-D-Glc D-G~c ~-D-GIC( l-6)-a-~-Glc( 1-4)- NA a-D-Glc NA {a-~-Glc( 1-4))2-a-D-GlcMeOH OMe P-D-Gal(I-'I)-P-D-GIC(1-4)18 P-~-3-deoxy-3-F-GlcP-~-3-deoxy-3-F-GkOMe OMe I P-D-GaI(1-3)-P-~-Gal-OMe

{a-D-GlC(I-4)}2-D-Glc

y-CD

Arabinogalactan

Arabinogalactan

Arabinogalactan

Arabinogalactan

P-Lac-F

(a-D-GlC(1-4))~-DGlc (a-D-GIC(I-4)}2-DGlc {a-D-GlC(l-4)}2-DGlc {a-D-GIC(1-4)}2-DGlc { a-D-GlC(1-4)}2-DGlc Pullulan

o\

P

b

69

68

67

66

65

64

63

62

61

p-~-Glc( 1-6)DNJ

Glycerol Glycerol [ 1-’4C]-Glc

{ p-D-Gal(1-4))~-D-

Gal Arabinogalactan Laminarin

Cellooligosaccharides DNJ

P-D-Gal-OMe

Arabinogalactan

P-D-Glc(1-4)-P-D-GIc(1-6)DNJ

P-D-Gal(1-4)-P-D-Gal-OGIyceryl p-D-Gal(1-4)-P-D-Gal-OGlyceryl {P-D-GlC(l-3)}!1-5,-D-[ Ii4C]-G1~

P-D-Gal(1-4)-o-~-Gal-OMe

PpNPGal

Arabinogalactan

I68

168

168

168

NA

171

40 to 169 75 40 to 169 75 26 170

16

P-D-Gal(1-4)-p-~-Gal-OpNP 14

PpNPGal

Arabinogalactan

11

P-D-Gal(1-3)-~-GlcNAc P-D-Gal(1-4)-P-u-Gal(1-3)D-GlcNAc

Arabinogalactan

Arabinogalactan

D-G~c

p-~-Gal(1-4)-P-~-Gal(l-4)11

P-D-Gal(1-4)-~-Glc

Arabinogalactan

P-Galactanase Penicillium citrinum P-Galactanase Penicillium citrinum 0-Galactanase Penicillium citrinum P-Galactanase Penicillium citrinum P-Galactanase Penicillium citrinum P-Galactanase Penicillium citrinum P-Galactanase Bacillus subtilus P-Galactanase Bacillus subtilus Endo-P(1-3)glucanase L-IV Spisula sachalinensis Cellulase Trichoderma sp.

(“A)

60

Yield Reference

Product

Acceptor

Donor

Enzyme/Source

~

Entry Linkage

~

Table 2 (continued)

~~~

80

79

p-~-Glc( 1-4)-8- Cellulase TrichoD-Glc derma viridae

P-D-GlC(I 4 - P - (I-3),(l-'I)-PD-Glc Glucanosynthase Bacillus licheniformis p-~-Glc( 1-4)-P- Cellulase TrichoD-G~c derma viridae

78

77

76

75

74

73

72

71

Cellulase P-D-GIc(1-4)Trichoderma sp. DNJ P-D-GIc(1-4)-P- Cellulase Trichoderma ~-(3-0Me)viridae Glc-0 P-D-G~c( 1-4)-8- (I-3),(1-4)-PGlucanase Bacillus D-G~c lichenformis P-D-GlC(1 4 - P - (l-3),(l-4)-PD-GIc Glucanase Bacillus licheniformis P-D-GIc(1-4)-P- (1-3),(1-4)+ Glucanase Bacillus D-Glc licheniformis 1-4)-P- (1-3),(1-4)-0P-D-G~c( Glucanase Bacillus D-Glc licheniformis (l-3),(1-4)-PGlucanosynthase Bacillus lichenformis Endoglucanase I Humicola insolens

70

P-D-GIc(1-4)-P-D-GICOMe

P-Lac-F

I -4)-a-D-GICP-D-G~c( OMe

P-D-GlC(1-4)-P-D-GICOMu

P-D-G~c( 1-3)-a-DGlc-F

0-Lac-F

P-D-G~c( 1 -4)-P-D-Glc-oindoyl ethanol

1-3)-P-D-GkP-D-G~c( OMu

P-D-G~c( 1-3)-U-DGlc-F

P-Lac-F

P-D-GIc(1-3)-O-D-GICOMe

P-D-Glc(I -3)-p-D-Glc-F

p-~-Glc( 1-3)-p-~-Glc-F

/3-~-(3-OMe)-Glc-OMe

1-3)-P-DP-D-G~c( Glc-F

P-D-GlC(I-3)-P-DGlc-F

P-D-GIc(1-3)-P-DGlc-F

P-D-GlC(1-4)P-DG k (1-3)-P-D-Gk-F

P-Lac-F

Cellooligosaccharides DNJ

173

29

174

40

NA

60

173

173

10

5

172

20

166, 167

171

1P-D-Gal(1-4)-{ P-D-G~c( 4))2-a-D-Glc-OMe

4))2-f3-D-GIC-OMU

60

36

166, 175, 167, 176 166, 167

NA 29 P-D-G~c(~-~)-{P-D-G~c(~-

1p-D-Gal(1-4)-{P-D-G~c( 4)}2-P-~-Glc-O-indoy1 ethanol

P-D-GlC(I-3)-P-D-GIC(1-4)P-D-GIc(1-3)-(3-~-GkOMe P-D-GIc(1-3)-P-~-Glc( 1-4)P-D-GIc(1-3)-P-D-GkOMu

4)}2-O-D-G1C(1-3)-D-Gk

{ P-D-GIC(1-3)-P-D-GlC(I-

P-D-Glc(1-4)-P-D-Gk(1-3){P-D-GlC(I-4))~-P-DGlc(1-3)-D-Gk P-D-G~c( 1-3)-P-D-Gk(1-4)P-D-GIc(1- 3 ) - ~ - G k

P-D-GIc(~-~)-P-D-GIC( 1-4)- NA DNJ P-D-Gal(1-4)-P-~-Glc(l-4)- 8 P-~-(3-0Me)-Glc-OMe

91

90

89

88

B-Lac-F

P-D-GlC(1-3)-a-DGlc-F

P-D-GlC(1-3)-P-DGk-F

P-Lac-F

p-D-Gal(I -4)-P-D-GIC(1-4)P-D-Glc-OMe

176

13

P-D-Glc-OMU

167

23

P-D-Gal(1-4)-{P-D-GIc(14)}2-P-~-Glc-SMe P-D-Gal(I -‘I)-{P-D-GIc(1~)}~-P-D-GIc-OAII { p-~-Glc( I-~)-P-D-G~C( 14)}2-P-D-Gk(1-3)-P-DGlc-OMe P-D-GIc(I-~)-P-D-GIC( 1-4)P-D-GIc-OMU P-D-GIc(1-4)-P-D-GkSMe P-D-GlC(1-4)-fl-D-GICOAll P-D-GlC(1-3)-P-D-GICOMe

P-Lx-F

87

29

166, 167, 176

88

51

0.6 173

177, 167

25 Cellulose

P-Cellobiosyl-F

P-Cellobiosyl-F

86

8

P-Cellobiosyl-F

P-Cellobiosyl-F

85

177, 167, 178 177; 167

166, 167, 176

166, 167

166, 167

Cellulose

P-Cellobiosyl-F

P-Cellobiosyl-F

Cellulase Trichodermu viridae Cellulase Aspergillus niger Cellulase Polyporus tulipiyerue Cellulase Trichoderma viridue Cellulase Trichodermu viridue (I-3),(1-4)-PGlucanase Bacillus lichenijormis P-D-GIc(1-4)-P- (1-3),(1-4)-PGlucanosynthase D-Glc-0 Bacillus licheniformis P-D-GIc(1-4)-P- Cellulase Trichodermu viridue D-Glc-0

84

P-D-GIC(I 4 - P D-GIC P-D-GIc(1-4)-PD-Glc P-D-GIc(1-4)-PD-Glc P-D-GIc(1-4)-PV-Glc fi-D-GIC(14-PD-GIC P-D-GlC(1 4 4 D-Glc

P-D-GIc(]-4)-P-~-Glc( 14)-P-~-Glc-oMe

P-Lac-F

P-D-GIc(1 -6)-P-~-GlcOMe

P-Lac-F

P-D-GIc(1-4)-P- Cellulase TrichoD-Glc derma viridue

82

83

23 P-D-Gal(l-4)-P-~-Glc( 1-4)P-D-GlC(I -3)-P-D-GICOMe 26 p-D-Gal(I - ~ ) - O - D - G 1-4)~( P-D-GIc(1-6)-P-~-GlcOMe P - D - G ~ I ( ~ - ~ ) - { P - D - G ~ ( 27 I4)}2-B-D-GIC(1-4)-P-DGlc-OMe 54 Cellulose

P-D-GIc(I -3)-P-D-GlCOMe

P-Lac-F

(“w

Yield Reference

Product

Acceptor

Donor

b-~-Glc( 1-4)-P- Cellulase D-Glc Trichoderma viridue p-~-Glc(1-4)-8- Cehlase TrichoD-Glc derma viridue

EnzymeJSource

81

Entry Linkage

Table 2 (continued)

Lo

-1 00

Cellulase Trichoderma viridae Endoglycoceramidase Cornybacterium sp. Endoglycoceramidase Cornybacterium sp. Endoglycoceramidase Cornybacterium sp. Endoglycoceramidase Cornybacterium sp. Endoglycoceramidase Cornybacterium sp. Endoglycoceramidase Cornybacterium sp. Endoglycoceramidase Macrobdella decora Endoglycoceramidase Cornybacterium sp.

P-D-GlC-0

P-D-GIc-O

P-D-Glc-0

102

103

104

106

P-D-GlC-0

P-D-Gk-0

101

105

P-D-GlC-0

100

99

97

98

Cellulase Trichoderma viridae Cellulase Trichoderma viridae Cellulase Trichoderma viridae Cehlase Trichoderma viridae

Cellulase Trichoderma viridae Xylanase Trichoderma viridae

P-D-GlC(1-4)-PD-G1C-S P-D-GlC(1-4)-Pu-Glc-S P-D-GlC(1 4 4 D-GlC-S P-D-GlC(1-4)-PD-G1C-S

96

95

94

93

92

166, 167 180

166, 167 181

52 58

25 22

P-D-Man-OMe P-D-xyl(I -4)-D-D-Glc-F

P-Lac-F p-u-Xyl(1-4)-8-uGlc-F

181 181 181 181 182

181

22 25 30 18 NA

35

2.7 181

179

41

P-D-GlC-S(1-4)-P-D-GIC-F

P-D-Glc-S(1-4)-P-DGlc-F

167

36

P-D-GlC-S-Ph

P-Lac-F

167

30

P-D-G~C-S-CI~H~~

P-Lac-F

167

43

P-u-Glc-S-CH2CONHPr

P-Lac-F

Q ..

5

a

P-D-GlC-0

0-D-GlC-0

fl-D-Gk-0

p-D-GlC-0

109

110

111

112

116

115

114

113

HO-(CH2)sCH3

Asialo-GM1

P-D-Gk-0

108

Endoglycoceramidase Macrobdella decora Endoglycoceramidase B-D-Gk-0 Macrobdella decora Endoglycoceramidase p-D-G1C-0 Macrobdella decora Endoglycoceramidase B-D-G~c-O Macrobdella decora (2-deoxy-21-4)- Lysozyme Hen egg D-D-GICN( white monochloroP-D-GkN acetamido-a-DGlc),

GM1

(2-deoxy-2chloroacetamido-fl-~Glc)3

(HO-CH2)3CNHCO(CH&COOMe

IV3NeuAcPGalNAc- HO-(CH2)5CH3 GbOse4-0-Cer Ceramide LacSer-support

MeOH

GMI

Endoglycoceramidase Cornybacterium sp. Endoglycoceramidase Cornybacterium sp. Endoglycoceramidase Cornybacterium sp. Endoglycoceramidase Leech Endoglycoceramidase Macrobdella decora

p-u-Glc-0

107

Acceptor

Donor

Enzyme/Source

Entry Linkage

Table 2 (continued)

17

GlcN

{ (P-D-GkN(1-4))(343)-D-

3NeuAcCgOse4-O-(CH2)gOH

3NeuAcCgOse4-O(CH2)6NHCOCO(CH3)3

3NeuAcCgOse4-OCH2(HOCH2)2CNHCO(CH2)4COOMe 3NeuAcCgOse4-O(CH2)6NHCOCF3

33

NA

NA

184

182

182

IV3NeuAcpGalNAc9.6 181 G~OS~~-O-(CH~)~CH~ GM3 61 18, 183

181

181

Yield Reference (%)

C ~ O S ~ ~ - O - ( C H ~ ) ~ C H33~

Product

m

g.

9 6 $

&?

0

Q

G-

l u v,

0

0

125

124

123

122

121

120

1 I9

1I8

117

Endo- P-N-acetylglucosaminidase Arthrohucter protophormiue P-D-GIcNAc(1- Endo-P-N-acetyl?)-a-Dglucosaminidase GIcNAc-0 Arthrohacter protophormiue P-D-GIcNAc(1- Endo- P-N-acetyl?)-~-2dGlc glucosaminidase Arthrobucter protophormiue 1- Endo-P-N-acetylP-D-G~cNAc( ?)-~-6-deoxy- glucosaminidase Glc Arthrohucter protophormiae P-D-GIcNAc(1- Endo-P-N-acetyl?)-D-F~ glucosaminidase Arthrohacter protophormiue P-D-GIcNAc(1- Endo-P-N-acetyl?)-D-Gk glucosaminidase Arthrobacter protophormiue P-D-GIcNAc(1- Endo-b-N-acetyl?)-D-xyl glucosaminidase Arthrobacter protophormiae 1- Endo-P-N-acetylP-D-G~cNAc( ~)-P-L-Fucglucosaminidase OMe Arthrohactrr protophormiue P-D-GIcNAc(1- Lysozyme Hen egg 3)-a-~-Glc white

P-D-GIcNAc(1?)-(3-OMe)D-Gk

36

36

Man&D-GlcNAc( 1-?)-~-6- 67 deoxy -G1c

GICNAC~

Man9GlcNAczAsn

Man9GlcNAc~Asn

Man9GlcNAc2Asn

Man9GlcNAc.1Asn

Man9GlcNAc2Asn

{a-D-Gk( I-4)}4-a-D-GkOpNP

0-L-Fuc-OMe

XYl

Glc

Fru

6-deoxy-Glc

36

36

185

61

16

34

5

Mans-P-D-GlcNAc( 1-?)-DClc

1-?)-DMan9-P-~-GlcNAc( XYl

Mans-P-D-GlcNAc(1-2)-0L-Fuc-OMe

p-~-GlcNAc(I-3)-{ a-DGIC(1-4))4-a-D-GIC-OpNP

186

36 4.7

Mans-P-~-GlcNAc(1-?)-DFru

Mang-P-~-GlcNAc( 1-?)-D2dGlc

61

2dGlc

36

l-?)-a-~-66 Mang-P-~-GlcNAc( GlcNAc-OMe

a-D-GlcNAc-OMe

ManqGlcNAczAsn

Mans GlcNAc2Asn

36

30

Man&~-GlcNAc( 1-?)-(3OMe)-D-Glc

(3-OMe)a-Glc

Man9GlcNAc2Asn

135

134

133

132

131

130

129

128

127

126

Man&~-GlcNAc( 1-4)-aD-GIc-O~NP

apNPGlc

Endo-J3-N-acetylglucosaminidase Arthrobacter protophormiae Lysozyme Hen eggwhite

GIcNAc(1-4)}3-PA

{ P-D-GkA(1-3)-P-D-

Hyaluronic acid

NA

P-D-GIc(1-6)-~-Glc

PpNPGlc

39

TIN(Glc)AS

MansGlcNAcz Asn

GICNAC~

188

P-D-GIcNAc(I -4)-D-D-GIC49 189 OpNP 72 191 { P-D-GlCA(I-3)-P-DGICNAC(~-~)}(~-~~)-PA

30, 31

190

3.4 189

75

PpNP2dGlc

Man6-P-~-GlcNAc( 1-4)-PD-Glc(1-6)-D-Glc

187 0.6 187

4.2

1.9 187

8.5 187

Yield Reference (“A)

GICNAC~

P-D-GIcNAc(1-4)-P-D2dGlc-OpNP

P-D-GIcNAc(1-4)-a-DGlc( 1-4)-a-~-Glc-OpNP P-D-GIcNAc(1-3)-a-~GlcNAc-OpNP

a-~-Glc( 1-4)-a-~-GlcOpNP apNPGlcNAc

Lysozyme Hen eggwhite Lysozyme Hen eggwhite

Product P-D-GlcNAC(1-3)-a-DG k (1 -4)-a-D-Glc-OpNP P-D-G~cNAc( 1-3)-a-nGlcNAc-OpNP

Acceptor ~-D-GIC( I-4)-a-D-GICOpNP apNPGlcNAc

Donor

Lysozyme Hen eggwhite Lysozyme Hen eggwhite

Enzyme/Source

Endo-P-N-acetylglucosaminidase Arthrobacter protophormiae P-D-GIcNAc(1- Endo-0-N-acetylglucosaminidase 4)-P-D-GIC Arthrobacter protophormiae P-D-GIcNAc(1- Lysozyme Hen eggwhite 4)-P-D-GIC-O P-D-GIcNAc(1- Hyaluronidase Bovine testes ~)-P-D-GIcA

P-D-GIcNAc(14)-P-D2dG 1c-0 P-D-GIcNAc(14)-P-D-GIC

P-D-GIcNAc(13)-a-D-GIC P-D-GIcNAc(13)-a-DGIcNAc-0 P-D-G~cNAc( 14)-a-~-Gk P-D-GIcNAc(14)-a-nGIcNAc-0 1S-D-GICNAC( 4)-a-n-Glc-O

Entry Linkage

Table 2 (continued)

144

143

CSN[GlcNAc]LST

GlcNAc2-0xazo1ine

GlcNAc2-0xazoline

MansGlcNAcz Asn

Endo-j3-N-acetylglucosaminidase Arthrobacter protophormiae P-D-GIcNAc(1- Chitinase Nocardia orientalis 4)-P-DGlcNAc 1- Endo-P-N-acetylP-D-G~cNAc( 4)-p-Dglucosarninidase GlcNAc Mucor hiemalis 1- Chitinase Bacillus sp P-D-G~cNAc( 4)-p-DGlcNAc

141

STF-GP

Chondroitin

P-D-G~cNAc( 1- Hyaluronidase 4)-p-~-GlcA Bovine testes

140

GlcNAcs

Chondroitin 4-0SO3H

P-D-GIcNAc(1- Hyaluronidase 4)-p-~-GlcA Bovine testes

139

100

19

8.5 193 CSN[(NeuAcGalG1cNAcMan)zManGlcNAcz]LST Chitin

192

23

32

191

191

191

191

191

G1cNAc.l

2.1 P-D-GIcA(1-3)-B-~-(6-0S03H)-GalNAc(1-4)-{0D-GIcA(1-3)-p-~GlcNAc)3-PA 45 { P-D-GlcA( I-3)-p-D-(4-O- { P-D-GIcA(1-3)-8-~GlcNAc(l-4)}(1-4)-{p-DS03H)-GalNAc(1-4))3 GkA( 1-3)-p-~-(4-0PA SO3H)-GalNAc)3-PA 35 1-3)-8-~{ p-D-GlcA(1-3)-8-D-(6-O- { p-~-GlcA( GICNAC(1-4))i 1-31 - { p-DS03H)-GalNAc(1-4))3 GlcA( 1-3)-p-~-(6-OPA S03H)-GalNAc)3-PA 10 { P-D-GlcA(1-3)-p-D-(4-O{ P-D-GlCA( 1-3)-p-DS03H)-GalNAc(1GIcNAc( 1-4))3-PA 4)) (1-21-tp-~-GlcA( 1-3)-pD-GIcNAc)~-PA 20 { p-~-GlcA( 1-3)-8-~{ p-D-GlCA( 1-3)-8-Dp-DGalNAc(1-4)}(1-5)-{ GIcNAc( 1-4)}3-PA GlcA( 1-3)-p-~GICNAC)~-PA NA EEKYN[Man6EEKYN[GlcGlcNAc21LTSVL NAcILTSVL GlcNAc( 1-4))3-PA

{ P-D-G~cA( 1-3)-p-D-

GlcNAcs

Hyaluronic acid

p-~-GlcNAc( 1- Hyaluronidase 4)-p-~-GlcA Bovine testes

138

142

Hyaluronic acid

p-~-GlcNAc( 1- Hyaluronidase 4)-p-~-GlcA Bovine testes

137

P-D-GICNAC( 14)-P-DGlcNAc

Chondroitin 6-0SO?H

P-D-GIcNAc(1 - Hyaluronidase 4)-p-~-GlcA Bovine testes

136

1P-D-G~cNAc( 4)-P-DGlcNAc

P-D-GIcNAc(14)-P-DGlcNAc

P-D-GIcNAc(14)-P-DGlcNAc

P-D-G~cNAc( 14)- P-DGlcNAc

P-D-GIcNAc(14)P-DGlcNAc P-D-GIcNAc(14)-P-DGlcNAc P-D-GIcNAc(14)-P-DGlcNAc

146

147

148

149

1so

152

151

P-D-GIcNAc(14)-P-DGlcNAc

145

Entry Linkage

Table 2 (continued)

Endo-P-N-acetylglucosaminidase Arthrobacter protophormiae Endo-P-N-acetylglucosaminidase Arthrobacter protophormiae Endo-P-N-acetylglucosaminidase Arthrobacter protophormiae Endo-P-N-acetylglucosaminidase Arthrobacter protophormiae Endo-P-N-acetylglucosaminidase Arthrobacter protophormiae Endo-P-N-acetylglucosaminidase Mucor hiemalis Endo-P-N-acetylglucosaminidase Mucor hiemalis Chitinase Nocardia orientalis

Enzyme/Source

26

NA

TIN[Man9GlcNAczCHz-IAS

ManhGlcNAcz-RNAseB

Man6GlcNAczAsn

NA Man6GlcNAc-p-DGIcNAc(1-4)-P-DGICNAC( 1-~)-D-GIcNAc

TIN[GIcNAc-CH~-]AS

GlcNAc-RNAse B

P-D-GlcNAc-Asn

GlcNAcz

Man9GlcNAczAsn

Man6GlcNAc2Asn

Man6GlcNAcz

Man6GlcNAczAsn

193

193

192

NA

NA

21

CSN[Man6GlcNAcz]LST

CSN[(GalG1cNAcMan)zManGlcNAcz]LST

CSN[GlcNAcILST

CSN[GlcNAc]LST

GICNAC~

Asn-GlcNAczMan6

ASTF-GP

GlcNAca

NA

2s

Po)

TIN[GlcNAc]AS

Yield Reference

Man9GlcNAczAsn

Product

Acceptor

Donor

52

00

162

161

160

159

158

157

156

155

154

153

P-D-GIcNAc(1- Lysozyme Hen egg white 4)-P-DGlcNAc P-D-G~cNAc( 1- Chitinase 4)-P-DTrichoderma reesei GlcNAc KDR-11 1- Lysozyme Hen egg ~-D-G~CNAC( 4)P-Dwhite GlcNAc S-D-GICNAC( 1- Lysozyme Hen egg 4)-P-Dwhite GlcNAc 1- Endo-0-N-acetylP-D-G~cNAc( 4)-P-Dglucosaminidase Arthrohactev GIcNAcprotophormiae CH21- Lysozyme Hen egg B-D-GICNAC( white 4)-P-DGlcNAc-0 1- Lysozyme Hen egg P-D-G~cNAc( 4)-P-Dwhite GlcNAc-0 P-D-GIcNAc(1- Lysozyme Hen egg4)-P-Dwhite GlcNAc-0 1- Lysozyme Hen eggP-D-G~cNAc( 4)-P-Dwhite GlcNAc-0 P-D-GIcNAc(1- Endo-P-N-acetyl4)- P-Dglucosaminidase GlcNAc-0 Arthrohacter protophormiae PpNPGlcNAc

PpNPGlcNAc

P-D-GlcNAc-0(CH2)3CH=CH2

GlcNAcz

MansGlcNAczAsn

OpNPGlcNAc

GlcNAcs

G~cNAc~

PpNPGlcNAc

TIN(GlcNAc-CHz-)AS

GlcNAc4

GIcNAc~

GIcNAc~

MangGlcNAc2Asn

GlcNAcl

GIcNAc~

GIcNAc~

GlcNAcz

14

14

35

17

3.3

26

33

84

Man&D-GlcNAc( 1-4)-PD-GlcNAc-0(CH2)3CH=CH2

189

195, 196

6.5 187

56

26

195, 196

14

40

18

14

12

P-D-G~cNAc( 1-4)-P-DGlcNAc-OpNP

P-D-GIcNAc(1-4)-P-DG k NAc-OpNP

GlcNAc-OpNP

{ P-D-G~cNAc( 1-4)}4-P-~-

GlcNAc-OpNP

{ P-D-G~cNAc( 1-4)}3-fi-D-

GlcNAcS

GICNAC~

GlcNAc7

169

168

167

166

165

164

163

Enzyme/Source

P-D-G~cNAc( 1- Endo-P-N-acetylglucosaminidase 4)-P-DArthrobacter GIcNAc-0 protophormiae P-D-GIcNAc(1- Endo-P-N-acetyl4)-P-Dglucosaminidase GlcNAc-0 Arthrohacter protophormiae P-D-GIcNAc(1- Endo-P-N-acetyl4)p-Dglucosaminidase GlcNAc-0 Arthrohacter protophormiae P-D-GIcNAc(1- Endo-P-N-acetyl4)-p-Dglucosaminidase GlcNAc-0 Arthrobacter protophormiae 1- Endo-P-N-acetylP-D-G~cNAc( glucosaminidase 4)-P-DArthrobacter GIcNAc-0 pro tophormiae P-D-G~cNAc( 1- Endo-P-N-acetylglucosaminidase 4)-P-DArthrobacter GlcNAc-0 protophormiae P-D-GIcNAc(1 - Endo-P-N-acetylglucosaminidase 4)-P-DArthrobacter GICNAC-S protophormiae

Entry Linkage

Table 2 (continued)

Man9-P-~-GlcNAc(1-4)-PD-GIcNAc-OMU

PMuGlcNAc

PpNPGlcNAc

Man9GlcNAczAsn

Man9GlcNAc2Asn

81

81

81

Man9-P-o-GlcNAc(1-4)-PD-GIcNAc-0CH2CH=CH2

P-D-GkNAc-0CH2CH=CH2

P-D-GIcNAc-S-CH~-Mans-P-~-GlcNAc( 1-4)-PCONHCH2CH(OMe)* D-GIcNAc-SCH2CONHCH2CH(OMe)2

Man9GlcNAc2Asn

Man9GlcNAczAsn

33

66

67

70

33

33

Yield Reference (“10)

Man9GlcNAc2Asn

Man9-P-~-GlcNAc(1-4)-PD-GlcNAc-OpNP

Man9-P-~-GlcNAc(1-4)-0D-GlcNAc-OBn

P-o-GlcNAc-OBn

MansGlcNAc2Asn

Product

P-D-G~cNAc-O-(CH~)~-Man9-P-~-GlcNAc(1-4)-pNHCOCH=CH2 D-G~cN Ac-O(CH2)3NHCOCH=CH2

Acceptor

Man9GlcNAczAsn

Donor

h,

\o

a

0

00

179

178

177

176

175

174

173

172

171

170

Endo-P-N-acetylglucosaminidase A rthrobacter protophormiue P-D-GICNAC( 1- Endo-0-N-acetylglucosaminidase 4)-P-DArthrobacter GlcNAc-S protophormiae P-D-GIcNAc(1- Lysozyme Hen egg4)-P-~-Man- white 0 b-~-GlcNAc( 1- Endo-P-N-acetylglucosaminidase 4)-~-Glc Arthrobacter protophormiae p-~-GlcNAc( 1- Endo-P-N-acetyl4)-~-GlcNAc glucosaminidase Mucor hiemalis P-D-GICNAC( 1- Endo-P-N-acetyl4)-~-GlcNAc glucosaminidase Mucor hiemalis 1 - Endo-P-N-acetylb-~-GlcNAc( glucosaminidase 4)-D-GlcNAc Mucor hiemalis P-D-G~cNAc( 1- Endo-P-N-acetyl4)-D-GlcNAc glucosaminidase Mucor hiemalis P-D-G~cNAc( 1- Endo-P-N-acetyl4)-D-GlcNAc glucosaminidase Mucor hiemalis P-D-GICNAC( 1- Endo-P-N-acetyl4)-~-GlcNAc glucosaminidase Arthrobacter protophormiae

P-D-G~cNAc( 14)p-DGlcNAc-S

GlcNAc-Asn-FMOC

GlcNAc

1-4)-Dp-~-GlcNAc( GlcNAc-PA

ASTTTN[GlcNAcIYT

p-~-GlcNAc( 1-4)-DGlcNAc-PA GlcNAc

ASTF-GP

ASTP-GP

STF-GP

MansGlcNAc2Asn

Man9GlcNAc2Asn

PpNPMan

STF-GP

GlcNAc2

30, 31

197

Man6-P-~-GlcNAc(l-4)-~-NA Glc

(NeuAcGalGlcNAcMan)2- 20 ManGlcNAcz-AsnFMOC (Ga1GlcNAcMan)zNA

NA

9

NA 91

(GalGlcNAcMan)2ManGlcNAc3-PA ASTTTN[( NeuAcGalG1cNAcMan)zManGlcNAczlYT Man6GlcNAc3-PA Man9GlcNAc2

ManGlcNAc2

187

b-~-GlcNAc( 1-4)-P-~-Man- 10 OpNP

185

198

199

198

198

33

33

83

P-D-GICNAC-S-CH~CN Man9-P-~-GlcNAc(1-4)-PD-G~cNAc-S-CH~CN

MangGlcNAc2Asn

Man9-P-~-GlcNAc(l-4)-p- 78 D-GIcNAc-S-(CH~)~CH~

P-D-GlcNAc-S(CH2)3CH3

Man9GlcNAclAsn

s

0 4

co

2

2.

2

b -.

2 co

24

ro

22

'h

%

2 c

2,

'5

? co

o\

9

ru

~~~~

187

186

185

184

IN[GlcNAc]ATL

IN[GlcNAc]ATL

DansylASTTTN[GlcNAc]YT

GlcNAc-Asn-FMOC

DansylASTTTN[GlcNAc]YT

STF-GP

183

P-D-GIcNAc(1- Endo-0-N-acetyl~)-D-G~cNAc glucosaminidase Mucor hiemalis P-D-G~cNAc(1- Endo-a-N-acetyl~)-D-GIcNAc glucosaminidase Mucor hiemalis D-D-GICNAC( 1- Endo-0-N-acetyl~)-D-G~cNAc glucosaminidase Mucor hiemalis P-D-GIcNAc(1- Endo-P-N-acety l~)-D-GICNAC glucosaminidase Mucor hiemalis

GlcNAc-Asn-FMOC

ASTF-GP

182

D-GIcNAc

Mans GlcNAcz Asn

CSN[GlcNAc]LSTCVL GKSNELHKLNTYPRT DVGAGTP

Acceptor

P-D-GIcNAc(1- Endo-P-N-acetyl4)-~-GlcNAc glucosaminidase A rthrobacter protophormiae P-D-GIcNAc(1- Endo-P-N-acetyl4)-~-GlcNAc glucosaminidase Mucor hiemalis P-D-G~cNAc( 1- Endo-P-N-acetyl4)-~-GlcNAc glucosaminidase Mucor hiemalis

Donor

181

~

STF-GP

~~

P-D-G~cNAc( 1- Endo-P-N-acetyl4)-~-GlcNAc glucosaminidase Mucor hiemalis

EnzymeJSource

180

Entry Linkage

~

Table 2 (continued)

~~~

IN[( NeuAcCalG1cNAcMan)zManGlcNAcz]ATL

Dansyl-ASTTTN[(GalGlcNAcMan)2ManClcNAc2]YT IN[Man6GlcNAc2lATL

(GalGlcNAcMan)2ManGlcNAc2-AsnFMOC DansylASTTTN[( NeuAcGalGlcNAcMan) 2 ManGlcNAcz]YT Man6GlcNAc2-Asn-FMOC

CSN[(NeuAcGalGlcNAcMan)zManGlcNAczILSTCVLGKSNELHKLNTYPRTD-VGAGTP Mans-P-~-GlcNAc(1- 4 ) - ~ GlcNAc

Product

-

197

4

197

199

15

10

197

199

197

33

8

16

15

85

8.5 200

Yield Reference

196

195

194

193

192

191

190

189

188

1- Endo- P-N-acetylb-~-GlcNAc( 4)-~-GlcNAc glucosaminidase Mucor hiemalis P-D-G~cNAc(1- Endo-P-N-acetylglucosaminidase 4)-~-Man Arthrobucter protophormiue P-D-GICNAC( 1- Endo-P-N-acetyl4)-~-Man glucosaminidase Arthrobacter protophormiae 1- Lysozyme Hen eggP-D-G~cNAc( 4)-~-Man white P-D-GIcNAc(1- Endo-P-N-acetyl~)-L-Fuc glucosaminidase Arthrobacter protophormiae P-D-GIcNAc(1 - Endo-P-N-acetylglucosaminidase 4)-L-Gal Arthrobucter protophormiue D-D-GICNAC-O Endo- P-N-acetylglucosaminidase Arthrobucter protophormiue P-D-GlcNAc-0 Endo-P-N-acetylglucosaminidase Arthrobacter protophormiue Endo-P-N-acetylglucosaminidase Arthrobucter protophormiue

185

33

33

33

83

21 26

19

64

8

47

I-4-DMan9-P-~-GlcNAc( Man

p-~-GlcNAc(1-4)-~-Man Mang-P-~-GlcNAc(l-2)-LFuc

Mans-P-~-GlcNAc(l-2)-LGal

Mans GlcNAc-OMe

Mang GlcNAc-OPr

Mans GlcNAc-OEt

Man

Man L-Fuc

L-Gal

MeOH

PrOH

EtOH

GlcNAcz Man9GlcNAc2Asn

Mans GlcNAczAsn

Mans GlcNAcz Asn

Man9GlcNAczAsn

Man9GlcNAc2Asn

185

187

185

30, 31

Man&b~-GlcNAc(I-~)-D- NA Man

197

Man

I

ManhGlcNAczAsn

IN[(GalGlcNAcMan)zManGlcNAcz 1ATL

IN[GlcNAc]ATL

ASTF-GP

199

198

197

Enzyme/Source

glucosaminidase Arthrobacter protophormiae P-D-GlcNAc-0 Endo-0-N-acetylglucosaminidase Arthrobacter protophormiae P-D-xyl( 1-4)-P- Xylanase TrichoD-xyl derma viridae

P-D-GlcNAc-0 Endo-P-N-acetyl-

Entry Linkage

Table 2 (continued)

P-D-xyl(1-4)-P-DXyl-F

MangGlcNAczAsn

Man9GlcNAc2Asn

Donor

P-D-Xyl(1-4)-P-D-Xyl-F

Glycerol

iPrOH

Acceptor

{ p-D-xyl(1-4))(2-12)-D-x)’l

Man9GlcNAc-(1-?)Glyceryl

Mans GlcNAc-OiPr

Product

72

56

9.6

(XI

201

33

33

Yield Reference

L? 0

2

0

R

$

-

?

9

0

&

Q

%-

0

b

0

+

00

a-D-Gal-0

a-D-Gal-0

a-D-Gal-0

a-D-Gal-0

a-D-Gal-0

a-D-Gal-0

a-D-Gal-0

a-D-Gal-0

a-D-Gal-0

a-D-Gal-0

1

2

3

4

5

6

I

8

9

10

Entry Linkage apNPGal

a-Galactosidase Asperg illus oryzae a-Galactosidase Asperg illus oryzae a-Galactosidase Coffee bean a-Galactosidase Trichoderma reesei a-Galactosidase Trichoderma reesei a-Galactosidase Trichoderma reesei a-Galactosidase Trichoderma reesei a-Galactosidase Trichoderma reesei a-Galactosidase Trichoderma reesei a-Galactosidase Asperg illus niger Gal

apNPGal

apNPGal

apNPGal

apNPGal

apNPGal

apNPGal

Raffinose

apNPGal

Donor

Enzyme/Source

Table 3. Enzymatically synthesised alkyl glycosides.

202

44 203

203

203

203

23

NA NA

NA

NA

NA

a-D-Gal-O-elymoclavine

a-D-Gal-OAll a-D-Gal-OMe

a-D-Gal-OEt

a-D-Gal-OPr

Elymoclavine

HOAll MeOH

EtOH

203

203

204

NA

NA

41

a-D-Gal-O-(CH2)3CH3

a-D-Gal-O-CH(CH3)CH*CH3

a-D-Gal-O(CH2)60H

HO-(CHz)jCH3

HO-CH(CH3)CHlCH3

HO-(CH2)60H

iPrOH

a-D-Gal-OiPr

202

24

a-D-Gal-0-chanoclavine

Chanoclavine

PrOH

Reference

Yield (%)

Product

Acceptor

3s

a-D-Gal-0

a-D-Gal-0

a-D-Gal-0

a-D-Gal-0

a-D-Gal-0

a-D-GalNAc-0

a-D-GalNAc-0

a-D-GalNAc-0

a-D-GalNAc-0

11

12

13

14

15

16

17

18

19

Entry Linkage

~~

Table 3 (continued)

204

205

205

206 206 207

207

207

207

~ - D - G ~ ~ - O ( C H ~ ) ~ C H = C H ~37

6

11

3

8 10

NA

NA

NA

a-D-Gal-O-lysergol

a-~-Gal-O-9,lO-dihydrolysergol

a-D-Gal-O-Ser-(N-Boc)-OMe a-D-Gal-O-Ser-(N-A1oc)-OMe a-D-GalNAc-0-L-Ser

a-D-GalNAc-0-L-Thr

a-D-GalNAc-0-D-Thr

HO-(CH2)3CH=CH2

Lysergol

9,lO-Dihydrolysergol

Ser-(N-Boc)-OMe Ser-(N-A1oc)-OMe L-Ser

D-Ser

L-Thr

D-Thr

Gal

apNPGal

apNPGal

Raffinose Raffinose GalNac

GalNac

GalNac

GalNac

a-Galactosidase Aspergillus niger a-Galactosidase Aspergillus oryzae a-Galactosidase Aspergillus oryzae a-Galactosidase Coffee bean a- Galactosidase Coffee bean a-N-acetylgalactosaminidase Bovine liver a-N-acetylgalactosaminidase Bovine liver a-N-acetylgalactosaminidase Bovine liver a-N-acetylgalactosaminidase Bovine liver

Yield (YO) Reference

Product

Acceptor

Donor

Enzyme/Source

30

29

28

27

26

25

24

23

22

21

20

a-D-GalNAc-0 a-Galactosaminidase Aspergillus oryzae a-D-GalNAc-0 a-Galactosaminidase Aspergillus oryzae a-D-GalNAc-0 a-Galactosaminidase Aspergillus oryzae a-D-GalNAc-0 a-Galactosaminidase Aspergillus oryzae a-D-GalNAc-0 a-Galactosaminidase Aspergillus oryzae a-D-GalNAc-0 a-Galactosaminidase Aspergillus oryzae Glucoamylase a-D-Glc-0 Rhizopus oryzae Glucoamylase a-D-Glc-0 Rhizopus oryzae a-Glucosidase a-D-Gk-0 Talaromyces duponti a-Glucosidase a-D-Glc-0 Talaromyces duponti a-D-Glc-0 a-Glucosidase Talarornyces duponti 208

10-50

a-D-GalNAc-0-Ser-(N-COCH3)OMe

210

21 1

21 1

NA

23

18

Maltose

Maltose

Maltodextrins

Glc

Glc

209

208

2

a-D-GalNAc-0-Ser-(N-CO(CH3)3)- 10-~50 OMe

a-D-GalNAc-0-Ser-(NCOCH*C(C1)3)-OMe

209

Ser-(N-COC(CH3)3)OMe

Ser-(N-COCH2C(C1)3)OMe

208

208

10-50

a-D-GalNAc-0-Ser-(NCOCH2CH=CH2)-OMe

25

apNPGalNAc

apNPGalNAc

apNPGalN Ac Ser-(NCOCH2CH=CH2)OMe apNPGalNAc Ser-(N-COCH3)-OMe

10-50

208

50

a-D-GalNAc-0-Ser-(N-CO0Me)OMe

Ser-(N-CO0Me)-OMe

apNPGalNAc

208

29

a-D-GalNAc-0-Thr-(N-Ac)-OMe

apNPGalNAc Thr-(N-Ac)-OMe

w

co

3F

Q

3

n

8 14

29 Glycosiduse-Cutulysed Oligosaccharide Synthesis

r0 N

2

+ L

$

$ c

Q

Q L

Y C

5

58

57

56

55

54

53

52

51

50

49

48

47

46

45

44

a-Mannosidase Jack bean a-Mannosidase a-D-Man-0 Almond a-Mannosidase a-D-Man-0 Jack bean a-Mannosidase a-D-Man-0 Almond a-Mannosidase a-D-Man-0 Almond a-D-Man-0 a-Mannosidase Almond a-L- Ara-O P-Galactosidase Bovine liver a-L-Fuc-0 a-L-Fucosidase Limpet Invertase P-D-Fruf-0 Bakers yeast b-D-Fruf-0 Invertase Bakers yeast P-D-Fruf-0 Invertase Succharomyces cervisiae P-D-Gal( 1-2)- P-Galactosidase Bacillus DNJ circulans P-Galactosidase P-D-Gal(1-3)DNJ Bacillus circuluns p-D-Gal(1-4)- P-Galactosidase DNJ Bacillus circuluns p-~-Gal( 1-4)-0 Rhodotorulu luctosa cells

a-D-Man-0

205 205 107 202 214 214 215

216

216

3 2 3 2 3 1

40

6

20

a-D-Man-0-lysergol

a-~-Man-O-9,lO-dihydrolysergol

p-~-F:ruf-O-(CH2)3CH3

p-~-Gal( 1-3)-DNJ

9,lO-Dihydrolysergol Monobenzyl ether 1,3,5- monobenzyl ether 1,3-(a-L-Ara-0)1,3,5-benzene-trimethanol benzene-trimethanol a-L-Fuc-0-chanoclavine Chanoclavine !~-D-FIx~-O-(CH~)~CH~

Lysergol

apNPMan a-L-pNPAra

216

217

26

10

P-D-Gal(1 -4)-DNJ

S-D-Gal(1-4)-O-calystegineB2

HO-(CH2)3CH3 HO-(CH2)3CH3

DNJ

DNJ

DNJ

CalystegineB2

Sucrose Sucrose

Lactose

Lactose

Lactose

Lactose

P-D-FN~-O-(CH~)~CH~

HO-(CH2)3CH3

Fru

apNPFuc

apNPMan

205

3

a-D-Man-0-elymoclavine

Elymoclavine

apNPMan

213

68

a-~-Man-O-5-phenyl1-pentyl

5-phenyl-1-pentanol

apNPMan

202

18

a-D-Man-0-chanoclavine

Chanoclavine

apNPMan

202

13

a-D-Man-0-ergometrine

Ergometrine

Man

ul

L

00

3F

a

3

R

68

67

66

65

64

63

62

61

60

59

P-D-Gal(1-6)-

Entry Linkage

Table 3 (continued)

P-Galactosidase Bacillus circulans P-Galactosidase Aspergillus oryzae P-Galactosidase Aspergillus oryzae P-Galactosidase Aspergillus oryzae P-Galactosidase Asperg illus oryzae P-Hexosaminidase Aspergillus oryzae P-Galactosidase Escherichia coli P-Galactosidase Escherichia coli P-Galactosidase Escherichia coli P-Galactosidase Escherichia coli

Enzyme/Source

212

63

13

10

P-D-Gal-0-p-bis(hydroxymethyl) benzene P-~-Gal-o-Ser

p-bis(hydroxymethy1) benzene Serine

PpNPGal

PoNPGal

Lactose

Propane-] ,2-diol

26 (R: 8 ee)

219

219 P-D-G~~-O-CH(CH~)CH~OH 20 (R: 13 ee)

P-D-Gal-O-CH2CHOHCH3

219

32 ( R)-P-D-G~I-O-CH(CH~)CH~CH~

(R)-(-)-Butan-2-01

Lactose

Lactose

219 P-D-G~I-O-CH(CH~)CH~CH~ 32 (R: 6 ee)

Butan-2-01

Lactose

218

212 12

P-~-Gal-O-pentaerythritol

Pentaerythritol

Pp NPGa1

23 P-D-GIcNAc-O-~-COOCH~-~-NO~ OH BpNPGlcNAc p-COOCH3-m-NO2-Bn-

212

20

P-D-Gal-0-cyanuricacid

Cyanuricacid

PpNPGal

Reference 216

(“/o)

7.2

Yield

P-D-Gal(1-6)-DNJ

Product

DNJ

Acceptor

Lactose

Donor

b v3

a

03

P-D-Gal-0

P-D-Gal-0

P-D-Gal-0

p-D-Gal-0

p-D-Ga1-0

P-D-Gal-0

P-D-Gal-0

P-D-Gal-0

P-D-Gal-0

P-D-Gal-0

B-D-Gal-0

69

70

71

72

73

74

75

76

77

78

79

P-Galactosidase Escherichia coli P-Galactosidase Escherichia coli p-Galactosidase Escherichia coli P-Galactosidase Escherichia coli P-Galactosidase Escherichia coli P-Galactosidase Escherichia coli P-Galactosidase Escherichia coli P-Galactosidase Escherichia coli P-Galactosidase Escherichia coli B-Galactosidase Escherichia coli P-Galactosidase Asperg illus oryzae 219

219

13

219 8 (R: 33 ee)

@)-(+)-Propane-1,2-diol

Butane-l,3-diol

Butane-I ,3-diol

Lactose

Lactose

Lactose

219

219

44

44

44

220

7

42

NA

NA

NA

20

(S)-(+)-Butane-l,3-diol

Propane-l,3-diol

HOAll

HOBn

Lactose

Lactose

HO-(CH2)2Si(CH3)3

HO-(CH2)2C6H5

Lactose

PPhGal

Lactose

Lactose

219 54

(S)-(+)-Butane-l13-diol

Lactose

219

38

(S)-(+)-Propane-l,2-diol

Lactose

3F

4

Donor

89

88

87

86

85

84

83

82

81

P-Galactosidase Lactose Escherichiu coli 0-Galactosidase Lactose Escherichiu coli P-Galactosidase Lactose Escherichia coli P-Galactosidase Lactose Escherichiu coli P-Galactosidase Lactose Escherichiu coli P-Galactosidase Lactose Asperg illus oryzue P-Galactosidase Lactose Asperg illus oryzue P-Galactosidase PPhGal Asperg illus oryzae P-Galactosidase PPhGal Aspergillus oryzue P-Galactosidase PPhGal Asperg illus oryzue

~~

80

~

EnzymeJSource

Entry Linkage

~~~~

Table 3 (continued) ~

P-D-Gal-O-3-digitoxigenin

P-~-Gal-0-3-16B,17P-epoxy-170.digitoxigenin

16B,17P-EPOXY-170.digitoxigenin

P-~-Gal-O-3-gitoxigenin

~~

13 (70 de)

36 (50 de)

28 (63 de)

24 (75 de)

10 (90 de)

30 (89 de)

64

38

26

222, 223

222, 223

222, 223

22 1

22 1

22 1

22 1

22 1

22 1

22 1

Yield (“4)Reference

P-D-Gal-O-cis-l,2-~ycIohexanediol 20 (38 de)

P-D-Gal-0-cis- 1,2-cyclopentanediol

P-D-Gal-0-cis- 1,2-cyclopentanediol

P-D-Gal-0-cis-norbornanediol

0-D-Gal-0-cis-norbornenediol

P-D-Gal-o-cis- 1,2-cyclohexanediol

P-D-Gal-0-cis-1,2-cyclopentanediol

Product

Digitoxigenin

Gitoxigenin

cis- 1,2-CycIohexanediol

cis-I ,2-Cyclopentanediol

4-Cyclopentane- 1,4-diol

cis-Norbornanediol

cis-Norbornenediol

cis- 1,2-Cyclohexanediol

cis- 1,2-Cyclopentanediol

Acceptor

819

29.6 Recent Developments and New Directions

d

N N

d

N N

d

N N

d

,?i

N N

N

-

0 A

0

59

6n

U

U

ci

0

d

0

9 e,

-

c u

0

4

Q

c

9

Q

2i

0 h

9

0

U

a

.e

63

4

b

ci

t3

e

110

109

108

107

106

105

104

103

102

coli

P-Galactosidase Escherichia coli P-Galactosidase Kluyveromyces lactis P-Galactosidase Kluyverornyces lactis P-Galactosidase Kluyveromyces lactis P-Galactosidase Kluy veromyces lactis P-Galactosidase Kluyveromyces lactis P-Galactosidase Escherichia coli P-Galactosidase Escherichia coli

P-D-Gal-08 224 C H ( C H ~ ) ( C H ~ ) Z C O O ( C H ~ )(R: ~ C11 H ~ee)

CH3CHOH (CH2)zCOO(CH2)3CH3 HO-(CHz)60H

PoNPGal

Lactose

P-D-Gal-0-1-Glyceryl

Glycerol

HOBn

EtOH

Lactose

Lactose

Lactose

~-D-G~I-O-(CH~)~OH

P-D-G~~-O-(CH~)~OH

PoNPGal

Lactose

P-D-Gal-OEt

P-D-Gal-OBn

P-D-G~I-O(CH~)~OH

Lactose

P-D-G~I-O-(CH~)~OH

P-D-Gal-0CH(CH3)(CH2)2COOCH2CH3

CH3CHOH(CH2)2COOCH2CH3

50

92

36

21

60

40

49

226

226

225

225

225

225

225

37 224 (R: 55 ee)

46 224 (R: 65 ee)

P-D-Gal-0CH(CH3)(CH2)2COOCH3

Reference

CH3CHOH(CH2)2COOCH3

P-Galactosidase PoNPGal Escherichia coli P-Galactosidase PoNPGal Escherichia

101

Yield (%)

Product

Acceptor

Donor

Enzyme/Source

Entry Linkage

Table 3 (continued)

121

120

119

118

117

116

115

114

113

112

111

P-Galactosidase Escherichia coli P-Galactosidase Kluyveromyces lactis P-Galactosidase Kluyveromyces lactis 0-Galactosidase Asperg illus oryzae P-Galactosidase Asperg illus oryzae P-Galactosidase Escherichia coli P-Galactosidase Escherichiu coli P-Galactosidase Escherichia coli P-Galactosidase Escherichia coli

coli

C

2

sa

3s

214

227 228, 229

228, 229

228, 230

228,230

23

14

52

23

66

82

Phenylethanol

HO-CH(CH3)CH2CH3

HO-C(CH3)3

5-Phenylpentanol

HO-(CH2)7CH3

Lactose

PpNPGal

PpNPGal

PpNPGal

PpNPGal

Lactose

g. 0 5

9 2

s.

8

2

5

c.i rp

5

o\

HO-(CH2)3CH3

\o

ru

214 23

HO-(CH2)3CH3

Lactose

214

18

HO-(CHZ)~CH~

226

27

Gal

226

50

HO-(CH2)7CH3

HO-(CHz)gCH3

226

78

HO-(CH2)loOH

PoNPGal

P-Galactosidase PoNPGal Escherichiu coli P-Galactosidase PoNPGal Escherichia

0-Galactosidase Escherichia coli P-Galactosidase Escherichia coli P-Galactosidase Escherichia coli P-Galactosidase Escherichia coli P-Galactosidase Escherichia coli P-Galactosidase Bacillus circulans P-Galactosidase Bucillus circulans 0-Galactosidase Bacillus circulans P-Galactosidase Bacillus circulans 0-Galactosidase Escherichia coli

122

131

130

129

128

127

126

125

124

123

Enzyme/Source

Entry Linkage

Table 3 (continued) ~

228, 231

41 16

36

<5

67

P-D-Gal-0-cholesteryl

P-D-Gal-O-(R)-1-phenylethyl P-D-Gal-o-(S)-I-phenylethyl

P-D-Gal-O-(R)-1-methyloctyl

P-D-Gal-O-(S)-I-methyloctyl

P-D-G~~-O-(CH~)~CH~

Cholesterol

(R)-1-Phenylethanol

(S)-I-Phenylethanol

(R)-I-Methyloctanol

(S)- 1-Methyloctanol

HO-(CH&CHj

PpNPGal

PpNPGal

PpNPGal

PpNPGal

PpNPGal

PpNPGal

228, 230

228, 230

228

228

228

228

P-D-Gal-0-1,2-di-O-dodecylglyceryl 52

1,2-di-0Dodecylglycerol

PpNPGal

7.2

228

P-D-Gal-O-1,3-di-O-dodecylglyceryl9

1,3-di-0Dodecy lglycerol

PpNPGal

%,

G.

2

9 E

P

5.

5

0, 8g

6.

+ -

Ba

f3

8-?

13

228, 230

15

P-D-Gal-O-(CHl)II C H ~

PpNPGal

2 228, 230

Yield (“h) Reference 13

Product P-D-G~~-O-(CH~)~CH~

Acceptor

PpNPGal

Donor

00 h,

w

29.6 Recent Developments and New Directions

z

rn 0

N

c\I

WN

00-

N

N

r J

N m

0 rn N

0

m

0 m N

0

N

w

00-

00-

w-

N

N

N N

00

m

3

N

w

N

v,

m

N N

m

N

N

2 N

m g r . 4N gg hl N

m

rN

W W

m

N

N

0

d

3

3

N

N

m

s-

rT)

N

N

d

s

0

cu

-

0

a ci

n u

0

d

+ 0

4

U

b

0

z

Ci

Ci

d

b

ru

m

m

0-

b

v

823

N

d

0-D-Gal-0

P-D-Gab0

p-D-Gal-0

P-D-Gal-0

p-D-Gal-0

P-D-Gal-0

0-D-Gal-0

P-D-Gal-0

P-D-Gal-0

B-D-Gal-0

144

145

146

147

148

149

150

151

I52

153

Entry Linkage

Table 3 (continued)

P-Galactosidase Bacillus circulans P-Galactosidase Escherichia coli P-Galactosidase Aspergillus oryzae P-Galactosidase Escherichia coli P-Galactosidase Escherichia coli 0-Galactosidase Escherichia coli P-Galactosidase Escherichia coli P-Galactosidase Escherichia coli 0-Galactosidase Aspergillus oryzae P-Galactosidase Aspergillus oryzae

Enzyme/Source

106

106

106

106

13

9

11

P-D-Gal-O-Ser-(N-Boc)-Gly-OMe

P-D-Gal-O-Ser-(N-Boc)-Ala-OMe

P-D-Gal-O-Ser-(N-Boc)-Ser-OMe

Gly-(N-BOG)-[ P-~-Gal]-0-Ser-OMe 7

5

P-o-Gal-O-but-3-en-2-yl

But-3-en-2-01

Ser-(N-Boc)-Gly-OMe

Ser-(N-Boc)-Ah-OMe

Ser-(N-Boc)-Ser-OMe

Gly-(N-Boc)-Ser-OMe

Ala-(N-Boc)-Ser-OMe

HOAll

HOAll

Gal

Lactose

Lactose

Lactose

Lactose

Lactose

PoNPGal

PVGal

93

93

62

50 /3-~-Gal-OAll

213

P-D-Gal-OAlI

Ala-(N-BOG)-[ P-~-Gal]-0-Ser-OMe

21 233 (R: 40 ee) 106

62

P-~-Gal-O-5-phenyl-1-pentyl

5-Phenyl-1-pentanol

PpNPGal

230, 231

35

P-D-Gal-0-pNP

PNP

PpNPGal

Yield (“YO) Reference

Acceptor

Product

Donor

P-u-Gal-0

0-D-Gal-0

P-D-Gal-0

0-u-Gal-0

P-r>-Gal-O

P-D-Gal-0

P-u-Gal-0

158

159

160

161

162

163

164

157

234

204

204

48

22

fi-D-Gal-O(CH2)60H

HO-(CH2)60H

Lactose

Gal

Gal

236

17

15

P-D-Gal-OAI1

HOAll

Gal

38

235

4

P-D-Gal-0-Thr-(N-Boc)

Thr-(N-Boc)

P-D-Gdl-OAll

235

4

P-D-Gal-0-Ser-(N-Boc)

Ser-(N-Boc)

Lactose

HOAll

234

28

P-D-Gal-O-Hyp-(N-Z)-OMe

Hyp-(N-Z)-OMe

Lactose

Lactose

234

10

P-D-Gal-0-Thr-(N-Z)-OMe

Thr-(N-Z)-OMe

107

107

Monoallyl ether (4,6-iso- monoally1 ether (P-~-Gal-0)-(4,6- 17 isopropylidine-P-D-Glc-0)-1,3,5propyhdine-P-D-Glcbenzene-trimethanol 0)-1,3,5-benzenetrimethanol monoally1 ether (p-D-Gal-O)-(p-D- 10 Monoallyl ether (0-DGlc-O)-l,3,5-benzeneGlc-0)-1,3,5-benzene-trimethanol trimethanol 35 P-u-Gal-0-Ser-(N-Z)-OMe Ser-(N-Z)-OMe

Lactose

Lactose

P-Galactosidase Achatina uchatinu P-Galactosidase Achatina achatina P-Galactosidase Achatina achatina 0-Galactosidase Asperg illus oryzae P-Galactosidase Asperg illus oryzae P-Glucosidase Almond P-Galactosidase Asperg illus oryzae P-Galactosidase Asperg illus oryzae P-Galactosidase Streptococcus therrnophilus

P-D-Gal-0

156

P-Galactosidase PoNPGal Bovine liver

P-Galactosidase PoNPGal Bovine liver

P-u-Gal-0

155

154

174

173

172

171

170

169

168

167

166

165

Entry Linkage

Table 3 (continued)

P-Galactosidase Escherichia coli P-Galactosidase Escherichia coli P-Galactosidase Escherichia coli 0-Galactosidase Achatina achatina P-Galactosidase Achatina achatina P-Galactosidase Achatina achatina P-Galactosidase Achatina achatina P-Galactosidase Achatina achatina P-Galactosidase Achatina achatina Rhodotorula lactosa cells

P-D-Gal-O-Ser-(N-Boc)-OMe

P-D-Gal-O-Ser-(N-A1oc)-OMe

P-D-Gal-O-Ser-(N-Z)-OMe

P-D-Gal-0-Ser-(N-Z)-Gly-OEt

P-D-Gal-0-Ser-(N-Z)-Ala-OEt

Gly-(N-2)-[P-~-Gal-o]-Ser-OEt

Ala-(N-Z)-[P-~-Gal-0]-Ser-OEt

Ala-(N-Z)-[P-~-Gal-0]-Ser-OMe

P-D-Gal-0-Thr-(N-Z)-Gly-OEt

Ser-(N-Boc)-OMe

Ser-(N-Aloc)-OMe

Ser-(N-Z)-OMe

Ser-(N-Z)-Gly-OEt

Ser-(N-Z)-Ala-OEt

Gly-(N-Z)-Ser-OEt

Ala-(N-Z)-Ser-OEt

Ala-(N-Z)-Ser-OMe

Thr-(N-Z)-Gly-OEt

CalystegineB 1

Lactose

Lactose

Lactose

Lactose

Lactose

Lactose

Lactose

Lactose

Lactose

Cellobiose

P-D-GIc(1-3)-O-calystegineB1

Product

Acceptor

Enzyme/Source Donor

0.9

9

7

8

29

15

16

9

15

15

Yield (“!o)

217

237

206

Reference b v3

m

N

03

827

29.6 Recent Developments and New Directions m m

m

2 N

% N

0

N

N

0 N N

oom "

m

m

N

m

N

N

00-

"

m

N

00-

m

N

m

m

N

oom

N

r

%

4

w-

m

%

%

00-

00-

N

m

N

m

m m

N

m m

N

d

z

5 ? P

82 a

P

sz

v

v

?

-8

?

'1 a

u

5a

u

L 8

v 3

?

?

h

h

2

-t

3

5

2

v

e:Pe

L

8 G

8

b ci

c i u

8

8

8 8

8 8 8 8

ci

& I

u

&L

Bn

5n

5a 59

c?

5a 5n 59 ci

cl

9

cl

8 8 59 5a a

&L

240

240 240 24 1

2 1.9 1.6 1.1

1 22

P-D-Glc-0-mHOBn P-D-Glc-0-pHOBn P-D-Glc-0-oHOBn P-D-Glc-0-mHOBn P-D-Glc-0-pHOBn P-D-Glc-OMe

P-D-Gk-OEt

P-D-Glc-OPr

P-~-Glc-O-3-gitoxigenin

P-~-Glc-O-3-digitoxigenin

mHOBnOH pHOBnOH oHOBnOH mHOBnOH pHOBnOH MeOH

EtOH

PrOH

Gitoxigenin

Digitoxigenin

Cellobiose Cellobiose Glc Glc Glc Cellobiose

Cellobiose

Cellobiose

PPhGlc

PPhGlc

P-D-G~c-O

P-D-G~c-O

P-D-G~c-O

P-D-G~c-O

P-D-Glc-0

P-D-G~c-O

P-D-G~c-O

P-D-GIc-O

P-D-G~c-O

P-D-GIc-O

191

192

193

194

195

196

197

198

199

200

222, 223

222, 223

74

24 1

24 1

240

240

43

33

10

240

Cellobiose

P-Glucosidase Almond P-Glucosidase Almond 0-Glucosidase Almond P-Glucosidase Almond P-Glucosidase Almond P-Glucosidase Almond P-Glucosidase Fusarium oxysporum P-Glucosidase Fusarium oxysporum P-Glucosidase Fusarium oxysporum P-Galactosidase Aspergillus oryzae 0-Galactosidase Aspergillus oryzae

P-D-GIc-O

190

239

P-D-G~c-O-CH~CH(CH~)NHAC 39 HOCH~CH(CH~)NHAC (R) ( R) P-D-Glc-0-oHOBn 3 oHOBnOH

PoNPGlc

P-Glucosidase Almond

P-D-G~c-O

Yield (“A) Reference

189

Product

Acceptor

Donor

Enzyme/Source

Entry Linkage

Table 3 (continued)

829

29.6 Recent Developments and New Directions

m

e

a"

b ci

9 B9 U

9

9

Bn

B9 Ba

0

ci

U

0

z

ci

Geraniol

3.4

246

246

Cellobiose

P-D-GIc-O

227

1.2

Geraniol

Cellobiose

P-D-GIC-O

226

214

44

HO-(CH2)3CH3

Glc

P-D-G~c-O

225

245

41

HO-(CH2)60H

Glc

P-D-Gk-0

224

245

40

HO-(CH2)50H

Glc

P-D-Glc-0

223

245

47

HO-(CH2)40H

Glc

P-D-G~c-O

222

24 5 55

HO-(CH2)30H

Glc

P-D-GIc-O

22 1

244

7.4

Cellobiose

P-D-Glc-0

220

244

12

Cellobiose

P-D-GIc-O

219

243

2.1

243

42

Glc

P-D-Glc-0

2 18

Reference

Yield ('%)

243

P-D-Glc-0

217

Product

3.2

PPhGlc

P-Glucosidase Sulfolobus solfataricus P-Glucosidase Almond P-Glucosidase Sulfolobus solfataricus P-Glucosidase Trichoderma viridae J3-Glucosidase Trichoderma viridae P-Glucosidase Almond P-Glucosidase Almond J3-Glucosidase Almond P-Glucosidase Almond P-Glucosidase Almond 0-Glucosidase Trichoderma reesei 0-Glucosidase Asperg illus niger

P-D-G~c-O

216

Acceptor

Glc

Donor

Enzyme/Source

Entry Linkage

Table 3 (continued)

0

w

00

P-D-GIc-O

P-D-GIC-O

P-D-Glc-0

P-D-G~c-O

P-D-G~C-O

P-D-GIc-O

p-D-G1c-0

P-D-GIC-O

P-D-G~c-O

P-D-G~c-O

P-D-GIc-O

P-D-G~c-O

229

230

23 1

232

233

234

235

236

237

238

239

240

228

P-Glucosidase Cundidu molischiana P-Glucosidase Almond P-Glucosidase Trichoderma reesei P-Glucosidase Asperg illus niger 0-Glucosidase Cundida molischiana P-Glucosidase Almond P-Glucosidase Trichoderrna reesei P-Glucosidase Asperg illus niger P-Glucosidase Almond P-Glucosidase Almond P-Glucosidase Almond P-Glucosidase Asperg illus niger 0-Glucosidase Agrobacterium tumejaciens 213

248

23

80

P-~-Glc-O-5-phenyl-1 -pentyl

5-Phenyl-1-pentanol

PpNPGlc

PpNPGlc

247 8 P-D-GIc-O-(CH~)~CH~

HO-(CH2)7CH3

247 17

2-Hydroxybenzylalcohol P-~-Glc-O-2-hydroxybenzyl

Glc Glc

246 0.1

246 0.3

P-D-Glc-0-citronellyl

246 0.3

Citronellol

246 0.1

Cellobiose

246

3.3

246

246

0.8

0.6

246

0.6

P-D-Glc-0-citronellyl

246

0.1

Citronellol

P-u-Glc-0-neryl

P-D-Glc-0-geranyl

Cellobiose

Citronellol

Nerol

Cellobiose Cellobiose

Nerol

Cellobiose

Nerol

Cellobiose

Nerol

Geraniol

Cellobiose

Cellobiose

Geraniol

Cellobiose

c

w

co

P-Glucosidase Almond

P-Glucosidase Almond P-Glucosidase Almond P-Glucosidase Almond P-Glucosidase Almond P-Glucosidase Almond P-Glucosidase Sulfolobus solfataricus

P-D-Glc-0

P-D-G1c-0

P-D-GIc-O

P-D-G~c-O

P-D-Glc-0

P-D-GIc-O

P-D-GIc-O

P-D-GIc-O

P-D-G~c-O

P-D-G1c-0

P-D-Glc-0

242

243

244

245

246

247

248

249

250

25 1

252

P-Glucosidase Almond P-Glucosidase Almond P-Glucosidase Almond P-Glucosidase Almond P-Glucosidase Almond

P-D-Gk-0

Enzyme/Source

241

Entry Linkage

Table 3 (continued)

Hex-1-en-3-01 Pent-3-en-2-01

Glc Glc

HO-CH2CHOHCH3

Pent-1-en-3-01

Glc

PPhGlc

But-3-en-2-01

HO-(CH&jOH

10

Glc

249

7

HO-(CH2)60H

1,2-3,5-DiphenylboronateGlcf 1,2-3,5-DiphenylboronateGlcf Glc

77

12 (R: 40 ee) 5 (R: 80 ee) 17 (R: 37 ee) 17

(R: 20 ee)

250

250

233

233

233

233

249

62

HOAll

Glc

27

249

17

HS-(CH& SH

249

249

7

HO-(CH2)6NHCOCF?

Glc Glc

249

Reference

Yield (%) 61

Product

HO-(CH2)60H

Acceptor

Glc

Donor

w

CO W

265

264

263

262

26 1

260

259

258

257

256

255

254

253

P-Glucosidase Sulfolobus souataricus P-Glucosidase P-D-GIGO Pyrococcus furiosus P-D-GIc-O P-Glucosidase Pyrococcus jiuriosus P-D-GIC-O P-Glucosidase Almond P-D-GIGO 0-Glucosidase Almond P-D-Glc-0 P-Glucosidase Almond P-D-Glc-0 P-Glucosidase Almond P-D-Glc-0 P-Glucosidase Almond P-D-G~c-O P-Glucosidase Almond P-D-G~cNAc-O P-Hexosaminidase Aspergillus oryzae P-D-GIcNAc-0 P-Hexosaminidase Aspergillus oryzae P-D-GIcNAc-0 0-Hexosaminidase Aspergillus oryzae B-D-GIcNAc-0 P-Hexosaminidase Aspergillus oryzae

P-D-G~c-O

250

250

63

15

Cellobiose

Cellobiose

17 204 204 204 212

212

208

208

40 61 50

11 29

12

10

10-50

HOBn HO-(CH2)60H HO-(CH2)3CH=CH2 HO-(CH2)2Si(CH3)3 Cyanuricacid

Glc Glc Glc Glc PpNPGlcNAc

PpNPGlcNAc Ser-(N-COCHzCH= CH2)-OMe

PpNPGlcNAc Ser-(N-CO0Me)-OMe

P-D-GlcNAc-0-Ser-(NCOCH2CH=CH2)-OMe

17 65

HOAll

Glc

PpNPGlcNAc p-bis(hydroxymethy1) benzene

25 1

NA

HO-(CH2)7CH3

Glc

P-D-GIcNAc-O-~ bis(hydroxymethy1) benzene P-D-GlcNAc-0-Ser-(N-CO0Me)OMe

250

HO-CH2CHOHCH3

20

PPhGlc

b

w w

cb

E

F:

(D

211

276

215

214

273

212

271

210

269

268

261

266

~~

Enzyme/Source

P-D-G~cNAc-O P-Hexosaminidase Bovine kidney P-Glucosidase P-D-GIc-O Almond 0-Mannosidase P-D-Man-0 Rhizopus niveus P-Mannosidase 0-D-Man-0 Rhizopus niveus P-Mannosidase P-D-Man-0 Rhizopus niveus P-Mannosidase P-D-Man-0 Snail 0-Mannosidase P-D-Man-0 Snail P-D-Man-0 P-Mannosidase Snail P-Mannosidase P-D-Man-0 Snail 0-Mannosidase P-D-Man-0 Snail 0-Mannosidase 0-D-Man-0 Snail P-Mannosidase P-D-Man-0 Snail

Entry Linkage

____

Table 3 (continued)

~~

Acceptor

206

8

76

45

P-D-Man-OMe P-D-Man-OEt P-D-Man-O-CH(CH3)l

P-D-Man-O-(CH2)20H P-D-Man-O-(CH2)40H P-D-Man-O-(CH&OH

MeOH EtOH HO-CH(CH3)2 HO-(CH2)3CH3 HO-(CH2)20H HO-(CH2)40H HO-(CH2)60H

PpNPMan PpNPMan PpNPMan PpNPMan PpNPMan PpNPMan PpNPMan

P-D-M~~-O-(CH~)~CH~

P-D-Man-OBn

P-D-Man(1-4)- HOBn D-Man

36

19

9

15

61

15

252

252

252

252

252

252

252

70

70

4

P-D-Man-O-(CH2)20H

70

1

P-D-Man(1-4)D-Man

P-D-Glc-Ser-(N-Aloc)-OMe P-D-Man-O-(CH2)3CH3

Ser-(N-A1oc)-OMe

Reference

213

Yield (%)

P-~-GlcNAc-O-5-phenyl-l-pentyl 33

Product

b-D-Man( 1-4)- HO-(CH2)3CH3 D-Man

Cellobiose

PpNPGlcNAc 5-Phenyl-1-pentanol

Donor

29 1

290

289

288

287

286

285

284

283

282

28 1

280

279

278

P-Mannosidase Snail 0-Mannosidase Snail P-Mannosidase Snail P-Mannosidase Snail P-Mannosidase Snail P-Mannosidase Snail P-Mannosidase Snail P-Mannosidase Snail P-Mannosidase Snail P-Mannosidase Snail P-Galactosidase Asperyillus oryzae (Impurity?) P-Galactosidase Aspergillus oryzae (Impurity?) P-Mannosidase Helix pomatia P-Xylosidase Aspergillus niger MeOH

HO-(CH2)60H

Man

XYh

0-D-Man-0-Thr-( N-Boc)

Thr-( N-Boc)

Man

253, 254

235

8

84

235

14

64 2

PpNPMan

P-D-Man-O-Ser-(N-Boc)

64 15

PpNPMan

Ser-(N-Boc)

64

1

PpNPMan

Man

64

5

PpNPMan

64

64

20

PpNPMan

3

64

50

PpNPMan

P-D-Man-0-cyclohexanol

64

75

PpNPMan

PpNPMan

64

65

PpNPMan

CH(CH3)(CH2)2CH3 Cyclohexanol

252

26

PpNPMan

UI

w

00

836

29 Glycosidase-Catalysed Oligosaccharide Synthesis

s N

W-

vr

N

00 N

-9u,z? u

0

v

39

U

5

u,

h

g u 0 z

v

-x"

N

m N N

W

m N

3N

vr

m

N

W

r-

00

N

N

N

m

Q\

m

m

N m

0

O m

-

X

29.6 Recent Developments and New Directions

z N

3 N

m

m"

m

N

N

N

m

m

m

m m

N

m m

N

m m N

Q

z

N

d

?

U

N

m 0

m m 0

Bm

omm

\o 0

m

r-

4

00

4

837

838

29 Glycosidase-Catalysed Oligosaccharide Synthesis

References E. Bourquelot, Ann. Chim., 1913, 29, 145. G. M. Watt, P. A. S. Lowden, and S. L. Flitsch, Curr. Opin. Struct. Biol., 1997, 7, 652-660. D. H. G. Crout and G. Vic, Curr. Opin. Struct. Biol., 1998, 2, 98-111. V. Kren and J. Thiern, Chem. SOC.Rev., 1997,26,463-473. C. H. Wong, R. L. Halcomb, Y. Ichikawa, and T. Kajimoto, Angew. Chem. Znt. Ed. Engl., 1995,34, 521-546. 6. J. Thiem, FEMS Microbiol. Rev., 1995, 16, 193-211. 7. K. G. I. Nilsson, Tibtech, 1988, 6, 256-264. 8. B. Henrissat and G. Davies, Curr. Opin. Struct. Biol., 1997, 7, 637-644. 9. D. E. Koshland, Biol. Reu., 1953,28, 416-436. 10. M. L. Sinnott, Chem. Rev., 1990, 90, 1171-1202. 11. J. D. McCarter and S. G. Withers, Curr. Opin. Struct. Biol., 1994, 4 , 885-892. 12. G . Davies and B. Henrissat, Structure, 1995, 3, 853-859. 13. G. Davies, M. L. Sinnott, and S. G. Withers, in Comprehensive Biological Catalysis, (Ed. M. L. Sinnott), Academic Press, London, Vol. I , 1997, pp. 119-208. 14. T. Usui, H. Matsui, and K. Isobe, Curbohydr. R e x , 1990, 203, 65-77. 15. K. Ajisaka, I. Matsuo, M. Isornura, H. Fujirnoto, M. Shirakabe, and M. Okawa, Curbohydr. Rex, 1995, 270, 123-130. 16. S. Suwasono and R. A. Rastall, Biotechnol. Lett., 1996, 18, 851-856. 17. G. Vic and D. H. G. Crout, Carbohydr. Rex, 1995,279, 315-319. 18. S. Nishimura and K. Yamada, J. Am. Chem. Soc., 1997, 119, 10555-10556. 19. S. Kobayashi, T. Kiyosada, and S. Shoda, J. Am. Chem. Soc., 1996, 118, 13113-13114. 20. P. Finch and J. H. Yoon, Carbohydr. Rex, 1997, 303, 339-345. 21. K. Ajisaka, H. Nishida, and H. Fujirnoto, Biotechnol. Lett., 1987, 9, 387-392. 22. G. F. Herrmann, Y. Ichikdwa, C. Wandrey, F. C. A. Gaeta, J. C. Paulson, and C.-H. Wong, Tetrahedron Lett., 1993, 34, 3091-3094. 23. Y. Ito and J. C. Paulson, J. Am. Chem. Soc., 1993,115, 7862-7863. 24. V. Kren and J. Thiem, Angew. Chem. Znt. Ed. Engl., 1995,34, 893-895. 25. T. Usui, S. Kubota, and H. Ohi, Curbohydr. R e x , 1993, 244, 315-323. 26. K. G. I. Nilsson, Carbohydr. R e x , 1987, 167, 95-103. 27. D. H. G. Crout, S. Singh, B. E. P. Swoboda, P. Critchley, and W. T. Gibson, J. Chem. Soc., Chem. Commun., 1992, 704-705. 28. L. F. Mackenzie, Q. P. Wang, R. A. J. Warren, and S. G. Withers, J. Am. Chem. Soc., 1998, 120, 5583-5584. 29. C. Malet and A. Planas, FEBS Lett., 1998, 440, 208-212. 30. K. Takegawa, S. Yamaguchi, A. Kondo, I. Kato, and S. Iwahara, Biochem. Znt., 1991, 25, 829-835. 31 K. Takegawa, S. Yamaguchi, A. Kondo, H. Iwamoto, M. Nakoshi, I. Kato, and S. Iwahara, Biochem. Znt., 1991,24, 849--855. 32 K. Takegawa, M. Tabuchi, S. Yamaguchi, A. Kondo, I. Kato, and S. Iwahara, J. Biol. Chem., 1995,270, 3094-3099. 33 J. Q. Fan, M. S. Quesenberry, K. Takegawa, S. Iwahara, A. Kondo, I. Kato, and Y. C. Lee, J. Biol. Chem., 1995,270, 17730-17735. 34 K. Yarnamoto and K. Takegawa, Trends Glycoscience Glycotechnol., 1997, 9, 339-354. 35 L. X. Wang, J. Q. Fan, and Y. C. Lee, Tetrahedron Lett., 1996, 37, 1975-1978. 36 J. Q. Fan, K. Takegawa, S. Iwahara, A. Kondo, I. Kato, C. Abeygunawardana, and Y. C. Lee, J. Biol. Chem., 1995,270, 17723-17729. 37 H. Nakano, M. Moriwaki, T. Washino, and S. Kitahata, Biosci. Biotechnol. Biochem., 1994, 58, 1435-1438. 38 H. Nakano, K.-I. Hamayasu, K. Fujita, K. Hara, M. Ohi, H. Yoshizumi, and S. Kitahata, Biosci. Biotechnol. Biochem., 1995, 59, 1732-1736. 39 H. Hashimoto, C . Katayama, M. Goto, T. Okinaga, and S. Kitahata, Biosci. Biotechnol. Biochem., 1995,59, 179-183.

I. 2. 3. 4. 5.

References

839

40. K. Ajisaka, H. Fujimoto, and H. Nishida, Curbohydr. Rex, 1988, 180, 35-42. 41. K. Koizumi, T. Tanimoto, Y. Okada, K. Ham, K. Fujita, H. Hashimoto, and S. Kitahata, Curbohydr. Rex, 1995, 278, 129-142. 42. K. Koizumi, T. Tanimoto, Y. Kubota, and S. Kitahata, Curbohydr. Rex, 1998,305, 393-400. 43. K. G. I. Nilsson and A. Fernandez-Mayoralas, Biotechnol. Lett., 1991, 13, 715-720. 44. K. G. I. Nilsson, Carbohydr. Rex, 1988, 180, 53-59. 45. K. G. I. Nilsson, Curbohydr. Rex, 1989, 188, 9-17. 46. A. Naundorf, M. Caussette, and K. Ajisaka, Biosci. Biotechnol. Biochem., 1998, 62, 13131317. 47. I. Matsuo, H. Fujimoto, M. Isomura, and K. Ajisaka, Bioorg. Med. Chem. Lett., 1997, 7, 255-258. 48. G. Vic, C. H. Tran, M. Scigelova, and D. H. G. Crout, J. Chem. Soc., Chem. Commun., 1997, 169- 170. 49. K. G. I. Nilsson, Tetrahedron Lett., 1997, 38, 4527-4527. 50. G. Vic, M. Scigelova, J. J. Hastings, 0. W. Howarth, and D. H. G. Crout, J. Chem. Soc., Chem. Commun., 1996, 1473-1474. 51. A. Millqvist-Fureby, D. A. MacManus, S. Davies, and E. N. Vulfson, Biotechnol. Bioeng., 1998, 60, 197-203. 52. S. Kitahata, H. Ishikawa, T. Miyata, and 0. Tanaka, Agric. Biol. Chem., 1989, 53, 29292934. 53. K. Hara, K. Fujita, N. Kuwahara, T. Tanimoto, H. Hashimoto, K. Koizumi, and S. Kitahata, Biosci. Biotechnol. Biochem., 1994, 58, 652-659. 54. S. Kitahata, K. Hara, K. Fujita, N. Kuwahara, and K. Koizumi, Biosci. Biotechnol. Biochem., 1992,56, 1518-1519. 55. K. Ajisaka and H. Fujimoto, Curbohydr. Rex, 1989, 185, 139-146. 56. K. G. I. Nilsson, Curbohydr. Rex, 1990, 204, 79-83. 57. K. Dax, M. Ebner, R. Peinsipp, and A. E. Stutz, Tetruhedron Lett., 1997, 38, 225-226. 58. T. Kasumi, C. F. Brewer, E. T. Reese, and E. J. Hehre, Curbohydr. Rex, 1986, 146, 39-49. 59. H. Fujimoto and K. Ajisaka, Biotechnol. Lett., 1988, 10, 107-112. 60. H. Fujimoto, H. Nishida, and K. Ajisaka, Agric. Biol. Chem., 1988, 52, 1345-1351. 61. S. V. Lobov, R. Kasai, K. Ohtani, 0. Tanaka, and K. Yamasaki, Agric. Biol. Chem., 1991, 55,2959-2965. 62. K. Ajisaka and H. Fujimoto, Curbohydr. Res., 1990, 199, 227-234. 63. B. Sauerbrei and J. Thiem, Tetrahedron Lett., 1992, 33, 201-204. 64. N. Taubken, B. Sauerbrei, and J. Thiem, J. Curbohydr. Chem., 1993,12, 651-667. 65. V. Laroute and R. M. Willemot, Enzyme Microb. Technol., 1992, 14, 528-534. 66. V. Laroute and R.-M. Willemot, Biotechnol. Lett., 1989, 11, 249-254. 67. R. A. Rastall, N. H. Rees, R. Wait, M. W. Adlard, and C. Bucke, Enzyme Microb. Technol., 1992,14, 53-57. 68. S. Suwasono and R. A. Rastall, Biotechnol. Lett., 1998, 20, 15-17. 69. E. Johansson, L. Hedbys, and P.-0. Larsson, Biotechnol. Lett., 1986, 8, 421-424. 70. H. Fujimoto, M. Isomura, and K. Ajisaka, Biosci. Biotechnol. Biochem., 1997, 61, 164-165. 71. K. Hamayasu, K. Hara, K. Fujita, Y. Kondo, H. Hashimoto, T. Tanimoto, K. Koizumi, H. Nakano, and S. Kitahata, Biosci. Biotechnol. Biochem., 1997, 61, 825-829. 72. K. Hara, K. Fujita, H. Nakano, N. Kuwahara, T. Tanimoto, H. Hashimoto, K. Koizumi, and S. Kitahata, Biosci. Biotechnol. Biochem., 1994, 58, 60-63. 73. T. J. Bartlett, R. A. Rastall, N. H. Rees, M. W. Adlard, and C. Bucke, J. Chem. Technol. Biotechnol., 1992, 55, 73-78. 74. A. Vetere, S. Ferro, M. Bosco, P. Cescutti, and S. Paoletti, Eur. J. Biochem., 1997, 247, 10831090. 75. Y. Makimura, H. Ishida, A. Kondo, A. Hasegawa, and M. Kiso, J. Curbohydr. Chem., 1998, 17, 975-979. 76. J. Thiem and B. Sauerbrei, Angew. Chem. Int. Ed. Engl., 1991, 30, 1503-1505. 77. A. Vetere and S. Paoletti, FEBS Lett., 1996, 399, 203-206. 78. K. Ajisaka, H. Fujimoto, and M. Isomura, Carbohydr. Rex, 1994, 259, 103-115. 79. S. C. T. Svensson and J. Thiem, Curbohydr. Rex, 1990, 200, 391-402.

840

29 Glycosiduse-Cutalysed Oligosucckuride Synthesis

K. Ajisaka and M. Shirakabe, Carbohydr. Rex, 1992, 224, 291-299. K. Ajisaka, H. Fujimoto, and M. Miyasato, Curbohydr. Rex, 1998, 309, 125-129. A. Vetere, C. Galateo, and S. Paoletti, Biochem. Biophys. Res. Commun., 1997,234, 358-361. K. G. I. Nilsson, H. F. Pan, and U. Larsson-Lorek, J. Curbohydr. Chem., 1997, 16, 459477. 84. K. G. I. Nilsson, A. Eliasson, and U. Larsson-Lorek, Biotechnol. Lett., 1995, 17, 717-722. 85. H. Prade, L. F. MacKenzie, and S. G. Withers, Carbohydr. Rex, 1998,305, 371-381. 86. K. Fujita, N. Kuwahara, T. Tanimoto. K. Koizumi, M. Iizuka, N. Minamiura, K. Furuichi, and S. Kitahata, Biosci. Biotechnol. Biochem., 1994, 58, 239-243. 87. M. Muramatsu and T. Nakakuki, Biosci. Biotechnol. Biochem., 1995, 59, 208-212. 88. M. Kurakake, T. Onoue, and T. Komaki, Appl. Microbiol. Biotechnol., 1996, 45, 236-239. 89. A. J. J. Straathof, A. P. G. Kieboom, and H. van Bekkum, Carbohydr. Rex, 1986, 146, 154159. 90. H. Ishikawa, S. Kitahata, K. Ohtani, and 0. Tanaka, Chem. Pharm. Bull., 1991, 39, 20432045. 91. H. Ishikawa, S. Kitahata, K. Ohtani, C. Ikuhara, and 0. Tanaka, Agric. Biol. Chem., 1990, 54, 3137-3143. 92. T. Usui, S. Morimoto, Y. Hayakawa, M. Kawaguchi, T. Murata, Y. Matahira, and Y. Nishida, Curbohydr. Res., 1996,285, 29-39. 93. V. Chiffoleau-Giraud, P. Spangenberg, and C. Rabiller, Tetrahedron: Asymmetry, 1997, 8, 20 17-2023. 94. S. Yanahira, T. Kobayashi, T. Suguri, M. Nakakoshi, S. Miura, H. Ishikawa, and I. Nakajima, Biosci. Biotechnol. Biochem., 1995, 59, 1021-1026. 95. E. Montero, J. Alonso, F. J. Canada, A. Fernandez-Mayorales, and M. Martin-Loma, Cavbohydr. Res., 1998,305, 383-391. 96. J. J. Aragon, F. J. Canada, A. Fernandez-Mayoralas, R. Lopez, M. Martinlomas, and D. Villanueva, Curbohydr. Rex, 1996,290, 209-216. 97. G. C. Look and C. H. Wona. Tetrahedron Lett., 1992, ?, 4253-4256. 98. K. Suzuki, H. Fujimoto, Y. ko, T. Sasaki, and K. Ajisaka, Tetrahedron Lett., 1997,38, 12111214. 99. U. Gambert and J. Thiem, Carbohydr. Rex, 1997,299, 85-89. 100. T. Murata, T. Itoh, Y. Hayakawa, and T. Usui, J. Biochem., 1996, 120, 851-855. 101. U. Gambert, R. G. Lio, E. Farkas, J. Thiem, V. V. Bencomo, and A. Liptak, Bioorg. Med. Chem., 1997, 5, 1285-1291. 102. K. G. I. Nilsson, Biotechnol. Lett., 1996, 18, 791-794. 103. T. Yasukochi, K. Fukase, Y. Suda, K. Takagaki, M. Endo, and S. Kusumoto, Bull. Chem. Soc. Jpn., 1997, 70, 2719-2725. 104. V. Dumortier, J. Montreuil, and S. Bouquelet, Carbohydr. Res., 1990, 201, 115-123. 105. S. Yanahira, Y. Yabe, M. Nakakoshi, S. Miura, N. Matsubara, and I. Nakajima, Biosci. Biotechnol. Biochem., 1998, 62, 1791-1794. 106. S. Attal, S. Bay, and D. Cantacuzene, Tetrahedron, 1992, 48, 9251-9260. 107. S. Menzler, H. Seker, M. Gschrey, and M. Wiessler, Biotechnol. Lett., 1997, 19, 269-272. 108. L. Hedbys, E. Johansson, K. Mosbach, P.-0. Larsson, A. Gunnarson, S. Swensson, and H. Lohn, GlycoconjugateJ., 1989, 6, 161-168. 109. T. Murata, S. Akimoto, M. Horimoto, and T. Usui, Biosci. Biotechnol. Biochem., 1997, 61, 1118-1120. 110. K. Fukase, T. Yasukochi, Y. Suda, M. Yoshida, and S. Kusumoto, Tetrahedron Lett., 1996, 37. 6763-6766. 111. G. Vie, J. J. Hastings, 0. W. Howarth, and D. H. Crout, Tetrahedron: Asymmetry, 1996, 7, 709-720. 112. L. Hedbys, E. Johansson, K. Mosbach, and P.-0. Larsson, Carbohydr. Res., 1989, 186, 217223. 113. J. H. Yoon and K. Ajisaka, Carbohydr. Res., 1996,292, 153-163. 114. T. Usui, K. Ogawa, H. Nagai, and H. Matsui, Anal. Biochem., 1992,202, 61-67. 115. K. Koizumi, T. Tanimoto, K. Fujita, K. Hara, N. Kuwahara, and S. Kitahata, Carbohydr. Res., 1993,238, 75-91. 80. 81. 82. 83.

References

841

116. S. Takayama, M. Shimazaki, L. Qiao, and C. H. Wong, Bioorg. Med. Chem. Lett., 1996, 6 , 1 123-1 126. 117. S. Singh, M. Scigelova, G. Vic, and D. Crout, J. Chem. SOC.,Perkin Trans. I , 1996,21, 19211926. 118. E. Bonnin and J. F. Thibault, Enzyme Microb. Techno/., 1996, 19, 99-106. 119. H. Fujimoto, M. Isomura, T. Miyazaki, I. Matsuo, R. Walton, T. Sakakibara, and K. Ajisaka, Glycoconjugate J., 1997, 14, 75-80. 120. T. Kimura, S. Takayama, H. M. Huang, and C. H. Wong, Angew. Chem. Znt. Ed. Engl., 1996, 35, 2348-2350. 121. W. S. Hays, D. J. VanderJagt, B. Bose, A. S. Serianni, and R. H. Clew, J. Biol. Chem., 1998, 273, 34941-34948. 122. J. W. Fang, W. H. Xie, J. Li, and P. G. Wang, Tetrahedron Lett., 1998, 39, 919-922. 123. J. Li and P. G. Wang, Tetrahedron Lett., 1997, 38, 7967-7970. 124. A. Vetere and S. Paoletti, Biochem. Biophys. Res. Commun., 1996, 219, 6-13. 125. G. F. Herrmann, U. Kragl, and C. Wandrey, Angew. Chem. Znt. Ed. Engl., 1993, 32, 13421343. 126. K. Sakai, R. Katsumi, H. Ohi, T. Usui, and Y. Ishido, J. Carbohydr. Chem., 1992, 11, 553565. 127. K. Suyama, S. Adachi, T. Toba, T. Sohma, C.-J. Hwang, and T. Itoh, Agric. Biol. Chem., 1986, 50, 2069-2075. 128. L. Hedbys, P.-0. Larsson, and K. Mosbach, Biochem. Biophys. Res. Commun., 1984, 123, 815. 129. S. Singh, M. Scigelova, P. Critchley, and D. H. G. Crout, Carbohydr. Rex, 1998, 305, 363370. 130. D. H. G. Crout, 0. W. Howarth, S. Singh, B. E. P. Swoboda, P. Critchley, and W. T. Gibson, J. Chem. SOC.,Chem. Commun., 1991, 1550-1551. 131. S. Singh, D. H. G. Crout, and J. Packwood, Carbohydr. Rex, 1995,279, 321-325. 132. C. Ravet, D. Thomas, and M. D. Legoy, Biotechnol. Bioeng., 1993, 42, 303-308. 133. K. Ajisaka, H. Nishida, and H. Fujimoto, Biotechnol. Lett., 1987, 9, 243-248. 134. T. Tanaka and S. Oi, Agric. Biol. Chem., 1985,49, 1267-1273. 135. K. Kuriyama, K. Tsuchiya, and T. Murui, Biosci. Biotechnol. Biochem., 1995,59, 1142-1143. 136. T. Murata, A. Tashiro, T. Itoh, and T. Usui, Biochim. Biophys. Acta G., 1997, 1335, 326-334. 137. Y. Matahira, A. Tashiro, T. Sato, H. Kawagishi, and T. Usui, Glycoconjugate J., 1995, 12, 664-671. 138. T. Murata, T. Itoh, and T. Usui, Glycoconjugate J., 1998, 15, 575-582. 139. S. Singh, J. Packwood, andD. H. G. Crout, J. Chem. SOC.,Chem. Commun., 1994,2227-2228. 140. S. Singh, J. Packwood, C. J. Samuel, P. Critchley, and D. H. G. Crout, Carbohydr. Res., 1995, 279, 293-305. 141. S. Singh, R. Gallagher, P. J. Derrick, and D. H. G. Crout, Tetrahedron: Asymmetry, 1995, 6 , 2803-28 10. 142. F. Nanjo, K. Sakai, M. Ishikawa, K. Isobe, and T. Usui, Agric. Biol. Chem., 1989, 53, 21892195. 143. F. Nanjo, M. Ishikawa, R. Katsumi, and K. Sakai, Agric. Biol. Chem., 1990, 54, 899-906. 144. S. Kobayashi, T. Kiyosada, and S. Shoda, Tetrahedron Lett., 1997, 38, 2111-2112. 145. S. Singh, J. Packwood, C. J. Samuel, P. Critchley, and D. H. G. Crout, Carbohydr. Rex, 1995, 279, 293-305. 146. E. Rajnochova, J. Dvorakova, Z. Hunkova, and V. Kren, Biotechnol. Lett., 1997, 19, 869872. 147. S. Kyosaka, S. Murata, Y. Tsuda, and M. Tanaka, Chem. Pharm. Bull., 1986, 34, 51405143. 148. T. Usui, M. Suzuki, T. Sato, H. Kawagishi, K. Adachi, and H. Sano, GlycoconjugateJ . , 1994, 11, 105-110. 149. S. Singh, M. Scigelova, and D. H. G. Crout, J. Chem. SOC.,Chem. Commun., 1996,993-994. 150. H. Kizawa, H. Shinoyama, and T. Yasui, Agric. Biol. Chem., 1991, 55, 671-678. 151. M. Kurakake, S. Osada, and T. Komaki, Biosci. Biotechnol. Biochem., 1997, 61, 2010-2014. 152. A. Vetere, M. Bosco, and S. Paoletti, Carbohydr. Rex, 1998, 311, 79-83.

842

29 Glycosiduse-Cutulysed Oligosucchuvide Synthesis

153. R. M. Bardales and V. P. Bhavanandan, J. Biol. Chem., 1989,264, 19893-19897. 154. T. Usui, T. Murata, Y. Yabuuchi, and K. Ogawa, Carbohydr. Res., 1993,250, 57-66. 155. R. Uchida, A. Nasu, K. Tobe, T. Oguma, and N. Yamaji, Carbohydr. Res., 1996,287, 271274. 156. K. Ogawa, 0. Uejima, T. Nakakuki, T. Usui, and K. Kainuma, Agric. Biol. Chem., 1990, 54, 581-586. 157. N. Payre, S. Cottaz, and H. Driguez, Angew. Chem. Znt. Ed. Engl., 1995, 34, 1239-1241. 158. T. Usui and T. Murata, J. Biochem., 1988, 103, 969-972. 159. K. Ogawa, T. Murata, and T. Usui, Curbohydr. Rex, 1991, 212, 289-294. 160. K. Ogawa, T. Nakakuki, andT. Usui, J. Curbohydr. Chem., 1991,10, 877-886. 161. Y. Yoshimura, S. Kitahata, and S. Okada, Agric. Biol. Chem., 1988, 52, 1655-1659. 162. S. Hizukuri, S. Kawano, J.-I. Abe, K. Koizumi, and T. Tanimoto, Biotechnol. Appl. Biochem., 1989,11, 60-73. 163. J.-I. Abe, N. Mizowaki, S. Hizukuri, K. Koizumi, and T. Utamura, Curbohydr. Rex, 1986, 154, 81-92. 164. T. Kuriki, M. Yanase, H. Takata, Y. Takesada, T. Imanaka, and S. Okada, Appl. Enoiron. Microbiol., 1993, 59, 953-959. 165. N. Shiomi and J. Yamada, Agric. Biol. Chem., 1988,52, 1861-1862. 166. S. I. Shoda, K. Obata, 0. Karthaus, and S. Kobyashi, J. Chem. Soc., Chem. Commun., 1993, 18, 1402-1404. 167. 0. Karthus, S. Shoda, H. Takano, K. Obata, and S. Kobayashi, J. Chem. Soc., Perkin Trans. I , 1994, 1851-1857. 168. H. Fujimoto, H. Nakano, M. Isomura, S. Kitahata, and K. Ajisaka, Biosci. Biotechnol. Biochem., 1997, 61, 1258-1261. 169. H. Nakano, S. Kitahata, H. Kinugasa, Y. Watanabe, H. Fujimoto, K. Ajisaka, and S. Takenishi, Agric. Biol. Chem., 1991, 55, 2075-2082. 170. P. W. Bezukladnikov and L. A. Elyakova, Carbohydr. Res., 1988, 184, 268-270. 171. T. Kawaguchi, K. Sugimoto, H. Hayashi, and M. Arai, Biosci. Biotechnol. Biochem., 1996, 60, 3444346. 172. J. L. Viladot, B. Stone, H. Driguez, and A. Planas, Curbohydr. Res., 1998, 311, 95-99. 173. J. L. Viladot, V. Moreau, A. Planas, and H. Driguez, J. Chem. Soc., Perkin Trans. 1, 1997, 2383-2387. 174. S. Armand, S. Drouillard, M. Schulein, B. Henrissat, and H. Driguez, J. Biol. Chem., 1997, 272, 2709-2713. 175. S. Kobayashi, T. Kawasaki, K. Obata, and S. I. Shoda, Chem. Lett., 1993, 685-686. 176. S. Shoda, T. Kawasaki, K. Obata, and S. Kobayashi, Carbohydr. Res., 1993,249, 127-137. 177. S. Kobayashi, K. Kashiwa, T. Kawasaki, and S. Shoda, J. Am. Chem. Sac., 1991,113, 30793084. 178. J. H. Lee, R. M. Brown, S. Kuga, S. Shoda, and S. Kobayashi, Proc. Nat. Acud. Sci. USA, 1994, 91, 7425-7429. 179. V. Moreau and H. Driguez, J. Chem. Soc., Perkin Trans. I, 1996, 525-527. 180. M. Fujita, S. Shoda, and S. Kobayashi, J. Am. Chem. Soc., 1998,120,6411-6412. 181. H. Ashida, Y. Tsuji, K. Yamamoto, H. Kumagai, and T. Tochikura, Arch. Biochem. Biophys., 1993,305, 559-562. 182. Y. T. Li, B. Z. Carter, B. N. N. Rao, H. Schweingruber, and S. C. Li, J. Biol. Chem., 1991, 266, 10723--10726. 183. K. Yamada, E. Fujita, and S. I. Nishimura, Curbohydr. Rex, 1998, 305, 443-461. 184. K. Akiyama, K. Kawazu, and A. Kobayashi, Curbohydr. Rex, 1995, 279, 151-160. 185. J. Q. Fan, L. H. Huynh, B. B. Reinhold, V. N. Reinhold, K. Takegawa, S. Iwahara, A. Kondo, I. Kato, and Y. C. Lee, Glycoconjugute J., 1996, 13, 643-652. 186. H. Matsui, H. Kawagishi, and T. Usui, Biochim. Biophys. Actu, 1990, 1035, 90-96. 187. Y. Matahira, K. Ohno, M. Kawaguchi, H. Kawagishi, and T. Usui, J. Curbohydr. Chem., 1995,14,213-225. 188. K. Takegawa, K. Fujita, J. Q. Fan, M. Tabuchi, N. Tanaka, A. Kondo, H. Iwamoto, I. Kato, Y. C. Lee, and S. Iwahara, Anal. Biochem., 1998,257, 218-223. 189. T. Rand-Meir, F. W. Dahlquist, and M. A. Raftery, Biochemistry, 1969, 8, 4206-4214.

References

843

190. I. L. Deras, K. Takegawa, A. Kondo, I. Kato, and Y. C. Lee, Bioorg. Med. Chem. Lett., 1998, 8,1763-1766. 191. H. Saitoh, K. Takagaki, M. Majima, T. Nakamura, A. Matsuki, M. Kasai, H. Narita, and M. Endo, J. Biol. Chem., 1995,270,3741-3747. 192. T. Usui, Y. Hayashi, F. Nanjo, K. Sakai, and Y. Ishido, Biochim. Biophys. Acta, 1987,923, 302-309. 193. K. Haneda, T. Inazu, M. Mizuno, R. Iguchi, K. Yamamoto, H. Kumagai, S. Aimoto, H. Suzuki, and T. Noda, Bioory. Med. Chem. Lett., 1998,8, 1303-1306. 194. L.X. Wang, M. Tang, T. Suzuki, K. Kitajima, Y. Inoue, S. Inoue, J. Q. Fan, andY. C. Lee, J. Am. Chem. Soc., 1997,119,11137-11146. 195. T. Usui, Y. Hayashi, F. Nanjo, and Y. Ishido, Biochim. Biophys. Acta, 1988,953,179-184. 196. T. Usui and H. Matsui, Agric. Biol. Chem., 1989,53, 383-388. 197. K. Haneda, T. Inazu, K. Yamamoto, H. Kumagai, Y. Nakahara, and A.Kobata, Carbohydr. Res., 1996,292,61-70. 198. K. J. Yamamoto, S. Kadowaki, J. Watdnabe, and H. Kumagai, Biochem. Biophys. Res Commun., 1994,203,244-252. 199. K. Yamamoto, K. Fujimori, K. Haneda, M. Mizuno, T. Inazu, and H. Kumagai, Carbohydr. Res., 1998,305,415-422. 200. M. Mizuno, K. Haneda, R. Iguchi, I. Muramoto, T. Kawakami, S. Aimoto, K. Yamamoto, andT. Inazu, J. Am. Chem. SOC.,1999,121,284-290. 201. S.Kobayashi, X.Wen, and S.-I. Shoda, Macromolecules, 1996,29,2698-2700. 202. M. Scigelova, V. Kren, and K. G. I. Nilsson, Biotechnol. Lett., 1994,16, 683-688. 203. E. V. Eneyskaya, A. M. Golubev, A. M. Kachurin, A. N. Savel’ev, and K. N. Neustroev, Carbohydr. Rex, 1998,305,83-91. 204. G. Vic, J. J. Hastings, and D. H. G. Crout, Tetrahedron: Asymmetry, 1996, 7, 1973-1984. 205. M. Scigelova, P. Sedmera, V. Havlicek, V. Prikrylova, and V. Kren, J. Carbohydr. Chem., 1998,17,981-986. 206. D. Cantacuzene, S. Attal, and S. Bay, Biomed. Biochim. Acta, 1991,50, S231LS236. 207. E. Johansson, L.Hedbys, and P. 0. Larsson, Enzyme Microb. Technol., 1991,13,781-787. 208. K. G. I. Nilsson, G. Ljunger, and P. M. Melin, Biotechnol. Lett., 1997,19,889-892. 209. V.Laroute and R.-M. Willemot, Biotechnol. Lett., 1992,14, 169-174. 210. M.-P. Bousquet, R.-M. Willemot, P. Monsan, and E. Boures, Enzyme Microb. Technol., 1998, 23, 83-90. 21 1. M.-P. Bousquet, R.-M. Willemot, P. Monsan, and E. Boures, Appl. Microbiol. Biotechnol., 1998,50, 167-173. 212. C.Kieburg, T.K . Lindhorst, andV. Kren, J. Carbohydr. Chem., 1998,17, 1239-1247. 213. T. Mori and Y. Okahata, Tetrahedron Lett., 1997,38,1971-1974. 214. A. Ismail and M. Ghoul, Biotechnol. Lett., 1996,18, 1199-1204. 215. A. J. J. Straathof, J. P. Vrijenhoef, E. P. A. T. Sprangers, H. van Bekkum, and A. P. G. Kieboom, J. Carbohydr. Chem., 1988,7, 223-238. 216. M. Kojima, T.Seto, Y. Kyotani, H. Ogawa, S. Kitazawa, K. Mori, S. Maruo, T. Ohgi, and Y. Ezure, Biosci. Biotechnol. Biochem., 1996,60, 694-696. 217. N. Asdno, A. Kato, H. Kim, K. Matsui, R. C. Griffiths, M. G. Jones, A. A. Watson, and R. J. Nash, Carbohydr. Res., 1997,304, 173-178. 218. T. Wiemann, N. Taubken, U. Zehavi, and J. Thiem, Carbohydr. R e x , 1994,257,CI-C6. 219. D.H.G. Crout, D. A. Macmanus, and P. Critchley, J. Chem. Soc., Perkin Trans. 1, 1990, 1865-1 868. 220. Y.Ooi, N.Mitsuo, and T. Satoh, Chem. Pharm. Bull., 1985,33,5547-5550. 221. H.-J. Gais, A. Zeissler, and P. Maidonis, Tetrahedron Lett., 1988,29, 5743-5744. 222. Y.Ooi, T. Hashimoto, N. Mitsuo, and T. Satoh, Chem. Phurm. Bull., 1985,33,1808-1814. 223. Y.Ooi, T. Hashimoto, N. Mitsuo, and T. Satoh, Tetrahedron Lett., 1984,25,2241-2244. 224. S. Matsumura, H. Yamazaki, and K. Toshima, Biotechnol. Lett., 1997,19,583-586. 225. D. E. Stevenson, R. A. Stanley, and R. H. Furneaux, Biotechnol. Bioeng., 1993,42, 657-666. 226. S. Matsurnura, H.Kubokawa, and S. Yoshikawa, Chem. Lett., 1991,945-948. 227. Y. Fortun and B. Colas, Biotechnol. Lett., 1991,13,863-866. 228. Y.Okahata and T. Mori, J. Chem. Soc., Perkin Trans. 1, 1996,2861-2866.

844 229. 230. 231. 232. 233. 234. 235. 236. 237. 238. 239. 240. 241. 242. 243. 244. 245. 246. 247. 248. 249. 250. 251. 252. 253. 254. 255. 256.

29 Glycosiduse-Cutulysed Oligosucchuride Synthesis T. Mori, S. Fujita, and Y. Okahata, Chem. Lett., 1997, 73-74. T. Mori, S. Fujita, and Y. Okahata, Carbohydr. Rex, 1997, 298, 65-73. Y. Okahata and T. Mori, J. Mol. Cat. B Enzym., 1998, 5, 119-123. T. MoriandY. Okahata, J. Chem. Soc., Chem. Commun., 1998, 2215-2216. R. R. Gibson, R. P. Dickinson, and G. J. Boons, J. Chem. Soc., Perkin Trans. I , 1997, 33573360. S. Leparoux, Y. Fortun, and B. Colas, Biotechnol. Lett., 1994, 16, 677-682. K. G. I. Nilsson and M. Scigelova, Biotechnol. Lett., 1994, 16, 671-676. D. E. Stevenson and R. H. Furneaux, Carbohydr. Rex, 1996,284, 279-283. S. Leparoux, M. Padrines, Y. Fortun, and B. Colas, Biotechnol. Lett., 1996, 18, 135-138. N. J. Turner and M. C. Webberley, J. Chem. Soc., Chem. Commun., 1991, 1349-1350. A. Baker, N. J. Turner, and M. C. Webberley, Tetrahedron: Assymetry, 1994, 5, 2517-2522. G. Vic and D. Thomas, Tetrahedron Lett., 1992, 33, 4567-4570. P. Christakopoulos, M. K. Bhat, D. Kekos, and B. J. Macris, Znt. J. Biol. Macromol., 1994, 16, 331-334. M. Gelo-Pujic, E. Guibe-Jampel, and A. Loupy, Tetrahedron, 1997, 53, 17247-17252. A. Trincone and E. Pagnotta, Biotechnol. Lett., 1995, 17, 45-48. H. Shinoyama, K. Takel, A. Ando, T. Fujii, M. Sasaki, Y. Doi, and T. Yasui, Agric. Biol. Chem., 1991, 55, 1679-1681. A. Millqvist-Fureby, 1. S. Gill, and E. N. Vulfson, Biotechnol. Bioeng., 1998, 60, 190-196. Z. Gunata, M. J. Vallier, J. C. Sapis, R. Baumes, and C. Bayonove, Enzyme Microb. Technol., 1994, 16, 1055-1058. G. Vie, D. Thomas, and D. H. G. Crout, Enzyme Microb. Technol., 1997,20, 597-603. D. K. Watt, H. Ono, and K. Hayashi, Biochim. Biophys., 1998, 1385, 78-88. G. Vie and D. H. G. Crout, Tetrahedron: Asymmetry, 1994, 5, 2513--2516. M. Gelo-Pujic, E. Guibe-Jampel, A. Loupy, and A. Trincone, J. Chem. Soc., Perkin Trans. 1, 1997, 1001-1002. G. Ljunger, P. Adlercreutz, and B. Mattiasson, Enzyme Microb. Technol., 1994, 16, 751-755. N. Taubken and J. Thiem, Synthesis ( C ) , 1992, 517-518. H. Shinoyama and T. Yasui, Agric. Biol. Chem., 1988, 52, 2375--2377. H. Shinoyama, Y. Kamiyama, and T. Yasui, Agric. Biol. Chem., 1988, 52, 2197-2202. P. Drouet, M. Zhang, and M. D. Legoy, Biotechnol. Bioeng., 1994, 43, 1075-1080. J. Sulistyo, Y. Kamiyama, H. Ito, and T. Yasui, Biosci. Biotechnol. Biochem., 1994, 58, 13111313.

Carbohydrates in Chemistry and Biology Editor Beat Ernst, Gerald W. Hart, Pierre Sinay copyright OWILEY-VCH Verlag GmbH, 2000

30 Production of Heterologous Oligosaccharides by Recombinant Bacteria (Recombinant Oligosaccharides) Roberto A . Geremia and Eric Samain

30.1 Introduction During recent years it has been firmly established that oligosaccharides, either free or conjugated, are involved in many biological recognition processes, such as those that occur during fertilization, embryogenesis, inflammation, symbiotic nitrogen fixation, and host pathogen adhesion [l]. These advances have led to increased interest in potential applications of oligosaccharides and derivatives in therapeutics and in the agricultural industry. Both chemical and enzymatic methods have been developed for production of relevant amounts of biologically active oligosaccharides. Chemical methods still need many protection and deprotection steps; as the number of steps increases with the size of the oligosaccharide, longer oligosaccharides are produced in lower yields. Enzymatic methods use purified glycosylhydrolases or glycosyltransferases to assemble the oligosaccharide. Glycosyltransferases (GTs), are enzymes that transfer a sugar from a donor to an acceptor. With Leloir GTs the sugar donor is a sugar-nucleotide (Figure 1). The most specific enzymatic methods use glycosyltransferases, consequently these methods suffer from the relatively poor availability of glycosyltransferases and from the need to regenerate in situ sugar-nucleotides. Taking into account that enzymatic methods basically mimic in vitro the natural pathways of oligosaccharide biosynthesis, a reasonable short-cut is the production of the oligosaccharide in a living recombinant organism (e.g. Escherichia coli). Unfortunately this task is not simple to accomplish, because of the characteristics of the biosynthetic machinery involved in the synthesis of the desired products. Production in E. coli of active eukaryotic GTs in suitable amounts is not always possible or requires considerable effort for reasons that are not well understood. This problem can be overcome by use of homologous enzymes from prokaryotic organisms. A second possible difficulty consists in the availability of sugar-nucleotides. Bacteria recycle these compounds, because they are necessary for the synthesis of their cell wall carbohydrate polymers (murein, lipopolysaccharide, capsular poly-

846

30 Production of Heterologous Oligosaccharides by Recombinant Bacteria

GT

Figure 1. Reaction catalyzed by 'Leloir' glycosyltransferases.

saccharides, etc. ) [2]. Occasionally, e. g. with GDP-fucose, the production of the sugar donor is probably coupled with the corresponding polysaccharide biosynthetic system, which is, in turn, induced by particular physiological or environmental conditions. For these reasons it is difficult to estimate the actual availability of the sugar donor. The solution consists either in adapting the culture conditions or genetically modifying the bacteria for the production of the desired sugarnucleotide. Finally, the most difficult problem to address is that of the acceptor. The physiological acceptor for glycosyltransferases involved in oligosaccharide synthesis is a mono- or oligosaccharide, which can be linked to a protein, a lipid, or another carrier molecule. The acceptor requirements are well studied for eukaryotic GTs, but little is known for the bacterial variety. Production of the oligosaccharide coupled to proteins or lipids requires engineering of the complete glycosylation pathway into the bacteria, which is an enormous task. The use of free sugars in vivo as acceptors is not feasible, because they are rapidly catabolized, thus their intracellular concentration is virtually zero. One possibility is to use lipid-linked intermediates as acceptors but they are present in small amounts [ 3 ] ,limiting the amount of oligosaccharide produced. The availability of an acceptor in sufficient amounts is, therefore, the limiting step for the production of oligosaccharides in bacteria. If the desired oligosaccharide is produced, it might be necessary to add nonglycosidic substituents, requiring the expression of the appropriate modifying enzyme and the presence of donors for the modification. Despite these drawbacks, this system remains attractive because once the genes are cloned and conditions established, the production is performed in vivo by the bacteria, without all the problems found in chemical or enzymatic synthesis. E. coli is commonly used for the production of heterologous DNA or proteins, and is a very well known organism. Its metabolism can be manipulated by varying the culture conditions. One of the most exciting features for our purposes is that it can be grown at high or very high cell density (OD optical density 100-200, in contrast with 2-5 under normal conditions). This means the amount of biomass obtained in a single culture can be more than 50 g L-' (dry weight). An exceptional opportunity for synthesizing oligosaccharides in E. coli, is provided by the rhizobial chitin oligosaccharide synthase NodC. The putative acceptor or primer for this enzyme has not yet been identified, but is present in E. coli, which enabled the production of chitin oligosaccharides in vivo. By use of the methodology developed by us and described below, E. coli strains expressing nodC produce 1-2 g L-' chitin pentasaccharide [4].Modification with glycosidic or non-glycosidic

30.2 Concept and Methodology of Heteroloyous Oligosuccharide Production

841

groups was also achieved by co-expression of the proper genes, with production yields between 0.2 and 1 g L-’ [4-61.

30.2 Concept and Methodology of Heterologous (‘Recombinant’) Oligosaccharide Production in E. cob Sugar-nucleotides and glycosyltransferases are already present in E. coli, because it produces a series of cell-wall polysaccharides. The enzymes for the biosynthesis of sugar-nucleotides are cytoplasmic, whereas most GTs involved in the biosynthesis of outer polysaccharides are peripheral membrane proteins whose catalytic domain is thought to face the cytoplasm [7, 81. Therefore, to use the pool of sugar-nucleotides, the recombinant GT should be targeted at the cytoplasm. An advantage of expressing recombinant bacterial glycosyltransferase in E. coli is that cytoplasm is their natural environment, and that having transmembrane domains, insertion into membrane in the E. coli host possibly occurs by similar mechanisms. Unfortunately, most eukaryotic GT are membrane-bound and present in the luminal side of endoplasmic reticulum or the Golgi apparatus, that is equivalent to the outside of cells. Because these organelles are not present in bacteria, it is necessary to remove the transmembrane domain (involved in targeting) to keep the protein in bacterial cytoplasm. It should be pointed out that the heterologous oligosaccharides are not secreted, and that they remain in the cytoplasm. The production of heterologous oligosaccharides uses recombinant DNA technology to produce recombinant GTs, and the E. coli cell machinery to produce the desired oligosaccharides. We have chosen to call the heterologous oligosaccharides obtained by this approach ‘recombinant oligosaccharides’.

30.2.1 Biosynthesis of Nod Factors One of the major sources of biological N2 fixation is the symbiosis relationship Rhizohium-leguminous. As part of this symbiosis a dedicated N2 fixation organ, called nodule, is formed. The nodule is a plant structure that is invaded by rhizobia, and its formation results from a ‘molecular dialog’ between the plant and the bacteria [9]. Briefly, the plant secretes specific flavonoids that induce the expression of nod genes, resulting in the synthesis of Nod factors. Upon secretion by specific proteins, the Nod factor will induce the deformation of hairy roots, and mitosis of plant cells. Eventually, the bacteria invade the root cell, migrate towards the cell division region, and are released into these cells. Finally, bacteria differentiate into a Nz-fixing form, the bacteroid. Nod factors consist of a chitin (tri-, tetra- or penta-) oligosaccharide, of which the non-reducing end N-acetyl group is substituted by a long-chain fatty acid. Several substitutions are found at the reducing- and non-reducing-end sugars, although substitutions on other residues are present in some species [lo, 111. The Nod factors

848

30 Production of Heteroloyous Oligosaccharides b-y Recombinant Bacteria

HO HO

0

*.

NodFE Figure 2. Structure of NodRmIV (Ac, S). Proteins involved in addition of each substituent are circled.

produced by R. meliloti are shown in Figure 2. The first committed step of Nod factor biosynthesis is the formation of the chitin oligomer by NodC using UDP-Nsatylglucosamine (UDP-GlcNAc) as the sugar donor [ 111. NodB specifically deacetylates the non-reducing terminus of the oligosaccharide. The NH2 group is finally acylated in a step that requires NodA. Other substituents are added before or after the acylation by specific enzymes (Figure 2). The final product, is secreted by a system that requires Nod1 and NodJ [ 1 I]. The nod genes were discovered as result of the identification of rhizobial nonnodulating mutants. These genes were isolated and sequenced [12, 131 but no function was assigned to them. Discovery of Nod factors [ 141 brought substance to the study of Nod protein function. The first hint of the function of NodC was the sequence similarity with cellulose and chitin synthases [ 151. The in vitro synthesis using rhizobial cell extracts or E. coli extracts containing recombinant NodC was then accomplished [ 161 using UDP-[ ''C]GlcNAc as sugar donor. The oligosaccharide formed was shown to bind to wheat germ agglutinin, and to be sensitive to chitinase, but the minute amounts of the chitin oligosaccharide available precluded unambiguous structural characterization. It was, nevertheless, discovered that the E. coli strain expressing NodC was able to produce, in vivo, small amounts of chitin oligosaccharides; these were characterized by mass spectrometry [ 171. The product consisted of a mixture of di-, tri-, tetra- and pentasaccharides, and purification involved several steps. The use of

30.2 Concept and Methodology of Heteroloyous Oligosucchuride Production

849

E. coli strains expressing several combinations of nod genes enabled the elucidation of the biosynthetic pathway and the function of these genes [17], by establishing the structure of the corresponding Nod metabolites. This was the first example of the production of a recombinant oligosaccharide and the starting point for the production of oligosaccharide in E. coli. Our further reasoning was that if a single batch culture (biomass representing OD = 2) can produce a few mg L-' Nod metabolites, in a culture of high density (OD = 100) it would produce a few hundred mg L-l. It was found by sequence analysis and tagging with reporter enzymes [ 181, that NodC is a membrane-bound protein with four transmembrane domains (one at the NH2 terminal end, and three at the COOH terminus) flanking a 300 amino acid cytoplasmic peptide. The cytoplasmic moiety of NodC shares sequence similarity with several glycosyltransferases and polysaccharide synthases [ 19-21] in two different domains (domain A towards the -NH2 end and domain B towards the -COOH end). Whereas both domains are conserved in polysaccharide synthases, only domain A is present in glycosyltransferases [21].This led to the hypothesis that domain B is involved in multiple sugar transfer. The suggested mechanism proposes the presence of one catalytic center in each domain. It was also proposed that the acceptor is formed by the enzyme [21].To date several authors have reported the in vitro formation of chitin oligosaccharides in the absence of the putative GlcNAc acceptor [22, 231.

30.2.2 Expression Systems and Cloning Strategy UDP-GlcNAc is a key metabolite of bacteria, because it is a precursor for the formation of murein and Lipid A, both necessaries for cell viability. UDP-GlcNAc is synthesized from fructose-6-phosphate in four steps. The two last reactions leading to UDP-GlcNAc from glucosamine-6-phosphate are catalyzed by a bifunctional enzyme encoded by glmU [24].The acetyltransferase activity of this enzyme is strongly inhibited by GlcNAc-1-P and by UDP-N-acetylmuramic acid, suggesting that this reaction is the main point of UDP-GlcNAc synthesis regulation. Because UDP-GlcNAc is substrate for NodC, the recombinant enzyme should be expressed at low levels, so as not to disturb cell wall synthesis. To regulate gene expression, the appropriate genes are cloned under the control of a given promoter. ~ ~ has the advantage that it Two promoters were used, P L and ~ ~PA^^. P L promoter allows a low-level expression in the absence of inducer, while PA^^ is a tight promoter [25], and can be used to tune the expression of the gene by use of an inducer (arabinose). For the production of a single oligosaccharide with glycosidic or nonglycosidic modifications, the expression of several genes is required. This can be achieved by cloning more than one gene into a given vector, or in two compatible vectors. With eukaryotic GT genes it is necessary to provide a bacterial ribosomal binding site, and to remove regions coding for transmembrane domains; one solution is to produce a fusion protein. Some of the plasmids used are shown in Figure 3; they are commonly available [4-61.

19 PLac@ PAra Neighbour non-translated DNA

Figure 3. Plasmids containing chitin oligosaccharide synthases and different modifying enzymes. Relevant genes are in black-filled arrows, promoters in hatched arrows. Because of the cloning strategy, DNA contiguous to the desired genes are present in the constructions (gray).

Vector DNA

-

9 a w.

00

30.2 Concept and Methodology of Heterologous Oligosaccharide Production

85 1

30.2.3 High Cell-Density Cultivation The cultivation of E. coli to high cell concentration is one way of maximizing the volumetric productivity of recombinant oligosaccharides. The main problem encountered in high cell-density culture (HCDC) is the risk of 0 2 limitation, which obliges the bacteria to shift to a fermentative metabolism, resulting in the production of growth-inhibiting metabolites such as acetic acid. Because the 0 2 delivery capacity of the fermenter is limited, the easiest way to prevent 0 2 shortage is to limit the bacterial growth by controlled feeding of the carbon source. Different feeding strategies have been proposed [26, 271. Although some strategies require sophisticated feedback control of substrate concentration to determine the feeding rate, reasonably high cell density can be also reached with simple fed-batch technique using pre-determined feeding rates to sustain carbon-limited growth [28]. This simple technique requires minimal fermentation equipment: a laboratory-scale fermenter with temperature, pH and 0 2 controls, and a peristaltic pump for the feeding. The protocol we routinely use to produce recombinant chitin oligosaccharides in high cell density cultures of E. coli includes the use of glycerol as carbon source because its catabolism does not provoke substantial accumulation of acetate, as does the use of glucose. The fed-batch strategy is outlined in Figure 4. During phase 1, cells grow exponentially until all the glycerol that was initially added in the starting media (17 g L-') had been consumed [4]. To prevent oxygen limitation,

phase 1

i phase 2

phase 3

100

80

-

60-

40

-

20 -

o+ 0

glycerol

I

5

10

15

20

25

30

35

Time (hours) Figure 4. High cell-density culture strategy. For details see Ref. [3].

40

45

50

852

30 Production of Heterologous Oligosaccharidesby Recombinant Bacteria

growth is then carbon limited, and it is supported by the continuous feeding of a glycerol solution. Glycerol is first supplied at a high feeding rate in phase 2 to promote a rapid increase in biomass. Finally growth is conducted at a low rate (phase 3 ) to increase in the intracellular concentration of recombinant oligosaccharide.

30.2.4 Purification of Recombinant Oligosaccharides The recombinant chitin oligosaccharides are produced and accumulated in the cytoplasm and are not excreted in the culture medium. The first purification step consists of harvesting the cells by centrifugation to remove all culture supernatant. The cells are then suspended in distilled water and permeabilized either by a heat shock (30 min at 100 "C) or by a series of freeze-thaw cycles. The latter method is employed for compounds which are susceptible to damage at high temperature. These treatments enable the oligosaccharides to diffuse outside the permeabilized cell. After a second centrifugation the oligosaccharides are recovered in the supernatant and are then further purified by adsorption on charcoal-celite and selective elution with aqueous ethanol. Positively charged N-deacetylated chitin oligosaccharides and negatively charged sulfated compounds can be advantageously separated by ion-exchange chromatography. These purification procedures enable gram-scale preparation of oligosaccharides, which are found to be more than 90% pure by HPLC and NMR.

30.3 Examples of Recombinant Oligosaccharides 30.3.1 Production of Chitin Oligosaccharides in E. cofi Expressing NodC

The production of chitin oligosaccharides is achieved in E. coli DH1 cells containing ~ C Aor , PCR~ (Figure 3). R. meliloti produces Nod factors in which the oligosaccharide consists of a mixture of tetra- and pentasaccharide [29], whereas A. caulinodans NodC produces pentasaccharide as the major product [30]. The final degree of polymerization of the oligosaccharides is determined by NodC [5, 231 but the concentration of UDP-GlcNAc influences the size of the oligosaccharide [22, 231, perhaps by acting as a chain terminator. Because the concentration of the sugar donor can change during cell growth, and under different metabolic conditions, the size of the oligosaccharides was assessed at the end of phase 1 and at the end of phase 3 . On both occasions the major products were the pentasaccharide for ~ C A , , and a mixture of tetra- (70%) and pentasaccharide (30%) for p c ~ , [ 5 ] . Whereas these oligosaccharides were almost pure at the end of phase 3 , other products were found at the end of phase 1. These products comprise mainly derivatives of the chitin oligosaccharides whose reducing end was found to be substituted with a glycerol moiety, linked in the configuration to either the primary or secondary hydroxyl groups of glycerol [31].

30.3 Examples of Recombinant Oligosaccharides

853

Glycerol-substituted oligosaccharides are no longer detectable at the end of phase 3. During phase 1 the intracellular concentration of glycerol is high, because its uptake is mediated by the glycerol facilitator GlpF [32]. In contrast, during phases 2 and 3 the added glycerol is immediately used to sustain growth, therefore the intracellular concentration is low. These results strongly suggest that high intracellular glycerol concentration promotes its incorporation at the reducing end. The incubation in vitro of E. coli membranes containing NodC in the presence of glycerol and UDP[ 14C]GlcNAc, leads to the formation of chitinase-sensitive radioactive oligosaccharides, that migrate more slowly than the corresponding chitin oligosaccharides 1311.These results suggest that NodC adds the glycerol moiety to the oligosaccharide. 30.3.2 Production of Nod Factor Precursors

As mentioned above, much work has been devoted to understanding the function of the proteins involved in Nod factor biosynthesis, but their mode of action is not fully understood. Future trends are related to Nod factor recognition and signalling [33-361. Because chemical synthesis is long and tedious, we have proposed a new strategy involving the production of Nod factor precursors in E. coli expressing appropriate biosynthetic genes [ 371. These oligosaccharides are provided to chemists for further modifications. In the first instance, when E. coli DH1 strains harboring the plasmid ~ B C Awere , employed, chitin pentasaccharides N-deacetylated at the non-reducing terminus were obtained in yields close to 1 g L-’ of culture [4]. This oligosaccharide was acylated by chemical means [37]. and used for characterization of Nod factor protein binding sites 1381. Further chemical modification will enable the production of Nod factor derivatives useful for affinity chromatography, fluorescence and/or photoactivatable probes. These molecules will be useful tools for isolation and biochemical characterization of the Nod factor binding sites. The substantial amounts obtained will also facilitate the study of the signal transduction pathway as well as the regulation of plant modulation genes expression. The biological activity of R. meliloti Nod factors requires a tetrasaccharide and the presence of sulfate and 0-acetate. We have used several combinations of the nod genes (NodB, NodC, NodH, NodL); these have enabled us to produce as many as eight different molecules (see Figures 5 and 6) in the recombinant oligosaccharide system. In R. meliloti, three genes (nodPQH) are necessary for the sulfation of Nod factors 1391. The protein NodH catalyzes the transfer of sulfate from PAPS (3’phosphoadenosine 5’-phosphosulfate) to the reducing terminal 6-0 position of Nod factors [40]. The enzyme is also active on chitin oligosaccharides and their Ndeacetylated derivatives. Kinetic analysis indicates that the affinity of NodH for chitin oligosaccharides is an order of magnitude lower than for lipo-chitin oligosaccharides, suggesting that sulfation occurs in vivo after the acylation [41]. The K, value of ca 150-200 p~ found in vitro for chitin oligosaccharides should, however, be sufficient to enable the sulfation of recombinant chitin oligosaccharides which have been shown to accumulate in the cytoplasm of E. coli at a concentration higher than 10 mM.

854

30 Production of Heterologous Oligosaccharides by Recombinant Bacteria

Figure 5. Metabolic pathway leading to the synthesis of Nod factor precursors in E. coli. (NodL and NodH can both act on unsubstituted oligosaccharides.)

30.3 Examples of Recombinant Oligosaccharides

855

The proteins NodP (ATP sulfurylase) and NodQ (adenosine 5'-phosphosulfate (APS) kinase) are associated in a sulfate-activating complex that enables the synthesis of the sulfate donor (PAPS) for Nod factor sulfation [42]. PAPS is also the sulfate donor used for the synthesis of cysteine and it is produced in E. coli by the normal household proteins CysC, CysD, and CysN [43, 441. The genes cysCDN are part of the cysteine regulon which is repressed in the presence of cysteine and other reduced sulfur compounds [45]. This means that the E. coli machinery of PAPS biosynthesis can be used for chitin oligosaccharide sulfation if the bacteria are cultivated with sulfate as the only sulfur source. The total PAPS demand for the synthesis of cysteine and methionine (87 and 146 pmol g-' dried cells, respectively) [46] largely exceeds the maximum demand that can be estimated for chitin oligosaccharide sulfation (around 20 pmol g-' dried cells). High cell density cultivation of strain DH1 (~BCHR,)leads to the production of 330 mg L-' sulfated chitin oligosaccharides. This represented a sulfation yield of ca 50%. Attempts were made to improve this yield by expressing nodH in trans under the control of the PBADpromoter in a separate plasmid compatible with ~BCR,. The sulfation yields were similar to that obtained with nodH expressed in cis. Moreover, this approach has another advantage, because it limits the number of plasmids that have to be constructed to obtain strains expressing different possible combinations of genes. Strain DHI was co-transformed with plasmid pBADH and different plasmids carrying the gene nodC or nodBC. Cultivation of these strains enabled the production of sulfated N-peracetylated and non-reducing end Nmonodeacetylated chitin oligosaccharides. The choice between the gene nodC from A . caulinodans or from R. meliloti provides the additional possibility of producing these molecules as pentamers or as tetramers. The protein NodL is a transacetylase that utilizes acetyl-CoA as a substrate to 0acetylate lipo-chitin oligosaccharides and both N-peracetylated and non-reducing end N-monodeacetylated chitin oligosaccharides at the 6 position of the nonreducing terminal sugar [46-491. Because acetyl-CoA is a metabolite normally present in the bacterial cytoplasm, the coexpression of nodL with nodC can result in the production of mono 6-0-acetylated chitin oligosaccharide. The gene nodL was sub-cloned in pBBR plasmid which is compatible with both pUC and pBAD derivatives. Analysis of chitin oligosaccharides produced by strain DH 1 carrying nodC and nodL indicated that ca 700/0of the total oligosaccharide was 0-acetylated. Cultivation of strains expressing the four genes nodBCHL enables the production of the direct precursor for the preparation of NodRm-IV (Ac, S, CI62), i.e. the natural Nod factor produced by R. meliloti. 30.3.3 Production of Derivatives of N-Acetyllactosamine

The production of chitin oligosaccharides in E. coli opened the possibility of adding glycosyl substitutions to it. One interesting possibility is the addition of a p(1,4)linked galactosyl residue at the non-reducing end (Figure 6), to form a terminal LacNAc, the core of several biologically important oligosaccharides. This step is catalyzed by N-acetyllactosamine synthases (EC 2.4.1.90). For this purpose, we

856

30 Production

of

Heterologous Oligosaccharides by Recombinant Bacteria

choose LgtB, a bacterial p- 1,4 galactosyltransferase (Swiss-Prot Acc. Number Q51116) [50].LgtB is involved in the synthesis of the lacto-N-neotetraose moiety of the lipo-oligosaccharide produced by Neisseria meningitidis [ 501. Because E. coli produces a P-galactosidase, namely LacZ, we have used a lacZ- strain to prevent degradation of the terminal LacNAc residue. Co-expression of NodC and LgtB leads to the production in vivo of the expected compound [6]. The total recombinant oligosaccharide was produced at 0.8 g L-' of culture. On the basis of NMR data from the crude recombinant oligosaccharide fraction, the extent of galactosylation of the chitin oligosaccharide is close to 90%. Expression of the two genes in cis ( ~ C ALgtB) , slightly increased the galactosylation rate and total amount of recombinant oligosaccharides. Analysis of the recombinant oligosaccharides produced after phase 3 showed that the major product was the galactosyl chitin pentaose (Figure 6); other minor galactosylated and nongalactosylated oligosaccharides were also detected. Spectroscopic characterization of the galactosylated pentasaccharide confirmed the proposed structure [61. We have tested whether the Gal-(GlcNAc)5 is an acceptor for the bovine a-1,3 galactosyltransferase in vitro using UDP[ 14C] Gal. The incorporation rate was similar to that of the physiological acceptor (LacNAc) [ 5 ] .This result is promising in E. coli. for the production of aGall,3-~Gal-l,4(GlcNAc)~

30.4 Conclusions and Future Perspectives In the last three years, we have produced more than 50 g oligosaccharide using two 2 L and one 10 L fermenters. Typically, purification of gram amounts of an end Ndeacetylated chitin oligosaccharide takes one week. Nine different compounds have been produced (Figure 6), with production yields between 0.25 g L-' and 2 g L-' depending on the product (Table 1). The work presented here represents the first few steps towards the production of tailored oligosaccharides in E. coli. Obviously, there are several critical points that can be improved, as well as the production of other oligosaccharides and other applications that can be envisaged. 30.4.1 Production of Labeled Chitin Oligosaccharides to Study Their Interactions with Proteins

The recognition of different sugars by specific proteins (glycosyl hydrolases, GTs, lectins) is a key factor of biological processes. Molecular probes are needed to enable better understanding of the molecular basis of sugar recognition. Chitin oligosaccharides can be used to study the interaction with chitinase, and recognition by lectins or Nod factor binding proteins. To this end we have produced 13C-enriched chitin pentaose (20% enrichment) for NMR study of the mechanism of action of

x , XI

OH

NodC,, or Rm

v , VI

OH

HO

XI1

I, I1

L V I I , VIII

Figure 6. Recombinant oligosaccharides obtained.

HO

NodH

OH

HO

2-3

HO HO

2-3

HO

2-3

OH

XIV

OH

HO

L&tB

2-3

2-3

NodH

2-3

H

ul 4

00

cp

3F

9R

Q

P

L2

858

30 Production of Heterologous Oligosaccharides by Recombinant Bacteria

Table 1. The recombinant oligosaccharides obtained. Product

Plasmid used

Estimated production yield (g L-')

I I1 I11 IV V VI VII VIII IX X XI XI1 XI11 XIV

PcRm PCAc PBCRm PBCA, PCR, pBADH pCac pBADH PCR, PBBRLl PCA, + pBBRLl ~ C R , PBBRLl + pBADH PBCHR, pBCac pBADH PBCR, PBBRLl PBCHR, + PBBRLl PCA,LgtB

1.o 2.0 0.5 1.2 0.4 0.6 0.8 1.4 0.3 0.25 0.25 0.4 0.2 1.o

+ +

+ +

+ +

chitinases [ 5 11. The enrichment might be improved by using specific bacterial strains or specific growth conditions. 30.4.2 Improvement of Oligosaccharide Production, and Metabolic Engineering In the first instance, the volumetric yield of recombinant oligosaccharides can probably be increased by genetically engineering the bacteria, either adding extra copies of the genes coding for the synthesis of sugar donors (gmlU: UDP-GlcNAc, galE: UDP-Gal epimerase), or using stronger promoters. In the second instance, sugar-nucleotides are expressed either under certain conditions (GDP-fucose) or in certain strains (CMP-Neurominic acid). The constitutive production of these sugar-nucleotides could also be engineered to enable synthesis of more diverse oligosaccharides. 30.4.3 Production of More Complex Oligosaccharides

An important limitation to this method is the availability of appropriate GTs. Although the production in E. coli of eukaryotic G T active in vitro was reported, the activity in vivo was not assessed [52].Production of fully active GTs in E. coli will broaden the range of possible recombinant oligosaccharides. Research on the mode of action of GTs is also necessary for eventual engineering of available bacterial GTs to meet specificity requirements. Alternatively, new bacterial glycosyltransferases whose activity resembles those of eukaryotic GTs should be found. In this regard the results are encouraging, because the bacterial genes can be used for the production of oligosaccharides in E. coli as we did with LgtB.

References

859

Here we have presented an original strategy to produce derivatives of chitin oligosaccharides by a simple living organism. The products were obtained in amounts and purity that enable their use as substrates to perform chemical modifications or study biological properties. It is conceivable that the recombinant oligosaccharides can also be used as substrates for enzymatic methods. We think that the limits of this method have not yet been achieved, and it is possible that in the next few years the production of more complex oligosaccharides will become routine.

Acknowledgments The authors acknowledge Drs Serge Perez and Annemarie Lellouch (Cermav, Grenoble) for helpful discussions, and Dr Hughes Driguez for reading the manuscript. This work was supported by Xenotransplantation Project BI04CT972242 of the Biotech program from the European Union, and the program PCV 97 061 from the CNRS (France).

References 1. Varki, Glycobiology, 1993, 3, 97. 2. 0. Gabriel, H. E. Umbarger, Esherichia coli and Salmonella cellular and molecular biology. ASM Press, 1987, p. 504. 3. P. D. Rick, G. L. Hubbard, M. Kitaoka, H. Nagaki, T. Kinoshita, S. Dowd, V. Sirnplaceanu, C. Ho, Glycobiology, 1998, 8, 557. 4. E. Samain, S. Drouillard, A. Heyraud A, H. Driguez, R. A. Geremia, Carbohydv. Rex, 1997 302, 35. 5. E. Samain, V. Chazalet, R . A. Geremia, J. Biotechnol., submitted. 6. E. Bettler, E. Samain, V. Chazalet, C. Bosso, A. Heyraud, D. H. Joziasse, W. W. Wakarchuk, A. Imberty, R. A. Gerernia. Submitted. 7. Whitfield, M. A. Valvano, Advances in Microbial Physiology, 1993, 35, 135. 8. I. Roberts, Annu. Rev. Microbiol., 1996, 50, 285. 9. P. van Rhijn, J. Vanderleyden, Microbiol. Rev., 1995, 59, 124. 10. J. Denarie, F. Debelle, J. C. Prome, Annu Rev Biochem, 1996, 65, 503. 11. P. Mergaert, M. Van Montagu, M. Holsters, Mol. Microbiol., 1997,25, 81 1. 12. T. W. Jacobs, T. T. Egelhoff, S. R. Long, J. Bacteriol., 1985, 162, 469. 13. I. Torok, E. Kondorosi, T. Stepkowski, J. Posfai, A. Kondorosi, Nucleic Acids Rex, 1984, 12, 9509. 14. P. Lerouge, P. Roche, C. Faucher, F. Maillet, G. Truchet, J. C. Prome, J. Denarie, Nature, 1990, 344, 781. 15. E. Bulawa, Mol. Cell. Biol., 1992, 12, 1764. 16. R. A. Geremia, P. Mergaert, D. Geelen, M. Van Montagu, M. Holsters, Proc. Natl. Acad. Sciences USA, 1994, 91,2669. 17. P. Mergaert, W. D’Haeze, D. Geelen, D. Prome, M. van Montagu, R. A. Geremia, J. C. Prome, M. Holsters, J. Biol. Chem., 1995,270, 29217. 18. M. A. Barny, E. Schoonejans, A. Economou, A. W. Johnston, J. A. Downie, Mol. Microbiol., 1996, 19, 443. 19. M. Atkinson, S. R. Long, Mol. Plant-Microbe Interact., 1992, 5, 439. 20. Debelle, C. Rosenberg, J. Denarie, Mol. Plant-Microbe Interact., 1992, 5, 443. 21. I. M. Saxena, R. M. Brown Jr., M. Fevre, R. A. Gerernia, B. Henrissat, J. Bacteriol., 1995, 177, 1419.

860

30 Production of Heterologous Oligosaccharides by Recombinant Bacteria

22. N. I. de Iannino, S. G. F’uepke, R. A. Ugalde, Mol. Plant-Microbe Interact., 1995, 8, 292. 23. E. Kamst, J. Pilling, L. M. Raamsdonk, B. J. Lugtenberg, H. P. Spaink, J. Bucteriol., 1997, 197, 2103. 24. D. Mengin-Lecreulx, J. J. van Heijenoort, J. Bacteriol., 1994, 176, 5788. 25. L. M. Guzman, D. Belin, M. J. Carson, J. Beckwith, JBacteriol, 1995, 177, 4121. 26. D. Riesenberg, Curr. Opin. Biotechnol., 1991, 2, 380. 27. S. Y. Lee, Trends Biotechnol., 1996, 14, 98. 28. D. J. Korz, U. Rinas, K. Hellmuth, E. A. Sanders, W. D. Deckwer. J. Biotechnol., 1995,39, 59. 29. M. Schultze, B. Quiclet-Sire, E. Kondorosi, H. Virelizer, J. N. Glushka, G. Endre, S. D. Gero, A. Kondorosi, Proc. Natl. Acad. Sci. USA, 1992, 89, 192. 30. P. Mergaert, M. van Montagu, J. C. Prome, M. Holsters, Proc. Natl. Acad. Sci. USA, 1993, 90, 1551. 3 1 . R. A. Geremia, personal communication. 32. C. Maurel, J. Reizer, J. I. Schroeder, M. J. Chrispeels, M. H. Saier Jr., J. Biol. Chem., 1994, 269, 11869. 33. J. J. Bono, J. Riond, K. C. Nicolaou, N. J. Bockovich, V. A. Estevez, J. V. Cullimore, R. Ranjeva, Plant J., 1995, 7, 253. 34. J. C. Prome, Curr Opin Struct Biol., 1996, 6, 671. 35. S. G. Puepke, Crit. Rev. Biotechnol., 1996, 16, 1. 36. K. van de Sande, T. Bisseling, Essays Biochem., 1997, 32, 127. 37. S. Drouillard, J. J. Bono, B. Henrissat, R. A. Geremia, F. Gressent, E. Samain, H. Driguez, 1997, 9th European Carbohydrate Symposium, 6-1 1 July, Utrecht, Nederlands. 38. J. J. Bono, F. Gressens, personal communication. See complete reference attached. 39. P. Roche, F. Maillet, C. Plazanet, F. Debelle, M. Ferro, G. Truche, J. C. Prome, J. Denarie, Proc. Natl. Acad. Sci. USA, 1996, 93, 15305. 40. D. W. Ehrhardt, E. M. Atkinson, K. F. Faull, D. I. Freedberg, D. P. Sutherlin, R. Armstrong, S. R. Long, J. Bacteriol., 1995, 177, 6237. 41. M. Schultze M, C. Staehelin, H. Rohrig, M. Joh, J. Schmidt, E. Kondorosi, J. Schell, A. Kondorosi, Proc. Natl. Acad. Sci, USA, 1995, 92, 2706. 42. J. S. Schwedock, C. Liu, T. S. Leyh, S. R. Long, J. Bucteriol., 1994, 176, 7055. 43. T. S. Leyh, J. C. Taylor, G. D. Markham, J. Biol. Chem., 1988,263, 2409. 44. T. S. Leyh, T. F. Vogt, Y. Suo, J. Biol. Chem., 1992, 267, 10405. 45. N. M. Kredich, Esherichia coli and Salmonella cellular and molecular biology. ASM press. Baltimore, 1996,pp. 514-527. 46. F. C. Neidhardt, H. E. Umbarger, Esherichia coli and Salmonella cellular and moleculur biology. ASM Press, 1996, p. 15. 47. P. Spaink, D. Sheeley, A. A. van Brussel, J. Glushka, W. S. York, T. Tak, 0 . Geiger, E. P. Kennedy, V. N. Reinhold, B. J. Lugtenberg, Nature, 1991, 354, 125. 48. V. Bloemberg, J. E. Thomas-Oates, B. J. Lugtenberg, H. P. Spaink, Mol. Microbiol., 1994, 11, 793. 49. G. V. Bloemberg, R. M. Lagas, S. van Leeuwen, G. A. Van der Marel, J. H. Van Boom, B. J. Lugtenberg, H. P. Spaink, Biochemistry, 1995, 34, 12712. 50. W. Wakarchuk, A. Martin, M. P. Jennings, E. R. Moxon, J. C. Richards, J. Biol.Chem., 1996, 271, 19166. 51. E. Samain, personal communication. 52. Fang, J. Li, X. Chen, Y. Zhang, J. Wang, Z. Guo, W. Zhang, L. Yu, K. Brew, P. G. Wang, J. Am. Chem. Soc., 1998, 120, 6635.

Part I Volume 2

IV Carbohydrate-Pr otein Interactions

Carbohydrates in Chemistry and Biology Editor Beat Ernst, Gerald W. Hart, Pierre Sinay copyright OWILEY-VCH Verlag GmbH, 2000

31 Protein-Carbohydrate Interaction: Fundamental Considerations Nikki F. Burkhalter, Sarah M. Dimick, and Eric J. Toone

31.1 Introduction A major-if not the major-rationale for the study of carbohydrate chemistry derives from the roles played by glycoconjugates in biology [l]. In almost all such roles a carbohydrate ligand must bind to a protein receptor. There exists, then, tremendous potential to modulate biological activity through the creation of high affinity mimics of native saccharide receptors; such compounds have potential therapeutic value in the treatment of viral, parasitic, mycoplasmal and bacterial infections, and the treatment of a range of human cancers [ 2 ] .The study of proteincarbohydrate interaction, then, forms a key focus of this discipline of carbohydrate chemistry and biology. The development of inhibitors of carbohydrate-mediated biological recognition is impeded by the exceptionally weak bindings that typify protein-carbohydrate interaction [ 3 ] . In this chapter we consider the energetic issues central to proteincarbohydrate interaction. We devote some space to a consideration of the general forces and interactions that provide both affinity and specificity during association in aqueous solution. A range of phenomena observed during protein-carbohydrate association appear to be dependent on the assay used to evaluate ‘binding’; we therefore consider the some of the more commonly utilized assays of protein carbohydrate interaction and delineate the microscopic events each assay is designed to evaluate. Finally, the last ten years has seen tremendous activity in the development of multivalent carbohydrate ligands; we thus explore the energetic issues surrounding multivalency in ligand binding. Throughout, we endeavor to relate energetic issues directly to protein-carbohydrate interaction. We stress, however, that a tremendous body of literature pertaining to the very issues central to protein-carbohydrate interaction exists in the context of other binding systems. Where appropriate, we draw on the lessons and experience of these adjacent disciplines in our consideration of the biophysical aspects of protein-carbohydrate interaction.

864

31 Protein-Carbohydrate Interaction: Fundumentul Considerations

31.2 Association in Aqueous Solution Two species combine to form a complex in water if the sum of the intermolecular forces between them more than offsets the sum of the loss of favorable interactions with solvent and any unfavorable interactions that develop between solutes during complex formation. Collectively the interactions between non-bonded species are referred to as cohesive forces, defined as those forces lost when the species are transferred to infinite separation in the gas phase. While it is common to classify chemical forces as covalent or non-covalent, the interactions are fundamentally the same; only the magnitude of the interactions varies. Cohesive, non-specific forces are weak compared to covalent interactions; typically we consider cohesive forces as those forces with strengths less than 10/0 of covalent bond strengths. We will see, however, that this definition is somewhat arbitrary and in fact a continuum of interaction energies exists. That non-covalent interactions between molecules are important has long been appreciated. By the middle of the 19th century failings in the ideal gas law were apparent, and by 1873 van der Waals had postulated his equation of state. Thus the functional dependence of intermolecular interactions on internuclear disvdnceattractive at long separations and repulsive at very short internuclear spacingswas clear long before there existed a complete understanding of the nature of the interactions involved. Association in aqueous solution is further complicated by the highly participatory nature of the solvent; as we shall see the contribution from solvation/desolvation processes to overall thermodynamic parameters is typically greater than that provided by interactions between the two associating solutes. We begin our discussion by considering the interactions that result during the approach of two species absent the effect of solvation. We later consider the effect of water on each of these terms.

31.2.1 Gas Phase Non Covalent Interactions

Dipole-Dipole Interactions Molecules in which the center of electron density differs from the center of mass possess a permanent dipole. The interaction of oppositely charged dipoles yields a coulombic attraction, while the analogous interaction of equivalently charged species provides a repulsive contribution to an intermolecular interaction energy..The product of the dipole separation and the partial charge of the dipoles yields the dipole moment p. Many species containing strongly electronegative elements show no dipole moments because of a symmetrical disposition of partial charge; examples include carbon dioxide and carbon tetrachloride. While this symmetry produces a net dipole moment of zero, the molecules have higher order moments. Thus carbon dioxide has a large quadrupole moment while carbon tetrachloride shows a large octopole moment. Dipole moments of groups commonly encountered in proteincarbohydrate interaction are listed in Table 1 [4].

31.2 Association in Aqueous Solution

865

Table 1. Dipole moments of representative small molecules [4].

H20 H2S CH3CH2F CHjCH2Cl CH3CH2Br CH3CH2I FCH2CH20H CH3CHzOH CH3CH20D CH3CH2SH HOCHzCHzOH HSCHzCH2SH CH30CH3 CH3SCH3 a

1.88 0.98 1.90 2.00 1.90 1.75 1.52 1.70 1.54 1.52 2.31 1.70 1.32 1.51

CH3CH20CH3 1.75 3.20 CH3CH2NO 3.50 CH3CH2CN 1.25 CH3CH2COOH 2.50 CH3CHzCHO CH3CH:!CONH2 3.60 CH3CH2NH2 1.30 NH2 CH2 CHI NH2 1.86 CH~CH~NOZ 2.25 3.95 CsH5NOz o - N O ~ C ~ H ~ N O ~5.90 M - N O ~ C ~ H ~ N O ~3.79 p-NO2C6HdN02 0.69

Debye

The contribution of interactions between permanent dipoles to the overall cohesive energy between two molecules depends on both the magnitude of the dipoles and their relative orientation. A stepwise consideration of each interactiondipole-dipole, dipole-quadrupole, quadrupole-quadrupole, etc.-provides the overall contribution of the interactions of partial charges to the total cohesive energy between two molecules. The strength of individual interactions drops off as a power function of the internuclear distance; the exact functional dependence varies with the interaction. In general, the interaction between multipoles of order n1 and 122 drops off as r - ( n l + n 2 + 1 ) .If we consider the interaction of two species each with both a dipole (p) and quadrupole (0)separated by an internuclear distance r and disposed relative to each other by the angles 0 ~ 0,, and c$ (Figure l), the intermolecular force is given by [ 5 ] :

866

31 Protein- Carbohydrate Interaction: Fundamental Considerations

where E , is the permitivity of free space. The dependence of each term on the respective dispositions of the multipoles is given by the expressions: f l ( 0 ~0, ~4), = sin 6, sin 0, cos $ - 2 cos 0~cos OB 3

f i ( e A , e B , $=) - [ C O S O ~ ( ~ C O S ~ O1)~ - 2sin0A sineBcosOBcos$] 2 3 f 3 ( e A , e B , $=) - [ I 4

-

5C0S2e~- 5 ~ 0 ~ ~~ ~ 0 ~c -o s ~ ~ ~ c o s ~ ~

+ 2(4 cos eAcos 0,

- sin eA sin 0,

cos $) 2 ]

The functional form of the relationship makes intuitive sense; f l reaches a favorable maximum when the dipoles are aligned antiparallel to each other, an unfavorable maximum when the dipoles are aligned in a parallel fashion, and zero when the dipoles are oriented perpendicular to each other. Averaged over all orientations, the terms sum to zero. Application of a Boltzman distribution provides an energy for dipole-dipole, dipole-quadrupole and quadrupole-quadrupole interactions of:

All interaction terms are negative (favorable) and inversely proportional to temperature. As the order of the multipole increases the distance dependence becomes increasingly severe, and interactions beyond quadrupole-quadrupole do not provide a significant contribution to overall interaction energies. Dipole-Induced Dipole A dipole is induced in a molecule with no net permanent dipole but with a polarizable electron cloud when it is brought into contact with a permanent dipole. The magnitude of the induced dipole and the strength of the resulting interaction with the permanent dipole depends on the magnitude of the permanent dipole and the polarizability of the second species: Pind = aE

where E is the electric field generated by the permanent dipole and a is the susceptibility, or polarizability, of the second species. The energy of the interaction is then given by:

31.2 Association in Aqueous Solution

867

The strength of the interaction again depends on the relative orientation of the two species; the effective field between a dipole and a polarizable species separated by a distance r and at an angle 8 to the normal is given by: E=pJ

1 + 3 cos2 8 47~5~3

and thus: U=

+

a p y 3 cos2 8 1) 2r6 ( 4 7 1 ~ ~ ) ~

As before, over all orientations the interaction sums to zero; application of a Boltzman correction yields an interaction energy given by:

Again, the magnitude of the term drops off as the sixth power of the internuclear distance, and is directional. Some values of polarizabilities of various groups commonly encountered in protein-carbohydrate interaction are shown in Table 2 [6]. Dispersive Interactions

The intuitive interactions described above--essentially Coulombic attractions between oppositely charged particles-cannot account for more than a fraction of the I Table 2. Group polarizabilities [6].

-H -F

-c1 -Br -I -0-H -S-H -NH2 -CN -NO2 -CH, -CH> -YO H I

-CI

a

10-25 cm3

a&!a

a?za

3.9 4.5 22.1 33.1 52.9 9.6 3.2 16.4 20.7 19.7 22.4 18.4 22.0

23.5

9.8

80.5

6.7 30.0 32.8 13.7 13.7 27.1 57.7 10.4 60.8 25.6

868

31 Protein-Carbohydrate Interaction: Fundamental Considerations

total cohesive energy between common materials in the condensed phase. By far the most important contributors to overall interaction energies are the dispersive forces first identified by Fritz London in 1930 [7, 81. Although it is impossible to properly consider dispersive interactions in a purely classical treatment, the concept is made intuitive by considering the structures of atoms and molecules on timescales shorter than electron reorganizations (i.e. the reciprocal of the dispersion frequencies). Frozen in time, all molecules possess instantaneous dipoles; interaction of these dipoles provides the London or dispersive interaction. The interaction energy between two species arising from London forces is given by the expression:

where VA and VB are the dispersion frequencies of A and B, and U A and U B the corresponding polarizabilities. Although individual dispersive interactions are weak, they have no orientational dependence. Summed over an entire molecule they are large, providing the bulk of the total cohesive energy. Specific Forces: Hydrogen Bonding and n-c Bonding

The forces described so far are weak, typically less than one percent the strength of a covalent bond. In many instances, interactions between non-covalently bonded species are significantly stronger than this limit. For example, the exothermic transfer of ethanol from the gas phase to triethylamine reduces the strength of the ethanol 0-H covalent bond by 13.5Y1: clearly the interactions formed between ethanol and triethylamine are significant on the scale of covalent bonds. These stronger interactions are referred to as specific forces, and result from the interaction of an electron lone pair with a permanent dipole (polarized o-bond). Such forces differ from the non-specific forces described above in several important ways: i) The interactions are directional and stoichiometric. Unlike other coulombic interactions that include a directional component to the force vector but that have no a priori defined stoichiometry, specific interactions have a discrete stoichiometry. Each lone pair of electrons forms a single interaction with an appropriately polarized n-G bond. It is thus correct to consider true equilibrium constants for the formation of two hydrogen bonds between water and acetone: H 0, H.

P-H ,H

Q O

K1 -

QoO'

K CH3

H3C

Scheme 1. Hydrogen bond formation.

K2

,H

P-H

Q0O'

KCH3

H3C

31.2 Association in Aqueous Solution

I

H 2.2

I Ca 1.o

cs

869

2.0 A1 1.6 sc

2.5 Si 1.9 Ge

3.0

P 2.2 As

3.5 S 2.6

4.0

Se

Br

c1 3.2

Ba

Figure 2. Representative Pauling electronegativities.

ii) The formation of spec$c interactions signiJicantly shortens the internuclear distance between the bound partners-typically to less than the sum of the van der Waal radii-while significantly lengthening the o-bond between the electropositive proton and its electronegative partner. This bond weakening is most readily observed as an increase in the IR stretch frequency. iii) The process is exothermic, despite the weakening of the polarized o-bond. The extent of polarization in an X-H bond that is required for effective activity as a hydrogen bond donor is unclear. A useful rule of thumb is that protons bonded to elements with Pauling electronegativities greater than three act as hydrogen bond donors. Most species with a lone pair of electrons act as hydrogen bond acceptors, including oxygen (alcohols and ethers, carbonyls, phosphine and amine oxides, sulfoxides), nitrogen (amines, pyridines, nitriles), and sulfur. Stable anions also form strong hydrogen bonds. The most reliable predictor of hydrogen bond strengths, at least in the gas phase, is the difference in proton affinity of the hydrogen bond donor and acceptor. By this criteria the FH-F- interaction is predicted as the strongest monomeric hydrogen bond; this prediction is borne out by a crystal phase heat of formation of -37 kcal mol-'. The value of hydrogen bond strengths in solvent, however, is greatly influenced by the interactions of both donor and acceptor with the solvent itself, and conclusions regarding the contributions of hydrogen bonds to overall net free energies of interaction in solution cannot be drawn from proton affinity data [9]. 31.2.2 The Effect of Water on Intermolecular Interactions Having considered intermolecular interaction in the absence of solvation, we now consider the consequence of addition of solvent to a system of interacting molecules. At the outset we note that the picture changes from one that considers the

870

31 Protein-Carbohydrate Interaction: Fundamental Considerations

Table 3. Proton affinities of representative small molecules [9]

a

PA"

PA"

226.8 222.3 222.2 218.0 214.9 194.1 198.6 189.3 203.5 189.8 165.4 187.4 196.4 183.7 180.3 165.2

168.5 166.4 173.2 165.7 163.3 209.4 142.5 209.4 195.0 188.7 185.6 179.6 209.1 194.9 190.5

kcal mol-'

interaction between solutes to one that considers the dzerence in interactions between solutes and solvent versus solute and solute. The transfer of all species from the gas phase to water is exothermic: by a variety of specific and non-specific mechanisms water interacts with all solutes such that the free energy of the solvated species is lower than that of the isolated gas-phase species. Binding two solutes thus necessarily involves the loss of favorable interactions with solvent, a loss that is ameliorated by the favorable interactions that arise during association in the absence of solvent. Below we consider the effects of solvation on each of the classes of interaction described above. Finally, we consider the 'hydrophobic effect' and its contribution to ligand binding. Coulombic Stabilization

We consider first the effect of aqueous solvation on all intermolecular stabilizations that derive from the interaction of charged or dipolar species. Because of its small size and significant dipole and quadrupole, water interacts strongly with all ionic and dipolar species. A binding event of charged or dipolar compounds thus proceeds with a significant loss of favorable cohesive interactions between solutes and water. The effect is most profound for ionic interactions; similar ameliorations of solute-solute interaction apply to multipole-multipole and dipole-induced dipole interactions. Gas phase ionic interaction energies are large, typically on the order of 50-200 kcal mol-'. The interaction of water with charged species is nearly as large; as a result solution-phase ionic interactions are weak, seldom more than five kcal mol-'

31.2 Associution in Aqueous Solution

871

and typically less than one kcal mol-'. Examples of this diminution, or leveling, of interaction energies are myriad. The binding constant for the association of the quaternary ammonium ion choline with acetyl cholinesterase differs from that of the uncharged analogue 3,3-dimethyl propanol by a factor of only 30. Likewise, a range of monovalent anions inhibit acetoacetate decarboxylase by interacting with the positively charged active site with binding constants ranging from 10 to lo5 M-', corresponding to binding free energies of only one to five kcal mol-'. The powerful effect of solvation is also apparent in the order of effectiveness of acetoacetate decarboxylase inhibition. The halogen anions bind with decreasing affinity in the order I- > Br- > C1- > F-, the opposite order to that predicted on the basis of electronegativities. Apparently the more effective solvation of smaller, harder anions leads to reduced binding to the protein: a decreasing favorable electrostatic interaction proceeding down the periodic table is more than compensated for by a more rapidly decreasing unfavorable desolvation energy. Similar arguments are readily apparent during the consideration of multipole-multipole and dipoleinduced dipole interactions. The effective strong favorable interaction of water with multipoles lost during binding precludes a significant energetic advantage from dipole-related interactions between solutes in the bound state relative to the solvated unbound state. It is important at this point to distinguish affinity from specificity, at least as it pertains to this discussion. Dipolar interactions are directional and, in an absolute sense, strong. Improperly satisfied dipolar interactions in a complex are strongly disfavored relative to the energy of the unbound solvated system. Thus dipolar and ionic interactions play important roles in determining the specificity of solute-solute interactions and the precise structure of the bound complex, even though they likely do not contribute strongly to the overall net negative free energy of complexation. Hydrogen Bonding The strongest known hydrogen bond is the FH-F- interaction. In the crystal phase this interaction is favorable by some 37 kcal mol-'. In aqueous solution the equilibrium constant for this hydrogen bond is 0.4 M-', corresponding to a free energy of 0.8 kcal mol-'. There is no evidence for the formation of any other monofunctional hydrogen bond in aqueous solution. The situation becomes somewhat more complex for intramolecular hydrogen bonds, which form without a loss of translational and rotational entropy. Most of the data collected in this regard involve pKa values of dibasic acids. Thus, for example, the two pKa values of fumaric acid, for which no intramolecular hydrogen bond is possible, are 3.03 and 4.54. From this reference, both pKa values are perturbed in the isomeric maleic acid, at 1.91 and 6.33. Presumably this larger 6pKa reflects the energetic contribution of an intramolecular hydrogen bond in the singly ionized species. Even with a loss of net favorable interaction energy, we stress again the important roles that hydrogen bonds play in determining both the specificity of solutesolute interactions and the structure of the bound complex. Because of the strong angular dependence of hydrogen bonds on the X-H+lone pair vector, hydrogen bonds provide an even more powerful orienting force than do dipolar interactions.

812

31 Protein-Carbohydrate Interaction: Fundamental Considerations

Dispersive Interactions Despite the fact that they are individually weak, dispersive forces contribute to the exothermicity of the interaction between solutes in aqueous solution. The magnitude of the dispersive component of the cohesive energy varies as the polarizability of the interacting solutes. Because oxygen polarizes very weakly, a binding cycle that replaces solute-water interactions with solute-solute interactions will be energetically favorable: the dispersive component of OH-OH self association is 470 kcal mol-’ A6, while that for CH2-CH2 interaction is 1160 kcal mol-’ A6 [ 5 ] . Additionally, the r6 dependence of the interaction is a powerful orienting force during ligand binding.

31.2.3 ‘Hydrophobic’ Interactions

It has long been appreciated that placement of most organic species into aqueous solution produces an energetically unfavorable perturbation to the structure of water. Ligand binding proceeds with the desolvation of some fraction of the interacting surfaces; to the extent that this desolvation relieves the perturbation of water structure the free energy of the bound system is lowered relative to the unbound state. This process is typically considered, at least philosophically, as a repulsion between water and ‘hydrophobic’ solutes. In reality, of course, the effect is the result of the strong attractive interactions between water molecues, providing a liquid with a boiling point of 100 “C and an enthalpy of vaporization of 540 cal g-’, despite a molecular weight of only 18 g mol-’ . The molecular origin of the hydrophobic effect remains obscure, despite nearly 100 years of intensive experimental and computational investigation. Water has a remarkably high cohesive energy density; a priori one might expect that the unfavorable free energy of transfer of ‘hydrophobic’ solutes from some non-aqueous condensed phase to water should come from the loss of water-water contacts as a cavity is formed to accommodate the solute. Such a process would lead to an unfavorable enthalpy of transfer; one would thus reasonably expect the unfavorable free energy of hydrophobic solvation to be dominated by an endothermic enthalpy of transfer. In fact, near room temperature the enthalpy of transfer of virtually all hydrophobic species is near zero or slightly exothermic. Rather, near room temperature the unfavorable free energy resulting from the transfer of ‘hydrophobic’ surface area to water derives from an unfavorable entropic term. Furthermore, both the enthalpy and entropy of transfer are strong functions of temperature: ‘hydrophobic’ dissolution proceeds with a large positive increment in the constant pressure heat capacity of the system, ACp. It is this term that most distinguishes water from all other solvents. In a seminal work Arnett and coworkers demonstrated that the heat capacity increment during aqueous dissolution was uniquely a property of water and is not found for any other solvent, regardless of polarity [ 101. During the past decades a variety of physical models designed to rationalize the observed thermodynamic behavior of nonpolar solutes in water have been proposed. It is beyond the scope of this work to describe each in detail; rather, the

31.2 Association in Aqueous Solution

813

reader is referred to any of several reviews and monographs on the topic [ 111. Here, we describe the most popular of the group-the ‘clathrate’ model of water-in enough detail to allow a qualitative, intuitive understanding of the energetic consequences of hydrophobic transfer to water. The transfer of a hydrophobic solute to water can be conceptually divided into two steps: the formation of a cavity of sufficient volume to accommodate the solute, and the transfer of the solute to that cavity followed by collapse of solvent onto the solute. The first of these steps must be endothermic; that the overall process is thennoneutral or exothermic near room temperature thus requires that the second conceptual step-placing the solute in the cavity and allowing the solvent to collapse on the solute-be exothermic. Water in bulk solution can form four hydrogen bonds in a tetrahedral orientation; in a solvation shell such bonding is impossible. The clathrate model of water asserts that water in a solvation shell orients to maximize favorable hydrogen bonds. In this structure a loss in the total number of hydrogen bonds is compensated for by an increase in the strength of the hydrogen bonds that can form. This increased strength comes at the price of reduced entropy, particularly the rotational entropy of water, in the clathrate solvation shell. The enthalpically enhanced hydrogen bonds of the clathrate shell also provide a mechanism for the positive increment in constant pressure molar heat capacity. The notion of clathrate water as a physical model to explain the unusual thermodynamics of hydrophobic hydration has been parameterized by both Muller and Ben-Naim to yield a semi-quantitative description of the event [12-141. In these constructs the enthalpy of transfer is given by the expression:

where f b and fhs is the fraction of hydrogen bonds in the bulk liquid and the hydration shell, respectively, and AHb and AHhs are the enthalpies of hydrogen bond formation in bulk solution and hydration shells, respectfully. While the clathrate model of water is intuitively accessible and explains both qualitatively and quantitatively many of the phenomena associated with hydrophobic hydration, many unanswered questions remain. Specifically, physical evidence for the existence of clathrates is lacking; indeed many studies specifically designed to observe such structures fail to note any increase in the order of water structure [ 15-20]. The notion of clathrate water is far from universally accepted and other models of water exist, especially a series of models based on scaled particle theory that explain the peculiar phenomenology of hydrophobic hydration based on the exceptionally small size of water [21-291. Putting aside issues of the molecular origin of ‘hydrophobic effect’, the more significant issue is the extent to which desolvation contributes to overall binding thermodynamics. A variety of experimental methodologies to evaluate the magnitude of this effect exist. The most commonly used tools are Hansch transfer parameters [30, 311. Briefly, a series of compounds are partitioned between water and watersaturated octanol. The value 7c is then derived for a substitutent according to the expression:

874

31 Protein-Carbohydrate Interaction: Fundamental Considerations

Table 4. Hansch parameters of representative functional groups [30, 311.

arom”

aliphb

0.15 0.70 1.02 1.26 -0.32 0.52 1.07 0.97 1.90 1.82 1.68 2.51 1.89 -0.61 -1.03 -0.04 0.62 - 1.23 0.18 0.24 -0.15 -0.72 -0.01

-0.73 -0.13 0.04

F CI Br I CN CH3 CF3 CH2CH3 n-Bu sec-Bu t-Bu cyclohexyl C6H5 OH CH20H OCH3 SCH3 NH2 N(CH3)2 NO2 COOH CH2COOH COOCH3 OCOCH3 COCH3 CONH2

-0.37 - 1.49

- 1.47

-1.80 -0.98 - 1.85 -0.95 - 1.26

-0.91 -0.91 - 1.26 -2.28

~~

“based on aromatic parent compound based on aliphatic parent compound

7t = log

P e l

where P and P, are the partition ratios of a substituted and parent compound, respectively. The value RTln7t is thus the incremental Gibbs free energy for the transfer of the substituent from water to water-saturated octanol, assuming at least rough group additivity. Hansch parameters for a range of commonly encountered substituents are shown in Table 4. Another approach to issues of solvation thermodynamics utilizes the differential hydrogen bonding properties of light and heavy water [32, 331. The use of thermodynamic solvent isotope effects, or the evaluation of the enthalpy of ligand binding in light and heavy water, allows evaluation of the extent to which desolvation aids

31.2 Association in Aqueous Solution

875

ligand binding. Again, our goal here is not to describe the technique exhaustively; the reader is referred to recent reviews of the subject for a more complete treatment. Briefly, the D-0 hydrogen bond is enthalpically favored relative to the H-0 analog by roughly 10%; an offsetting decrease in the entropy of bond formation leads to a free energy difference near zero between the two. Assuming a clatbrate model of water, the transfer of a solute from light to heavy water will be exothermic by an amount related to the enthalpy of clathrate formation. Straightforward extension of the Muller formalism above quantifies the enthalpy of transfer as [13, 341:

A schematic representation relating ligand binding in isotopic solvents can be created as:

AHt,"

AHt,b

\'

Figure 3. Cycle relating binding in isotopic waters. The subscripts represent light (1) and heavy (h) waters, and unbound (u) and bound (b)systems.

Thermodynamic parameters are state functions. Thus the difference in the two horizontal processes of Figure 3-binding in isotopic solvents-is equivalent to the difference in the two vertical processes-transfer of the free and bound systems from light to heavy water. From this formalism it is apparent that the difference in enthalpy of binding in isotopic waters is equivalent to the enthalpy of transfer of the jraction of the ligand and protein desolvated during binding. Because this enthalpy is related to the enthalpy of solvation, the thermodynamic solvent isotope effect is a measure of the enthalpy of binding that derives from removal of the binding site and ligand from solution. Thermodynamic solvent isotope effects have been utilized to evaluate the extent to which desolvation aids ligand binding [35,361. While the assumptions implicit in the calculations likely make precise determination of AHsolvimpossible, the relative role of desolvation for a series of bindings is clearly accessible (Table 5). This exercise facilitates two observations that are likely general. First, a significant fraction

876

31 Protein-Carbohydrate Interaction: Fundamental Considerations

Table 5. Solvation-associated enthalpies of ligand binding [ 351. Binding pair

AG

AH

Dioclea/3,6-di-(0-aMan)aManOMe Concanavalin A/aManOMe Concanavalin A/3,6-di-(0-aMan)aManOMe Vancomycin/diAcLys-D-Ala-D- Ala RNAse/CMP FK506/FKBP

-8.2 -5.0 -7.5 -7.2 -7.5 -12.3

-13.0 -7.1 -10.7

A& -4 -5

-5 -1

-11.5 -14.1

-14

-17.2

-18

of overall binding enthalpy derives from solvent reorganization; this trend is likely repeated in binding free energies. Secondly, the enthalpy derived from solvent reorganization is lower for protein-carbohydrate interaction than for any other interacting system studied. This observation hardly comes as a surprise, given the hydrophilic nature and high aqueous solubility of carbohydrates. On the other hand, the relatively small enthalpy of binding available from saccharide desolvation is the most probable cause of the low binding constants that typify protein-carbohydrate interaction. This limitation is fundamental, severe, and likely insurmountable. Together, the forgoing discussion leads one to a picture of binding that is largely driven by ‘hydrophobic’ effects. In this model, a range of polar interactions are largely responsible for the specificity of the interaction: if key hydrogen bonds, salt bridges and dipolar interactions are satisfied by solute-solvent interactions prior to association but lost during binding, an unfavorable contribution to the overall binding free energy results. If all interactions lost to solvent are adequately replaced in the complex, the net favorable free energy that derives from the desolvation of non-polar surface area drives formation of the complex. Lemieux described such notions two decades ago in the concept of a ‘polar gate’ [37]. More recently, several researchers have shown that a significant fraction of the total surface area of an oligosaccharide ligand can be replaced by non-carbohydrate surface [ 38-40],

31.3 The Evaluation of Protein-Carbohydrate Binding Discussion of the effect of ligand structure on protein-carbohydrate affinity requires an evaluation of complex stability constants. A number of biophysical techniques are appropriate for the study of protein-carbohydrate interaction; many of the more enlightening strategies are the topics of separate chapters elsewhere in this volume. We describe below three techniques used extensively in glycobiologyinhibition of hemagglutination, enzyme-linked lectin assay (ELLA), and isothermal titration microcalorimetry-and we consider the types of information provided by each technique in order to facilitate appropriate interpretation of the data.

31.3 The Evaluation of Protein-Carbohydrate Binding

811

31.3.1 Precipitin Assays

By far the most common of the techniques utilized to evaluate proteincarbohydrate interaction are a group of assays based on the hemagglutination process. Anticoagulated red blood cells spontaneously separate on standing in microtitre plate wells, as the denser erythrocytes settle to the bottom of the wells. Addition of a polyvalent lectin to a suspension of erythrocytes prevents this segregation and results in the formation of a gel-phase cross-linked lattice. This crosslinking process is referred to as hemagglutination, and provides a straightforward read-out for inhibition assays. Assays of protein-carbohydrate binding function by inhibiting this hemagglutination event. Addition of a soluble saccharide sets up a competition between glycoconjugates on erythrocyte surfaces. Typically serial dilutions are performed across a microtitre plate, and the minimum concentration of soluble ligand required to inhibit 50% agglutination is reported as an ZC50. Because the assay is run as a serial dilution errors are typically assumed at one well, or a factor of two. Absolute ZCso values are frequently not reproducible from lab to lab, although the order of inhibitory potency through a series of ligands is generally robust. Agglutination assays evaluate a Straightforward physical process, namely the ability of a soluble ligand to prevent a cross-linking aggregation process. The danger of agglutination assays lies in the assumption that ZC50 values are simply related to Keq for a reversible protein-carbohydrate binding event. At a minimum, the activity of a soluble ligand in an agglutination assay is the inhibition of an irreversible or quasi-reversible phase transition, from a soluble lectin to an ordered cross-linked gel matrix. Ligand binding is surely requisite for the initiation of this event, but all events that follow serve to produce a complicated set of coupled equilibria, each with a distinct microscopic set of kinetic and thermodynamic constants. The situation becomes significantly more complicated when the soluble ligands are themselves multivalent, since now there exists a competitive set of crosslinking/aggregation processes. In the limit, a protein-carbohydrate binding followed by an irreversible aggregation/precipitation will appear in this assay as an exceptionally high affinity association, since the binding equilibrium is coupled with the aggregation/precipitation event. Hemagglutination assays have been and will remain a mainstay of the study of protein-carbohydrate interaction. In many respects the assay is highly relevant to the study of biological processes; Nature doubtlessly uses the high valency of both carbohydrate ligands and binding proteins for a range of functions. On the other hand, ZC50 values from agglutination assays cannot be interpreted in terms of protein-carbohydrate affinity. In instances where protein-carbohydrate interactions have been evaluated by agglutination assay in addition to some other biophysical technique, hemagglutination ZC50 values do not correlate with K, values for the equivalent process. Recently we reported a study of the binding of multivalent saccharide ligands to concanavalin A [41]. While ZC50 values varied by a factor of 30 on a valence corrected basis, binding constants for the same ligands evaluated by titration microcalorimetry varied by less than a factor of two. Rather, hemaggluti-

+

818

31 Protein-Carbohydrate Interaction: Fundamental Considerations

nation IC50 values correlate with entropies of ligand binding, again demonstrating that hemagglutination evaluates an aggregation, rather than a binding, process. 31.3.2 Enzyme-Linked Lectin Assay (ELLA) Roy and coworkers have reported a modification of the precipitin assay [42-501. This assay evaluates competitive binding between soluble and immobilized ligands to a lectin-enzyme conjugate and is essentially a variant of quantitative ELISA. In the enzyme-linked lectin assay a high molecular weight polyvalent ligand is adsorbed to the surface of microtitre plates. The assay has been utilized most frequently for the evaluation of binding to mannose-specific proteins; in such cases yeast mannan serves as the immobilized ligand. Following blockage of non-specific binding with BSA, a lectin-horseradish peroxidase (HRP) conjugate is incubated in the microtitre plate wells with serial dilutions of a soluble ligand. Following an incubation period the microtitre plates are evacuated and the binding mixture replaced with hydrogen peroxide and a pro-dye substrate. Color formation, proportional to the amount of lectin bound to the microtitre plates, is read quantitatively with a UV plate reader. ELLA is designed to obviate the most significant complication of agglutination assays, namely the irreversible formation of aggregates. In theory, the competition between immobilized and soluble ligand should not involve the formation of kinetically trapped aggregates. On the other hand, data from ELLA assays do show evidence of kinetic contributions. Plots of inhibitor concentration versus fractional inhibition frequently asymptote to values less than 100% inhibition. Often different ligands asymptote to different fractional inhibitions, The most obvious explanation for this behavior is the formation of some irreversibly bound lectin species on the surface of the microtiter plate wells. The exact nature of this species is somewhat difficult to imagine, although the use of microheterogeneous polydisperse polysaccharide as an immobilized ligand likely presents a range of binding epitopes for the lectin-HRP conjugate. In summary, it is now clear that precipitin assays do not measure ligand binding. ELLA is a technique designed to overcome the issues that render agglutination unsuitable for evaluating ligand binding. However, until IC50 values obtained through this technique are verified by comparison to binding constants measured by other biophysical techniques, results from the assay should be regarded with some caution. 31.3.3 Isothermal Titration Microcalorimetry Isothermal titration calorimetry (ITC) has long been recognized as a useful tool for the evaluation of binding constants [51]. The field of calorimetry changed markedly during the early 1990s with the introduction of commercial titration microcalorimeters [52].These devices, available from several suppliers, operate with volumes near one to two milliliters. Virtually all the data reported to date on the thermody-

31.3 The Evaluation of Protein-Carbohydrate Binding

819

namics of protein-carbohydrate interaction has been obtained with one of several generations of the Microcal ITC; these instruments have sample cell volumes near 1.3 ml (Figure 4). The sensitivity requirements of the technique mandate a continuous power compensation design rather than the simpler and more traditional passive thermal conductivity experimental approaches. In this design, a sample and reference cell are heated at a constant rate, typically less than 100 microcalories per second. This heating rate alters the temperature of the cell contents by less than 0.1 "C during the course of a titration. The temperature between the two cells is evaluated and a second compensating voltage is applied to bring the two cells into precise thermal equilibrium. Addition of an aliquot of ligand through a syringe into the sample cell disrupts this equilibrium; in turn, the compensating voltage is adjusted to return the cells to equilibrium. The raw data is thus power as a function of time. Integration over time yields the more familiar plot of enthalpy per injection versus ligand concentration. The enthalpy evolved on each injection is a function of the concentrations of the two binding partners, the stoichiometry of the interaction, the binding constant (Keg)and the enthalpy of binding ( A H ) . A simple non-linear least squares fit to the data provides estimates of the binding stoichiometry, binding constant, and enthalpy of binding. The binding constant is related to the free energy of binding by the relationship:

AG

=

-RTInKeq

The entropy of binding is thus available by subtraction. The partial differential of enthalpy with respect to temperature defines the partial change in constant pressure heat capacity. Over short temperature ranges this derivative is approximated as

i3AH AC -AT

'-

AAH AT

-

The required concentrations of both binding partners are determined by the binding constant. The shape of the binding curve is determined by the unitless parameter c, equivalent to the product of the binding constant and the concentration of binding sites within the calorimeter cell. Figure 5 shows curves that result from values of c ranging over 1-1000. While in principle it is possible to fit curves anywhere within this range, values of c of 10-100 give optimal results. Practically, titration microcalorimetry operates to furnish estimates of both enthalpies and free energies of binding for systems with binding constants in the range of lo3 to lo7 M-'. At low affinity, protein solubility becomes limiting; a binding constant of lo3 M-' requires at least millimolar protein in solution. At high affinity, sensitivity becomes limiting. A binding constant of lo7 M-' limits the concentration of binding sites to 0.1 micromolar, which in turn limits the size of individual aliquot enthalpies. Across a complete titration, peak areas (i.e. enthalpies) should average no less than 100 microcalories. The range of lo3 to lo7 M-' thus assumes stable soluble protein and significant enthalpies of binding. The concentration of titrant in the syringe is typically 20-times the concentration of binding sites.

880

31 Protein- Carbohydrate Interaction: Fundamental Considerations

Sensor

?$

Lead Screw

Sensor Injector N

lie--

Plunger

Stirring

Syringe

Outer Shield Inner Shield

-

Reference Cell

Figure 4. The Microcal Omega titration microcalorimeter.

-

Sample Cell

31.3 The Evaluation of Protein-Carbohydrate Binding

-500

!

I

I

I

I

I

0

10

20

30

40

88 I

I

injection number Figure 5. The effect of c on calorimetric curve shape. Data calculated for AH = -6 kcal mol-' and 1 mM binding sites. Circles Keq = lo6; triangles Keq = lo5; diamonds Keq = lo4; squares Keq = 10'.

Even beyond the high affinity limit, calorimetry remains a powerful technique for the study of intermolecular interactions. With stoichiometric binding, enthalpies of binding are still accessible in a straightforward fashion. In this instance, the total enthalpy evolved on each injection is simply divided by the quantity of titrant added. As before, evaluation of the molar enthalpy of binding as a function of temperature provides ACp. Determination of the entropy of binding requires an independent determination of Keq;this protocol offers the additional advantage of avoiding correlated errors. That thermodynamic parameters are state functions provides another powerful methodology for determining both enthalpies and free energies of binding for very high affinity systems through a competitive cycle. If two species bind a single binding site with differing affinities, titration of the higher affinity ligand to binding sites loaded with a lower affinity ligand yields the difference in binding enthalpy and free energy between the two ligands. Thermodynamic parameters characterizing binding

882

31 Protein-Carbohydrate Interaction: Fundamental Considerations

of the lower affinity ligand can be determined independently and the corresponding values for the high affinity ligand are then equivalent to the sum of the two values. An elegant demonstration of this methodology was provided by Sigurskjold and coworkers in a series of papers describing the binding of acarbose to glucoamylase (vide infra) [53-561.

31.4 The Interpretation of Calorimetric Data While the measurement of thermodynamic parameters by ITC is straightforward, interpretation of the derived values is not. It is important to recall that thermodynamic properties represent the sum of all molecular processes that occur during a binding event. A range of intermolecular interactions must be considered carefully before interpreting thermodynamic properties in terms of solute-solute interaction. Below we consider some of the processes that contribute to net measured thermodynamic processes and, where possible, describe experimental techniques for the evaluation of the contribution of each set of microscopic events to overall ligand binding thermodynamics.

31.4.1 Solvation/Desolvation It has always been fashionable to reach conclusions regarding the role of solvation during association processes based on measured enthalpies and entropies of binding. Processes with favorable entropic components are frequently referred to as ‘entropy driven’-indicative of a ‘hydrophobically driven’ binding-while processes with large favorable enthalpies of binding are referred to as ‘enthalpy driven’; such a pattern is often interpreted in terms of favorable hydrogen bonding or salt bridge formation. There is no basis for this distinction. The predicted magnitude and sign of enthalpies and entropies arising from desolvation of both ligand and binding site are unclear. Indeed, the thermodynamic parameter most uniquely associated with solvation processes in aqueous solution is ACp. The compensating dependence of both enthalpy and entropy on this term requires that over a sufficient temperature range both the magnitude and sign of a measured enthalpy and entropy will change. Surely, however, the same physical processes aid and oppose binding over this range. Two methodologies allow independent evaluation of the role of solvation in ligand binding processes. The first requires measurement of thermodynamic solvent isotope effects, the differential enthalpy of binding in light and heavy water [33]. The fraction of the enthalpy of binding arising from solvent reorganization is directly related to this quantity. In principle, an absolute solvation-associated enthalpy is obtainable through this approach; in practice, derivation of such a value requires the use of a proportionality constant not known with either precision or accuracy. Thus the absolute value of solvation-associated enthalpies derived in this

31.4 The Interpretation of Calorimetric Data

883

way likely have substantial error associated with them. On the other hand, the error is a constant, deriving from a poor understanding of the precise difference in enthalpies between the H-0 and D-0 interaction. Comparative thermodynamic solvent isotope effects are thus free of this error. TSIE is thus most useful for evaluating the effect of ligand modification on binding thermodynamics [36]. The fraction of an overall enthalpy of binding attributable to solvation/desolvation for two ligands is readily compared by comparing solvent isotope effects. In this way assignment of changes in the enthalpy of binding in response to a change in ligand structure to either changes in solute-solute interaction or to changes in ligand solvation is straightforward and unambiguous. We have previously shown that thermodynamic solvent isotope effects correlate linearly with the change in molar heat capacity that accompanies binding [35].This correlation demonstrates that ACp is exclusively a measure of solvation, and comparison of AC, values for related ligands allows a similar interpretation of differences in binding enthalpies in terms of protein-carbohydrate interaction versus differences in solute-solvent interactions prior to association [ 571. The relative random errors of the two methodologies are unclear; in the one instance small values must be measured accurately while in the other errors in individual determinations are propagated during curve fitting. In any event, a meaningful discussion regarding the molecular origin of measured thermodynamic parameters is impossible without some evaluation of the contribution of solvation to overall measured enthalpies of binding. The entropy of binding arising from solvation is also available with knowledge of the change in molar heat capacity accompanying binding. Overall entropies of binding can be conceptually dissected to describe the contributions of changes in translational and rotational entropy, solvation-associated entropy, and changes in entropy from restriction of conformational degrees of freedom during binding:

Separation of overall entropies of binding into component segments in this fashion again allows relatively sophisticated conclusions to be drawn regarding the molecular origin of measured thermodynamic parameters. In this formalism, entropies of binding are divided into three contributions and knowledge of any two of the three provides the third by subtraction. Evaluation of ASsolv and AS,,, are both feasible. Solvation Entropy It has long been recognized from efforts to understand protein folding that solvation associated entropies are related to changes in molar heat capacity accompanying binding. Thus,

where T is the temperature at which the entropy of binding was measured and T * is the so-called entropy convergence temperature, or the temperature at which all

884

31 Protein- Carbohydrate Interaction: Fundamental Considerations

solvation associated entropies approach zero [ 58, 591. AS*solv includes contributions to entropy from proton transfer and electrostatic effects; for most proteincarbohydrate interactions this term is negligable. Evaluation of ACp thus allows extraction of the fraction of the total entropy of binding attributable to solvent reorganization. Note that this term is temperature dependent; again, this dependence makes intuitive sense. As the temperature of the system is raised the residual order of water slowly dissipates, and at 385.15 K disappears entirely. Translational/RotationalEntropy

A major fraction of the entropy of a particle derives from its ability to translocate in three dimensions and rotate on three axes. The magnitude of this entropy varies as the logarithm of the particle mass and the principle moment arms describing the distribution of mass. Because of the logarithmic relationship ligand binding proceeds with a loss of entropy roughly equivalent to the translational and rotational entropy of the smaller particle. The magnitude of this term has long been a contentious issue. Evaluation of A&+R in the gas phase is straightforward, and follows from the Sakur-Tetrode equation. Transfer to condensed phase, however, greatly complicates the matter. All solutions show greatly reduced molecular motion relative to the gas phase. Water is especially severe with regard to the restriction of molecular mobility: as a highly associated liquid with a tremendous cohesive energy density, water interacts strongly with both adjacent water molecules and with dissolved solutes. A wide range of values of the contribution of AST+R to complexation entropies has been suggested; these are values of 7-50 eu (Table 6 ) . The entropy values reported in Table 6 correspond to the best estimates of the transla-

Table 6. Reported values of AST+R. Investigators

AST+R~

TAS~

Doty & Myers Dunitz Erickson Horton & Lewis Janin Janin & Chothia Jencks Murphy et al. Page & Jencks Searle & Williams Searle et al. Spolar & Record Tidor & Karplus Williams et al.

122 7 (0-7) 23-37 21 100 (50) 57-74 45 (35) 8 40-50 (25-35) 40 (0-40) 40 40-60 100 (77) 39

36.4 2.1 7-1 1 6.2 30 (15) 17-22 13.4 (10.4) 2.4 12-15 (7.5-10.4) 12 12 11.9-17.9 30 (23) 11.5

ae.u. bkcal mol-' at 298K

xnta

0-7

50 10

15 0-40

23

Ref. 60 61 62 63 64 65-67 68 58 69 70 71 72 73 74

31.4 The Interpretation of Calorimetric Data

885

tional and rotational entropy loss during association. Many of these values refer to total immobilization, i.e. total loss of translational and rotational entropy with no account for residual internal entropy. In those cases for which an estimate of the residual internal entropy was made, the corrected translational and rotational entropy loss has been placed in parentheses after the value for total immobilization, and the compensating internal energy (Sint)has been noted. In order to interpret changes in solution thermodynamics in response to changes in ligand structure, we require ASsolv and ASconf.With a methodology for evaluation of ASsolv in hand knowledge of AST+R would provide ASconf by subtraction. Although accurate knowledge of AST+R remains elusive, it is important to recall that values of AST+R are relatively insensitive to molecular size and shape. Accordingly, an arbitrary value of AST+R can be selected for the purpose of this exercise; this value will remain constant across a set of ligands. Changes in entropies of binding arising from ligand modification can thus confidently be ascribed to changes in solvation-associated entropies, or interactions of solvent and solute prior to association, or to changes in conformational degrees of freedom in ligand and protein during binding. Again, the range of values in Table 6 precludes extraction of absolute contributions to overall measured entropies arising from the loss of conformational degrees of freedom. 31.4.2 Other Contributions to Thermodynamics of Association Proton Transfer Many ligand binding events proceed with the transfer of a proton, either from or to buffer; in such instances the enthalpy of buffer ionization will be included in measured values of AHB. Enthalpies of buffer protonation range from near zero to greater than 10 kcal mol-' (Table 7) [51]. Evaluation of enthalpies of binding as a function of buffer ionization enthalpy is a straightforward method of determining the extent of protonation contributions to overall binding enthalpies. A plot of measured enthalpies of binding versus buffer ionization enthalpy provides a plot with a slope equal to the number of protons transferred during binding, and a yintercept equal to the enthalpy of binding in the absence of buffer ionization. Salt Effects/Binding Site Reorganization In some instances, a binding site undergoes reorganization during association. An equilibrium can be written describing the enthalpy, entropy and free energy difference between the two states. Ligand binding thus involves a coupled equilibrium, and the reported values of thermodynamic properties are sums of those values for binding site reorganization and ligand binding to the preformed site. If environmental conditions affect the position of this equilibrium, the precise conditions under which binding is measured will affect the measured enthalpy of binding. An example of such an effect is observed in the binding of mannosides by concanvalin A [36]. Here, a salt bridge between Asp16 and Arg288 in the unbound form of the protein is ruptured during ligand binding, a process that proceeds with an unfa-

886

31 Protein-Carbohydrate Interaction: Fundumentul Considerations

Table 7. Enathalpy of ionization of common buffers [51]. Trivial Name

Formula

AHB"

MES

O/ N L +J H C H ~ C H ~ S O ~ -

12.68

bis-tris ACES ADA

(HOCH2)3CN+H(CH2CH2OH)2 H2NCOCHzNfH2CH2CH2SO3H2NCOCH2N+H(CH2COO-)l

29.25 30.12 11.51

MOPS

O~+HCH~CH~CH~SO~-

19.0

PIPES

Na03SCH2CH2NnN+HCH2CH2S03LJ

8.70

BES

(HOCH2CH2)2N+HCH2CH2SO3-

23.10

HEPES

HOCH2CH2HN+r\NCH2CH2SOju

16.40

TES Ethyl glycinate Glycinamide

(HOCH2)3CNtH2CH2CH2S03H3NfCH2COOCH2CH3 H3N+CH*CONH2

29.25 46.32 44.77

HEPPS

HOCH2CH2 HtNC\NCH2 CHzCH2SO3

17.95

Tricine THAM Glycylglycine Bicine TAPS N,N-Dimethylglycine

(HOCH2)3CN+H2CH2COO(HOCH2)3CN'H3 H3N+CH2CONHCH2COO(HOCH2CH2)2N+HCH2COO(CH3)2N+HCHzCOO-

30.50 47.28 43.72 26.23 40.12 31.51

CAPS

cNtH2CH2CH:CH2S03-

48.53

(HOCH2)3CNtH2CH2CH2CH2S03-

"kcal mol-'; enthalpy of removing the H from the quaternary nitrogen atom

vorable enthalpy. Increasing ionic strength cleaves this intramolecular salt bridge prior to ligand binding, diminishing the unfavorable contribution to the overall enthalpy of binding. As a consequence, enthalpies of mannoside binding vary by a factor of two over a salt range from 5 mM to 1 M.

31.4.3 van't Hoff versus Calorimetric Enthalpies A measured difference between van't Hoff and calorimetric enthalpies of binding has been offered as a possible test of contributions other than ligand binding to calorimetric enthalpies of binding [75-771. van't Hoff enthalpies of binding are determined from the temperature dependence of the binding free energy according to the expression: AHVH

1nK = 1nA - RT where A is a pre-exponential term related to the entropy of binding. A plot of 1nK

31.5 The Thermodynamics

of'Protein

Carbohydrate Interaction

887

versus 1/T thus gives a slope equivalent to AHU"/R, This relationship has been utilized extensively by the protein folding community to observe coupled equilibria unrelated to folding. There is, however, a range of problems associated with its use in ligand binding. First, associations in aqueous solution universally proceed with a change in molar heat capacity, i.e. ACp # 0. The expression above then is predicted to be nonlinear, since A H is itself a function of temperature. The fit required to extract A H Uis~thus a non-linear regression for two variables using values that have themselves been extracted from a non-linear regression for three variables. Appropriate permutation of errors almost certainly precludes observation of all but extraordinarily large differences between the calorimetric and van? Hoff enthalpies with any degree of confidence. Secondly, it is somewhat difficult to devise a physical model that corresponds to a measured difference. van't Hoff enthalpies will differ from calorimetric enthalpies only to the extent that calorimetric enthalpies contain contributions from processes that do not contribute to the free energy of binding. Because thermodynamic processes are state functions, this condition can hold only for situations where perfect enthalpy-entropy compensation exists, and even then only if the compensating terms have equivalent temperature dependencies. Because ACp affects both A H and A S , this condition can be true only if A S for the compensating process is large compared to ACp. Thus while in principle differences between van't Hoff and calorimetric enthalpies contain important information regarding the precise molecular processes that occur during binding, such differences should be regarded cautiously.

31.5 The Thermodynamics of Protein-Carbohydrate Interaction Having considered those intermolecular interactions that contribute to overall net measured thermodynamic parameters, we turn now to a review of thermodynamic measurements of protein-carbohydrate binding reported during the last five years. Values reported prior to this time can be found in earlier reviews. Calorimetrically derived changes in enthalpies, entropies, free energies and molar heat capacity that occur during protein-carbohydrate binding are shown in Table 8. Extension of the database of known protein-carbohydrate thermodynamic parameters continues to reinforce the basic concepts identified some time ago. Virtually all of the examples in Table 8 show binding free energies of 4-8 kcal mol-', corresponding to association constants of 103-105 M-'. Typically, enthalpies of binding are more negative than the corresponding free energies of binding and the entropic contribution to AG is unfavorable. The database of changes in molar heat capacity accompanying protein-carbohydrate binding remains small, but continues to grow. Again, in almost all cases exceptionally small values of ACp are recorded, almost always less than 200 cal mol-' deg-', and usually less than 100 cal mol-' deg-'. A small number of values in Table 8 are positive; this observation is unusual. In most instances the errors in individual values of both A H and the A H versus T fits are large and the validity of these data is unclear. There is at least some correlation between ACp and AG. Thus, for example, the binding of acarbose to glu-

fragment free concanavalin A concanavalin A

Succinyl concanavalin A

CeNumonasJimi CBD

Heparin Heparin Heparin Heparin Heparin

Antithrombin 111 K'2'-A'34 Antithrombin 111 K"'AA Antithrombin I11 K'25AA Antithrombin 111 R'29AQ Antithrombin I11 K'33AA CBDNl

-AHb

34.5 14.9 36.0 48.6 57.5 9.7 PG~c(~-~~)[PG~c(~-+~)]~G~cOH 12.7 PGk( 1--+4)[PGl~(l-+4)]3Gl~OH 13.0 PGlc(l-+4)[PGl~( I+4)]4GlcOH Barley a-glucan 13.7 14.4 Oat P-glucan 13.4 Hydroxyethyl Cellulose Acid Swollen Cellulose 7.7 Cellulose I 1.2 Cellulose I1 0.2 14.5 1 2 13.8 5.0 3 14.3 Mana(l+6)[Mana(l-3)]-Man aMeGlc 5.3 6.6 6.7 6.8 aMeMan 8.2 8.4 aMe2dMan 7.2 5.9 C-allylglucose 5.9 C-allylmannose 1 8.5 2 11.1 3 6.2

Ligand"

Protein

Table 8. Thermodynamics of protein-carbohydrate binding.

-28.0 -7.6 -31.2 -45.8 -53.9 -4.1 -6.6 -6.9 -7.4 -8.2 -7.3 -1.8 9.7 8.2 -9.7 -8.7 0.2 6.7 -0.7 -2.1 -2.1 -1.5 -2.9 -2.8 -1.9 -0.2 -0.8 -3.5 -5.0 -1.0

TASb 10.1 4.5 4.3 4.1 4.8 5.0 6.1 6.1 6.3 6.2 6.1 5.9 10.8 8.4 4.8 5.1 5.2 31.6 4.6 4.5 4.6 5.3 5.3 5.6 5.2 5.0 5.1 5.0 6.1 5.2

-AG~

-103

-50.0 -61.9

ACpc

81

78 78 78 78 78 79 79 79 79 79 79 79 80 80 81 81 81 82 81 82 83 81 82 84 83 81 81 81 81

Ref.

aGlcOPh 1dGlc aGlcF 2dGlcOH

3,4dMana(1+3)[Mana(1+6)]2,4dManOMe

Mana( 1+3)[Mana(l-6)]ManOMe 2dMana(l+3)[Mana( lk6)lManOMe

3dMana(l+3)[Mana(l+6)]4dManOMe 3dMana(l+3)[Mana(l+6)]2,4dManOMe

3dMana( 1+3)[Mana( 1+6)]2dManOMe

6dMana(l+6)[Mana(l-+3)]ManOMe Mana(l+3)[Mana(l+6)]2dManOMe Mana(l+3)[Mana(1+6)]4dManOMe

2dMana(1+6)[Mana( 1+3)]ManOMe 4dMana(l+6)[Mana(1+3)]ManOMe

2dMana(l+6)[Mana(l+3)]-ManOMe

4dMana( 1+3)[Mana( 1+6)]-ManOMe

aGlc( 116)GlcOH aMe2dGlc Mana( 1i 2 ) M a n Mana( 1j 2 ) M a n O M e Mana( 1+2)Mana( l i 2 ) M a n Mana( 1i 3 ) M a n Mana( 1+3)ManOMe Mana( 1-6)Man Mana( lh6)ManOMe GlcNAcP(1-2)Man Mana(1+6)[Mana(l+S)]-Man 4

aMeGlcNAc aGlc( 114)GlcOH 6.2 6.2 6.2 6.7 7.3 9.9 10.5 10.7 10.2 10.7 9.4 8.4 5.3 14.1 10.6 10.6 12.3 14.9 14.0 11.2 11.7 11.6 13.4 12.1 10.6 9.7 8.7 14.7 14.1 8.9 3.5 4.2 7.1 4.8 -2.1 -1.9 -1.9 -1.6 -2.6 -3.6 -3.5 -3.1 -4.5 -4.5 -3.8 -3.1 -0.1 -6.6 -2.2 -2.2 -5.5 -7.1 -6.2 -4.9 -5.7 -5.5 -6.5 -5.7 -4.4 -3.8 -3.2 -7.1 -6.9 -3.3 1.8 0.7 -4.2 -1.4 4.1 4.3 4.3 5.1 4.7 6.3 7.0 7.6 5.7 6.2 5.6 5.3 5.2 7.5 8.4 8.4 6.8 7.8 7.1 6.3 6.0 6. I 6.9 6.4 6.2 5.9 5.5 7.6 7.2 5.6 5.3 3.5 2.9 3.4 -19.4 -6.7 41.1 1.8

82 82 83 83 82 82 82 82 82 82 82 82 82 82 82 84 85 85 84 85 85 85 85 85 85 85 85 84 84 84 76 76 76 76 W

00

00

z

Y

Ni-concanavalin A

Co-concanavalin A

Protein

Table 8 (continued)

2-OMe aManOMe 2-OEt aManOMe 2-OPr aManOMe 2-OBn aManOMe 3-OMe aManOMe 2,3-diOMe aManOMe 3-OMe aMan(l-3)-2-OBn aManOMe aGlcOMe 5 aMan (lk2)aManOMe aMan(1-3)aManOMe aMan(144)aManOMe aMan(1-6)aManOMe aGlc(1-4)GlcOH aGlc(1+6)GlcOH aMan(l+3)[aMana(l+6)]-aManOMe aManOMe ManOH aGlcOMe GlcOH aManOMe ManOH aGlcOMe GlcOH

2FGlcOH 3FGlcOH aManOMe

Ligand" 6.5 3.8 6.8 6.8 4.8 2.8 4.2 6.2 5.3 6.8 7.5 5.3 5.9 7.0 7.4 7.4 6.9 3.7 4.6 10.2 6.4 5.3 4.8 3.6 6.2 4.0 4.5 3.4

AH^ -3.0 -0.7 -1.5 -1.5 0.4 3.4 1.2 -0.2 -0.3 -2.0 -1.7 -0.7 -0.7 -0.2 -1.4 -2.3 -1.6 0.7 -0.2 -2.8 -0.9 -1.0 -0.1 0.1 -0.6 0.6 0.0 0.3

TAS~

3.5 3.1 5.3 5.3 5.2 6.2 5.4 6.0 5.0 4.8 5.8 4.6 5.2 6.8 6.0 5.0 5.3 4.4 4.4 7.4 5.5 4.4 4.6 3.1 5.6 4.6 4.5 3.6

-AGb

-93.0

-44.0

-110

40.8 -9.6 -50.0

ACpc

76 76 36 86 86 86 86 86 86 86 86 36 36 36 36 36 36 36 36 36 87 87 87 87 87 87 87 87

Ref.

10

co

g'

'

5

$

9

& $

'$

%'

Z

2

.1"

$

1

3 g.

2

0

Dioclea grandi3ora

concanavalin A (pH 5.2)

Cd-concanavalin A

a M a n ( l i 3 ) [ a M a n (lh6)laManOMe

aMan(1-4)ManOMe aMan(1+6)ManOMe

aMan(143)ManOMe

aGlcCally1 aManCal1yl a2dGlcOMe aMan (1+2)ManOMe

aManOMe

aManOMe ManOH aGlcOMe GlcOH ManOH aMan(1-3)ManOH aMan (146)ManOH aMan(l+3)[aMan (1+6)]aManOMe aGlcOMe

7.5 5.5 4.6 3.9 5.0 8.4 7.2 14.2 4.4 5.0 4.4 7.8 8.2 7.8 2.4 6.3 7.5 9.9 6.4 10.1 11.4 8.4 8.4 8.6 16.2 13.0 15.2 11.0 14.6 15.1 12.8 10.0 10.4 9.8

-2.2 -1.0 -0.1 -0.1 -0.5 -2.7 -1.6 -6.8 -0.2 -0.3 -0. I -1.5 -3.3 -2.9 2.0 -1.7 -2.9 -3.8 -0.5 -4.3 -5.1 -3.5 -3.3 -3.6 -7.9 -4.8 -7.2 -4.5 -6.8 -7.1 -5.5 -4.0 -4.5 -4.0

5.3 4.5 4.5 3.8 4.5 5.6 5.7 7.4 4.2 4.7 4.2 4.8 4.9 4.8 4.4 4.6 4.6 6.1 5.9 5.8 5.5 5.0 5.1 5.1 8.3 8.2 8.0 6.5 7.8 8.0 7.3 6.0 5.9 5.8 -96.0

-22.0

-40.0

-56.0

87 87 87 87 88 88 88 88 81 84 36 81 84 36 81 81 84 84 36 84 36 36 84 36 84 36 84 84 84 84 84 84 84 84

Erythrina corallodendron

Protein

Table 8 (continued)

7.3 6.8 5.7 6.4 4.7 4.7 5.2 4.3 4.2 3.6 3.9 3.9 4.2 4.4 4.3 4.4 4.4 4.2 4.3 4.3 4.4 4.7 4.5 5.0 5.8 5.5 5.4 5.0 5.0 4.9

-7.5 -6.0 -4.2 1.8 -1.4 -1.6 -1.7 -0.1 -0.2 -0.8 -1.0 -0.5 -2.9 -1.2 -1.2 0.9 1.1 -1.1 -0.9 -2.5 -1.9 -5.7 -5.4 -5.9 -6.8 -5.5 -6.0 -6.3 1.1 0.6

14.8 12.8 9.9 4.6 6.1 6.3 6.9 4.4 4.4 4.4 4.9 4.4 7.1 5.6 5.5 3.6 3.3 5.3 5.2 6.8 6.3 10.4 9.9 10.9 12.6 11.0 11.3 11.3 3.9 4.3

aMan (1 -+3)[aMan(1+6)]-2daManOMe aMan (1+3)[aMan (1+6)]-4daManOMe 3,4daMan (1+3)[aMan(l+6)]2,4daManOMe

pGal(1+4)PGlcNAcOMe aFuc( 1+2)PGal( 144)GlcOH

PGa1(1-+4)GlcNAcOH

PGalNAcOMe PGa1(1-+4)GlcOH

aGalOMe

GalOH

GalNAcOH

aGlc ( I -4)GlcOH aGlc (1 +6)GlcOH 5 aGlc (144)GlcOH aGlc( 1+6)GlcOH PGalOMe

4

-AGb

TAP

-AHb

Ligand"

93.2

ACp'

84 84 84 84 83 83 36 36 36 89 90 90 89 90 YO 90 90 90 90 89 89 90 90 89 89 90 90 89 90 90

Ref.

5'

f

9 $ $

2

$ & 2

o

3

'2.

3

2

2

$

$

1

g.

2

2

\D h,

co

Gal- 1 (dimer)

FGF-1 cyclic mimic FGF-1 Native Binding Site FGF- 1: D-Pro 136 Gal-1 (Bovine Spleen)

Erythrina indica

Erythrina cristagalli

bGal(1+4)PGlcNAcOMe Heparan Sulfate Heparan Sulfate Heparan Sulfate PGdl (1-4)Gk pGal(1-4)pGlcOMe pGa1(1-+4)FrucOH pGd( 144)ManOH pGal(1t3)AraOH pGal( 1-3)GlcNAcOH 2-OMe pGal (lh4)GlcOH Galp( 144)GlcNAcOH Thio Galp(l+l)PGal OGal(1+4)GlcNAcOH Dithiogalactoside

PGal(1+4)PGlcNAcOMe PGalOMe GalNAcOH PGalNAcOMe pGa1(114)GlcOH

PGalOMe GalNAcOH PGalNAcOMe pGa1(1+4)GlcOH pGal(1+4)GlcNAcOH

FucOH

aMe-N-dansylcalaminide

1.1 4.7 7.3 6.8 6.0 10.9 12.8 10.3 5.9 4.9 5.2 10.4 7.4 13.7 10.9 10.4 24.4 23.2 22.9 5.0 7.1 8.2 8.4 8.8 7.1 7.3 7.8 11.1 6.6 3.8

1.o

4.4 5.5

2.9 2.1 2.6 2.5 -0.7 -2.9 -2.5 -1.1 -5.4 -6.5 -4.8 -2.1 -0.7 -0.8 -5.6 -2.7 -7.8 -5.1 -4.8 -16.9 -15.9 -15.9 0.2 -2.0 -2.9 -3.2 -3.7 -1.3 -1.5 -2.3 -5.6 -1.4 0.9

7.3 7.6 3.6 3.6 4.1 4.4 4.3 4.9 5.6 6.3 5.5 3.8 4.2 4.4 4.8 4.7 5.9 5.9 5.7 7.5 7.3 7.1 5.2 5.0 5.4 5.5 5.1 5.7 5.8 5.5 5.5 5.2 4.8 66.3 6.7 12.5 -48.6 -20.0 -35.1 -29.1 -57.1 -82.4 -88.4

-244

-200

105

89 89 89 89 89 91 91 91 75 75 75 75 75 75 75 75 75 89 89

90 90 90 90 89 89 89 89 89 89 89 89 89 89

w

\o

co

=. 2

f:

3

%

9

.1"

8.A

2

q

$,2

'

rp

3 3

$

GI : Gly 174Cys

G1: Thr173Gly

G 1: Gln 172Asn

GI: Asnl71Ser

GI: Trpl20Phe

GI: Arg54Lys GI: Arg54Leu GI: Serll9Tyr

GI: Trp52Phe

GI: Tyr5OPhe

Gal-] triple mutant Glucoamylase G1

Gal-I C2S mutant

Protein

Table 8 (continued)

~~

~

Acarbosed 1-Deoxynojirimycin Acarbosed 1-Deoxynojirimycin Acarbosed Acarbosed Acarbosed 1-Deoxynojirimycin Acarbosed 1 -Deoxynojirimycin Acarbosed 1-Deoxynojirimycin Acarbosed 1-Deoxynojirimycin Acarbosed 1-Deoxynojirimycin Acarbosed 1-Deoxynojirimycin

9

pGal(1-+4)GlcNAcOH Dithiogalactoside PGal(1+4)GlcNAcOH 1-Deoxynojirimycin Acarbosed Methyl a,&acarviosinide P-cyclodextrin 6 7 8

Ligand" 2.8 2.6 0.6 1.7 7.9 7.4 12.9 20.2 21.4 22.5 16.5 9.5 2.4 12.7 4.5 9.9 8.6 7.2 2.2 6.9 6.9 12.3 3.6 11.3 3.3 12.8 4.1 8.4 2.4

-AHb 1.9 2.1 4.8 4.8 8.6 2.1 -6.5 -10.2 -11.2 -11.8 -5.6 7.3 3.9 2.9 0.6 -0.4 0.4 9.3 4.5 2.7 -0.7 4.3 2.9 5.8 2.8 4.0 2.4 8.8 4.0

TASb 4.8 4.7 5.4 6.4 16.5 9.5 6.4 10.0 10.2 10.7 11.0 16.8 6.2 15.6 5.1 9.5 9.0 16.4 6.7 9.6 6.2 16.7 6.5 17.0 6.1 16.8 6.5 17.2 6.4

-AG~ ACPC

~

89 89 89 53 53 53 55 55 55 55 55 56 56 56 56 56 56 56 56 56 56 56 56 56 56 56 56 56 56

Ref.

5.

@

2

,"

-

2 g.

&

6'

'2

&

2

3

2

$

g

Q

1

2 2 g.

2

P

v,

m

Lentil Lectin

Hevein

Gyrase B

G2: Tyrl75Phe

Glucoamylase C 2

G1: Trp317Phe GI Starch Binding Domain

G1: Arg305Lys G1: Asp309Glu

G1: Serl85His

G1: Glul80Gln

G 1: Asp 176Asn

8 9 Acarbose' 1-Deoxynojirimycin Clorobiocin Novobiocin PGlcNAc( 1-4)GlcNAcOH PGlcNAc( 1+4)PGlcNAc( 1-+4)GlcNAcOH aGluOPh ldGlu aGluF

7

6

Acarbosed 1-Deoxynojirimycin Acarbosed 1-Deoxynojirimycin Acarbosed 1-Deoxynojirimycin Acarbosed Acarbosed 1-Deoxynojirimycin Acarbosed GlcSGlc2 GlcSGlc3 GlcSGlCd P-cyclodextrin 1-Deoxynojirimycin Acarbose' D- Gluco-dihydroacarbose L-Zdo-dihydroacarbose aSGlc( 1+4)aGlcOMe cLSGlc(1+4)BGlcOMe aS,SGlc( lh4)aGlcOMe

9.4 3.2 4.7 3.4 13.0 5.4 7.3 1.4 2.8 8.5 9.1 17.7 11.1 13.1 2.7 9.7 7.1 2.3 1.4 3.3 1.4 8.0 7.6 6.5 9.9 8.9 2.9 8.8 9.3 4.9 6.0 2.6 3.6 5.9

6.0 2.8 1.4 2.4 2.2 0.4 -0.5 6.1 3.7 2.4 3.4 13.2 6.4 6.6 3.4 6.6 3.2 5.0 4.8 1.9 3.4 >3.0 >3.4 >4.3 >1.1 9.2 2.5 1.o 0.8 -1.2 -0.2 0.7 -1.1 -3.4

15.4 6.1 6.1 5.8 15.2 5.0 6.8 14.2 6.5 10.9 4.3 4.5 4.7 6.5 6.2 16.3 10.3 7.3 6.2 5.2 4.8 >11.0 >11.0 >I03 >11.0 18.0 5.3 9.8 10.1 3.8 5.8 3.3 2.4 2.5 -23.4

-64.5 -83.7 - 12.1 -9.0 -44.7

-200

56 56 56 56 56 56 56 56 56 56 54 54 54 54 53 53 53 53 53 53 53 55 55 55 55 56 56 92 92 93 93 76 76 76 s

4 g.

3

?

e

$

Y

Ricin cornrnunis Agglutinin

Peptide: YYWIGIR-NH2

Pea Lectin

MBP (E. coli)

MBP-C (rat liver)'

MBP-A (rat serum).

Protein

Table 8 (continued)

ThiodiGal

PGal(1+4)GlcOH

OMeGlcNAc NAcYD(G-(CH&-Man)2 aGlc( 1+4)GlcOH 0-cyclodextrin aGlcOPh 1dGlu aGluF 2dGluOH 2FGluOH 3FGluOH 30MeGluOH sLeX LeX aNeuAc(2-.3)PGal( 1-t4)GlcNAcOH pGa1(I +4)GlcNAcOH

2dGluOH 2FGluOH 3FGluOH 30MeGluOH aManOMe OMeGlcNAc NAcYD(G-(CH2)6-Man)2 aManOMe

Ligand" 4.3 3.7 3.5 2.9 4.7 5.2 5.2 5.1 0.0 4.1 7.4 2.8 3.0 3.0 4.3 4.3 5.0 3.7 3.0 3.4 73.7 18.6 3.2 6.9 6.6 7.6 8.1 7.5 12.2 12.0

AH^ -1.7 -0.6 -1.3 1.6 -0.9 -1.4 -1.4 -1.3 0.4 -0.9 -1.5 10.9 4.4 0.9 -1.4 -1.4 -1.9 0.0 0.0 1.2 -67.3 -12.4 0.9 -0.89 -0.6 -1.7 -1.6 -2.1 -6.8 -6.5

TAS~

2.6 3.1 2.2 4.5 3.8 3.8 3.8 3.8 0.4 3.8 5.9 8.1 1.4 3.9 2.9 2.9 3.0 3.6 2.8 4.6 6.4 6.3 4.1 6.0 6.1 6.0 6.0 6.0 5.4 5.5

-AGb

-207 162 -4.5 11.0 -5.0 -40.3 -46.3 8.7 -1.8

-4.8 -33.4 -17.4 -53.5

ACpc

76 76 76 76 94 94 94 94 94 94 94 95 95 76 16 76 76 76 76 76 96 96 96 97 97 91 97 91 97 91

Ref.

Soybean Agglutinin

Snowdrop Lectin

Sheep Spleen Lectin (L-14)

Se 155.4

PGalOMe

aMan(l+3)[aMan(l+6)]uManOMe

GalOH aGalOMe PGalOMe aMan(1 i 3 ) u M a n O M e aMan(1+6)aManOMe

MumaGal

MumSLac

pGal(144)GlcNAcOH

ThiodiGal

14

12 13

11

MumbPGalNAc GalOH 10

SGalOMe

uGalOMe

MumbPGal

MumbaGal

5.8 5.2 9.5 10.4 5.3 5.3 5.4 5.5 7.2 5.8 6.8 8.8 8.2 8.6 7.0 12.9 11.5 10.0 10.5 8.6 10.3 8.8 8.6 2.4 2.2 0.6 0.7 0.5 3.1 2.1 3.9 10.6 8.9

-1.4 -0.8 -3.6 -4.6 -0.9 -1.2 -0.3 -0.8 -1.7 -1.4 0.3 -1.8 -1.2 -1.5 0.7 -1.4 -5.5 -3.8 -4.3 -2.9 -4.5 -3.6 -3.4 2.4 2.9 -1.7 -1.7 -1.7 1.7 2.1 0.9 -6.9 -4.9 4.4 4.5 5.9 5.8 4.3 4.4 5.0 4.8 5.5 4.3 7.1 1.0 7.0 7.0 7.6 5.5 6.0 6.2 6.2 5.7 5.7 5.3 5.3 4.8 5.0 2.4 2.4 2.2 4.8 4.2 4.8 3.1 4.0 -93.2

99 99 99 99 99 36 36 36 89 89

99

98 99 99 99 99 99

91 97 97 97 97 97 97 97 97 97 98 98 98 98

W 4

00

'3

$

is9 %

g.

2

q

$,2

3

'2

R

9

9.5 13.9 12.1 5.5 8.2 1.5 4.7 6.3 5.1 5.1 6.2 5.9 3.9 5.0 6.5 5.8 6.1 7.0

GalNAcOH PGalNAcOMe

PGal( 1-4)GlcOH PGal( 1-4)GlcNAcOH pGal( 1-14) PGlcNAcOMe PGlcNAc( 1+4)GlcNAcOH PGlcNAc( I -+4)PGlcNAc( 1+4)GlcNAcOH PGlcNAc( 1+4)[ PGlcNAc( 1+4)]2GlcNAcOH PGICNAC(1+4)[ PGlcNAc( 1 +4)]3GlcNAcOH GalOH 1 HGal 2HGal FucOH alFGal 2FGalOH 6FGalOH 20MeGalOH

-AHb

Ligand" -4.1 -8.0 -5.9 -2.4 -4.3 -3.8 -0.8 -1.2 0.5 0.8 -1.8 -1.6 -0.7 -1.5 -2.1 -1.6 -1.9 -1.8

TASb

3.9 5.1 5.6 5.9 4.4 4.3 3.2 3.5 4.4 4.1 4.3 5.2

3.1

5.4 6.0 6.2 3.2 3.9

-AGb

-36.4 8.4 -23.2

-13.1 -3.0 -6.0 24.9

- 100

ACp'

100 100 101 101 101 101 101 101 101 101

100 100

89 89 89 89 89 89

Ref.

"structures of ligands 1-14 are shown in Attachment to Table 8; d = deoxy; reducing sugars are assumed to be equilibrating mixtures of anomers bkcal mol-' cal mol-' deg-' by displacement data obtained by fixing n

Winged Bean Agglutinin I

Urtica dioica Agglutinin'

Protein

Table 8 (continued)

zri

6' 3

5-9

$.

0

n

E ??

2

g.

2

6'

2

g

2

B

.$

g

1

2 g.

2

w \o w

899

31.6 The Role of Multivalency in Protein-Carbohydrate Interaction Attachment to Table 8: Ligands 1-14. OH

0

0

"3 0

0

HO&OEH

HO

OH

H z + H o o p ~ H

0

0

p

OH HO OH

1

2

p1-2

GlcNAc -M a n q l - 6 ManPI-2 / GlcNAc -Man a?-3

OH

5

4 -OH

HyA:

HO

3

900

31 Protein-Carbohydrate Interaction: Fundamental Considerations

Table 9. Acarbose binding by displacement method [53]. Enzyme

Ligand

K"

G2

1-deoxynojirimycin acarbose (apparent)" acarbose 1-deoxynojirimycin acarbose (apparent)" acarbose methyl a$-acarviosinide acarbose (apparent)d acarbose

3.30 2.70 x 8.80 x 4.70 x 2.00 x 9.40 x 7.80 x 3.80 2.90 x

G1

G1

104 107 10" 104 107 10"

lo6 104 10"

-AGb

-AHh

TAS~

6.2 10.1 16.3 6.4 10.0 16.5 9.5 6.3 15.8

2.7 7.0 9.7 1.7 6.2 7.8 7.4 1.7 9.1

3.5 3.1 6.6 4.8 3.9 8.6 2.1 4.6 6.6

aM-1

kcal mol-' Inhibited by 1-deoxynojirimycin Inhibited by methyl a$-acarviosinide

coamylase G2 proceeds with a free energy of binding of 16.3 kcal mol-' and a ACp of -200 cal mol-' deg-' [53].Similarly, the binding of maltose to maltose binding protein shows a free energy of binding of 8.14 kcal mol-' and an accompanying change in molar heat capacity of -207 cal mol-' deg-' [95]. It seems likely that the source of enhanced affinity is highly efficient removal of solvent from poorly solvated surfaces. Sigurskjold and coworkers have utilized a displacement strategy to evaluate the binding thermodynamics of the highest affinity protein-carbohydrate complexes known, that of acarbose and Aspergillus glucoamylase (Table 9) [53]. The high affinity of this complex precludes direct calorimetric titration of the protein with ligand. Instead, this group takes advantage of the additivity of binding thermodynamic parameters and displaces the moderately binding ligand 1-deoxynojirimycin with the high affinity ligand acarbose. In an appropriate concentration domain, the enthalpy of acarbose binding is roughly equivalent to the sum of the enthalpy of 1-deoxynojirimycin binding and the enthalpy of 1-deoxynojirimycin displacement by acarbose. Similarly, the free energies of binding are additive, requiring that the complex stability constants are mulitiplicative. The effect of metal ion identity on the thermodynamics of concanavalin Aoligomannoside binding has been investigated [87].The role of metal ion in legume lectin binding has long been questioned; although metal ions are required for binding they do not appear to contact saccharide ligands. Through the series cobalt nickel and cadmium only minor alterations in enthalpies and free energies of binding were observed, suggesting a primary role for metal ion in the maintenance of protein structure rather than in ligand binding. Previously, we and others have commented on the similarities and differences between patterns of saccharide association for lectins and antibodies [3]. Unfortunately, the database of thermodynamic parameters for antibody-carbohydrate complexation remains small. While important structural and energetic differences may exist that provide clues to the origin of both affinity and association in aqueous solution, it is impossible to reach any conclusions at this time.

31.6 The Role of Multivalency in Protein-Carbohydrate Interaction

901

31.6 The Role of Multivalency in Protein-Carbohydrate Interaction The tremendous potential of saccharide ligands for the treatment of a wide range of human diseases continues to drive much of contemporary glycoscience. The major impediment to the use of carbohydrate or carbohydrate mimetics continues to be the low affinity of saccharide ligands for their lectin receptors. As we note above, the inherently low enthalpy and free energy available from desolvation during protein-carbohydrate complexation likely places fundamental limitations on the affinity that might be achieved through modification of monomeric ligands. Lectins are seldom found in vivo as monomeric species; rather, they are aggregated into higher order oligomeric structures. A reasonable conclusion drawn from this observation is that nature takes advantage of the remarkable diversity of carbohydrate structures to achieve high selectivity, while overcoming the low inherent affinity through the use of multivalency. From this concept an enormous array of multivalent carbohydrate ligands has been synthesized and evaluated in a range of binding assays. Many such ligands show remarkable enhancements in activity compared to an equivalent concentration of the corresponding monovalent ligand; indeed some show affinity enhancements as large as lo9 on a valence corrected, or per mole of saccharide, basis. In most instances the physical basis of the observed effects is unclear. Here, we review briefly the phenomonology of multivalency effects in protein-carbohydrate interaction, then consider the thermodynamic consequences of tethering recognition domains. The goal of this exercise is to provide a molecular basis for the ‘cluster glycoside effect’ and evaluate its potential for therapeutic application.

3 1.6.1 Phenomenology The ‘cluster glycoside effect’, as defined by Y. C. Lee, was first observed using the hemagglutination assay [ 11. Evaluating several hepatic lectins against several monosaccharide types tethered by amino acids, Lee noted the ZCso of these multivalent neoglycoconjugates was lower than expected based on the saccharide content of the ligand (Figure 6). Since this time an enormous variety of multivalent ligands have been prepared and bound (Table 10) [ 102, 1031. The magnitude of the cluster glycoside effect varies over nine orders of magnitude across the range of neoglycoconjugate structures that have been evaluated. Some of the first multivalent ligands reported are a group of glycosylated polyacrylamides. These ligands are prepared by copolymerization of acrylamide and an acrylic acid ester [ 1041. Two conceptual routes have been reported, varying in the nature of the acrylic acid ester. In one instance the ester contains the saccharide recognition domain; in the other the acryloyl ester incorporates an Nhydroxysuccinamide ester that is later displaced by a saccharide recognition domain tethered through an amine spacer (Figure 6). The molecular weight range achieved with these methods spans 10,000-450,000 Da, with larger polymers typically formed by incorporating the saccharides during

precipitation

4.7 x 10' 2.6 x 103 5.1 x 10' 1.8 x lo2 2.6 x 10' 6.6 x 10' 1.4 103 1.0 107 5.6 107 6.0 x 10' 1-2 x 100 8.3 x lo2 4.9 x 100 1.4 107 3.3 103 1.7 103 2.6 105 2.0 x 10' 1.2 x 102

2 2 3 12 4 16 16 6 6 4 2 2 2 270 7 7 3 105 25 4

Chicken hepatic lectin Chicken hepatic lectin LimaxJiavus lectin Linzax Jiauus lectin Concanavalin A Pea lectin Human mannose receptor Human mannose receptor Strepococcus suis Viral hemagglutinin Intact influenza virus Immobilized E-selectin Viral hemagglutinin Cholera toxin B subunit Heat labile enterotoxin Concanavalin A P-selectin Concanavalin A

"calculated by dividing K, for multivalent ligand by K, for monovalent refrence compound Hemagglutination Assay Enzyme Linked Immunosorbent Assay Enzyme Linked Lectin Assay

precipitation precipitation ELLAd ELLA ELLA ELLA Displacement Displacement HA HA HA ELLA Fluorescence ELLA variation ELLA variation HA Inhibition HA

47 41 42 42 106 106 107 108 108 109 110 111 111 112 113 41

1 1

1

104 104 104 104 105

HA^ HA ELISA" ELISA precipitation

5.0 x lo6 5.0 x 105 1.0 x 106 1.6 x 10' 1-2 x 100

~

200-2000 200 2000 200-2000 200- 2000 2-3

Viral hemagglutinin Viral hemagglutinin Viral hemagglutinin Viral hemagglutinin Serum-type mannose binding protein rat hepatic lectin

Sialylated glycopolymer; pre-P Sialylated glycopolymer; co-P Sialylated glycopolymer; pre-P Sialylated glycopolymer; co-P Bivalent and trivalent mannosylated glycoclusters Bivalent glycopeptide based NAcYD(GG-ah-GlyC) Bivalent glycopeptide based YEE (ah-GlyC) Tris-GlcNAc glycocluster Sialylated PAMAM glycodendrimer Sialylated PAMAM glycodendrimer Mannosylated PAM AM glycodendrimer Mannosylated PAMAM glycodendrimer Lysine-based mannosylated glycoclusters Lysine-based mannosylated glycoclusters Galabioside glycoclusters Bivalent sialosides Bivalent sialosides sLeXdimer Lyso-GM3-PGA glycopolymer oligo-GM1 propylene-imine glycodendrimer oligo-GM1 propylene-imine glycodendrimer C-mannosyl. Glycopolymer 3', 6' sulfated galactosyl glycopolymer Mannosylated glycodendrimer

Ref.

Assay Method

p"

Ligand valency

Lectin

Ligand

Table 10. Representative binding enhancements through multivalency.

31.6 The Role of’A4ultiualency in Protein-Carbohydrate Interaction

903

Figure 6. Representative multivalent ligands. From top left: mannosylated ROMP polymers; acrylamide/acryloyl ester copolymers; peptide-based dendritic ligands. See text for descriptions.

the copolymerization, rather than adding the saccharides to an activated backbone. Several other groups have reported a similar strategy [ 114-1 191. The performance of glycopolymers in agglutination assays is remarkable; indeed, the largest cluster glycoside effects observed to date are for the acrylamide polymers. Spaltenstein and Whitesides observed a reduction of 1O5 in IC50 for polyacrylamide glycopolymers bearing a-sialic acid residues compared to monosaccharide assayed against influenza viral hemagglutinin, placing minimum inhibitory concentrations in the nanomolar range [ 1201. Remarkably the efficacy of polymeric inhibitors varied both as the content of ligand and as the method by which it was incorporated: IC50 values for glycopolymers prepared by the activated backbone method were 75 times lower than equivalent sized glycopolymers prepared by copolymerization of sialic acid acrylamide monomers [ 1211. Apparently the mechanism by which these polyvalent ligands inhibit hemagglutinin-mediated agglutination is dependent on both the content and spatial orientation of saccharide epitopes. Polyacrylamides bearing N-acetyllactosamine, prepared by Tsuchida and co-workers, showed ZC50 values enhanced by roughly lo3 relative to unmodified monovalent ligand in agglutination assays against plant lectins [ 1181.

904

31 Protein-Carbohydrate Interaction: Fundamental Considerations

Kiessling and coworkers have prepared a range of glycopolymers using a ringopening metathesis polymerization (ROMP) strategy (Figure 6) [ 1121. Again, enhancements in affinity near lo3 on a valence-corrected basis are observed in agglutination assays. Significantly, greatly reduced enhancements in apparent affinity are observed when the ligands are assayed by other methodologies. A surface plasmon resonance study of the binding of mannosylated ROMP ligands to concanavalin A showed enhancements of five to 40-fold [122]. Kiessling and coworkers have also demonstrated a size dependence during glycopolymer inhibition [ 1231. Mannosylated glycopolymers showed enhancements in the range of 50-3,000 compared to monosaccharide, increasing exponentially with polymer chain lengths up to 143 units in agglutination assays against concanavalin A. Increasing chain lengths beyond this value failed to provide a continued valence corrected enhancement, although overall ligand affinities apparently continue to increase. With very large polymers (MW lo6) enhancements near lo5 were observed. The cell surface presents a multivalent display of carbohydrate exquisitely suited to maximize protein-carbohydrate binding while minimizing entropic penalties. On the one hand, carbohydrate epitopes are free to orient themselves in two dimensions in a configuration that optimizes favorable contacts with the lectin. On the other hand, the bulk of the translational entropy and all of the rotational entropy has already been lost during placement of the ligand in the lipid bilayer. This situation is mimicked by a variety of synthetic glycosylated liposomes and, like glycopolymers, glycosylated liposomes show remarkably enhanced performance in agglutination assays compared to an equivalent concentration of the corresponding monovalent ligand. The general method for synthesis of glycosylated liposomes is: i) conjugation of the glycosyl moiety to a lipid chain; ii) incorporation of some amount of this glycolipid into a mixture of phosphatidyl choline and cholesterol; and iii) sonication or extrusion through a membrane to form the liposome [124]. Kingery-Wood and co-workers produced a series of sialic acid containing liposomes with mole fractions of sialic acid (compared to phosphatidyl choline and cholesterol) of 2-25%. By hemagglutination assays, a maximum inhibition was observed at 5% sialic acid incorporation, with ZC50 values in the nanomolar range. DeFrees and co-workers repeated this procedure with the sialyl Lewis' epitope [ 1251. Ketis et al. incorporated the glycoprotein glycophorin into liposomes as a 33% component for binding assays with wheat germ agglutinin [126]. Charych and co-workers have developed an innovative methodology for glycoliposome production [ 127- 1291. Incorporation of diacetylenic lipids followed by UV (254 nm) irradiation yielded a highly colored (blue) polymerized liposome species. The incorporated chromophore provides a unique binding read-out: upon lectin binding the liposome turns red, presumably as the lipid bilayer adopts a conformation optimal for ligand binding. A much larger group of multivalent ligands include the so-called glycoclusters and glycodendrimers. These ligands are typically smaller in size than the polymers, and require significantly greater synthetic effort. On the other hand, the resulting species are homogenous in nature and more easily characterized than polymeric

31.6 The Role of Multivalency in Protein-Carbohydrate Interaction

905

ligands; additionally, greater control is achieved over the spacing and orientation of glycosyl moieties. A complete description of this group of ligands is beyond the scope of this document; rather, the reader is directed to several recent reviews of the field [102, 1031. In contrast to the remarkable performance of glycopolymers, glycoclusters and glycosylated dendrimers provide much more modest enhancements in affinity. With valences ranging to 36, typical valence-corrected enhancements in affinity are on the order of 10-103, although some exceptional enhancements have been reported. Hansen and co-workers reported increases of 102-103 in ZCso values for several low-valency galabiose inhibitors relative to unmodified galabiose against the Grampositive bacterium Streptococcus suis [ 1071. Nonetheless, this enhancement raised the approximate range of binding efficacy from micromolar to nanomolar, approaching ranges required for therapeutic utility. Knowles and coworkers observed a 100-fold increase in the relative potency for a series of dimeric sialosides compared to unmodified sialic acid when evaluated against the influenza virus in agglutination assays [ 1081. A peptide-based mannosylated glycocluster reported by Biessen and coworkers showed a cluster glycoside effect of lo6 relative to mannose when evaluated against human mannose receptor; this dramatic increase in relative inhibitory potential is one of the largest seen with non-polymeric systems [ 1061. Again, the effect is also a function of the analytical methodology utilized, and apparent enhancements in binding measured by ELLA are much smaller than those seen in hemagglutination assays 142-501. For example, many of the glycosylated PAMAM dendrimers, with valencies ranging over 2-1 6, show enhancements in apparent affinity relative to the unmodified saccharides ranging from 2- to 15-fold when considered on a per saccharide basis 147, 1301. Similarly, Ashton and coworkers report only modest enhancements in affinity during evaluation of glycodendrimers based on a combination aliphatic/aromatic core [ 1311. 31.6.2 The Energetic Consequence of Ligand Linkage

Having considered the reported phenomonology of multivalency effects in proteincarbohydrate interaction, we now aim to understand the origin of the cluster glycoside effect at a molecular level. To the extent that a set of design principles relevant to high affinity can be extracted, multivalency is among the most promising strategies towards the development of therapeutically useful carbohydrates. We consider here only a brief treatment of multivalency effects in aqueous association; a more complete treatment of the topic has recently appeared [ 1031. A meaningful discussion of multivalency in ligand binding first requires a precise definition of the terms involved. Consider the binding of N monovalent ligands versus a ligand of valency N, LN, to an N-valent lectin. Assuming the binding sites are equivalent and non-interacting, thermodynamic parameters describing the association of each monovalent ligand can be represented as AJN. The corresponding terms for association of the multivalent ligand are related to those for binding of the monovalent ligand by the expression:

A J N = NAJN + AJi

906

31 Protein- Carbohydrate Interaction: Fundamental Considerations

where AJi is an interaction term, describing the energetic consequences of physical linkage [68]. A discussion of multivalency effects next requires distinction be drawn between an overall affinity for the multivalent ligand and affinity on a per saccharide basis. These concepts have been variously termed functional afinity and intrinsic afinity, or avidity and ufinity, respectively [103, 1321. Whitesides and coworkers provided a more precise rendering of these concepts by defining the quantities a and b as:

or, in free energy terms:

Defined in this way, c1 is a measure of what is traditionally referred to as positive cooperativity; in the analysis of Jencks, values of a greater than unity correspond to favorable (i.e. negative) interaction free energies. Determination of a values require knowledge of the number of ligands bound. Since in most instances this information is not known, the quantity fl describes simple phenomenology:

At its simplest level enhanced affinity from multivalency requires only that jAGNI > IAGI, where AGN represents the change in Gibbs’ free energy upon binding of the N-valent ligand and A G represents the change in Gibbs’ free energy upon binding of each of the constituent monomeric ligands to an individual lectin binding site. Note that in this construct, the contribution of each monovalent recognition epitope to the overall binding free energy need not be greater than-or even equivalent to-that of monovalent ligand. The only requirement for the observation of enhanced affinity of the ligund is that IAJil be less than I(N - l)A&I; in such instances p will be greater than or equal to one. At a second level, a multivalency effect requires that the interaction free energy be favorable. This case describes what is normally referred to as positive cooperativity; here lAGNI > INAGNI. The key value describing the effect of ligand multivalency on binding thermodynamics is the interaction energy, AGi. A range of physical processes contribute to these interaction free energies; the concepts are intuitively more accessible if considered as separate enthalpic and entropic terms. Enthalpic Contributions to AGi The addition of a linker region has enthalpic consequences to overall binding free energies. First, a linker must be long enough to facilitate optimal placement of ligands within binding sites. Linkers too short or too rigid to facilitate optimal placement of recognition epitopes within binding sites have the effect of contributing unfavorably to interaction enthalpies. Assuming the recognition epitopes

31.6 The Role of Multivalency in Protein-Carbohydrate Interaction

901

can in fact bind successfully, the linker will almost certainly contact the surface of the protein at the periphery of the binding site, on the surface of the protein beyond the binding sites, or both. These contacts can be either favorable or unfavorable, contributing postively or negatively to interaction energies. The a priori prediction of favorable or unfavorable contributions to AHi is virtually impossible. Combinatorial approaches using, for example, peptide spacers may be a useful method of sampling wide ranges of linker-peptide interaction. Finally, the interaction of the linker region with solvent prior to binding leads to an energetic consequence during binding if the linker region is desolvated: this desolvation could make a favorable or unfavorable contribution to AHi, depending on the precise molecular details of the linker surface. Again, association in solution requires consideration both of the interactions between solutes and the interactions of both solutes with solvent prior to association. Entropic Contributions to AGi

Energetic consequences of ligand tethering are typically considered primarily in entropic terms; there is little doubt that there will be a significant contribution to binding free energies attributable to an entropic interaction term. Predicting the magnitude-or even sign-of the entropic consequence of tethering ligands, however, is not straightforward. As described above, an overall entropy of binding can be conceptually separated into a translational and rotational entropy, a conformational entropy, and a solvation-associated entropy. A similar exercise facilitates consideration of entropic contributions to ASi. In this construct, A& is considered as the sum of the differences in each entropic term for binding of the N-valent ligand and N monovalent ligands; that is:

where MNLNrepresents multivalent and monovalent complexes, respectively. The first term describes the entropic ‘savings’ that derives from minimizing the loss of translational and rotational entropy-the ability of a molecule to translocate in three dimensions and rotate on three axes-by converting N-ligands into a single species. Because translational and rotational entropies scale as the logarithm of the molecular size, AST+R for a multivalent ligand will be approximately equivalent to that of the corresponding monovalent species. As previously noted, the magnitude of this term is largely unknown. Table 6 above shows the wide range of values currently assigned to this term. The second term of the entropy decomposition describes the loss in conformational degrees of freedom during binding, typically the restriction of flexible dihedrals. In the context of interaction entropies, this term primarily arises from losses in degrees of freedom in a flexible linker region during binding. Although the magnitude of this term is somewhat unclear, the situation is considerably less opaque than for A S T + R . Some agreement seems to have emerged for a value near 0.5

908

31 Protein-Carboh ydrate Interaction: Fundamental Considerations

kcal mol-' near room temperature for complete restriction of a bond previously free to equilibrate among three energetically equivalent staggered conformers [ 133, 1341. Obviously not all bonds restricted during binding possess such freedom prior to complexation, and not all are completely constrained following binding. Linkers designed to span binding sites separated by 20-70 8, possess considerable conformational flexibility; the loss of this flexibility will unquestionably lead to a significant unfavorable contribution to ASl. Based on an entropic analysis, the prescription for the design of linker regions is clear. Linkers must be long enough to facilitate optimal placement of the saccharide recognition domains within the binding site, but with enough rigidity to minimize a conformational entropy penalty. An alternative strategy might be to make linkers of sufficient length that residual flexibility following binding minimizes the unfavorable contribution of ASconf to ASl. A number of studies on the role of flexibility in linker regions have been reported. Glick et al. evaluated a series of bivalent sialic acid ligands for the viral hemagglutinin, varying in length and linker flexibility [ 1081. A linker of intermediate flexibility provided the tightest binding as evaluated by hemagglutination assay. Unfortunately, without the required thermodynamic data, molecular interpretation of the result is impossible. Bundle and coworkers also explored the role of ligand flexibility, albeit in a somewhat different context [98, 1351. Two decades of intense study on small molecule recognition in organic solvent clearly demonstrated that preorganization and complementarity are key requirements of ligand structure for the observation of high affinity. From that observation, ligands with diminished conformational entropy, or preorganized ligands, should show enhanced affinity relative to ligands with freely rotating bonds. Monoclonal antibody Sel55.4 binds the trisaccharide methyl 2-(a-~-ga~actopyranosyl)-3-(a-~-3,6-dideoxyxy~opyranosy~)-~-~ mannopyranoside. This binding buries the dideoxyxylopyranose (abequose) moiety completely, and consequently freezes rotation about the AbeMan glycosidic bond. A series of ligands were prepared that preorganize the ligand by freezing this rotation (Figure 7). Binding of each restricted ligand proceeded with thermodynamic parameters remarkably unchanged from those of the native-presumably less ordered-ligand. The origin of this effect is unclear. On the one hand, the inherent flexibility of the localized bond is low; the exo-anomeric effect likely limits the entropic gain available. On the other hand, the tether used to preorganize the ligand may interact unfavorably in the complex. The role of ligand flexibility/preorganization in determining affinities has also been probed through the use of C-glycosides [ 136, 1371. Absent the anomeric oxygen and consequent exo-anomeric effect, C-glycoside ligands should show enhanced flexibility around the glycosidic bond relative to the corresponding 0-glycoside. This additional flexibility in the unbound ligand should enhance the conformational entropy loss upon binding. Again, the results to date fail to unambiguously demonstrate general design principles with regard to anomeric flexibility. A series of ligands that sequentially replace both the anomeric and pyranosidic oxygen with a methylene and thioether failed to yield a clear picture of structure-activity relationships. As with Bundle's work, however, the alteration in flexibility predicted to arise from such modification is small. Assuming an exo-anomeric effect of roughly

31.6 The Role of Multivalency in Protein-Curbolzydrute Interaction

909

R1=Oor S R2=(CH2), or CH2C6H4CH2

HO

Figure 7. Ligands designed to assess the role of conformational entropy in ligand binding. See text for description.

1-2 kcal mol-' and that the gg-configuration about the glycosidic linkage is precluded in both the C- and 0-glycoside, deletion of the anomeric oxygen alters the predicted rotomeric distribution about the anomeric linkage by roughly a factor of two. Furthermore, the uncertain role of changes in ligand solvation renders a molecular basis for changes in affinity opaque; further calorimetric study of such ligands may illuminate the relevant issues. Beyond concepts of linker flexibility, unpredictable values of ASsOlv and AHi

9 10

31 Protein-Carbohydrate Interaction: Fundamental Considerations

produce intractable design problems. The large potential energetic consequences of even small ligand modifications, coupled with the relatively weak interaction free energies that typify protein-carbohydrate interaction require a careful and stepwise exploration of the relationship between ligand structure and binding energetics. There likely do exist a simple set of rules to govern intelligent construction of multivalent ligands. Finally, it is imperative that the analytical methodology used to observe cluster glycoside effects be carefully considered before a molecular explanation of phenomenology is offered. 31.6.3 A Molecular Basis for the Cluster Glycoside Effect

Having considered the phenomenology of the cluster glycoside effect and the thermodynamic considerations surrounding aqueous association of multivalent ligands, we now turn briefly to possible molecular explanations for the affinity that apparently derives from multivalency. At this time, there are no recorded definitive examples of cooperativity in protein-carbohydrate binding deriving from favorable interaction energies. It seems most likely, then, that most cluster glycoside effects reported to date are in fact measures of the ability of multivalent saccharide ligands to drive the formation of some form of higher order aggregate. While formation of this aggregate might be favorable, providing a coupled equilibrium that would, in turn, enhance protein-carbohydrate affinities, there is no evidence to suggest that this is in fact true. Alternatively, Mammen, Choi, and Whitesides postulated a theory for the remarkable performance of glycopolymers in agglutination assays [ 1031. After binding several sialic acids to the viral hemagglutinin, these large, sterically demanding molecules may form a water-swollen, gel-like layer between the lectin and the erythrocyte surface. This barrier prevents additional binding from proceeding, but does not require thermodynamically enhanced saccharide-lectin binding. In some respects, protein-carbohydrate interaction is a poor system with which to study multivalency effects in aqueous association. The low monovalent affinities, poor analytical methodology for evaluation of binding, and the propensity of lectins to aggregate in the presence of multivalent saccharide ligands provides a daunting series of complications to the interpretation of binding data. More recently the study of protein-carbohdyrate interaction using bacterial two-component toxins suggests that some enhancement in affinity may be possible through multivalency [ 1381. These proteins, which do not aggregate polyvalent saccharides and show relatively small distances between binding sites, are more amenable to study. The enhancements in affinity available in these systems is not yet clear. Finally, we note that the appropriate analytical methodology and model of binding is dependent on the question asked. It may well be the case that in the context of biological systems, ligands especially capable of driving aggregation/ precipitation processes might act as useful agents. The concern, rather, is a precise definition of the issue at hand. With regard to ligand binding, in the commonly accepted definition, it appears unlikely that multivalency can afford high affinity through positive cooperativity.

Rejerences

9 11

Acknowledgments

The authors gratefully acknowledge the assistance of Mr. J. Lundquist and Ms. Aimee Butler for preparation of figures. Financial support for this work was provided by the National Institutes of Health (GM57179). References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.

12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39.

Lee, Y.C.; Lee, R.T. Acc. Chem. Res. 1995,28, 321. Dwek, R.A. Chem. Rev. 1996,96,683-720. Toone, E.J. Current Opinion in Structural Biology 1994, 4, 719-728. McClellan, A.L. Tables of Experimental Dipole Moments; Rahara Enterprises: El Cerrito, 1974. Rigby, M.; Smith, E.B.; Wakeham, W.A.; Maitland, G.C. The Forces Between Molecules; Oxford: New York, 1986. Zeegers-Huyskens, T.; Huyskens, P. Intevmolecular Forces; Huyskens, P. L.; Luck, W. A. P.; Zeegers-Huyskens, T., Ed.; Springer-Verlag: New York, 1991, pp 1-30. London, F. Z. Physik 1930, 63, 245. London, F. Z. Physik Chem. Leipzig 1930, 11, 222. Hunter, E.P.L.; Lias, S.G. J. Phys. Chem. Ref: Data 1998. 27, 413-656. Mirejovsky, D.; Arnett, E.M. J. Am. Chem. Soc. 1983, 105, 1112. Water: A Comprehensive Treatise, Vol. 1-7; Franks, F., Ed.; Plenum Press: New York, 1974. Muller, N. J. Solution Chem. 1988, 17, 669. Muller, N. Acc. Chem. Res. 1990,23, 23-28. Ben-Naim, A.M., R. J. Phys. Chem. B 1997, 101, 11221-11225. Soper, A.K.; Luzar, A.J. J. Phys. Chem. 1996, ZOO, 1357. Turner, J.; Soper, A.K. J. Chem. Phys. 1994, 101, 6116. Turner, J.Z.; Soper, A.K.; Finney, J.L. J. Chem. Phys. 1995, 102, 5438. Turner, J.; Soper, A.K.; Finney, J.L. Molecular Phys. 1990, 70, 679. Finney, J.L.; Soper, A.K.; Turner, J.Z. Pure Appl. Chem. 1993, 65, 2521. Leberman, R.; Soper, A.K. Nature 1995,378, 364. Reiss, H.; Frisch, H.L.; Lebowitz, J.L. J. Chem. Phys. 1959, 31, 369. Reiss, H.; Tully-Smith, D.M. J. Chem. Phys. 1971, 55. 1674. Reiss, H. Adv. Chem. Phys. 1966, 9, 1. Frisch, H L. Adv. Chem. Phys. 1963, 6, 229. Pierotti, R.A. A. Chem. Rev. 1976, 76, 711. Stillinger, F.H. J. Solution Chem. 1973, 2, 141. Ben-Naim, A,; Tenne, R. J. Chem. Phys. 1977, 67, 627. Lee, B. Biopolymers 1991, 31, 993. Lee, B. Methods Enzymol. 1995,259, 555. Fujita, T.; Iwasa, J.; Hansch, C. J. Am. Chem. Soc. 1964, 86, 5175. Iwasa, J.; Fujita. T.; Hansch, C. J. Med. Chem. 1965, 8, 150. Arnett, E.M.; McKelvey, D.R. in Solute-Solvent Interactions Coetzee, J.F.; Ritchie, C.D., Ed.; Marcel Dekker: New York, 1969, pp. 342. Oas, T.G.; Toone, E.J. Adv. Biophys. Chem. 1997, 6, 1-52. Muller, N. J. Solution Chem. 1991,20, 669. Chervenak, M.C.; Toone, E.J. J. Am. Chem. Soc. 1994, 116, 10533-10539. Chervenak, M.C.; Toone, E.J. Biochemistry 1995,34, 5685-5695. Lemieux, R.U. Chem. Soc. Rev. 1989, 18, 347-374. Simanek, E.E.; McGarvey, G.J.; Jablonowski, J.A.; Wong, C.H. Chem. Rev. 1998, 98, 833862. Huwe, C.M.; Woltering, T.J.; Jiricek, J.; Weitz-Schmidt, G.; Wong, C.H. Bioorg. Med. Chem. 1999, 7, 773-788.

9 12

31 Protein-Carbohydrate Interaction: Fundamental Considerations

40. Du, M.; Hindsgaul, 0. Synlett 1997, 395-397. 41. Dimick, S.M.; Powell, S.C.; McMahon, S.A.; Moothoo, D.N.; Naismith, J.H.; Toone, E.J. J. Am. Chem. Soc. 1999, in press. 42. PagC, D.; Zanini, D.; Roy, R. Bioorganic & Medicinal Chemistry 1996, 4, 1949-1961. 43. PagC, D.; Roy, R. Bioconjugate Chem. 1997, 8, 714-723. 44. PagC, D.; Roy, R. Glycoconjugate J. 1997, 14, 345-356. 45. Roy, R.; Tropper, F.D.; Romanowska, A. Bioconjugate Chem. 1992,3,256-261. 46. Roy, R. PagC, D.; Perez, S.F.; Bencomo, V.V. Glycoconjugate J. 1998, 15, 251-263. 47. Zanini, D.; Roy, R. J. Am. Chem. SOC.1997,119,2088-2095. 48. Tropper, F.D.; Romanowska, A.; Roy, R. Methods in Enzymology 1994,242,257-271. 49. Arya, P.; Kutterer, K.M.K.; Qin, H.; Roby, J.; Barnes, M.L.; Kim, J.M.; Roy, R. Bioorg. Med. Chem. Lett. 1998,8>1127-1132. 50. Garcia-Lopez, J.J.; Hernandez-Mateo, F.; Isac-Garcia, J.; Kim, J.M.; Roy, R.; SantoyoGonzalez, F.; Vargas-Berenguel, A. J. Org. Chem. 1999, 64, 522-531. 51. Grime, J.K. Analytical Solution Chemistry; John Wiley & Sons: New York, 1985; Vol. 79. 52. Wiseman, T.; Williston, S.; Brandts, J.F.; Lin, L.-N. Anal. Biochem. 1989, 179, 131. 53. Sigurskjold, B.W.; Berland, C.R.; Svensson, B. Biochemistry 1994, 33, 10191-10199. 54. Sigurskjold, B.W.; Svensson, B.; Williamson, G.; Driguez, H. Eur. J. Biochem. 1994, 225, 133-141. 55. Sigurskjold, B.W.; Christensen, T.; Payre, N.; Cottaz, S.; Driguez, H.; Svensson, B. Biochemistry 1998, 37, 10446--10452. 56. Berland, C.R.; Sigurskjold, B.W.; Stoffer, B.; Frnadsen, T.P.; Svensson, B. Biochemistry 1995, 34, 10153-10161. 57. Sturtevant, J.M. Proc. Natl. Acad. Sci. USA 1977, 74, 2236-2240. 58. Murphy, K.P.; Privalov, P.L.; Gill, S.J. Science 1990, 247, 559-561. 59. Murphy, K.P. Biophys. Chem. 1994,51, 311-326. 60. Doty, P.; Meyers, G.E. Discussions Faraday Soc. 1953,13, 51. 61. Dunitz, J.D. Science 1994, 264, 670. 62. Erickson, H.P. J. Mol. Biol. 1989, 206, 465-474. 63. Horton, N.; Lewis, M. Protein Science 1992, 1, 169-181. 64. Janin, J. Proteins Struct. Funct. Genet. 1995,21, 30-39. 65. Janin, J.; Chothia, C. J. Mol. Biol. 1976, 100, 197-211. 66. Janin, J.; Chothia, C. Biochemistry 1978, 17, 2943-2948. 67. Janin, J.; Chothia, C. J. Biol. Chem. 1990, 265, 16027-~16030. 68. Jencks, W.P. Adu. Enzymol. 1975, 43, 219-410. 69. Page, M.I.; Jencks, W.P. Proc. Natl. Acad. Sci. 1971, 68, 1678-1683. 70. Searle, M.S.;Williams, D.H. J. Am. Chem. Soc. 1992, 114, 10690-10697. 71. Searle, M.S.;Williams, D.H.; Gerhard, U. J. Am. Chem. Soc. 1992,114, 10697-10704. 72. Spolar, R.S.; Record, M.T., Jr. Science 1994,263, 777-784. 73. Tidor, B.; Karplus, M. J. Mol. Biol. 1994, 238, 405-414. 74. Williams, D.H.; Searle, M.S.; Mackay, J.P.; Gerhard, U.; Maplestone, R.A. Proc. Natl. Acad. Sci. USA 1993, 90, 1172-1 178. 75. Schwarz, F.P.; Ahmed, H.; Bianchet, M.A.; Amzel, L.M.; Vasta, G.R. Biochemistry 1998, 37, 586775877, 16. Schwarz, F.P.; Misquith, S.; Surolia, A. Biochem J. 1996, 316, 123-129. 77. Schwarz, F.P. J. Solution Chem. 1996,25,471-484. 78. Tyler-Cross, R.; Sobel, M.; McAdory, L.E.; Harris, R.B. Arch. Biochem. Biophys. 1996,334. 206-21 3. 79. Tomme, P.; Creagh, A.L.; Kilburn, D.G.; Haynes, C.A. Biochemistry 1996,35, 13885-13894. 80. Creagh, A.L.; Ong, E.; Jervis, E.; Kilburn, D.G.; Haynes, C.A. Proc. Natl. Acad. Sci. USA 1996, 93, 12229-12234. 81. Weatherman, R.V.; Mortell, K.H.; Chervenak, M.; Kiessling, L.L.; Toone, E.J. Biochemistry 1996,35, 3619-3624. 82. Mandal, D.; Kishore, N.; Brewer, C.F. Biochemistry 1994, 33, 1149-1 156. 83. Dam, T.K.; Oscarson, S.; Sacchettini, J.C.; Brewer, C.F. J. Biol. Chem. 1998, 273, 3282632832.

References

9 13

Dam, T.K.; Oscarson, S.; Brewer, C.F. J. Biol. Chem. 1998,273, 32812-32817. Gupta, D.; Dam, T.K.; Oscarson, S.; Brewer, C.F. J. Biol. Chem. 1997, 272, 6388-6392. Chervenak, M.C.; Toone, E.J. Bioorg. Med. Chem. 1996, 4, 1963-1977. Sanders, J.N.; Chenoweth, S.A.; Schwarz, F.P. J. Znory. Biochem. 1998, 70, 71-82. Swaminathan, C.P.; Surolia, N.; Surolia, A. J. Am. Chem. Soc. 1998,120, 5153-5159. Gupta, D.; Cho, M.; Cummings, R.D.; Brewer, C.F. Biochemistry 1996, 35, 15236-15243. Surolia, A,; Sharon, N.; Schwarz, F.P. J. Biol. Chem. 1996,271, 17697-17703. Fromm, J.R.; Hileman, R.E.; Weiler, J.M.; Linhardt, R.J. Arch. Biochem. Biophys. 1997,346, 252-262. 92. Tsai, F.T.F.; Singh, O.M.P.; Skarzynski, T.; Wonacott, A.J.; Weston, S.; Tucker, A,; Pauptit, R.A.; Breeze, A.L.; Poyser, J.P.; O’Brien, R.; Ladbury, J.E.; Wigley, D.B. Proteins Struct. Funct. Genet. 1997, 28, 41-52. 93. Garcia-Hernandez, E.; Zubillaga, R.A.; Rojo-Dominguez, A,; Rodriguez-Romero, A,; Hernandez-Arana, A. Proteins: Struct., Funct., Genet. 1997,29, 467-477. 94. Quesenberry, M.S.; Lee, R.T.; Lee, Y.C. Biochemistry 1997,36, 2724-2732. 95. Thomson, J.; Liu, Y.; Sturtevant, J.M.; Quiocho, F.A. Biophys. Chem. 1998, 70, 101-108. 96. Briggs, J.B.; Larsen, R.A.; Harris, R.B.; Sekar, K.V.S.; Macher, B.A. Glycohiology 1996, 6, 831-836. 97. Sharma, S.; Bhardwaj, S.; Surolia, A,; Podder, S.K. Biochem. J. 1998, 333, 539-542. 98. Bundle, D.R.; Alibes, R.; Nilar, S.; Otter, A,; Warwas, M.; Zhang, P. J. Am. Chem. Soc. 1998,120, 5317-5318. 99. Ramkumar, R.; Surolia, A,; Podder, S.K. Biochem J. 1995, 308, 237-241. 100. Lee: R.T.; Gabius, H.J.; Lee, Y.C. Glycoconjugate J. 1998, IS, 649-655. 101. Swaminathan, C.P.; Gupta, D.; Sharma, V.; Surolia, A. Biochemistry 1997, 36, 13428-1 3434. 102. Roy, R. Curr. Opin. Struct. Biol. 1996, 6 , 692-702. 103. Mammen, M.; Choi, S.K.; Whitesides, G.M. Angew. Chem. Znt. Ed. 1998, 37, 2754-2794. 104. Sigal, G.B.; Mammen, M.; Dahmann, G.; Whitesides, G.M. J. Am. Chem. SOC.1996, 118, 3789-3800. 105. Lee, R.T.; Ichikawa, Y.; Kawasaki, T.; Drickamer, K.; Lee, Y.C. Arch. Biochem. Biophys. 1992,299, 129-136. 106. Biessen, E.A.L.; Noorman, F.; vanTeijlingen, M.E.; Kuiper, J.; BarrettBergshoeff, M.; Bijsterbosch, M.K.; Rijken, D.C.; vanBerkel, T.J.C. J. Biol. Chem. 1996,271,28024-23669. 107. Hansen, H.C.; Haataja, S.; Finne, J.; Magnusson, G. J. Am. Chem. SOC.1997, 119, 69746979. 108. Glick, G.D.; Toogood, P.L.; Wiley, D.C.; Skehel, J.J.; Knowles, J.R. J. Biol. Chem. 1991, 266,23660-23669. 109. Wittman, V.; Takayama, S.; Gong, K.W.; Weitz-Schmidt, G.; Wong, C.H. J. Ory. Chem. 1998,63, 513775143, 110. Kamitakahara, H.; Suzuki, T.; Nishigori, N.; Suzuki, Y.; Kanie, 0.;Wong, C.H. Angew. Chem. Int. Ed. Engl. 1998, 37, 1524-1528. 111. Thompson, J.P.; Schengrund, C.L. Glycoconj. J. 1997, 14, 837-845. 112. Mortell, K.H.; Weatherman, R.V.; Kiessling, L.L. J. Am. Chem. Soc. 1996, 118, 2297-2298. 113. Manning, D.D.; Hu, X.; Beck, P.; Kiessling, L.L. J. Am. Chem. Soc. 1997, 119, 3161-3162. 114. Nishida, Y.; Dohi, H.; Uzawa, H.; Kobayashi, K. Tet. Let. 1998, 39, 8681. 115. Cao, S.; Roy, R. Tet. Let. 1996, 37, 3421. 116. Bovin, N.V.; Gabius, H.J. Chem. Soc. Rev. 1995, 413. 117. Bovin, N.V. Glycoconj. J. 1998, IS, 431. 118 Tsuchida, A,; Akimoto, S.; Usui, T.; Kobayashi, K. J. Biochem. 1998,123, 715. 119 Park, K.H.; Takei, R.; Goto, M.; Maruyama, A,; Kobayashi, A,; Kobayashi, K.; Akaike, T. J. Biochem. 1997, 121, 997. 120. Spaltenstein, A.; Whitesides, G.M. J. Am. Chem. Soc. 1991, 113, 686. 121. Lees, W.J.; Spaltenstein, A,; Kingery-Wood, J. E.; Whitesides, G. M. J. Med. Chem. 1994, 37, 3419. 122. Mann, D.A.; Kanai, M.; Maly, D.J.; Kiessling, L.L. J. Am. Chem. Soc. 1998, 120, 1057510582. 123. Kanai, M.; Mortell, K.H.; Kiessling, L.L. J. Am. Chem. Soc. 1997, 119, 9931. 84. 85. 86. 87. 88. 89. 90. 91.

9 14

31 Protein-Carbohydrate Interaction: Fundamental Considerations

124. Kingery-Wood, J.E.; Williams, K.W.; Sigal, G.B.; Whitesides, G.M. J. Am. Chem. SOC.1992, 114, 7303. 125. DeFrees, S.A.; Phillips, L.; Guo, L.; Zalipsky, S. J. Am. Chem. SOC. 1996, IZ8,6101. 126. Ketis, N.V.; Girdlestone, J.; Grant, C.W. M. Proc. Natl. Acad. Sci. USA 1980, 77, 3788. 127. Charych, D.; Cheng, Q.; Reichart, A.; Kuziemko, G.; Stroh, M.; Nagy, J.O.; Spevak, W.; Stevens, R.C. Chern. & Biol. 1996,3, 113. 128. Spevak, W.; Foxall, C.; Charych, D.H.; Dasgupta, F.; Nagy, J.O. J. Med. Chern. 1996, 39, 1018.

129. Pan, J.J.; Charych, D. Langmuir 1997,.13, 1365. 130. Zanini, D.; Roy, R. J. Org. Chem. 1998, 63, 3486. 131. Ashton, P.R.; Hounsell, E.F.; Jayaraman, N.; Nilsen, T.M.; Spencer, N.; Stoddart, J.F.; Young, M. J. Org. Chem. 1998, 63, 3429. 132. Hornick, C.L.; Karush, F. Immunochem. 1972, 9, 341. 133. Mammen, M.; Shaknovich, E.I.; Deutch, J.M.; Whitesides, G.M. J. Org. Chem. 1998, 63, 3821-3830. 134. Mammen, M.; Shaknovich, E.I.; Whitesides, G.M. J. Org. Chern. 1998, 63, 3168-3175. 135. Alibes, R.; Bundle, D.R. J. Org. Chern. 1998,63, 6288-6301. 136. For example: Espinosa, J.F.; Bruix, M.; Jarreton, 0.;Skrydstrup, T.; Beau, J.M.; JimenezBarbero, J. Chem. Eur. J. 1999, 5, 442-448, and references therein. 137. Aguilera, B.; Jimenez-Barbero, J.; Fernandez-Mayoralas, A. Carbohydr. Res. 1998, 308, 1927. 138. Toone et al., Unpublished Data.

Carbohydrates in Chemistry and Biology Editor Beat Ernst, Gerald W. Hart, Pierre Sinay copyright OWILEY-VCH Verlag GmbH, 2000

32 Structural Analysis of Oligosaccharides: FAB-MS, ES-MS and MALDI-MS Anne Dell, Howard R. Morris, Richard Easton, Stuart Haslam, Maria Panico, Mark Sutton-Smith, Andrew J. Reason, and Kay-Hooi Khoo

32.1 Introduction In the glycobiology field no structural technique can match mass spectrometry for the breadth of problems that can be addressed, the complexity of samples that can be successfully analyzed, and for the amount of structural information that can be obtained from sub-nanomolar amounts of material. The introduction of fast atom bombardment-mass spectrometry (FAB-MS) at the beginning of the 1980s [ l ] revolutionized the structure determination of a very wide range of carbohydratecontaining biopolymers [2, 31 and this revolution has continued with the newer techniques of electrospray ionization (ES-MS) [4] and matrix assisted laser desorption ionization (MALDI-MS) [ 5 ] . A summary of each of these techniques is given below in order to facilitate understanding of their problem-solving capabilities. All three technologies permit the direct ionization/desorption of non-volatile substances and are applicable to intact glycoconjugates as well as fragments.

32.2 Fast Atom Bombardment-Mass Spectrometry (FAB-MS) Double focusing sector mass spectrometers are employed in the majority of biopolymer FAB-MS studies. These consist of a source where the ions are generated, an analyzer comprising an electric sector for energy focusing and a magnetic sector for separating ions of different mass to charge ratios, and a detector. In the FAB experiment, an accelerated beam of atoms (usually xenon) or ions (usually cesium) is fired from an atom or ion gun towards a small metal target attached to the end of a probe (see Figure la). Prior to insertion of the probe into the FAB source, it is loaded with a viscous liquid called the matrix in which is dissolved the sample to be analyzed. When the atom or ion beam collides with the matrix, kinetic energy is

9 16

32 Structural Analysis of Oligosacclzarides

Source Slit

+To

Analyser

Laser attenuator Variable-voltage

Flight

detector

source chamber Ground Grid

\

Collision

Sampling Quadrupole Analyser

Sample in Solution

+

Drying Gas Atmospheric Pressure

1

1

-1 'mbar

-

10-4 10-5 m bar

Figure 1. (A) Fast Atom Bombardment (FAB): the sample is dissolved in a liquid matrix and ionization/desorption is effected by a high energy beam of particles fired from an atom or ion gun. @) Matrix Assisted Laser Desorption Ionisation (MALDI): the sample is dried on a metal target in the presence of a chromophoric matrix and sample ions are produced by energy transfer from matrix molecules that have absorbed energy from the laser pulse. (C) Electrospray (ES): a stream of liquid containing the sample of interest enters the source through a capillary interface, where the sample molecules are stripped of solvent, leaving them as multiply charged species.

32.3 Matrix Assisted Laser Desorption Ionization-Time

9 17

transferred to the surface molecules, many of which are sputtered out of the liquid into the high vacuum of the ion source. A significant number of these molecules are ionized during the sputtering process. Thus gas-phase ions are generated without prior volatilization of the sample allowing the analysis of polar, involatile, and thermally labile compounds. Both positive and negative ions are produced during the sputtering process, and either can be recorded by an appropriate choice of instrumental parameters. Molecules are ionized by protonation or by the addition of a cation such as sodium, potassium, ammonium (positive ion formation), or by the loss of a proton, or addition of an anion such as chloride, thiocyanate (negative ion formation). During ionization, some internal energy is imparted to the molecule resulting in fragmentation of labile bonds. The sputtered molecular and fragment ions are accelerated to about 8,000 eV prior to passage through the analyzer sectors to a detector where the mass to charge ratios are recorded by a computer to give the mass spectrum. Magnetic sector analyzers separate ions on the principle that charged molecules are deflected by strong magnetic fields. Ions of larger mass are deflected by the magnetic field less than ions of smaller mass according to the equation m / z = B2r2/2V where m / z is the mass to charge ratio of the ion, B is the strength of the magnetic field, r is the radius of the circular path through which the ion is travelling through the magnet and V is the accelerating voltage. Ions of different m / z values are brought into focus at the detector by scanning B with V and r being kept constant. Modern high field magnetic sector mass spectrometers are capable of focusing ions up to 15,000 Da at full sensitivity, but the working mass range of FAB-MS for carbohydrate polymers is limited to about 6,000 Da. Higher molecular weight samples are either too difficult to desorb from the matrix or do not produce a sufficient abundance of molecular ions to allow detection above background. The FAB experiment often yields structurally informative fragment ions as well as molecular ions but the quantity and quality of fragmentation can be very variable depending on factors such as the purity of the sample and whether it has been derivatized. To ensure that sufficient fragment ions are produced to allow structural assignments it may be necessary to employ tandem MS techniques (MS/MS) in which ions produced in the FAB source pass through the mass analyzer into a chamber containing an inert gas. Collisions with the gas provide sufficient energy for bond cleavages and the resulting fragment ions (called daughter ions) are detected after passing through a second (tandem) analyzer. Tandem MS is especially powerful when mixtures are being analyzed because it allows unambiguous attribution of fragment ions to individual components of the mixture.

32.3 Matrix Assisted Laser Desorption Ionization-Time of Flight-Mass Spectrometry (MALDI-TOF-MS) In the MALDI experiment (Figure lb) the sample is embedded in a low molecular weight, UV absorbing matrix which enhances sample ionization. When the matrix

9 18

32 Structural Analysis of Oligosucchurides

absorbs the laser pulse enough energy is transferred to the sample, via mechanisms that are not well understood, to enable the formation of molecular ions. The matrix is present in a vast excess over the sample and therefore isolates individual sample molecules. This results in the volatilization of predominantly monomeric molecular ions, although dimers, trimers etc are observed in some MALDI spectra. Ions produced in MALDI-MS are usually analyzed by Time of Flight (TOF) instrumentation. TOF analyzers work on the principle that when ions are accelerated with the same potential from a fixed point and at a fixed initial time and are allowed to drift, the ions will separate according to their mass/charge ratios. Lighter ions drift more quickly to the detector and heavier ions drift more slowly. The time required for ions to reach the detector can therefore be related to their mass. Following acceleration, ions exhibit a broad energy distribution. This energy spread can be minimized using Delayed Extraction (DE) which involves applying a high voltage pulse at a pre-determined time following ion generation in a weak electrical field leading to enhanced resolution and mass accuracy. Most modern MALDI instruments allow both linear and reflector mass analysis. In the linear mode, ions travel without interference down the flight tube to the detector which records the m / z ratio and signal intensity. A reflector is a single-stage gridded mirror that focuses energy. In reflector mass analysis mode, a uniform electric field is applied to the mirror to reflect ions. This filters out neutral molecules, corrects dispersion of the ions occurring in the flight tube and thus provides greater mass accuracy and resolution. In contrast to the magnetic sectors used for analysis in FAB-MS, the TOF analyzer has an almost unlimited mass range. MALDI is also considerably more efficient than FAB in ionizing large biopolymers and is the method of choice for defining the molecular weights of carbohydrate polymers, particularly at high mass.

32.4 Electrospray-Mass Spectrometry (ES-MS) ES-MS is a method by which a stream of liquid containing the sample of interest is introduced into the atmospheric pressure ion source of a mass spectrometer (Figure lc). This can be achieved by direct injection into buffer which is pumped at the rate of a few microliters per minute through a metal-tipped glass capillary. Alternatively the eluent of a microbore LC, after stream splitting, can be passed via the capillary into the ES source. Many newer instruments have high sensitivity nanospray sources capable of operating at flow rates of as low as a few nanoliters per minute. With this type of source, samples are introduced using a probe with an attached nanospray needle, the latter being pre-loaded with about a microliter of a solution of the sample (see 3.9.2, Protocol 2). Irrespective of the introduction method, an aerosol of microdroplets is generated in the source which then traverses a series of skimmers, encountering a drying gas, the net effect of which is the creation of charged molecular species, devoid of solvent. These gaseous ions, whose charge depends on the number of ionizable groups

32.5 Appearance o j Mass Spectra Obtained in FAB-MS

919

in the molecule, are passed into the mass analyzer which is commonly a quadrupole mass filter. This type of analyzer is compact and relatively inexpensive. It comprises four parallel rods which act as mass filters when rf and dc fields are applied to diagonally opposed rods. By sweeping the rf and dc voltages in a fixed ratio, ions of successive mass to charge ratios follow a stable path to the detector. Ions with m / z values above about 4,000 cannot be detected with a quadrupole analyzer. However this is not too problematical for a significant portion of biopolymer ES-MS applications because of the presence of multiple functional groups capable of carrying charge. For example proteins and glycoproteins in excess of 100 kDa can be amenable to ES-MS analysis. Uncharged or poorly charged polysaccharides are not, however, amenable to ES-MS on a quadrupole instrument. The ionization process in ES-MS is very gentle thus resulting in very few fragment ions and little or no sequence information. To overcome this problem many ES instruments have triple quadrupole analyzers for collisional activation MS/MS experiments. The first quadrupole is used to select the parent ion, the second is the collision chamber and the third separates the resulting fragment ions. Whilst triple quadrupole instruments are very powerful, their sensitivity is limited by a number of factors including scanning ion detection and poor fragment ion resolution. These problems have been overcome in a novel mass spectrometer, called the Q-TOF, which was introduced in the mid-1990s [6, 71. The Q-TOF has both quadrupole and TOF analyzers which are arranged orthogonally with a collision cell between them (Figure 2). For normal mass spectra, the quadrupole is used in the rf-only mode as a wide-bandpass filter to transmit a wide mass range. The collision cell is not pressurized, and ions are transmitted to the TOF for mass analysis. In the MS/MS mode the quadrupole operates in the normal resolving mode and is able to select precursor ions up to m / z 4,000 for collisional activation in the hexapole gas cell. The fragment ions are transmitted to the TOF for mass analysis. The orthogonal geometry, and parallel rather than sequential detection of the ions, leads to a significant improvement in sensitivity over scanning instruments when used to acquire full spectra. The Q-TOF is an immensely powerful instrument. The good signal/noise ratios in MS/MS on the Q-TOF correspond to low femtomole/attomole sample consumption. Definitive unambiguous sequence assignment is facilitated by daughter ion resolutions of greater than 3,000 (which allows easy assignment of the z value of m / z ) and daughter ion mass accuracies of 0.05 Da.

32.5 Appearance of Mass Spectra Obtained in FAB-MS, MALDI-MS and ES-MS Experiments FAB spectra are characterized by a high level of chemical “noise”, arising from the matrix and sample, giving a peak at every mass upon which are superimposed the signals for molecular and fragment ions of the sample and matrix. The chemical noise is much lower in MALDI-MS which results in better signal to noise for sam-

920

32 Structural Analysis of Oligosaccharides

SAMPLING CONE QUADRUFQLE IN NARROW BANDPASS MODE

HDV\POLE GAS COLLISION CELL

-

Q TOF MSIMS MODE

REFLECTRON

7

Figure 2. Schematic illustration of the Q-TOF in the MS-MS mode of operation. The quadrupole is set to transmit the parent ion of interest and the hexapole collision chamber contains the collision gas. The orthogonal TOF separates the daughter ions.

ple ions. Most of the molecular and fragment ions observed in FAB and MALDI spectra are singly charged, e.g. [M+H]+, [M-HI-, [M+Na]+, [M-H+2Na]+ etc. In contrast to FAB-MS and MALDI-MS, electrospray mass spectra are characterized by multiply charged ions. The raw data obtained following ES-MS analyses is complex because ions can carry a range of charges due to the different ionisable sites in the molecule. Multiple m / z signals are therefore recorded for each mass value. Fortunately these data can be processed using a simple computer algorithm to produce a mass spectrum of comparable appearance to a FAB or MALDI spectrum. Ions are observed as clusters in all types of mass spectra because of the existence of isotopes. The contribution of I3C (1.1% natural abundance) means that for every carbon atom in the molecule there is a 1.1% chance that it will be a 13Catom and so have an atomic mass of 13 instead of 12. The relative intensity of the different peaks in the cluster reflects this probability distribution, i.e. the height of the signal at m / z X + 1 (where X corresponds to the molecular ion that contains only 12C)reflects the probability of finding one I3C atom in the molecule. The individual signals in a

32.7 Derivatisation

92 1

cluster will be observed when the instrument resolution is set to resolve nominal masses. At lower resolutions the clusters are present as an unresolved envelope. For very high molecular weight ions these envelopes may embrace more than one cluster of molecular ions, e.g. [M+H]+ and [M+Na]+, resulting in a very broad peak which cannot be accurately mass-assigned.

32.6 Assignment of Mass Values There are four principal ways of denoting the mass of an ion in the mass spectrum: i) Nominal mass-this is the sum of the integer atomic weights of the isotopes comprising the ion, giving the specific peak whose mass is being assigned. ii) Accurate mass-this is the sum of the accurate atomic weights of the isotopes comprising the ion giving the specific peak whose mass is being assigned. Accurate masses are obtained when spectra are computer assigned by comparison with a calibration spectrum containing ions whose accurate masses are known (the calibration standard is often an alkali halide such as CsI). iii) Average mass (chemical mass)-this is the average of the accurate isotope mass of each element, weighted by the relative abundance of the isotopes, i.e. it corresponds to the sum of the chemical atomic weights. Average masses are assigned when an isotopic cluster is recorded at such low resolution that the cluster is completely unresolved. The average mass is the centre of gravity of the unresolved cluster. iv) Peak top mass-this is the accurate mass of the top of an unresolved cluster. At high masses, above about m l z 5,000, the peak top mass is very close to the average mass because the isotopic distribution results in a Gaussian shape for the cluster. ES-MS assignments are normally based on peak top masses.

32.7 Derivatisation Although native samples are amenable to FAB-MS, MALDI-MS and ES-MS, it is often desirable to prepare derivatives prior to analysis. As a general rule glycans whose hydroxyl groups are protected by functional groups, such as methyl or acetyl, fragment more reliably than their native counterparts. Also derivatives are easier to obtain free from salt impurities which may prejudice the MS experiment and sensitivity is significantly improved when hydrophobic moieties are present. Derivatization methods can be broadly divided into two categories: (i) “tagging” of reducing ends, and (ii) protection of most or all of the functional groups. Commonly used tagging reagents include p-aminobenzoic acid ethyl ester (ABEE) and 2-aminopyridine (2-AP). This type of derivatization facilitates chromatographic

922

32 Structural Analysis of Oligosaccharides

purification and enhances reducing-end fragment ions in MS and MS/MS experiments. Protection of functional groups by permethylation or per(deuter0)acetylation is by far the most important type of derivatization employed in carbohydrate MS. These derivatives fragment reliably to yield abundant A-type fragment ions (see below) which are extremely useful for sequencing.

32.8 Fragmentation Pathways The following general rules apply to the fragmentation behavior of oligosaccharides and glycoconjugates [3]: i) The most abundant fragment ions are formed by cleavages at glycosidic linkages. Glycosidic cleavage is accompanied by a hydrogen transfer to the glycosidic oxygen if the charge on the fragment ion is not specifically located at the point of cleavage. ii) Ring cleavage, when it occurs, is best rationalized as arising from the sequential movement of electron pairs around the ring resulting in the breakage of single bonds and the formation of double bonds (see Figure 3). iii) Fragment ions can sometimes be formed by two or more cleavage events occurring in different parts of the molecule. This phenomenon is more frequently observed in native samples than in derivatives. Unambiguous sequencing of native samples is often not possible if “double cleavage” ions are formed in abundance. iv) The fragment ions produced by glycopeptides are derived predominantly from cleavage of the glycosidic linkages. Fragment ions resulting from cleavage of the peptide bonds are often of low abundance. v) If a permanent charge is present in the molecule (e.g. a sulfate moiety) then fragment ions produced by cleavage of labile bonds in the vicinity of the charge dominate the spectrum. The major fragmentation pathways relevant to carbohydrate MS are summarized below: i) A-type cleavage-Glycosidic cleavage (Figure 3) yields an oxonium ion thereby locating the charge at the point of cleavage on the non-reducing fragment. The term A-type cleavage is normally applied to this type of fragmentation based on nomenclature used in electron impact mass spectrometry. A-type cleavage is the major mode of fragmentation of permethylated and per(deuter0)acetylated oligosaccharides. If HexNAc residues are present in the sequence, cleavage occurs predominantly (and sometimes exclusively) at the amino sugar residues. A secondary fragmentation associated with A-type cleavage is elimination of the substituent at the 3-position of the HexNAc oxonium ion thus defining whether the 3-position is occupied by a sugar such as fucose.

32.8 Fragmentation Pathways

923

0

A-type ion

B-cleavaae and 6-elimination

J.

J.

p-cleavage ion

0-elimination ion

0-elimination ion

b-cleavage ion

(non-reducing)

(reducing)

(non-reducing)

(reducing)

Rina cleavaae

CH20R

I

bR bR

bR

bR

Figure 3. Key fragmentation pathways in glycopolymer mass spectrometry are shown in this figure. The terms A-type cleavage, P-cleavage, ring cleavage etc are useful descriptors of these pathways. The reader should consult [ 191 for information on how to systematically name fragment ions arising from each of these pathways.

924

32 Structural Analysis of Oligosacchavides

ii) P-cleavage-and p-elimination-This involves cleavage on either side of the glycosidic linkage with hydrogen transfer to the glycosidic oxygen resulting in one fragment having a hydroxyl group at the position of glycosidic linkage and the other having a double bond (Figure 3 ) . These are referred to as P-cleavage and 0-elimination products respectively. Note that, unlike A-type cleavage, a charged moiety is not produced as a result of cleavage. Depending on where the charge is located in the fragmenting molecule, either or both reducing and nonreducing P-cleavage and/or P-elimination ions will be observed. This type of cleavage is favored by native oligosaccharides and glycopeptides and is especially prominent in samples modified with a reducing end “tag”. iii) Ring cleavages-These are frequently observed in MS/MS experiments especially when cationised molecular ions are subjected to collisional activation. They can provide useful information on attachment sites of functional groups and glycosidic linkages. Examples of ring cleavages are shown in Figure 3.

32.9 Protocols for MS Analysis In this section we describe some of the MS protocols employed in our laboratories. These are given for illustrative purposes in order to facilitate understanding of the case studies presented in later sections of this Chapter. The protocols exemplify general strategies underlying each technique and are not intended to be exclusive. 32.9.1 Protocol 1-Sample

Loading for FAB-MS Analysis

i) Dissolve the sample in either 50/0 acetic acid (for underivatized samples) or methanol (for derivatives) to such a concentration that 1 pl aliquot of the sample contains the desired amount of sample for FAB-MS analysis. ii) Smear about 1 pl of the matrix onto the metal target which is attached to the end of the FAB probe. iii) Load about 1 pl of the sample on the surface of the matrix using a micropipette or syringe. iv) Introduce the probe into the FAB source and collect data immediately. v) When a wide mass range of scanning is required, e.g. mlz 4,000 to 200, it may be necessary depending upon instrument type to acquire scans covering the high and low mass range separately with two separate sample loadings. vi) Use different matrices and matrix additives to improve the quality of the data or alter the molecular and/or fragment ion patterns: use monothioglycerol as the matrix for general analysis - use rn-nitrobenzyl alcohol matrix for salty samples. add 1 pl of dilute aqueous (100-200 mM) HC1 to the matrix to minimize cationization - other matrix additives, e.g. sodium acetate and ammonium thiocyanate can also be used in specific cases [3]. ~

~

32.9 Protocols for M S Anulysis

925

32.9.2 Protocol 2-Sample Loading for NanoES-MS and MS-MS Analysis on the Q-TOF

i) Samples should be solubilized in an ES-MS compatible solvent such as acetonitrile/O.l% aqueous TFA (1 : 1, v:v). It should be noted that when using nanospray, it is observed that increased beam stability is achieved sometimes when using methanol. 1-2 pl of sample can be loaded to the end of the metal coated capillary using either plastic gel loader tips or a microlitre syringe. Care must be taken in handling the capillary because the 1-2 pm tip through which the sample is sprayed is very fragile and easily broken. ii) Place the metal coated capillary (square end first) into its holder. This consists of a knurled nut, a 5 mm length of conductive elastomer and a Swagelok union. Once secured, screw the capillary/holder onto the probe and place into the source. iii) Having inspected the MS spectrum, and tuned the quadrupole to, for example, a double or triply charged ion of interest, the argon collision gas pressure is adjusted to approximately lop4 (this will vary for optimum collision-induced fragmentation) and the collision energy is set between 10 and 60 eV relative to the size of the molecule under study, and its charge state. iv) Collect data until summation of scans produces a good quality spectrum. Collection times may vary depending on sample concentration, its propensity for ionization and state of capillary tips. However, for a 100 femtomole per yl solution, 0.5 min should be adequate for a high quality spectrum. v) Manipulation of the capillary tips by fracturing may be needed during ionization to maintain the spray without air bubbles and particulate blockages. 32.9.3 Protocol 3-Sample Q-TOF

Loading for LC-ES-MS and LC-ES-MS-MS on the

i) Dissolve the sample in an appropriate volume by using starting condition buffer and load into a 20 pl sample loop. Elute using 0.1% aqueous trifluoroacetic acid (buffer A) and a gradient up to 100% acetonitrile in 0.1% aqueous trifluoroacetic acid (buffer B) at a flow rate of 50 pl/min. The elution is monitored by UV absorbance at 214 nm in a nanoflow cell. ii) After the column a methoxyethanol :isopropanol (1 : 1, v :v) solvent mixture is added at a flow rate of 50 pl/min; this helps to counteract the signal suppressive effects of trifluoroacetic acid. The flow is then stream-split to allow collection of approximately 85 pl fractions from the column at 1 min intervals, with the remaining 15 pl/min directed on-line to the electrospray source of the Q-TOF instrument. 32.9.4 Protocol 4-Sample

Loading for MALDI-MS Analysis

i) Dissolve the sample to be analyzed in pure water (for native carbohydrates) or 80 :20 methanol :pure water (for derivatized carbohydrates) to produce a sample concentration of around 1 10 pmoles/yl. ~

926

32 Structural Analysis of Oligosaccharides

ii) Prepare the appropriate matrix: - 2,5-Dihydroxybenzoic acid (DHB) is used as general matrix; 10 mg/ml in 90 : 10 water :ethanol (for native carbohydrates) or 80 :20 methanol :water (for derivatized carbohydrates) - 2-(4-hydroxyphenylazo)-benzoicacid (HABA) is used for enhancement of polar carbohydrates; 1.3 mg/ml in 50 : 50 water :acetonitrile (for native carbohydrates) or 80 :20 methanol :water (for derivatized carbohydrates) iii) Pipette 1 p1 of the sample solution (using a Gilson P2 pipette or similar) to the metal target followed by 1 pl of the matrix solution. iv) Dry under vacuum. v) Introduce into the MALDI source and operate the instrument - DE-MALDI-TOF-MS analysis in the reflector mode is the method of choice for carbohydrate samples in the mass range 100-10,000 Da. - DE-MALDI-TOF-MS analysis in the linear mode is the method of choice for carbohydrate samples in the mass range >10,000 Da.

32.10 Applications of FAB-MS, MALDI-MS and ES-MS in Glycobiology Broadly speaking FAB-MS, MALDI-MS and ES-MS can be exploited in two general ways in the glycobiology field: i) detailed characterization of purified individual glycopolymers or mixtures of glycopolymers which usually requires acquisition of a considerable body of structural information, and ii) rapid screening of cell-, tissue- and whole animal-extracts where a limited number of MS experiments on relatively crude extracts are often sufficient to answer the questions addressed. Below we have selected examples from recent glycopolymer research which exemplify the versatility of problems amenable to current technology. 32.10.1 Case Study l-Molecular MALDI-MS

Weight Profiling of Polysaccharides by

The following study of laminarin polysaccharides illustrates the potential of MALDI-MS for rapidly and sensitively defining the degree of polymerization (d.p.) of carbohydrate polymers and for revealing structural modifications which result in altered molecular weight. Laminarins are a class of low-molecular-weight storage fi-glucans of brown algae consisting of (1-3)-linked P-D-glucopyranose residues in which some 6-O-branching in the main chain and some fi(1-6)-intrachain links are present. The majority of

32.10 Applications of FAB-MS, MALDI-MS and ES-MS in Glycobiology

927

laminarins contain polymeric chains of two types, polymeric glucopyranose only (G-chains) and polymeric glucopyranose terminated with 1-@substituted Dmannitol residues (M-chains). Both native and permethylated laminarins give excellent MALDI data [8]. The permethylated derivatives were examined in order to facilitate discrimination between M- and G-laminarins which differ by only 2 mass units in their native form but by 16 mass units after permethylation. Typical spectra are shown in Figure 4a for native laminarin from Laminaviu cichorioides (a mannitol-containing laminarin) and in Figure 4b for permethylated laminarin from Chorda crinitu (a mannitol-free laminarin). The L. cichorioides sample shows a Gaussian-like d.p. profile peaking at about d.p. 27 and ranging over about d.p. 9-40. The C. crinitu sample has a bimodal d.p. profile with abundant oligomers at both low and high mass. Interestingly, close examination of Figure 4b reveals two regions of the spectrum where clusters of ions suggest additional components, namely d.p. 5-10 and d.p. 16-17. The latter show abundant signals 46 mass units lower than the corresponding Goligomer which is consistent with cyclic components of d.p. 16 and 17. The former exhibit satellite peaks 41 mass units higher than the corresponding G-oligomer. This mass interval is consistent with the presence of a HexNAc residue replacing a hexose residue in a minority of the sample. This was an unexpected discovery because aminosugars had not previously been found in laminarins and illustrates the power of the MALDI technique for revealing the presence of novel minor constituents in complex mixtures. 32.10.2 Case Study 2-Analysis of Glycoproteins by LC-ES-MS and FAB-MS

The following study of glycodelin A illustrates how the strengths of LC-ES-MS and FAB-MS can be exploited in a complementary manner to obtain structural information on low micromolar quantities of novel glycoproteins [9]. Glycodelin-A (GdA), is a human amniotic fluid-derived glycoprotein that has potent contraceptive and immunosuppressive activities. GdA has 162 amino acids and there are three potential N-linked glycosylation sites at Asn-28, Asn-63 and Asn-85. Together LC-ES-MS and FAB-MS gave sufficient data to define the site occupancy and the sequences of the N-glycans present on GdA. The strategy for analyzing GdA is shown in Figure 5. The key experimental steps were as follows: i) LC-ES-MS and LC-ES-MS-MS analyses of tryptic and cyanogen bromide digests of GdA confirmed the protein sequence, established the non-occupancy of Asn-85 and showed that Asn-28 and Asn-63 were glycosylated. ii) Glycans were released from LC fractions containing the glycopeptides identified by ES-MS analysis. After permethylation these were analyzed by FAB-MS yielding abundant molecular ions and A-type fragment ions. The latter define the non-reducing sequences in the antennae of complex-type and hybrid Nglycans whilst the former define the overall compositions of each type of glycan including high mannose structures. FAB-MS is the method of choice for this

928

32 Structural Analysis of Oligosuccharides

d

8 In

d

8 P

$ %

:: siuno3

d

d

Counts

N

8 0

E a

0

I

a

3 v

L C

930

32 Structural Analysis of Oligosaccharides

PNGase F

PNGase F

$1SeP-Pak -:

4

Void permethylation

$1-

Void exoglycosidase

4

Hydrolysis

4 4 Acetylation 4 Reduction

$1c=( Permethylation

Void Permethylation

$1-

$1-

Sep-Pak

Sep-Pak

$1-

$1-

Figure 5. The experimental strategy used to characterize glycodelin A.

type of analysis because it reliably produces A-type ions without the need for MS/MS experiments. iii) The products of various exoglycosidase digests were analyzed by FAB-MS after permethylation. iv) Linkage analysis by GC-MS was used to define sites of glycosyl attachment in the N-glycan sequences. From these experiments the structures of the majority of the oligosaccharides present in GdA were defined as shown in Figure 6. The Asn-28 site was shown to carry high mannose, hybrid and complex-type structures, whereas Asn-63 is exclusively occupied by complex-type glycans. 32.10.3 Case Study 3-Characterization of a Novel N-Glycan by FAB-MS and FAB-MS-MS The following study of Haemonchus contortus N-glycans demonstrates how novel glycans present in complex mixtures can be rigorously identified by strategies based on FAB-MS and FAB-MS-MS analyses [lo].

32.10 Applications of FAB-MS, MALDI-MS and ES-MS in Glycobiology

931

Key

V

= Fucose

0

= Mannose = Galactose

0 = Sialic acid 0 = N-acetylglucosamine = N-acetylgalactosamine

Figure 6. This cartoon structure shows the major glycoforms of glycodelin A. Further information can be found in [9].

932

32 Structural Analysis of Oligosaccharides FUC a14

I

M a n p1-4GlcNAc p14GlcNAc

I FUCal-3

FUc aI ~ - 3

Figure 7. Structure of the novel highly fucosylated core of some H. contortus N-glycans.

Adult Haernonchus

contortus

+ + +

Reduction I Carbnxymethylation Trypsin digest

Glycopeptides I peptides PNGase F digest I Sep-pak pnrifreation

J FAB-MS

PNGase A digest I Sep-pak puritiultion

1..

Exoglycosidye digestion /

J

FAB~MS

Linkage analysis

,

+

rdasedigestion

Linkage analysis FAB-MS

+ +

MS-MS Mild methmolysis

4 .

Linkage analysis

f

Linkage analysis

Figure 8. The experimental strategy used to characterize the H. contortus N-glycan shown in Figure 7.

H. contortus is an economically important nematode that parasitizes domestic ruminants. In a programme of work aimed at identifying carbohydrate antigens that could be targets for vaccine development, the structures of N-glycans present in H. contortus extracts have been investigated. Novel N-linked glycans with trifucosylated cores (Figure 7) have been identified in adult animals using the strategy outlined in Figure 8. The key experimental steps were as follows:

i) FAB-MS analysis of permethylated glycans released from H. contortus glycopeptides by peptide N-glycosidase (PNGase F) and PNGase A (the latter releases glycans with fucose attached to the 3-position of the proximal GlcNAc of the core which are resistant to PNGase F) established the presence of high mannose structures, minor amounts of complex structures and unusual truncated glycans substituted with up to three fucose residues. ii) Digestion of released glycans with a- and (3-mannosidase greatly reduced

32.10 Applications of FAB-MS, MALDI-MS and ES-MS in Glycobioloyy

933

the complexity of the mixture of glycans facilitating subsequent MS-MS experiments. iii) Selected molecular ions were subjected to FAB-MS-MS which provides sequence and linkage informative daughter fragment ions. iv) Linkage analysis by GC-MS was used to confirm sites of glycosyl attachment. An especially important element of the above structural strategy is the MS-MS component because the novel glycans could not be isolated in sufficient quantities for individual structural studies. Some of the MS-MS data which helped to establish the sites of fucosyl attachment are shown in Figure 9.

32.10.4 Case Study 4-High Sensitivity Sequencing of a Novel Glycopeptide by Q-TOF ES-MS-MS and MALDI-MS This study of a novel glycopeptide derived from Dictyostelium sp. SKPl exemplifies the power of the Q-TOF nano-electrospray mass spectrometer for high sensitivity sequencing [ 1 I]. The efficacy of MALDI-MS for screening enzyme digests is also demonstrated. SKPl is a cytoplasmic protein which has been identified in a variety of eukaryotes including yeast, mouse and man. It is found as part of a multiprotein complex which is involved in the ubiquitination of certain cell cycle and nutritional regulatory proteins thus condemning them to proteasomal degradation. Prior to the Q-TOF study described below, metabolic labelling and other experiments had indicated that, unusually for a cytoplasmic protein, SKPl from Dictyostelium is modified by an oligosaccharide containing Fuc and Gal. To investigate the glycosylation of SKPl , cells were metabolically labelled with '[HIFuc, radioactive SKPl was purified, reduced and alkylated and finally digested with endo-Lys-C. Components of the digest were separated by gel filtration and reverse phase HPLC yielding a single radioactive peak which was likely to be the expected fucoglycopeptide. Parallel experiments with non-radioactive material were performed and the fraction eluting at the position of the radioactive peak was subjected to nanospray-ES-MS-MS on the Q-TOF. Analysis of a few picomoles in the MS-only mode gave a major [M+3H]+++ signal at m / z 829.42 which was subjected to collisional activation yielding a spectrum which was remarkably rich in doubly and singly charged daughter ions (Figure 10). A portion of the spectrum is shown at high magnification (Figure 11) to illustrate the excellent signal to noise and resolution of even very minor fragment ions. This is an important strength of Q-TOF instrumentation especially since information such as sugar attachment sites in glycopeptides is often carried by low abundance fragment ions. Detailed interpretation of the MS-MS data allowed assignment of the carbohydrate sequence, Hex-Hex-deoxyHex-Hex-HexNAc, and the complete peptide sequence, N-D-F-T-P(0H)-E-E-E-E-Q-I-R-K and also the site of attachment of the carbohydrate, which was unexpectedly found to be hydroxyproline. Some of the key fragment ions used in the structure assignment are shown in Figure 12. MALDI-MS of the glycopeptide gave an [M+H]+ signal at m / z 2487 thus providing corroborative evidence for the structure proposed from the Q-TOF data. To

934

32 Structural Analysis of Oligosaccharides

a)

Figure 9. Data from a FAB-MS-MS experiment which provided evidence for the trifucosylated structure shown in Figure 7. The [M+H]+ and [M+Na]+ ions at rn/z 1263 (A) and 1285 (B) respectively were selected for collisional activation. Note that because of the different internal energies of these two molecular ions, the fragmentation pathways are different. Structurally useful fragment ions, produced via fragmentation pathways outlined in Figure 3, are shown on the inset. These ions provide important information on the attachment sites of the fucosyl residues.

assign linkages, and to rigorously establish the types of sugar present, the glycopeptide was treated with a variety of exoglycosidases with the products being analysed by MALDI-TOF-MS. Representative data are shown in Figure 13 which shows the evidence for the non-reducing sequence being Gala 1-6Galal -?Fucal -2. Similar experiments showed that the proximal disaccharide in Figure 12 is GalP13GlcNAc.

935

32.10 Applications of FAB-MS, MALDI-MS and ES-MS in Glycobiology

1

-+-

I

259 c-) 1049

CHzOMe

CHzOMe

0-

4 *- :*M a'

OMe

AcMeN

676

0 OMe

NMeA!?

1. 3

OMc

li

845

OMe

1095

'I' 25

259

I

426

b)

Figure 9 (continued)

32.10.5 Case Study 5-FAB-MS Content

i

Screening of Biological Samples for Glycan

Although detailed structural analysis is fundamental to a full understanding of structure/function relationships, there are many situations where the rapid acquisition of partial structural information is of greater value to the program of research than a time-consuming strategy leading to complete structure assignments. For example, many months of painstaking research was required to characterise the novel N-glycan described in Case Study 3 . Once its structure was known it was of interest to determine whether its synthesis was developmentally regulated and whether it was found in other parasites occupying a similar niche to H. contortus. These issues can be addressed by screening experiments [ 121 in which tissue extracts are analysed for their glycan content using FAB-MS or MALDI-MS to screen for molecular ions

?I0

2

[M+2H]2’ 9: 32

927.4

826.41 825.90

[M+ZH]

26.92

1

HexNAc, Hex

I

Peptide

& 10

,1008.46 ,1009.43

[M+2H]

[M+2HI2+

HexNAc

HexNAc, Hex, Fuc

I

Peptide

’+28.42,1008.94’+

Peptide

I

Peptide HexNAc

1651.79 1650.78

[M+H]

Peptide

HexNAc

I

Peptide

Figure 10. Q-TOF analysis of a glycopeptide from SKPI: MS-MS collision activated decomposition spectrum of m / ; 829.42’++

I

163.08

+

m

W

W

.

382.91

175.63

174.54

1274.64

y" I0

y" 0 I

HexNAc 1376.61

y"11

Figure 11. Part of the MS-MS spectrum shown in Figure 10 expanded to show fragment ions more clearly.

1081.4

D82.41

[M+2H] 2+ 1c 98

HexNAc, Hex, Fuc

I

Peptlde

57 y" 9

I I

HexNAc, Hex

y-9

y" 10

mlz 0

938

32 Structural Analysis of Oligosaccharides

N-D-F-

PEEEEQ IRK

I?-

l - t n -

-

-

-

HexNAc

tA-

!Hex

-

-- -

+ 1173.57 1376.61

+ 1538 84

deoxyHex Hex Hex

Figure 12. Structure of the SKPl glycopeptide deduced from the Q-TOF data showing some of the diagnostic fragment ions used to assign the position of attachment of the sugar.

and FAB-MS to screen for non-reducing structures via A-type fragment ions. This type of experiment can be completed in a few weeks in contrast to the months or years required to fully characterize individual glycans in complex mixtures. Screening methods which are applicable to a wide range of biological material, including organs, cell lines and whole parasites, are exemplified by the following profiling study of a variety of mouse organs in which the FAB-MS data are interpreted in the context of previous knowledge of murine glycosylation. This work is part of a study using knockout mice to address fundamental issues of glycan function in which FAB-MS is being used to identify changes in glycosylation occurring when particular glycosyltransferase and glycosidase genes are ablated. The experimental strategy employs the following steps:

i) Detergent extraction of the tissue/organ followed by detergent removal ii) Reduction/carboxymethylation, tryptic digestion and Sep-pak separation of the peptide/glycopeptide pool from salts, free sugars etc iii) Release of N-glycans by peptide N-glycosidase F digestion iv) Sep-pak separation of N-glycans from the peptidelo-linked glycopeptide pool v) Permethylation of N-glycans and FAB-MS analysis vi) Reductive elimination of 0-glycans from peptidelo-glycopeptide pool and Dowex purification vii) Permethylation of 0-glycans and FAB-MS analysis These experiments provide compositional information (via the molecular ions) on the majority of neutral and sialic acid containing glycans in the tissueslorgans together with information on the types of non-reducing structures present (via the A-type ions). This is illustrated by the data in Figures 14 and 15. Figure 14 shows the molecular ion region of the N-glycan pools from mouse brain, liver and lung. A unique composition can be ascribed to each molecular ion as shown. Figure 15 shows the A-type fragment ions formed from these molecular ions. The A-type ions are assigned using a variety of information including their masses, the presence or absence of associated fragment ions and prior knowledge of glycosylation in the mouse. The data show that all organs are rich in both high mannose and complextype glycans. The brain and kidney have complex-type glycans that are rich in Lewis x antennae (note the A-type ion at m / z 638 in Figures 15A and B) whilst

32.10 Applications of FAB-MS, MALDI-MS and ES-MS in Glycobiology

939

A. Hex- Hex- deoxyHex- Hex- HexNAc- OHPro 2487

B. [Galal,6) Hex- deoxyHex- Hex- HexNAc- OHPro 2325

-b C. [Galal,6- Gala1 )deoxyHex- Hex- HexNAc- OHPro

0. [Galal,6-Galal ,-Fuccrl,2).Hex-HexNAc-WPro

201 7

4 Figure 13. MALDI-TOF analysis of the SKPl glycopeptide: (A) spectrum from the native glycopeptide, (B) spectrum after treatment with X. manihotis a 1-3/6 galactosidase, (C) spectrum after treatment with green coffee bean a-galactosidase, (D) spectrum after treatment with green coffee bean a-galactosidase and X. manihotis a 1-2 fucosidase.

those of the lung are characterised by sialylation (note the A-type ions at m/z 825 and 855 in Figure 15C) or alpha-Gal capping (note the A-type ion at m / z 668 in Figure 15C). The above screening strategy is remarkably reproducible with very little variation in relative molecular ion abundance occurring between data from different experiments on the same organ or on organs from different animals. Hence

940

32 Structural Analysis

of

Oligosaccharides

A Fuc,HeqHexNAc.

.--

NeuAc,Fuc,He~Hm~NAc.

03 2dOO 2200 2200 23100 24'00 .2+00 2dOO 27'00 2dOO 2dOO 30100 31100 32'00 33'00 34'00

'm/z

Figure 14. Molecular ion region of FAB spectra obtained from screening experiments (see text) on murine (A) brain, (B) liver and (C) lung. Compositions of major signals are shown.

32.10 Applications of FAB-MS, MALDI-MS and ES-MS in Glycobiology 1

941

A

+

HexNAc

/

+

FuoHexHaxNAc

0

,

+

+

B

FucHexHaxNAc

HexNAc

\

/ +

I

+

NeuGc

HaxHexNAc

\ 374 4

I

100 953

+

NeuGcHsxHexNAc

+

d

C

T+Ac Y O C 4

lu

+

NeuAcHexHexNAc

x

Figure 15. A-fragment ion region of the spectra giving the molecular ion data shown in Figure 14: (A) brain, (B) liver, (C) lung.

942

-

32 Structural Analysis of Oligosaccharides

surface glycollpidd llpooligosaccharides

uoutermembrane"

- mycollc aclds

nner plasma membrane Figure 16. A chemical model of the mycobacterial cell wall. The two major components are the mycolyl arabinogalactan-peptidoglycan complex and the lipoarabinomannan/lipomannan. The amount and exact chemical nature of the surface glycolipids are species- and/or strain-specific.

quantitative as well as qualitative differences can be readily explored in experiments on knockout mice.

32.10.6 Case Study 6-MS

Analysis of Mycobacterial Glycoconjugates

The carbohydrate-rich bacterial cell wall presents a structural challenge of a very different nature to the case studies above. Although the available sample amount for analysis is less of a problem in comparison to the situation with bioactive mammalian or helminth glycoproteins, the extremely diverse range of glycoconjugates, as well as novel saccharide residues and non-saccharide substituents that may be present, exerts a high premium on detailed structural characterization. Studies of mycobacterial cell wall components (Figure 16) exemplify recent trends in mass spectrometric analysis in this area of biopolymer research. Sequencing of the species- and strains-specific surface glycolipids, including the glycopeptidolipids and the acyltrehalose-containing lipooligosaccharides, requires the kind of sequence-informative fragment ions afforded by FAB-MS analysis of permethyl and/or peracetyl derivatives, in addition to precise molecular weight determination [13, 141. On the other hand, structural elucidation of the two highly heterogeneous homo-polymeric components, namely the mycolyl arabinogalactan (AG) and the lipoarabinomannan (LAM), requires different strategic approaches. In the case of AG, initial efforts were directed towards GC-MS and FAB-MS analysis of the small fragments generated by partial hydrolysis. The deduced structural motifs were then pieced together to give the model shown in Figure 17 based

I

t

A&

I

Rha

Q

GICNAC

I

-1

Peptldoglycan

ddacidhyddysis

I Arabinogalactan I

I

~n

r%F!

Phosphatldyllnositol mannosldes

I

Mannan Core

I

1

I

Figure 17. Structural model of LAM and AG. A G is drawn without the mycolic acids attached to the terminal Ara residues. The molecular weight of LAM has been determined by MALDI-TOF on both native and permethylated samples. The arabinan circle in both LAM and AG represents structural details not yet defined. Detailed characterizations of the various arabinan motifs were based on MS/chemical analysis of fragments obtained through acetolysis, mild acid hydrolysis and endoarabinase digestion as indicated. The branched Aras motif is common to both AG and LAM whereas the linear Arao terminal motif is found only on LAM. An Arazzmotif as drawn is deduced to be present on A G [ 151.

Ads

;-

Galactan

I

Arabinan

Lipoarabinomannan

J

Arablnan

I

acetolysis I mild acid hydrolysis -I

Cap

944

32 Structural Analysis of Oligosaccharides

on FAB-MS analysis of larger pieces generated by very mild hydrolysis of permethylated samples [ 151. Importantly, any branching position in addition to the reducing end linkage is “visualised” by the mass difference between a methyl group and exposed hydroxyl group (which can be further tagged by deuteromethyl or deuteroethyl) as a consequence of hydrolyzing off the branch(es). Thus, as illustrated, arabinan oligomers with no additional branched point were detected as Ara 5, 6, 7, 8 and then Ara 17 since Ara 9-16 cannot be generated by partial hydrolysis with only one clip at the reducing end. The success of this approach relies on the ability of MS in detecting small mass difference at high mass range, i.e. high resolution. This important information cannot at present be obtained via direct MS analysis of non-derivatized samples. Similarly, although the approximate molecular weight distribution of intact LAM can be determined by MALDI-TOF and has been used effectively to confirm the truncation in size of LAM due to ethambutol drug inhibition [16], detailed structures of LAM can only be determined through analysis of fragments derived through both chemical and enzymatic degradation ~71. The capability of MS for examining crude mixtures (see for example Case Study 6 ) , affording precise mass values for each component present, enables rapid screening of novel modifications otherwise cryptic to conventional analysis such as the inositol cap on some LAMS [17]. Likewise, whilst the introduction of ES and MALDI has made the determination of the intact molecular weights of lipopolysaccharides (LPS), lipooligosaccharides, and other polysaccharides feasible, these techniques merely complement more detailed studies of fragments generated via partial degradation which are usually most effectively examined by FAB-MS. Thus, complete structural characterisation of LPS still involves painstaking and dedicated MS analysis at the level of the repeating units (0-antigen), core units, and lipid A with and without deacylation, just as is the case with the various structural motifs of LAM and AG (Figure 17). Finally, it should be mentioned that the LC-ES-MS methodology now widely used for detecting eukaryotic glycopeptides in proteolytic digests (see Case Study 2) has proven successful in analyzing mycobacterial glycoproteins [ 181. In this study, which provides the first firm chemical evidence for the presence of protein glycosylation in mycobacteria, the neutral loss and daughter ion scanning in ES-MS provided a very sensitive means of detection of minor glycopeptides in proteolytic digests.

32.11 Concluding Remarks As exemplified by the case studies described above, FAB-, MALDI- and ES-MS are powerful techniques for glycopolymer structure analysis. Each has its own strengths. For example, MALDI-MS .would be the method of choice for molecular weight profiling of polysaccharide mixtures (Case Study l), FAB-MS is ideally suited to screening for non-reducing epitopes in biological samples (Case Study 5),

References

945

whilst nano-ES-MS-MS on the Q-TOF provides the most sensitive means of sequencing peptides and glycopeptides (Case Study 4). However it is important to bear in mind that, with an appropriate choice of experimental strategy, many structural problems can be successfully addressed with any one of the three ionization methods. Thus, for many researchers, access to instrumentation and expertise might be more important than factors such as relative sensitivities or whether collisional activation is required to generate fragment ions. Advances in biopolymer mass spectrometry in the past 20 years have been truly breathtaking. We hope that the information provided in this Chapter will encourage readers to delve more deeply into the vast recent literature of mass spectrometric glycopolymer analysis and to use their new-found knowledge to tackle the next generation of structural problems. References 1. M. Barber, R.S. Bordoli, R.D. Sedgwick and A.N. Tyler, J. Chew. Soc. Commun., (1981) 325327. 2. A. Dell, H.R. Morris, H. Egge, G. Strecker and H.V. Nicolai, Curbohydr. Rex, 115 (1983) 4152. 3. A. Dell Adv. Curbohvdr. Chem. Biochem., 45 (1987) 19-72. 4. J.B. Fenn, M. Mann, C.K. Meng, S.F. Wong, d.M. Whitehouse, Mass Spectrom. Rev., 9 (1990) 37. 5. M. Karas, A. Ingendoh, U. Bahr. and F. Hillenkamp Biomed. Environ. Muss Spectrom., 18 (1989) 841. 6. H.R. Morris, T. Paxton, A. Dell, J. Langhorne, M. Berg, R.S. Bordoli, J. Hoyes, R.H. Bateman. Rupid Commun. Mass Spectrom., 10 (1996), 889-896. 7. H.R. Morris, T. Paxton, M. Panico, M. McDowell, A. Dell. Muss Spectrometry ojBioloqica1 Materials Vol. 2 (eds B. Larsen & C. McEwan) Marcel Dekker, New York, (1998), pp 53-80. 8. A.O. Chizhov, A. Dell, H.R. Morris, A.J. Reason, S.M. Haslam, R.A. McDowell, 0,s.Chizhov. and A.I. Usov, Carbohydr. Rex, 310 (1998), 203-210. 9. A. Dell, H.R. Morris, R.L. Easton, M. Panico, M. Patankar, S. Oehninger, R. Koistinen, H. Koistinen, M. Seppala, M. and G.F. Clark J. Bid. Chem., 270 (1995) 24116-241266. 10. S.M. Haslam, G.C. Coles, E.A. Munn, T.S. Smith, H.F. Smith, H.R. Morris, H.R. and A. Dell J. Biol. Chem., 271 (1996) 30561-30572 11. P. Teng-umnuay, H.R. Morris, A. Dell, M. Panico, T. Paxton, and C.M. West, J. Biol. Chem., 273 (1998) 18242-18249. 12. S.M. Haslam, G.C. Coles, A.J. Reason, H.R. Morris and A. Dell Mol. Biochem. Parasitol., 93 (1998) 143-147. 13. K.-H. Khoo, R. Suzuki, H.R. Morris, A. Dell, P.J. Brennan, and G.S. Besra, Carbohydr. Res., 276 (1995) 449-455. 14. K.-H. Khoo, D. Chatterjee, A. Dell, H.R. Morris, P.J. Brennan, P.J. and P. Draper, J. Bid. Chem., 271 (1996) 12333-12342. 15. G.S. Besra, M. McNeil, M., K-H. Khoo, A. Dell, H.R. Morris and P.J. Brennan, Biochemistry 34 (1995) 4257-4266. 16. K.-H. Khoo, E. Douglas, P. Azadi, J.M. Inamine, G. Besra, P.J. Brennan and D. Chatterjee, J. Bid. Chem., 271 (1996) 28682-28690. 17. K.-H. Khoo, A. Dell, H.R. Morris, P.J. Brennan, D. Chatterjee, J. Biol. Chem., 270 (1995) 12380- 12389. 18. K.M. Dobos, K.-H. Khoo, K. Swiderek, P.J. Brennan, P.J. and J.T. Belislie, J. Bucteriol., 178 (1996) 2498-2506. 19. B. Domon and C.E. Costello Glycoconj. J . , 5 (1998) 397-409.

Carbohydrates in Chemistry and Biology Editor Beat Ernst, Gerald W. Hart, Pierre Sinay copyright OWILEY-VCH Verlag GmbH, 2000

33 Conformational Analysis in Solution by NMR S. W. Homans

33.1 Introduction Of the functions that have been ascribed to oligosaccharides [ 11, their role in molecular recognition appears to be a repetitive theme in many biological processes. Recent work on the molecular genetics of glycosylation has indicated that oligosaccharides exhibit a particularly prominent role in the development of multicellular organisms [2-91. Morever, it has been known for many years that oligosaccharides are important for the invasion, infectivity and survival of parasites [ 101 and pathogens [ 111 in host cells. There is therefore compelling evidence to suggest that a role in cell-cell communication might be the primary function of oligosaccharides. This might explain the apparent lack of function of these moieties when examined at the level of a protein or indeed an individual cell. In order to understand the molecular basis of oligosaccharide-mediated recognition phenomena, we need to have knowledge of the structure and dynamics of the free carbohydrate ligand in addition to the ligand-receptor complex. In this Chapter, I shall review current knowledge on the former. A discussion of ligand-receptor complexes can be found elsewhere in this volume.

33.2 Solution Conformations of Oligosaccharides 33.2.1 The NMR Technique

Since oligosaccharides in general fail to crystallize, the only technique that is able reliably to offer information on the structure of oligosaccharides in solution at atomic resolution is nuclear magnetic resonance (NMR) spectroscopy [ 12-1 51. There are three NMR parameters that are relevant to the structural analysis of oli-

948

33 Conformational Analysis in Solution by N M R

gosaccharides. The chemical shift refers essentially to the frequency at which each discrete resonance line in the spectrum can be found. This frequency is dependent upon the precise chemical environment (hence the term chemical shift) of the nucleus that gives rise to the resonance, since this environment has the capacity to shield or deshield the nucleus from the full effects of the applied magnetic field. The chemical shift contains a wealth of information on molecular structure, but in view of the complexity of the factors that contribute to it, it is only possible to make practical use of this information in a few simple systems. Nevertheless, chemical shifts can offer useful information on carbohydrate structure when interpreted in a semi-quantitative sense [16-281. In addition to the chemical shift, each NMR resonance line typically exhibits a multiplet structure or ‘splitting’. This arises from the sensitivity of a given nuclear environment to the spin-state of nearby covalently bonded nuclei and is transmitted via the bonding electrons, becoming essentially unmeasurable over more than three bonds, at least in saturated systems. The threebetween nuclei is of considerable practical imporbond coupling (denoted 35nn’) tance, since the dihedral angle formed by the outermost pair of bonds influences the magnitude of the splitting. It was shown by Karplus [29] many years ago that an = A cos2 8 + Bcos 8 C (where A , B and C empirical relationship of the form 35nn’ are constants) can be used to derive angular information from three-bond couplings. A third parameter is known as the nuclear Overhauser effect (NOE). This parameter is not manifest in the NMR spectrum directly, but can be observed as a change in the intensity of the resonance corresponding to one nucleus when the spin-state populations of another resonance are perturbed by selective application of radio frequency energy [30]. This change in intensity is related, inter alia, to the inverse sixth power of the distance between the nuclei. Under the appropriate circumstances the NOE can therefore be used to measure internuclear distances up to -0.5 nm.

+

33.2.2 Conformational Parameters in Oligosaccharides Oligosaccharides are comprised of linear or branched monosaccharide units that are invariably linked from the C-1 position of one monosaccharide unit (the ‘glycon’) to one of several positions on the neighboring monosaccharide unit (the ‘aglycon’). With few exceptions, the chemical nature of the coupling between the monosaccharide units is an ether linkage (C-0-C), which is termed the glycosidic linkage. Therefore two rotatable bonds exist about the glycosidic linkage (Figure 1). In pyranoses (the types of monosaccharide that will almost exclusively be discussed here) the geometry of each monosaccharide ring can essentially be thought of as being fixed on the timescale of the NMR experiment, usually in the ‘chair’ (4C1) configuration. Therefore the primary sites of conformational variation are the two glycosidic torsion angles defined by the atoms H-1-C-1-0-1-C-X and C-1-0-1C-X-H-X, where C-X and H-X represent aglyconic atoms. These torsion angles are denoted +H and $H in IUPAC convention. In addition there is a third dihedral angle about the rotatable C-5-C-6 bond in each monosaccharide residue defined by H-5-C-5-C-6-0-6. This dihedral angle is of particular importance for conforma-

33.2 Solution Conformations of Oligosaccharides

949

Figure 1. Conformational torsion angles in oligosaccharides. For NMR studies, these angles are usually defined as Cp = H1-C1-01-CX, I) = Cl-01-CX-HX and o = H5-C5-C6-06, where CX and HX are aglyconic atoms.

tional studies when substitution occurs at the C-6 position and is given the symbol OH. Henceforth, the subscripts on +H, QH and OH will be dropped for convenience. 33.2.3 Conformational Restraints Since the internal geometry of each monosaccharide ring is essentially fixed, the solution conformation of an oligosaccharide can be defined by the torsion angles 4, and O. These angles can in principle be determined experimentally by measurement of conformational restraints from one monosaccharide across the glycosidic linkage to its neighbour. Thus, one or more NOEs can usually be measured from H1 of the glycon to aglyconic protons proximal to the glycosidic linkage, and two 13C-' H three-bond coupling constants can be measured across the glycosidic linkage. Unfortunately, even when three such NOEs can be measured (usually the maximum number that can be observed) together with two coupling constants, the number of restraints is such that it can be impossible to distinguish between models involving a single conformation about the glycosidic linkage rather than a model involving substantial conformational flexibility [3 11. Additional conformational restraints are required to distinguish between these two models, and much effort has been expended in recent years to achieve this goal.

+

33.2.4 I3C Isotopic Enrichment

In order to obtain additional conformational restraints in oligosaccharides, it is useful or even essential to work with the glycan in 13C-enriched form. This is by no means a trivial task. Although stable isotopic enrichment of proteins is now commonplace, this is often relatively straightforward by use of the relevant organism that over-expresses the protein in question together with isotopically enriched media. While it is possible to grow mammalian cells using such media and thus to

950

33 Conformational Analysis in Solution by N M R

prepare isotopically enriched glycoproteins [32], the cleavage of the glycans from the latter is not an efficient way to generate isotopically enriched material. It is usually more efficient to use a synthetic or chemoenzymatic strategy [33-361. The latter is particularly attractive in view of its simplicity on the milligram scale which is ideal for NMR studies. In particular, once the relevant isotopically enriched building blocks have been obtained, it is straightforward to prepare oligosaccharides with a variety of different labelling regimes to suit a particular NMR technique. In general, the protocols for the chemoenzymatic synthesis of 13C-enriched oligosaccharides can mirror those used for the synthesis of their natural-abundance 13C counterparts. However, it is of course necessary not only to have available the relevant I3C-enriched monosaccharide precursors, but also the nucleosidediphosphate derivatives that are the correct donor substrates of the relevant glycosyltransferase. Since the only monosaccharide that is available in fully 13C-enriched form at reasonable cost is glucose, a good deal of synthetic work might be necessary in order to obtain the relevant monosaccharides and their derivatives for the chemoenzymatic synthesis. As an example Figure 2 shows the protocol used in our own laboratory for the chemoenzymatic synthesis of sialyl-Lewisx,NeuSAca2-3Galpl4(Fucal-3)GlcNAc, which required 38 steps [34].

33.2.5 Additional Conformational Restraints Exchangeable Protons Conventional H NMR experiments on oligosaccharides are typically recorded in D20 solvent, in order to avoid swamping proton resonances from the oligosaccharide with the much higher concentration of protons (-1 10 M) in pure water. Under these circumstances, the hydroxyl and amide protons exchange rapidly with the deuterated solvent and become invisible in proton NMR spectra. While highly efficient solvent suppression pulse sequences have been developed recently which enable oligosaccharide proton NMR spectra to be recorded in H20 without substantial interference from the solvent resonance [37], it is still impossible to observe the hydroxyl protons at ambient temperature since they exchange too rapidly with the solvent. Pioneering work by Poppe and van Halbeek [38-401 showed that the hydroxyl proton exchange rate could be slowed sufficiently to enable the observation of discrete resonances in aqueous solution at temperatures lower than - 15 "C. This important development has enabled the measurement of NOEs to and from exchangeable protons, thus increasing dramatically the number of available conformational restraints in certain instances. One of the earliest applications of such NOEs concerns the study of the solution conformation of Neu5Aca2-6Gal~l-4Glc [41]. These authors also demonstrated that the observation of hydroxyl proton resonances can be very useful in demonstrating the existence of long-lived hydrogen bonds in solution. Such hydrogen bond restraints can be very powerful in the conformational analysis of oligosaccharides, since unlike the NOE, the presence of a hydrogen bond restrains the relevant atoms to a narrow range in the region of 0.25

-

33.2 Solution Conjiwmutions of Oligosuccharides OH OH

951

OH

O-"DP

OH

OH OH

OH

OH

I

O

NHAc H

W

OHE & O *

Galactosyltransferase

NHAc

=

O

OH

/

*

OH

OH

06

Milk Fucosyltransferase

I

OH

NHAc

OH

NHAc

OH

C H ; F - G D P OH

b *

0 -

H O

NHAc

NHAc

on

OH

C H ~ @ ~ ~ OH

Figure 2. Scheme for the chemoenzymatic synthesis of SLe" in 13C-enrichedform. The donor substrates UDP-Gal, PNP-NeuSAc and GDP-Fuc are typically prepared synthetically from uniformly "C-enriched glucose, giving rise to full 13C-enrichment in the glycan moiety of the donor. This scheme works efficiently on the milligram scale providing adequate material for NMR studies.

nm. Hydrogen-bonded hydroxyl protons present significantly smaller exchange rates in comparison with those that are not hydrogen bonded, and also possess significantly smaller temperature coefficients (i.e. the change in chemical shift with temperature). In addition, hydrogen bonded hydroxyl protons often exhibit scalar coupling constants to the adjacent non-exchangeable proton that differ substantially from the -5.5 Hz coupling that is observed when the hydroxyl group is free to rotate. In the case of Neu5Aca2-6Gal~1-4Glc,Poppe et al. [41] observed anomalous values for these parameters for NeuSAc OH8, NeuSAc OH7 and Glc OH3, confirming earlier reports of a hydrogen bond between Neu5Ac OH8 and the ring oxygen or carboxyl group of the Neu5Ac residue [38], and suggesting the presence of a hydrogen bond between Glc OH3 and the ring oxygen of the Gal residue. One

952

33 Conformational Analysis in Solution by N M R

further approach for defining hydrogen bond connectivities is to make use of the deuterium isotope effect on I3C chemical shifts. At low temperatures (248-268 K) and neutral pH values, the exchange of hydroxyl protons in water is slow enough to observe these short-range effects, which arise from the influence of the O-D group on the chemical shift of the parent carbon resonance. This effect is most pronounced for hydroxyl groups involved in long-lived hydrogen bonds, and can readily be detected as a 'splitting' of the carbon resonance into two signals in 50%H20/50%D20 solution [42-441. Isotopic enrichment is clearly not a prerequisite for the observation of hydroxyl protons in oligosaccharides. However, use of enriched material does simplify the attenuation of the strong H20 resonance and moreover permits spectral editing in a third I3C dimension that can be very useful in overcoming resonance overlap that is typically observed in the hydroxyl region of the 'H spectrum [45]. Heteronuclear Overhauser Effects It has long been recognized that useful structural information can be obtained from heteronuclear 13CC1H}Overhauser effects [46-48, 30, 49, 501, and it follows that such measurements might be of value in the conformational analysis of oligosaccharides. A principal difficulty with this approach is that NOES from H to 13C are of significant intensity only for quaternary carbons [41]-protonated carbons do not exhibit a substantial NOE by virtue of the efficient dipolar relaxation by the attached proton and consequent leakage of the NOE. A further complication is that an indirect NOE often exists from the source proton through the attached proton to the carbon, and since this NOE is opposite in sign to the direct effect and often larger, I3C{'H} NOEs in carbohydrates are usually negative [51]. This effect can readily be demonstrated from full-relaxation matrix simulations on a simple disaccharide (Figure 3 ) . A suitable means by which the l3CC1H}NOE intensity can be increased to more useful levels is to perdeuterate the aglycon. The indirect effect via protons is thus abolished, and permits the direct trans-glycosidic 13C{'H} NOE to be observed without interference. Given the substantial potential values of 13C{'H} NOEs for the conformational analysis of oligosaccharides, it is worthwhile considering the appropriate experimental regime for their measurement. Na'ively, one would choose steady-state NOE experiments where the source proton is saturated by an irradiating field, since theoretically it can be shown that these NOEs reach a value more than ten times greater than the transient NOE method. However, steady-state methods involve detection of 13C, whereas by comparison H detection offers approximately eightfold enhancement of signal-to-noise [50]. Given the difficulties with selective saturation of protons in very crowded regions of carbohydrate spectra, the transient method, which is at the heart of two-dimensional methods [46,47], is the method of choice. The applications of 13CC1H}NOEs for the conformational analysis of oligosaccharides have thus far been quite limited. However, they have been measured at natural abundance in Neu5Aca2-6Galp1-4Glc [41] and in I3C, 2H enriched Galpl-4Glc [51].

'

33.2 Solution Conformations of Oligosaccharides

-0.5

953

9 0

1

2

3

4

5

6

Mixing time (s)

0.6 h . c1I

B

c1

0.4

E

. I

0.2

0.0

Mixing time (s) Figure 3. Theoretical full-relaxation matrix simulations of H{ I3C} NOES across the glycosidic linkage in GalP1-4[U-'3C]Glc (top panel) and GalP1-4[U-13C,2H]Glc(bottom panel). Traces show Glc C-4, (0) Glc the NOE intensities to Gal H-1 vs mixing time in a HOESY experiment for (0) C-5 and (0)Glc C-6.

1 3 ~ - ' 3 Coupling-Constants ~

In recent years the value of long-range 13C-13C coupling constants for the conformational analysis of oligosaccharides has been realized [ 52-55, 56-58]. A principal difficulty with the measurement of these parameters is the requirement for 13C-enriched material-unlike heteronuclear 13C-' H measurements where the -1% natural abundance of I3C is not a substantial sensitivity barrier, 0.01% of

954

33 Conformational Analysis in Solution by N M R

molecules will contain two 13Cnuclei. For this reason most applications of 13C-13C coupling constant measurements have concerned bacterial polysaccharides or small oligosaccharides where isotopic enrichment can be achieved either biosynthetically or by chemoenzymatic synthesis. The most convenient approach for the measurement of these couplings appears to be the 'LRCC' method developed by Bax and co-workers for application to proteins [ 591, and several Karplus parametrizations have been reported with which to convert the relevant coupling constant into angular information for conformational analysis [56, 57, 601. Dipolar Couplings Recently, work on the inherent magnetic alignment of proteins in very high magnetic fields (>14 T) has permitted the measurement of residual dipolar couplings between NMR active nuclei [61-631. These couplings, which average to zero for an isotropically tumbling macromolecule, present a small but finite value due to the small net alignment in the magnetic field. Importantly, residual dipolar couplings provide long-range structural information since their magnitude depends, inter aka, on the inverse cube of the distance between the two nuclei and the function (3 cos20 - l), where 0 is the angle between the relevent bond vector and the principal axis of the alignment tensor [61]. The long-range structural information derives from the fact that 0 for each bond vector is referenced to the same axis system, and hence the angles in disparate regions of the molecule are directly related. Since the original observation of magnetic alignment in proteins, more recent work has demonstrated that it is possible to obtain alignment two orders of magnitude greater by use of dilute liquid-crystalline solvents [ 641. These solvents comprise dihexanoylphosphatidylcholine (DHPC) and dimyristoylphosphatidylcholine (DMPC) in aqueous solution, which form disc-like micelles ('bicelles') above the transition temperature (35 "C) that align in a magnetic field [65]. Macromolecules that are not spherically symmetric, when dissolved in these solutions, become partially aligned by virtue of their hydrodynamic properties (Figure 4). Below the transition temperature (27 "C) the solution becomes isotropic, and the dipolar coupling is no longer observed. Thus it is straightforward to measure dipolar couplings for each bond vector by comparison of the splittings at 35 "C, which will be the sum of the scalar and dipolar couplings, with those at 27 "C, which will exhibit only the scalar couplings. An example of the measurement of 13C-'H dipolar couplings from heteronuclear single quantum correlation spectra is shown in Figure 5. The dipolar couplings can be measured to reasonable accuracy in this manner, but for higher accuracy J-modulated or IPAP methods can be used [62, 661. The measured dipolar couplings are related to 0 by the following expression:

where S is the generalized order parameter for internal motion of the bond vector PQ, po is the magnetic permeability of vacuum, yp and yQ are the magnetogyric ratios of P and Q, h is Planck's constant, Y ~ Qis the distance between P and Q, A , and A , are the axial and rhombic components of the alignment tensor A, and 0 and

33.3 Experimental Restraints in Conformational Analysis

955

C

I

Bo

Figure 4. Diagrammatic illustration of the partial alignment of an anisotropic macromolecule in a liquid crystalline solution of phospholipid bicelles. The bicelles align spontaneously with respect to the applied magnetic field (Bo),and the macromolecule gains a small net degree of alignment by virtue of its physicochemical properties in the spaces between the bicelles.

4 are cylindrical coordinates describing the orientation of the vector P-Q in the principal axis system of A [64]. It is straightforward to extract the relevant angular information from this dependence, which gives rise to knowledge of the orientations of the relevant bond vectors with respect to a single fixed axis, thus providing longrange structural information that hitherto has been lacking in NMR studies.

33.3 Experimental Restraints in Conformational Analysis 33.3.1 Restraining Protocol

The principal purpose of the collection of the parameters described above is as restraints in conformational analysis. These restraints are incorporated in molecular mechanical simulations as pseudo-energy functions which apply an energy penalty if the theoretical parameter deviates from the experimental value according to some prescribed formula. However, care must be exercized in applying experimental restraints in conformational analysis. This is because in general it is impossible to exclude motional averaging about the glycosidic linkages (see Section 33.4). Under these circumstances, the relevant NMR parameter represents an average over a number of possible different configurations of the molecule, and if the restraints are applied without consideration of this possibility (i.e. assuming the molecule is ‘rigid’), there is every possibility that the resulting structure will be a ‘virtual conformation’ that bears little resemblance to reality [67, 681. Fortunately this problem has been recognised very early in structural studies of macromolecules, and protocols have been devised to deal with it. Biharmonic Restraints

In the case of distance restraints derived from NOE measurements or from hydrogen-bond connectivities, it is usual to apply a restraining function in the form of a

956

33 Conformational Analysis in Solution by N M R 654

66

a

/i

73

77

78 79 80 q,

, . , I , , . , I , , . . I , , , , I , , . . , . , . . I , . , , I , , , , I , , , , I , . , , I , , , . I . , ,

4.5

4.4

4.3

4.2

4.1 4.0

3.9

3.8

3.7

3.6

3.5

F2 (PPm)

0

4.5

4.4

4.3

4.a

4.1

4.0

3.9

0

3,s

3;7

3,6

3i5

F2 ( m m )

Figure 5. Proton coupled ‘H-I3C HSQC spectra of uniformly I3C-enrichedGalal-4Gal~1-4Glcin a 7.5% solution of DHPC:DMPC (1 :2.9) at (a) 27 “C (isotropic phase) and (b) 35 “C (liquid crystalline phase). The difference in ‘Jc,H for Gala H-5 is shown in each case.

biharmonic potential [69]. That is, there is no energy penalty if the theoretical distance lies within prescribed distance bounds, whereas an increasing penalty function is applied if the distance is outwith these bounds. Typically, the NOE restraints are semi-quantitatively categorized into ‘weak’, ‘medium’ and ‘strong’, with corre-

33.3 Experimental Restraints in Conformational Analysis

957

sponding distance bounds of 1.SA < r < 2.7A, 1.8A < r < 3.3A and 1.SA < r < 5 A respectively. In this manner the molecule is free to adopt any conformation without additional energy penalty provided all theoretical distances are within the relevent bounds. The relatively loose bounds are supposed to take account of the fact that the measured NOE may correspond to a motionally averaged distance. Incorporation of angular restraints is more difficult. Unlike the NOE, where the measurement of this parameter demonstrates at least that the relevant atoms must be close in space for some of the time, a given spin-coupling constant may correspond with a single, fixed angle in one extreme, or may represent an average over 360" rotation at the other extreme. Although methods have been proposed to overcome this limitation, in general it is undesirable to apply angular terms as primary restraints in conformational analysis. Time-Dependent Restraints

A principal limitation of conventional biharmonic restraints is that their presence can mask transitions between conformational states that are widely separated on the potential surface. Thus, if during the course of a transition from one low-energy state to another, distance restraint is violated (the distance between two restrained atoms exceeds 5 A, for example), then a substantial energy barrier to this transition will be created. In order to overcome this problem, restraints can be imposed in a so-called 'time-dependent manner' [70]. The basic concept is that the theoretical average distance over a defined part of the simulation, rather than the theoretical instantaneous distance, is required to satisfy certain bounds during the simulation. By use of this protocol, violations of conventional distance bounds are permitted, as long as the running average satisfies these bounds. Thus transitions between different low-energy minima are more likely to occur in simulations using the time-averaging protocol. Remarkably, to our knowledge very few studies on glycoconjugates have utilized time-averaged restraints [71, 721.

33.3.2 Dynamical Simulated Annealing

In principle, conformational restraints can be incorporated into conventional energy minimization algorithms in order to obtain a single, fixed, 'minimum energy' conformation that is consistent with available experimental data. However, this approach is not very efficient, and moreover by virtue of the local minimum problem, it is very likely that the resulting structure will not be the global minimum energy configuration. A much more efficient approach is termed dynamical simulated annealing [69]. This can be thought of as a hybrid between a molecular dynamics simulation and energy minimization. The idea is to simulate the dynamics of the molecule at a relatively high temperature (typically 750 K) over a period of time in order efficiently to sample a substantial region of conformational space. The simulation is continued while the system is then slowly cooled to a very low temperature. Unlike conventional minimization, the system has sufficient thermal energy so there is a tendency for the molecule to escape local minima (provided the cooling is slow

958

33 Conformational Analysis in Solution by N M R

enough) and to ‘fall’ to a lower energy minimum. Once the system has cooled to around 5 K, the simulation is terminated with a conventional energy minimization step. Throughout the simulation the experimental restraints are applied, such that the molecule is constantly driven to a minimum energy conformation that is hopefully consistent with all experimental data. In order that the restraints do not limit the accessible regions of conformational space in the early stages of the simulation at high temperature, their strength is typically scaled for a period of time until the temperature falls to around 300 K. Moreover, a series of simulations is usually performed with a randomized starting structure as input, in order further to explore conformational space and to ensure that the final structure is not dependent on the starting geometry. This typically results in one or more families of structures that are essentially consistent with the applied distance restraints.

33.4 Analysis of Oligosaccharide Dynamics Oligosaccharides do not exist as ‘rigid’ entities in solution. While each monosaccharide residue can be thought as being fixed in a certain conformation (at least for pyranoses) on the NMR timescale, there may be substantial torsional oscillations about the glycosidic linkages. In extended structures, these oscillations will be additive, and the manifold of conformations is solution can span considerable regions of conformational space. Characterization of the solution dynamics is very important from the point of view of binding affinity to receptors. To appreciate why this is the case it is convenient to consider a simple ligand-protein association involving a flexible ligand (A) in comparison with a rigid ligand (B). The configurational entropy (i.e. number of available degrees of freedom) of the former is greater than the latter, and hence the Gibbs free energy of the former will be lower by virtue of the relation G = H - TS (where H is the enthalpy and T is the absolute temperature). Thus, the combined free energy of the uncomplexed components will be lower in the case of the flexible ligand and, since the free energy of binding A& is the difference between the free energy of the complexed (which will be the same irrespective of the dynamics of the ligand, assuming it is rigidly held in the binding site) and the uncomplexed components, the free energy of binding of the flexible ligand will be lower than that of the rigid ligand. This in turn will give rise to a lower affinity by virtue of the relation AGO = -RTlnK,, where AGO is the standard free energy of binding, R is the gas constant and K, is the association constant. Thus in terms of inhibitor design, it is important to consider the dynamics of the ligand, i.e. a rigid ligand that adopts the bound-state conformation in solution is likely to have optimal affinity. The characterization of oligosaccharide dynamics by NMR is not trivial. All of the conformation-sensitive NMR parameters are averaged by the motion, and it is not possible directly to deconvolute the nature or number of conformational states that contribute to this average. NMR relaxation parameters are directly sensitive to motion if on the appropriate timescale [73, 741, but again cannot be interpreted in

33.5 A Case Study on NeuSAcd-3Gal~I-4Glc

959

the absence of a model for the motion [75]. Thus we are forced to rely on methods that simulate the regions conformational space that an oligosaccharide can access in solution, and two methods are in general use. 33.4.1 Monte-Carlo Simulations In this method [76-781 the starting point is an arbitrary conformation of the system which is perturbed slightly by, for example, a torsional rotation. The internal energy of the system is calculated and compared with the energy before the perturbation. If the energy has decreased then the new conformation is accepted. If the energy has increased, the increase is compared with the thermal energy k T is order to decide whether the new conformation is accepted. The value of exp(-AU/kT) is compared with a random number between 0 and 1, where AU is the increase in energy. If exp(-AU/kT) is larger than the random number, the new conformation is accepted. Otherwise a new conformation is generated, and the process is repeated. In this manner the system under investigation explores conformational space. Since perturbations that raise the energy are sometimes allowed, the procedure permits the escape of the molecule from local minima. 33.4.2 Molecular Dynamics Simulations The molecular dynamics method directly simulates the motions of all the atoms in the system, by using Newton’s laws of motion [79]. Since the intenal energy of a given conformation of the system can be derived from a conventional molecular mechanics approach, the force on each atom can be determined from the derivative of the energy along different directions of space. The motion of the atom can in turn be determined from the acceleration which is a known quantity given the force on each atom and its mass. The system is usually started in an arbitrary conformation near 0 K, and the velocities of the atoms are then increased so that the average kinetic energy corresponds with a desired temperature. One of the advantages of the molecular dynamics method is that the temperature can be raised or lowered at will by coupling to a temperature bath, which thus permits energy minimization by simulated annealing (Section 33.3.2). Moreover, since the time-scale of the internal motions is known from the simulation, it is possible to back-calculate time-averaged theoretical NMR parameters for comparison with experiment (Section 33.5.4).

33.5 A Case Study on NeuSAca2-3Galp1-4Glc As an example of the use of some of the techniques described above in the conformational analysis of oligosaccharides, a case study of the solution properties of the trisaccharide NeuSAca2-3GalS 1-4Glc will be described.

960

33 Conformational Analysis in Solution by N M R

33.5.1 Resonance Assignments in NeuSAca2-3Galj31-4Glc The first step in the conformational analysis of the oligosaccharide is the determination of complete proton and carbon resonance assignments. These have been reported previously for this particular trisaccharide [ 801. In systems with unknown assignments, the presence of uniform 13C-enrichment permits the application of conventional HCCH-COSY and HCCH-TOCSY experiments [81, 821, which invariably give complete resonance assignments in an efficient manner.

33.5.2 ROE Connectivities Although NeuSAca2-3Galfil-4Glc contains only 29 non-exchangeable protons, the proton NMR spectrum is remarkably complex, with most resonances concentrated within 0.4 ppm. Despite the fact that complete 'H and 13C resonance assignments are available, this overlap renders impossible analysis of the conventional H-' H NOESY spectrum. Isotopic 13C-enrichment permits the acquisition of threedimensional I3C-edited spectra at high sensitivity. Since the rotational tumbling time of the trisaccharide is close to the point where the homonuclear 'H-'H NOE is zero, it is necessary to acquire three-dimensional 13C edited ROESY-HSQC spectra. It is convenient to examine such spectra as a series of two-dimensional planes. In NMR of proteins, it is usual to examine the Fl/F3 ('H/'H) plane, but due to severe overlap of proton resonances, it is more useful to examine F2/F3 (I3C/'H) planes in oligosaccharide spectra. A typical F2/F3 plane from the ROESY-HSQC spectrum of NeuSAca2-3Galp1-4Glc is shown in Figure 6 [56]. This illustrates all the ROE connectivities derived from Gal H-1. Many of these correspond with intraresidue ROEs that do not contain conformational information across the glycosidic linkages. The remainder (crosspeaks plotted in boldface) represent ROE connectivities across the GalPl-4Glc glycosidic linkage. The ROE connectivities to Glc H-4 and Glc H-6 are anticipated on the basis of the known conformational preferences about the glycosidic linkage. However, substantial ROEs are also observed to Glc H-3 and H-5. These ROEs are extremely difficult to observe in conventional homonuclear H NMR spectra due to extensive overlap, and given their substantial intensity, they cannot arise from a single conformation about the Galb1-4Glc linkage that simultaneously gives rise to ROEs to Glc H-4 and H-6. Thus these measurements alone indicate that a degree of flexibility exists about the Galpl-4Glc glycosidic linkage. In particular, the ROEs to Glc H-3 and H-5 demonstrate the existence of the 'anti' conformation about the glycosidic linkage, i.e. where the glycosidic torsion angle $ adopts a value of -180" [83-861. By examination of other F2/F3 planes in the ROESY-HSQC spectrum, a total of seven transglycosidic ROE connectivities is observed for this glycan. These can be used as conformational restraints in the determination of the 'global minimum' energy configuration of the glycan .

'

'

33.5 A Cuse Study on Neu5Acc12-3Gul/lI-4Glc

961

F2

(PP4 62 64 66 68

3c

70 72 74

,

76

78

Gal H3

80 wpGlc H4

82 I""I""I""I""I""I""I""I""I~~"I""I""~

4.3

4.1

3.9

3.7

3.5

3.3

l H F3 (PPm) Figure 6. Two-dimensional F2/F3 ( I3C/'H) plane derived from three-dimensional ROESY-HSQC spectrum of NeuSAca2-3Gal~1-4Glc.This plane shows all intra- (normal face) and inter-residue (boldface) ROE connectivities to Gal H-1 .

33.5.3 'Global Minimum' Conformation of NeuSAcaZ3Gal~l-4Glc The 'global minimum' energy conformation of the glycan can be determined by dynamical simulated annealing calculations with the ROE connectivities described above as time-averaged conformational restraints. In general it is desirable to compute a series of such calculations with different (pseudo-random) conformations of the glycan as input. In principle, it is possible to begin with random coordinates, but this introduces a substantial complication since it is necessary to ensure that the correct chirality is preserved, and moreover the correct ring geometry must be defined. In practice, the chirality and ring geometry is known, and hence it is more convenient to generate a series of random structures with defined chirality and ring geometry but with pseudo-randomized torsion angles about each glycosidic linkage [87]. A set of such pseudo-random structures can be generated by use of a dynamical quenching procedure.

962

33 Conformational Analysis in Solution by N M R

NeuNA

Figure 7. Stereo view of the family of structures derived from a dynamical simulated annealing calculation on NeuSAca2-3Galfi1-4Glc. The three families are labelled ‘A’, ‘B’ and ‘C’ respectively.

The result of the dynamical simulated annealing calculations is a set of structures most of which satisfy the applied conformational restraints. Those that do not are discarded, and the remaining structures typically form ‘families’ with similar conformations (Figure 7). These data therefore indicate that different conformations satisfy the experimental ROE restraints simultaneously, and thus provide a crude picture of motional dynamics. A more detailed analysis of dynamics can be obtained from molecular dynamics simulations.

33.5.4 Conformational Dynamics of NeuSAca2-3Galp1-4Glc In order to probe the conformational dynamics of the glycan, one of the low-energy structures derived from the simulated annealing procedure is chosen as input to a restrained molecular dynamics simulation. The conformer with the lowest overall energy can arbitrarily be chosen for this purpose. In principle, it would be desirable to perform free dynamics simulations with explicit inclusion of solvent water in order to obtain an accurate picture of the solution dynamics of the glycan. However, the accuracy of current forcefield parametrizations for oligosaccharides is insufficient over the complete potential surface to generate an accurate model of solution dynamics. Moreover, the inclusion of solvent water molecules severely restricts the length of time over which the molecular dynamics simulation can simulated due to the very large number of computations that are required per step. Thus it is practical only to simulate the effects of solvent water in part by use of an appropriate dielectric constant. In adopting this approach, it must be remembered that a restrained molecular dynamics simulation will be performed which will include data from solution NMR studies, including hydrogen-bonding restraints where appropriate. It is therefore not necessary for the simulation conditions to reproduce accurately these solution properties. It can be argued that the application

33.5 A Case Study on Neu5Aco12-3Gu1~1-4Glc

2 O

O 2

1

‘-180

180

I

Phi (deg)

963

i

-180

180

Phi (deg)

+

Figure 8. Instantaneous values of vs $ for (a) NeuSAca2-3Gal and (b) GalP1-4Glc glycosidic linkages derived from a 5 ns in vacuo MD simulation of Neu5Aca2-3Gal~l-4Glc.

of experimental restraints will of course modify the available conformational space explored by the glycan during the simulation. However, in the absence of wellparametrized forcefields for oligosaccharides there is little option but to modify the behavior of the forcefield in this manner, and the use of time-dependent restraints minimizes the influence of such restraints in e.g. preventing excursions to other local minima via a substantial energy barrier. The result of a 5 ns restrained MD simulation for Neu5Aca2-3GalP1-4Glc is shown in Figure 8. It can be seen that each glycosidic linkage exhibits considerable conformational freedom, and in particular the GalPl-4Glc glycosidic linkage is seen to adopt the ‘anti’ conformation for a considerable period of time. The validity of this simulation can be further assessed by back-calculating the theoretical ROES from the MD simulation for comparison with experimental values, using the appropriate formalism for the computation of the time-averaged ROE. The backcalculation of theoretical parameters that are used as restraints in the simulation is not a circular argument, since such restraints are typically applied as a biharmonic function with lower and upper bounds that differ substantially, such that the theoretical ROE can in principle vary over a very wide range. Moreover, other conformation-sensitive parameters described above such as long range I3C- ‘H and I3C13Ccoupling constants can be back-calculated from the simulation for comparion with experimental values that are not used as restraints in the simulation. As seen in Table 1, the theoretical results for Neu5Acu2-3Gal~l-4Glcagree well with experimental values, suggesting that the dynamical properties defined by the simulations are a good approximation to the true solution dynamical behavior. 33.5.5 Short-range vs Long-range Restraints

Despite the good agreement between theoretical and experimental parameters in Table 1, a valid criticism of these studies (or indeed all NMR studies on macromolecular conformation until recently) is that all of the conformational parameters are short-range in nature. Thus by virtue of the inverse sixth power distance dependence of the NOE(ROE), distance restraints are typically limited to 0.5 nm or

964

33 Conformational Analysis in Solution by N M R

Table 1. Experimental ROE intensities and long-range coupling constants for I3C Neu5Aca23Galpl-4Glc vs theoretical values computed from a 5 ns MD simulation. ROE Connectivity

ROE Neu5Ac H-3ax-Gal H-3 Neu5Ac H-8-Gal H-3 Gal H-1-Glc H-4 Gal H-1-Glc H-6 Gal H-1-Glc H-6’ Gal H-1-Glc H-3 Gal H-1-Glc H-5

Jcc Gal C-3-NeuSAc C-3 Gal C-2-NeuSAc C-2 Gal C-4-Neu5Ac C-2 Gal C-l-GlcP C-5 Gal C-2-GlcaIP C-4 Gal C-l-Glca/P C-3

ROE Intensity

(%)l

Expt.

Theor.’

1 .o 0.4 4.2 0.4 0.5 0.4 0.5

1.5 0.6 3.9 0.6 0.1 0.4 0.6

Coupling (Hz) Expt.3

1.9 1.8

Theor.

1.9 2.6 <1

1.6 1.8 1.1 1.6 2.6 1.1

3.5 5.1 4.7

3.1 4.9 4.5


JHC

Gal H- 1-Glc C-4 Gal C- 1-Glc H-4 Neu5Ac C-2-Gal H-3

Experimental and theoretical intensities shown are for the Glcp anomer only, after correction for the mole fraction of this anomer. Calculated with a rotational correlation time of 0.13 ns. 3Error in these measurements estimated as k0.5 Hz.

less, and the long-range coupling constants provide conformational information on two bond vectors separated by a single covalent bond. It could be argued that such restraints satisfy the local order but do not define long-range order in the molecule. Thus, for extended structures of which polysaccharides are a prime example, the overall geometry of the molecule could be in error. This difficulty has recently been addressed by measurement of residual dipolar couplings as described above (Section 33.2.5). Typical such couplings, measured for NeuSAca2-3Gal~1-4Glcusing the J-modulated HSQC method [88], are given in table 2. These couplings can be applied as restraints in dynamical simulated annealing calculations, in order to obtain a family or families of structures which are consistent with the long-range order imposed by these restraints. The result for Neu5Aca2-3Gal~1-4Glcis two families of structures that differ in the conformation about the GalPl-4Glc linkage (Figure 9). Importantly, both of these structures have glycosidic torsion angles that map into low-energy regions of conformational space predicted from MD simulations involving ‘local’ restraints, therefore further suggesting that the latter provide an

33.5 A Cuse Study on N e u S A c d - 3 G u l ~ l - 4 G k

965

Table 2. Residual 'H-I3C Dipolar couplings for Neu5Aca2-3Galp1-4Glc in a 7.5% (w/v) solution of DHPC:DMPC (1 :2.9 w/w)in D20, pD 7.2, containing 100 mM KC1. Bond Vector

Residual dipolar coupling (Hz)'

GlcP H-1-C-1 Glcp H-2-C-2 Glcp H-3-C-3 GlcP H-4-C-4 GalP H-1-C-1 GalP H-2-C-2 GalP H-3-C-3 GalP H-4-C-4 GalP H-5-C-5 Neu5Acu H-4-C-4 Neu5Acu H-5-C-5 Neu5Acu H-7LC-7 Neu5Aca H-8-C-8

$7.4 +9.5 +9.6 +7.7 +11.5 $11.9 $9.2 -1.4 +9.5 f3.2 +4.8 -13.4 -11.1

' Values obtained by non-linear least-squares fitting of experimental intensities from J-modulated HSQC experiments. Estimated average error in the measurements is k 0 . 5 Hz.

NeuNAc

Gal

8 r

L

'-180

J 180

Phi (deg)

O ' L 2

k

-

'-180

l 180

Phi (deg)

Figure 9. (top) Family of structures derived from dynamical simulated annealing calculation on Neu5Aca2-3Ga1fi1-4Glc with residual dipolar coupling restraints given in Table 2. (bottom) Resulting values of and $ (open circles) for the NeuSAca2-3Gal and Galpl-4Glc glycosidic linkages indicated on the plots of Figure 8.

966

33 Conformational Analysis in Solution by N M R

adequate representation of the dynamic behavior of the glycan in solution. Clearly, further work is required to calculate time-averaged residual dipoar couplings from MD simulations for comparison with experimental values.

33.6 Conclusions Since early studies on oligosaccharides which treated these moieties as essentially ‘rigid’ bodies [89-931, a wealth of more recent data has indicated that oligosaccharides enjoy considerable motional freedom about the glycosidic linkages in solution. These conclusions are derived from improved computational procedures and better molecular mechanical forcefields, but the availability of additional experimental conformational restraints has also had a particularly significant impact. In particular the possibilities for the synthesis of oligosaccharides in 13C-enriched form using chemoenzymatic methods simplifies considerably not only the task of defining the solution behavior of the free glycan, but also the conformation while bound to a protein receptor. A description of techniques for the study of the latter can be found elsewhere in this volume.

References A. Varki, Glycobiology, 1993, 3, 97-130. E. Ioffe, P. Stanley, Proc. Natl Acad. Sci. U.S.A., 1994, 91, 728-732. P. Stanley, E. Ioffe, Faseb Journal, 1995, 9, 1436-1444. E. Ioffe, Y. Liu, P. Stanley, Proc. Nut1 Acad. Sci. U.S.A., 1996, 93, 11041-11046. P. Stanley, T. S. Raju, M. Bhaumik, Glycobiology, 1996, 6, 695-699. D. Chui, M. OhEda, Y. F. Liao, K. Panneerselvam, A. Lal, K. W. Marek, H. H. Freeze, K. W. Moremen, M. N. Fukuda, J.D. Marth, Cell, 1997, 90, 157-167. 7. J. J. Priatel, M. Sarkar, H. Schachter, J. D. Marth, Glycobiology, 1997, 7, 45-56. 8. M. Bhaumik, T. Harris, S. Sundaram, L. Johnson, J. Guttenplan, C. Rogler, P. Stanley, Cancer Rex, 1998, 58, 2881-2887. 9. T. S. Raju, P. Stanley, J. Bid. Chem., 1998, 273, 14090-14098. 10. M. A. J. Ferguson, Phil. Trans. R. Soc. Lond. Ser. B, 1997, 352, 1295-1302. 11. K. Aktories, Curr. Topics Microbiol. Immunol., 1992, 175, 1-143. 12. J. F. G. Vliegenthart, L. Dorland, H. van Halbeek, Adv. Carbohydr. Chem. Biochem., 1983, 41, 209-374. 13. S. W. Homans, Progr. N M R Spectr., 1990,22, 55-81. 14. H. van Halbeek, Meths. Enzymol., 1994,230, 132-168. 15. H. van Halbeek. Curr. Ouinion Str. Biol.. 1994. 4. 697-709. 16. J. F. G. Vliegenthart, L. Dorland, H. van Halbeek, Adv. Carbohydr. Cliem. Biochem., 1983, 41, 209-374. 17. P. E. Jansson, L. Kenne, G. Widmalm, Carbohydr. Res., 1987, 168, 67-77. 18. G. M. Lipkind, A. S. Shashkov, Y. A. Knirel, E. V. Vinogradov, N . K. Kochetkov, Carbohydr. Res., 1988, 175, 59-75. 19. A. S. Shashkov’, G. M. Lipkind, Y. A. Knirel, N. K. Kochetkov, Mag. Res. Chem., 1988, 26, 735-147. 20. P. E. Jansson, L. Kenne, G. Widmalm, Pure Appl. Chem., 1989, 61, 1181-1192. 1. 2. 3. 4. 5. 6.

References

967

P. E. Jansson, L. Kenne, G. Widmalm, Curhohydr. Rex, 1989, 193, 322-325. P. E. Jansson, L. Kenne, G. Widmalm, CLzrhohydr. Res., 1989, 188, 169-191. G. M. Lipkind, A. S. Shashkov, N. K. Kochetkov, Curhohydr. Res., 1990, 198, 399-402. H. Baumann, P. E. Jansson, L. Kenne, G. Widmalm, Carhohydr. Res., 1991, 211, 183-190. P. E. Jansson. L. Kenne, G. Widmalm, A n d . Bichenz., 1991, 199, 1 1 17. P. E. Jansson, L. Kenne, G. Widmalm, Actu Chem. Scandin., 1991, 45, 517-522. N. K. Kochetkov, E. V. Vinogradov, Y. A. Knirel, A. S. Shashkov, G. M. Lipkind, Bioorg. Khim., 1992,18, 116-125. 28. N. E. Nifantev, A. S. Shashkov, G. M. Lipkind, N. K . Kochetkov, B. Jann, K. Jann, Glycoconj. J.; 1993, 10, 335-335. 29. M. Karplus, J. Am. Chem. Soc., 1963, 85, 2870-2871. 30. D. Neuhaus, M. P. Williamson, The Nwlear Overliuuser EfSect in Structurul und Conformationul Analysis, Verlag Chemie, New York 1989. 31. T. J. Rutherford, J. Partridge, C. T. Weller. S. W. Homans, Biochemistry, 1993, 32, 1271512724. 32. C. T. Weller, J. Lustbader, K. Seshadri. J. M. Brown, C. A. Chadwick, C. E. Kolthoff, S. Ramnarain, S. Pollak, R. Canfield, S. W. Homans, Biochemistry, 1996, 35, 8815-8823. 33. Y. Ichikawa, Y.-C. Lin, D. P. Dumas, G.-J. Shen, E. Garcia-Junceda, M. A. Williams, R. Bayer, C. Ketcham, L. E. Walker, J. C. Paulson, C.-H. Wong, J. Am. Chem. Soc., 1992, 114, 9283-9298. 34. M. A. Probert, M. J. Milton, R. Harris, S. Schenkman, J. M. Brown, S. W. Homans, R. A . Field, Tetruhed. Lett., 1997, 38, 5861-5864. 35. A. K. Misra, J. M. Brown, S. W. Homans, R. A. Field, Curhohydr. Lett., 1998, 3, 217-222. 36. H. Shimizu, J. M. Brown, S. W. Homans, R. A. Field, Tetrahedron, 1998, 54, 9489-9506. 37. T. L. Hwang, A. J. Shaka, J. Magn. Reson., 1995, A 112, 275-279. 38. L. Poppe, H. vanHalbeek, J. Am. Chem. Soc., 1991, 113, 363-365. 39. L. Poppe, H. van Halbeek, J. Mugn. Reson., 1992, 96, 185-190. 40. L. Poppe, H. van Halbeek, Nuture Struct. B i d , 1994, 1, 215-216. 41. L. Poppe, R . Stuikeprill, B. Meyer, H. van Halbeek, J. Biomol. N M R , 1992,2, 109-136. 42. J. Reuben, J. Am. C’hem. Soc., 1984, 106, 6180-6186. 43. J. C. Christofides, D. B. Davies, J. A. Martin, E. B. Rathbone, J. Am. Chem. Soc., 1986, 108, 5738-5743. 44. E. Alvarado, T. Nukuda, T. Ogawa, C. E. Ballou, Biochemistry, 1991, 30, 881--886. 45. R. Harris, T. J. Rutherford, M. J. Milton, S. W. Homans, J. Biomol. N M R , 1997, 9, 47-54. 46. P. L. Rinaldi, J. Am. Clzem. Suc., 1983, 105, 5167-5168. 47. C. Yu, G. C. Levy, J. Am. Chem. SOC.,1984,106, 6533-6537. 48. K. E. Kover, G. Batta, J. M a p . Reson., 1988, 79, 206-210. 49. D. Neuhaus, C. P. M. Vanmierlo, J. Mugn. Reson., 1992, 100, 221-228. 50. K. Stott, J. Keeler, Mug. Res. Chem., 1996. 34, 554-558. 51. G. R. Kiddle, R. Harris, S. W. Homans, J. Biomol. N M R , 1998, 11, 289-294. 52. J. M. Duker, A. S. Serianni, Carbohydr. Rex, 1993, 24Y, 281-303. 53. T. Church, I. Carmichael, A. S. Serianni, Curbohydr. Res., 1996, 280, 177-186. 54. A. S. Serianni, Abstr. Am. Chem. Soc., 1996, 211, 10-CARB. 55. T. J. Church, I. Carmichael, A. S. Serianni, J. Am. Chem. Soc., 1997, 119, 8946-8964. 56. M. J. Milton, R. Harris, M. Probert, R. A. Field, S. W. Homans, Glycuhiology, 1998, 8, 147153. 57. Q. W. Xu, C. A. Bush, Carhohydr. R e x , 1998, 306, 335-339. 5 8 . S. K. Zhao, G. Bondo, J. Zajicek, A. S. Serianni, Curhohydr. Res., 1998, 309, 145-152. 59. A. Bax, D. Max, D. Zax, J. Am. Chen?. Soc., 1992, 114, 6923-6925. 60. B. Bose, S. Zhao, R. Sterutz, F. Clovar, P. €3. Bordo, G. Bordo, B. Hertz, I. Carmichael, A. S. Seriarri J. Am. Chem. Soc., 1998, 120, 1 1 158-1 1173. 61. J. R. Tolman, J. M. Flanagan, M. A. Kennedy, J. H. Prestegard, Proc. Natl. Acud. Sci., 1995, 92, 9279-9283. 62. N. Tjandra, S. Grzesiek, A. Bax, J. Am. Chern. Soc., 1996, 118, 6264-6272. 63. N. Tjandra, A. Bax, J. M a p . Reson., 1997, 124, 512-515. 64. N. Tjandra, A. Bax, Science, 1997,278. 1 1 11-1 114.

21. 22. 23. 24. 25. 26. 27.

-

968 65. 66. 67. 68. 69. 70. 71.

33 Conformational Analysis in Solution by N M R

C. R. Sanders 11, J. P. Schwonek, Biochemistry, 1992, 31, 8898-8905. M. Ottiger, F. Delaglio, A. Bax, J. Mugn. Reson., 1998, 131, 373-378. 0. Jardetzky, Bioch. Biophys. Actu, 1980, 621, 227-232. D. A. Cumming, J. P. Carver, Biochemistry, 1987, 26, 6664-6676. M. Nilges, A. M. Gronenborn, A. Briinger, G. M. Clore, Protein Eng, 1988, 2, 27-38. A. E. Torda, R. M. Scheek, W. F. van Gunsteren, J. Mol. Biol., 1990,214, 223-235. D. G. Low, M. A. Probert, G. Embleton, K. Seshadri, R. A. Field, S. W. Homans, J. Windust, P. J. Davis, Glycobiology, 1997, 7, 373-381. 72. R. Harris, G. R. Kiddle, R. A. Field, B. Ernst, S. W. Homans, J. Am. Chem. SOC.,1999, 121, 2546-2551. 73. M. Vignon, F. Michon, J. P. Joseleau, K. Bock, Macromolecules, 1983, 16, 835-838. 74. Q. W. Xu, C. A. Bush, Biochemistry, 1996,35, 14512-14520. 75. G. Lipari, A. Szabo, J. Am. Chem. SOC.,1982, 104,4546-4559. 76. N. Metropolis, A. W. Rosenbluth, M. N. Rosenbluth, A. H. Teller, E. Teller, J. Chem. Phys., 1953,21, 1087-1092. 77. R. Stuike-Prill, B. Meyer, Eur. J. Biochem., 1990, 194, 903-919. 78. T. Peters, T. Weimar, J. Biomol. NMR, 1994, 4, 97-116. 79. H. J. C. Berendsen, J. P. M. Postma, N. F. van Gunsteren, A. DiNola, J. R. Haak, J. Chem. Phys., 1984,81, 3684-3690. 80. L. Lerner, A. Bax, Curbohydv. Res., 1987, 166, 35-46. 81. A. Bax, G. M. Clore, P. C. Driscoll, A. M. Gronenborn, M. Ikura, L. E. Kay, J. Muyn. Reson., 1990, 87, 620-627. 82. L. Yu, R. Goldman, P. Sullivan, G. F. Walker, S. W. Fesik, J. Biomol. N M R , 1993, 3, 429441. 83. G. M. Lipkind, A. S. Shashkov, N. K. Kochetkov, Curbohydr. Res., 1985, 141, 191-197. 84. L. Poppe, C.-W. von der Lieth, J. Dabrowski, J. Am. Chem. SOC.,1990, 112, 7762-7771. 85. K. Bock, J. 0. Duus, S. Refn, Curbohydr. Res., 1994, 253, 51-67. 86. J. Dabrowski, T. Kozar, H. Grosskurth, N. E. Nifantev, J. Am. Chem. SOC.,1995, 117, 55345539. 87. S. W. Homans, M. Forster, Glycobiology, 1992,2, 143-151. 88. G. R. Kiddle, S. W. Homans, FEBS Lett., 1998, 436, 128-130. 89. K. Bock, J. Arnarp, J. Lonngren, Eur. J. Biochem., 1982, 129, 171-178. 90. S. W. Homans, R. A. Dwek, D. L. Fernandes, T. W. Rademacher, FEBS Lett., 1982, 150, 503-506. 91. J . R. Brisson, J. P. Carver, Biochemistry, 1983, 22, 3680-3686. 92. J. R . Brisson, J. P. Carver, Biochemistry, 1983, 22, 3671-3680. 93. S. W. Homans, R. A. Dwek, D. L. Fernandes, T. W. Rademacher, FEBS Lett., 1983, 164, 231-235.

Carbohydrates in Chemistry and Biology Editor Beat Ernst, Gerald W. Hart, Pierre Sinay copyright OWILEY-VCH Verlag GmbH, 2000

34 Oligosaccharide Conformations by Diffraction Methods Serge Pkrez, Cutherine Gautier, and Anne Imherty

34.1 Introduction

Diffraction by single crystals is by far the most powerful experimental method for the characterization of the atomic arrangements in molecules. X-ray and neutron, are used since their wavelengths are of the same order of magnitude as the interatomic distances, typically around an Angstrom unit. The electron and nuclei are the scatterers of the X ray and neutron incident radiation, respectively. The single crystals are usually grown by slow evaporation of saturated solution under controlled environments. Ideally their dimensions should be in the order of 0.2 to 0.5 mm and over 1.0 mm, in the respective case of studies from X-ray and neutron diffraction. Irradiation of the crystal by a monochromatic beam leads to constructive interferences, known as diffraction, of the scattered waves in specific directions. The diffraction patterns, that are recorded on electronic devices, consist of a series of Bragg reflections. Their positions and intensities, respectively, contain the details on the unit cell dimensions, space group symmetry, and atomic positions. Unit cells are the building block of the crystals. In order to determine the crystal structure, both the amplitude and phase of the diffracted wave in every Bragg reflection are required. While the measured intensities are proportional to the square of the amplitude, the phase of the reflection relative to that of the incident beam is unknown as it cannot be recorded experimentally. There are several ways to calculate the phases of some important reflections, among them direct methods which are now used routinely to solve crystal structures. Using Fourier transform, the amplitude and phase information is used to synthesized electron density maps that reveal the atomic content of the unit cell. Least square refinement methods are then used to minimize the discrepancy between the measured and calculated structure amplitudes. From a knowledge of the final atomic coordinates, the molecular shape is revealed and such structural features as bond lengths, bond angles, torsion angles, hydrogen bonds, intermolecular distances are directly computed.

970

34 Oligosaccharide Conformations by Diffruction Methods

Carbohydrates were among the first organic compounds to be investigated by X-ray crystal structure analysis: pentaerythritol [ 11 and a-D-glucosamine hydrochloride and hydrobromide [2]. The work developed slowly as only eight crystal structures were described over the next fifteen years; among them to elucidation of sucrose.NaBr.2H20, by Beevers and Cochran in 1947 [3]. The number of X-ray diffraction studies of carbohydrate increased slowly thereafter and the crystal structures of sucrose and a-D-glucose were analyzed by neutron diffraction in the sixties by Brown and Levy [4, 51. Very recently, the crystal structure elucidation of the cellodextrins [6-8] were the first example of the use of synchrotron radiation, and exemplifies how single crystals having minute dimensions can be successfully investigated.

34.2 General Analysis The Cambridge Structural Data Base (CSDB) contains over 160,000 entries in the form of structural data related to geometry, configuration, conformation, and packing of molecular crystals for organic and metal-organic compounds [9]. As a consequence of the automation of experimental measurements along with the development of crystal structure analysis for non-centrosymmetric space groups, there has been a significant increase in the number of reported crystal structures of carbohydrates between the years 1980-1 995. The number of entries dealing with carbohydrate crystal structures in the CSDB reaches almost 4000. About 1000 among these entries do not list any atomic coordinates and cannot be considered as structurally informative, in terms of conformations and configurations. Despite the fact that it is possible to determine the absolute configuration of an optically active molecule from the effects of anomalous scattering, such a level of structural characterization is not performed routinely. Determination of absolute configuration generally involves scattering from relatively heavy atoms. It can nevertheless be determined from molecules containing a significant fraction of oxygen atoms providing that efforts are given to obtain accurate diffracted intensities. Among the 4000 structural investigations of carbohydrates, only 150 dealt with the determination of the absolute configuration. Caution should be taken to the few crystal structures that have been published with the wrong enantiomer. Most of the recently reported crystal structures use the a priori knowledge of the configuration of the constituting monosaccharide units. Crystalline data can be readily found for many members within the classification of the carbohydrates that encompasses (i) the aldose and kestoses which are the monomers, (ii) the alditol, or sugar alcohol which differ from the aldoses in having members of the series being in meso-configurations, (iii) the cyclitols which make up a family of nine compounds, the configurations of which differ by the axial or equatorial disposition of the hydroxyl groups around the chair-shaped cyclohexane ring, (iv) the anhydro sugars which are formed by the elimination of water between hydroxyls of the pyranoses and furanoses thereby forming fused bicylcic or tricyclic

34.2 Generul Analysis

971

ring structures, (v) the carbohydrate acids which are related to the alditols and aldoses, (vi) the oligosaccharides. The disaccharides result from the condensation of a reducing group [C-1 OH in aldose, C-2 OH in ketose] with another hydroxyl. The family of cyclodextrins is characterized by excellent crystals and therefore many structures analyses have been reported. Typical examples of these compounds are cyclic 1-4 linked a-D-glucopyranosides with 6 , 7 and 8 monosaccharides residues. Being shaped like truncated cones, with the inside voids occupied by solvent or guest molecules, these cyclic oligosaccharides are amenable to crystallization, and exhibit only three types of packing modes. A very large number of crystals structures of the cyclodextrins and their complexes have been reported, including some neutron diffraction analysis. Larger cyclodextrins, namely with 10 glucose and 14 glucoses in the ring, have also been crystallized and display distorted ring shape. Very recently, crystal structure of large cycloamylose compound, consisting of 26 glucose residues, have been solved [lo]. The large ring is folded into a structure similar to the one of V-amylose. This cyclodextrin field has been extensively reviewed several times [ 111 and it will now be covered in the present article. There are 55 crystal structures of unsubsituted disaccharides and a dozen of crystal structures of the fully (or almost fully) acetylated disaccharides. When looking at larger compounds, there are 17 trisaccharide crystals structures and only 4 tetrasaccharides (Table 1). One linear oligomaltose has been crystallized and analyzed as a poly-iodide complex, that of p-nitrophenyl-a-D-maltohexopyranoside [ 121. The malto-hexaose molecules form a double helical anti-parallel stranded structure which enclosed the poly-iodide chain in away similar to that found in cyclodextrin inclusion compounds. Both the conformation and the relative orientation of the single stranded chains are completely different from those found in the crystalline arrangement of a small fragment of amylose, the structure of which has been determined from the combined used of fiber X-ray diffraction and electron single crystal diffraction [13]. Crystal structure have been solved for most of the pentoses and hexoses, in one or more of the isomeric forms or as a 1-0-methyl derivatives. A significant amount of crystal structures of carbohydrates have been determined from neutron diffraction experiments. From these highly accurate structural determination, standard molecular dimensions of the constituting units have been established (Table 2). These data are an update of the ones published previously by Jeffrey & Taylor [ 141 and, in a similar way, they can serve as a basis for parameterizations of molecular mechanics force fields. There is an obvious reluctance of carbohydrates to crystallize in a form suitable for X-ray or neutron diffraction studies. This is particularly true for aldose and ketose containing carbohydrates as a configurational mixture of four isomers, aand P-pyranoses and a- and p-furanose, is likely to occur. Configurational heterogeneity in solution tends to inhibit crystallization. For this reason 1-0-methyl derivatives, which cannot epimerize, are likely to crystallize more readily. In some instances, the a and epimers can co-crystallize, and more than 10 examples of such co-crystallization can be found. The a and P ratio may be dependent on the temperature and solvent of crystallization, and may not be reproducible between independent investigators. One extreme example is provide by the crystal struc-

912

34 Oligosuccharide Conformations by Diffruction Methods

Table 1. Crystal structures of trisaccharides and tetrasaccharides. Trisaccharide Name

Common name

Code

aFuc( I-2)PGal( I-3)GlcNAc aGlcA-4-O-Me( 1-2)PXyl( 1-4)PXyl PGal( 1-4)[ aFuc( 1-3)] PGlcN Ac- 0-Me Manu(l-3)ManP( 1-4)GlcNAc uGlc(l-4)aGlc(l-2)PFru

Blood group B Aldotriuronic acid Lewix X Fragment N-glycan Erlose (mono et trihydrate)

aGlc(1-4)aGlc( 1-4)OGlc-0-Me aGlc(1-6)aGlc( 1-4)a/fiGlc aGal( 1-6)aGlc( l-2)PFmf fiMan(l-4)PMan(l-4)aMan PGlc(1-4)PGlc( I-4)PGlc-0-Me aGlc( 1-2)PFruf( 1-2)PFruf aGlc( 1-2)PFruf( I-6)PFruf aGlc( 1-3)PFruf(2-l)aGlc

Ma1totrioside Panose Raffinose Mannotnose Cellotrioside I-Kestose 6-Kestose Melezitose (2 forms) Planteose Cellotriose

nd GURXPX 10 nd MPYAGL HAHXUJ HEGXOG DUDXOP KOYZAZ RAFINOOI COFMEPIO TAQYAL KESTOS CELGIJ MELEZTO1 MELEZT02 PLANTEIO ACCELLlO

Common name

Code

Ref

STACHYOI PEKHESOI WIMNOV

(481 [49]

ZILTUJ

PI

_ _ _ _ _ _ _ ~ ~ ~

Ref.

~

aGal( 1-6)PFruf(2- 1)aGlc PGIc( 1-4)PGlc( 1-4)PGlc peracetylated Tetrasaccharides Name aGal( 1-6)aGal( I-6)ctGlc( I-2)PFruf aGlc(1-2)fiFruf( l-2)PFruf(l-2)PFruf

Stachyose Nystose Schyzophyllan fragment PGIC(~-~)[PG~C(~-~)]PGIC(~-~)PG~C peracetylated PGlc(I-4)PGlc( 1-4)PGlc( 1-4)PGlc Cellotetraose

[5OJ

171

ture of the disaccharide lactulose, as the crystals contains an mixture of isomers in which the fructose moiety is p-D-fructofuranose, a-D-fructofuranose and p-Dfructopyranose in the ratio 74.5/10/15.5 [ 151. The reluctance of carbohydrate to crystallize in a form suitable for X-ray diffraction studies, is more pronounced for compounds having molecular weights ranging from 1000 to 5000. This is true but for the exception of cyclic compounds such as cyclodextrins and cycloamyloses. One of the reasons may be the lack of sufficient amount of material available for crystal growth. The other reason is that the techniques of growing organic crystals of medium size bio-molecules has not parallel the revolution of growing protein and viruses crystals. Since the beginning of the 90s an increasing number of crystal structures have been reported for glycoproteins and protein-carbohydrate complexes. The resolu-

34.3 Crystalline Conjormations of Disaccharide Moieties

973

Table 2. Standard molecular dimensions for pyranosides. Distances are given in A and angles in '.

'

U-D-4C1

p-D-4c~

u-L- c4

p-L- I cq

379 1.52(2) 1.520(17) 1.520(18) 1.526(17) 1.412(17) 1.413(17) 1.440(16) 113.9(14) 11 l.3(11) 116(3) (3x4) 108 (7) 185 fvug 72( 10) 194 fvug

184 1.5 17(19) 1.522( 15) 1.521(16) I.S26(16) l.394( 17) 1.422( IS) 1.433(14) 111.9(13) 107.4(12) I I5(2) 177(3) -80(11)

35 1.519(16) 1.5 16( 17) 1.514(19) 1.523(16) 1.408(15) 1.413( 13) 1.440(13) 113.8(16) 112.0(13) 115(3) -62(4) - 104(8) 6 fvug -73(12) 2 9 f r ~ g

12 1 .5 1S( 12) 1 .5 1 8(9) 1.529(9) 1.521( 14) l.389( 11) 1.428(9) 1.437(7) 11 1.2(11) 107.7(6) 115(2) 177(2) 79P)

tion of the first reported structures was rarely sufficient to provide reliable conformational information. Significant and rapid progresses arising from the use of synchrotron radiation are providing, now, access to highly resolved structures.

34.3 Crystalline Conformations of Disaccharide Moieties 34.3.1 The Disaccharides

Important in their own right, disaccharides take great importance as the shortest components of the family of oligo, polysaccharides and complex carbohydrates. As such a particular attention is given to their conformational properties because their molecular shapes are considered to be important determinants of their properties and those of their larger parents. The disaccharides result from the condensation of a reducing group [C-1 OH in aldose, C-2 OH in ketose] with another hydroxyl. For hexose in the pyranose shape, this linkage may be 1+n where n is 1 to 6, except 5. If two reducing groups are involved, the disaccharide is non-reducing as in sucroses and trehaloses. For disaccharide moieties, the main conformational determinants are: the ring shapes, the orientations of the hydroxyl groups, and the relative orientations of the monosaccharide units at the glycosidic linkage. Ring shapes can be defined in terms of reference conformations (chair, C; twist T; boat, B; envelope, E; skew, S) or by the so-called puckering parameters [ 161. The exocyclic primary alcohol groups can adopt a number of low-energy conformations. They are usually in staggered ar-

974

34 Oligosaccharide Conformations by Difraction Methods

axial-axial 1-1

1-2

1-1

+

1-2

axial-equatorial 1-3

1-4

1-6

1-4

=4

=4

h

0

axial-hanose

equatorial-equatorial 1-2

1-1

qoq0+ 1-3

w 0 J a o

1-2

1-6

equatorial-furanose

hanose-furanose

1-4

2-1 and 2-6

Figure 1. Schematic representation of the different type of linkages observed in crystal structures of disaccharides and oligosaccharides. Sugars with 'Cd ring shape such as L-fucose have not been represented.

34.3 Crystalline Confbrmations of Disaccharide Moieties

975

rangements that correspond to local minima. In the case of pyranoses, primary hydroxyl groups most frequently occupy two positions, avoiding interactions between 0 - 4 and 0-6. However, each of the secondary hydroxyl groups can rotate almost freely. The relative orientation of two consecutive monosaccharide units in a disaccharide is customarily described by the torsion angles CD and Y around the glycosidic bonds. @ represents the torsion angle about the C(anomeric)-0 bond, whereas Y represents the torsion angle about the 0-Cx bond. The sign of the torsion angles is given in accordance with the IUPAC-IUB Joint Commission of Biochemical Nomenclature [ 171. The consideration of the axial/equatorial nature at the glycosidic linkage provides a useful framework for a classification of the disaccharide moieties, independently of the remaining and of the surrounding of the oligosaccharidic molecule. Using such a classification, all the unsubstituted components of the linear oligosaccharide structures have been reported in Table 3 . For each class, the nature of the glycosidic linkage: l+n is indicated, along with the REFCODE of the corresponding oligosaccharidic structure, its trivial/usual name, description of the disaccharide, the magnitude of the angles [(o)), Y ,z] at the glycosidic linkage, and the occurrence (if any) of inter-residual hydrogen bonds. As for six membered ring containing disaccharides, axial-axial, axial-equatorial and equatorial-equatorial are found. The families involving axial-furanose, and equatorial-furanose linkage to a hexopyranose are also indicated, along with the furanose-furanose cases. Proper references to the original (or most recent) crystal structure work is also provided. In order to have a comprehensive vision of the spatial occurrence of all these conformations as a function of their belonging to a given class, a schematic representation of their crystallographic conformations at the glycosidic linkage, has been set with a superimposition onto the low energy contours that have been computed for a prototypical motif using molecular mechanics calculations (Figure 2). Those contours correspond to conformation of the prototypical disaccharide having energy values of 2 kcal/mol and 6 kcal/mol with respect to the lowest energy minimum. As clearly indicated in Fig. 2a, b, and c, the chosen classification and representations appear to be quite relevant to describe the ensemble of solid state conformations found in crystalline oligosaccharides and their analogs. It is worth noticing that there is no representative of the equatorial-axial class, even though such type of glycosidic linkage can be found in several carbohydrate containing molecules and macromolecules. Whereas all the crystalline conformations lie within the 5 kcal/mol energy contour, it is noteworthy to observe a limited dispersion about the glycosidic 0 angles compared to that observed about Y , as an expression of the influence of the exo-anomeric effect on the establishment of preferred conformation in crystalline oligosaccharides. The exo-anomeric effect arises due to the particular bonding sequence C-5-0-5C- 1-0-g-C-x' in glycopyranosides and disaccharides and influences the conformation about the glycosidic torsion angle @ [ 181. In the case of axial type of linkage (typically as in 4 C D-a ~ configuration) only one staggered conformation is preferred with @ being in the vicinity of 60". As for the equatorial type of linkage (typically as in 4C1 D-p configuration); two staggered conformations are preferred (CD = 60" and -60") of which that corresponding to @ = -60" is favored due further stabilization occurring from non-bonded interactions. The values of @ generally varies between

a,a-Allo-trehalose, CaC12, 5 H 2 0 3,3’ deoxy-Arabino-trehalose. H 2 0 a,a-Trehalose, 2 H 2 0 a,a-Trehalose a,a-Trehalose, 1/2(CaBr2) H 2 0 a,a-Galacto-trehalose a,a-tetrachloro-Galacto-trehalose

[52] (531 [54] [55] [56] [ 571

ALTRCA LETTEJ TREHALlO DEKYEX TRECAB YOXFUM YODSOZ

[58]

a,a-Allo-trehalose, 6H20

[51]

YOXFOG

Molecule

1-1

Ref

Code

=4

1-4

1-2

Atom

1-1

axial-axial

a/Axial-axial linkage between two pyranose rings

aAllo(1- 1)aAllo aAra(I-1)aAra aGlc( 1- 1)aGlc aGlc( 1-1)aGlc aGlc( 1-1)aGlc aGal( 1-1)aGal aGal(1-1)aGal

aAllo(1-1)aAllo

Disaccharide

110.4 110.2 115.3 114.4 115.7 113.3 113.2 116.2 114.8

z

74.7 73.1 45.1 64.5 75.1 60.8 77.0 66.3 77.3

(D

Table 3. Details of the conformations at the glycosidic linkages for disaccharides and oligosaccharides in crystalline state.

75.5 74.8 57.1 75.6 61.6 60.1 77.0 62.4 77.3

Y

H-bond

%

0

$ s

6’ s

2

2

b

4-

m

-4

W

LEUCROOI

CITSTHlO

1-4

[61]

[60]

[59]

Galabiose

P-Leucrose, 2H20

Man, a (1-2) Mana-0-Me

a,P-Trehalose, H 2 0

Blood group B trisacch. H type 1, 1/2 H 2 0

[25]

[33] [62]

[63] Mana (1-2) PGlc [34] Aldotriuronic acid trisacch., 3H20

TIQDUS

nd TIYYOP

RESMOR GURXPXlO

1-1

1-2

Molecule

1-6

Ref

1-3

axial-equatorial

Code

1-4

1-2

Atom

1-1

blAxia1-equatorial linkage between two pyranose rings

FABYOW10

1-2

1- 5

aFuc(1-2)PGal aFuc( 1-2)SGal aFuc(1-2)PGal aMan( 1-2)PGlc aGlcA-4-O-Me( 1- 2)PXyl

aGlc(l-1)PGlc

Disaccharide

115.8 116.7 116.6 118.0 116.0

113.7

z

117.5

aGal(1-4)aGal

114.0 115.4

aMan(1 -2)aMan aGlc(1-5)PFrup

-66.0 -92.7 -92.6 59.0 78.8

68.8

-91.3 -174.8 -174.8 -148.6 -81.9

-93.4

Y

157.7

98.1

0

136.1 -94.0

64.2 68.8

0 - 6 . . .O-6

H-bond

HO-3.. .O-5’

0-2g. . .HO-2f

n

1-6

1-4

1-3

aGlc( 1-4)Glc aGal(lL6)Glc aGlc( 1-6)aGlc

alP-Panose

a/P-Melibiose, H 2 0 a/P-Panose

KOYZAZ

76.4 70.7

aGlc( 1-4)aGlc aClc(I-4)PGlc aGlc( 1 -4)uGlc

6‘-iodo phenyl a-Maltose Methyl 0-Maltoside Methyl P-Maltrotrioside. 3 H 2 0

IPMALT MMALTS DUDXOP

111.7 112.0

aGlc( 1-4)aGlc aGlc( 1-4) PGlc aGlc( 1-4)aGlc

a-Ma1tose P-Maltose, 1 H 2 0 Phenyl a-Maltoside

MALTOT MALTOSll PHMALT

,,

-63.4 74.7

aGlc( 1-4)aGlc

p-nitrophenyl-Maltohexaoside, BaI2 27H20

FOXSUG20-b

MELIBMlO KOYZAZ

aGlc( 1-4)aGlc aGlc( 1-4)aGlc

Erlose, H 2 0 Erlose 3 H 2 0

HAHXUJ HEGXOG

0

107.2 68.7 51.1 112.2 95.7 93.2 118.2 103.1 45.4 94.0 104.7 79.9 106.5 116.1 121.7 108.5 110.0 72.5 109.9 82.3 82.7 96.8

117.0 115.0 116.7 113.3 119.2 119.6 114.1 113.3 119.2 125.5 112.1 112.3 115.7 120.1 117.9 116.9 116.5 112.7 117.6 114.7 115.5 113.9

aMan( 1-3)PMan wGlc(1-3)aGlc aGlc( I-3)bFrup

Mana( 1-3)ManP( 1-4)GlcNAc Methyl a-Nigeroside P-Turanose

MPYAGL MOGLPR TURANS

,

56.2 -72.5 -76.7 60.5 99.9 98.3

116.4 115.1 115.1 114.1 116.1 116.3

aGal(1-3)PGal uFuc( 1-3)GlcNAc

Blood group B trisacch. Lewis X trisacch., 9 H 2 0

nd nd

Table 3 (continued

167.3

- 174.2

-163.5 -120.8 -117.3 -126.1 -142.6 -128.4 -164.8 -129.5 -133.7 -151.8 - 129.6 -118.0 -107.7 -139.4 -138.9 -155.0 -109.1 -148.9 -151.8 -134.8

0 - 3 . . .HO-2’

0 - 3 . . .O-2’ 0 - 3 . . .O-2’ 0 - 3 . . .O-2’ 0 - 3 . . .O-2’ 0 - 3 . . .O-2’ 0 - 3 . . .O-2’ 0 - 3 . . .O-2’ 0 - 3 . . .O-2’ 0-3.. .O-2’ 0 - 3 . . .O-2’ HO-3.. .O-2’ HO-3.. .O-2’ 0 - 3 . . .HO-2’ 0 - 3 . . .HO-2’

HO-3.. .O-2’

- 129.9 - 150.0

HO-4.. .O-2’ HO-4.. .O-2’

60.0 139.2 139.0 97.0 104.2 111.6

YEPNUC

1711

[40] [48]

KDO disacchac., 2Na+ H 2 0

Raffinose, 5 H 2 0 Stachyose, 4 H 2 0

WACHOX

SOPROS

LAMB10 WAGBOV VIZFUF CHONDM

1-2

1-3

Code

[74] [75] [76] [77]

[73]

[72]

Ref

equatorial-equatorial

1-1

1-2

1-1

P-Laminarabiose, H 2 0 Methyl P-Laminarabioside, H 2 0 Neocarrabiose beta monohydrate Chondrosine monohydrate

a-Sophorose, H 2 0

P,P-Trehalose, 2 H 2 0

Molecule

c/Equatorial-equatorial linkage between two pyranose rings

*Two other torsion angles at the linkage: 56.3 63.3

2-8

RAFINOOI STACHYOI

-93.6 -85.7 94.5 -87.8

118.2 117.5 116.5 116.5 PGlc( I-3)PGlc PGlc( 1-3)PGlc a-L-3,6AnGal( 1-3)PGlc PGlcA( 1-3)aGalNAc

-78.9

113.7

PGlc( I -2)aGlc

-75.6

CD

71.8 85.2 64.9 58.1

115.2

z

112.0 113.1 111.1 115.50

PGlc(l-l)PGlc

Disaccharide

aGal(1-6)ctGlc -63.1 aGal( 1 -6)aGal 87.1 nGal( 1-6)aGlc -62.1 aKD0(2-8)aKDO*

77.7 76.0 141.9 55.4

-139.8

-75.5

Y

-170.8 -172.4 -175.1 131.9

HO-4.. .05’ HO-4.. . 0 5 ’ HO-2. . .O-5‘ HO-4.. .05’

H-bond

HO-7.. .OIA’

\D

\D 4

2

s:

2. R

1-4

PGlc(l-4)PGlc

P-Cellobiose P-Cellobiosyl-nitromethane Methyl P-Cellobioside, methanol p-Cellotetraose, 1/2 H20

Methyl P-Cellotrioside, EtOH,H20

a-Chitobiose P-Chitobiose, 3H20

CELLOB02 WEHTEI MCELOB ZILTUJ

TAQYAL

ACHITM10 BCHITTIO

[87] [88]

PGal(1-4)aGlcNAc PGa1(1-4)aGlc PGal(1-4)aGlc pGal( 1-4)PGlc PGaI(1-4) PGlc pMan( I-4)aMan PMan(1-4)PMan PMan(1-4)aMan PGlc(1-4)PGlc PGk( 1-4)PGlcPGlc(1-4) PGlc-0-Me PGlc(I-4)PGlc

a-N-acetyllactosamine, H 2 0 a-Lactose, CaC12 7H20 a-Lactose, 1H20 P-Lactose P-Lactosylurea, 2H20 a-Mannobiose a-Mannotriose, 3H20

ACLACT LACCCB LACTOS03 BLACTO REMVUA DIHTUJ COFMEPlO

PGlcNAc(1-4)aGlcNAc PGlcNAc(1-4)PGlcNAc

,f

PXYV 1-4)PXYI PMan(1-4)aGlcNAc pGal( 1-4)GlcNAc

Aldotriuronic acid trisacch., 3H20 Mana(l-3)ManP(l-4)GlcNAc Lewis X trisac., 9H20

GURXPXIO MAYAGL

Table 3 (continued)

-

113.8 114.6 117.0 117.7 116.3 115.9 117.2 116.5 117.2 115.0 117.4 115.6 116.1 116.5 115.8 116.7 115.9 116.2 116.3 117.5 116.3 117.3 117.2 116.6 116.8 117.6 117.3 117.2 117.6 116.3 117.1

-145.5 -75.7 -80.0 -70.5 -88.1 -76.9 -93.5 -70.9 -88.1 -96.0 -71.9 -93.9 -76.3 -93.0 -91.1 -95.3 -94.8 -93.2 -98.3 -93.2 -91.9 -93.7 -96.7 -96.8 -93.4 -96.3 -91.5 -91.7 -95.5 -76.9 -90.3 -

-79.6 -130.7 -104.6 -107.7 -139.5 -136.7 -143.5 -131.5 - 159.4 -148.1 -131.6 - 150.4 - 132.3 -139.8 -160.7 -143.9 -142.9 -141.6 150.8 - 149.0 - 152.4 -141.5 -141.2 -141.9 -141.5 - 150.6 -151.8 - 152.0 - 150.6 -106.9 -162.3

HO-3.. .O-5’

HO-3.. .O-5’ HO-3.. .O-5’ HO-3.. .O-5’ HO-3.. .O-5’ HO-3.. .O-5’ HO-3.. .O-5’ HO-3.. .O-5’ HO-3.. .O-5’ HO-3.. .O-5’ HO-3.. .O-5’ HO-3.. .O-5’ HO-3.. .O-5’ HO-3.. .O-5’ HO-3.. .O-5’ HO-3.. .O-5’ HO-3.. .O-5’ HO-3.. .O-5’ HO-3.. .O-5’ HO-3.. .O-5’ HO-3.. .O-5’ HO-3.. .O-5’ HO-3.. .O-5’ HO-3.. .O-5’ HO-3.. .O-5’ HO-3.. .O-5’

HO-3.. .05’

GENTBS

[90]

[89]

P-Gentiobiose

Gal P(1-4) Man, 2H20

axial-furanose

1-2

1-6

1-3

KESTOS CELGIJ MELEZTOl MELEZTO2 PEKHESO 1 PLANTEI 0 HAHXUJ HEGXOG

Code

Ref

1-Kestose 6-Kestose, H20 Melezitose, H 2 0 form I Melezitose, H20 form I1 Nystose, 4H20 Planteose, 2H20 Erlose, H20 Erlose, 3H20

Molecule

q0e

1-2

dllinkage between a pyranose ring axial and a furanose ring

1-6

DICMEH

aGlc( 1-2)PFru aGlc( 1-2)PFru aGlc( 1-2)PFru aGlc(1-2)PFru aGlc( 1-2)PFru aGlc( l-2)PFru aGlc( 1-2)PFru aGlc( 1-2)PFru

Disaccharide

PGlc(lL6)PGlc

PGa1(1-4)aMan 60.7

0

0

84.6 89.6 99.8 109.6 102.3 108.2 104.5 98.0 109.9

119.4 121.3 119.4 115.3 118.3 118.8 116.5 117.7 116.2

58.6

-83.2

7

113.6

117.1

H-bond 0-2g.. .O-lg 05-g.. .HO-6f

Y

-65.9 -168.0 -30.6 -43.4 -18.6 -26.2 -32.6 -55.7 -39.8

155.1

-109.9

00

2

53.

e&

S-

ai;

E;.

a k

k

IMATUL PLANTE10

1-6

[97] [46]

[44] 1451

1-4

1-4

BOBKUYlO PAJNUJ

Code

equatorial-h a n o s e

[15] [98]

Ref

P-Lactulose 0-Lactulose, 3 H 2 0

Molecule

PGal( 1-4)PFru PGa1(1-4)PFru

Disaccharide

aGlc(1 -6)PFru aGal(1-6)PFru

aGlc( I-3)PFru aGlc( 1-3)PFru

Melezitose monohydrate form I Melezitose monohydrate form I1

P-Isomaltulose, H 2 0 Planteose, 2H20

aGlc( 1-2)PFru aXyl( l-2)PFru

-64.6 63.5

0

113.1 116.1

z

115.5 111.2

115.9 116.9

118.6 116.6 118.1

120.9 117.1 4-CI-Gal(l-2)1,6di-Cl-Fru 119.2 aGlc( 1-2) PFru 114.3 aGlc( 1-2)PFru 116.7 aGlc( I-2)PFru

aGlc(1-2)PFru aGlc( 1 -2)PFru

Sucrose, sarcosine Sucroxyl, H 2 0

Raffinose, 5H20 Stachyose, 4 H 2 0 Tri-chloro-galactosucrose Sucrose Sucrose, NaBr2 2 H 0 Sucrose, i(2 NaI 3H20)

e/linkage between a pyranose ring equatorial and a furanose ring

MELEZTOI MELEZT02

[95] [96]

NEHCAE HAHYUK

1-3

[40] [48] 1911 [92] [93] [94]

RAFINOOl STACHYOl KANJOY SUCROS04 DINYOOlO n.d

Table 3 (continued)

~

-82.1 -79.9

Q

76.9 58.5

78.4 91.6

82.1 109.3 91.4 107.8 99.8 79.9 79.6 84.0 108.2 96.9

-164.4 -170.3

Y

143.6 172.5

143.1 153.9

12.0 -47.3 -162.2 -44.8 -46.1 -66.6 -61.6 -20.6 -46.9 -37.2

0 - 5 g . . .HO-3f

H-bond

0-2x.. .HO-lf 0 - 2 ~, ..HO-lf

HO-2g.. .0-3f

0-2g.. .HO-lf

3

8g.

5

b

z

2

F'

5

3

5

9

8-

2L

'.

8

0, 2

G.

Q

h

h,

00

Q

[42] [49]

[43]

KESTOS PEKHESO1

CELGIJ

2-6

Ref

1-2

Code

2-1 and 2-6

fhnose-furanose

6-Kestose, H 2 0

1-Kestose Nystose, 3H20

Molecule

f/Linkage between two furanose rings

w

179.2 -63.3 -175.9 67.9

Disaccharide

PFm(1-2)PFru PFru( 1-2)PFru PFru(l-2)PFru PFru(2-6) PFru 116.4 117.1 115.3 118.0

t

-41.1 55.9 -47.1 -55.8

a,

-169.6 -133.3 -165.5 145.7

Y

H-bond

n

%

s

2E

0

i,

984

34 Oligosaccharide Conformations by Diflraction Methods 360 300

240


o ! -120 -60

0

60

120

180

240

1

I

-120 -60

Y

,

0

I

I

I

60

,

,

,

,

120

180

24(

6o Y I2O

I8O

240~

Y

360

120

300

60

0

240

CD

0

0

180

-60

120

-120

60

-1 80 -240

0

-240 -180 -120

I

-60

Y

0

60

120

-60

Figure 2. Iso-energy maps for the linkage between two pyranose linked in an axial-axial configuration (a), in a axial-equatorial configuration (b), in a equatorial-equatorial configuration (c) or linkage between a furanose and a pyranose with an axial glycosidic oxygen. The iso-contours represent approximately 2 and 6 kcal/mol as calculated using the MM3 software for aMan(1-2)Man (a), aMan(1-3)Man (b), cellobiose in (c) and sucrose (d). Conformations observed in the crystal structures have been superimposed on the maps using the following code: circle (I-l), square (l-2), triangle (1-3), diamond (1-4) and cross (1-5). The black symbols represent the conformation of an analog with an hetero atom at the position of the glycosidic oxygen. Appropriate transformation has been carried on the value of the @ and Y angle for the aim of comparison: the inverse of the value was taken for L-sugar, for 1-3 linkages, +120" or -120" was added to the value.

34.4 Hydrogen Bonding in Crystalline Oligosaccharides

985

40" and 120" for 1-x axial linkages, and between -100" and -65" for l - + x equatorial linkages. Obviously, the magnitude of the exo-anomeric effect is not strong enough to drive the CD angle to one single conformation. It can be counterbalanced by the occurrence of inter-residue hydrogen bond or other favorable interatomic interactions. Not unexpectedly, the dispersion of the observed conformations about the torsion angle Y is much wider. A somehow similar set of observations can be made in the case of disaccharide segments belonging to the axial-furanose family (Fig. 3d), which encompasses most of the sucrosyl containing molecules. In such cases, a double exo-anomeric effect, occurring from the C-S-O-5-C- 1-O-g-C-x'-O-5' sequence could be even more influential in the establishment of glycosidic conformations. Whereas the dispersion in CD angles goes from 60" to loo", the Y torsion angles span more than 200". Most of these observations have been rationalized throughout molecular modeling investigations. 34.3.2 The Analogs (S, C, N , . . ..) The replacement of the interglycosidic atom in a disaccharide by S, C, NH, . . . generates a class of interesting analogues, namely thio-disaccharides, C-disaccharides, diglycosylamines, which all constitute potential non-hydrolyzable epitopes and substrates analogues. Similarly, sulfur-analog can be obtained by substitution of the ring oxygen. The crystal structures of those analogs that have been reported, are listed in Table 4. The X-ray structure of methyl a-thiomaltoside showed that the effect of replacing the glycosidic oxygen atom by sulfur on the geometry and conformation of the pyranose rings is minimal [ 191. However, the magnitude of glycosidic valence angle 7 (100.3") is larger than the value usually found for C-S-C fragments; as such it follows the general trend observed in the corresponding O-glycosides. The same value has been observed when both the glycosidic oxygen and the ring oxygen of the non-reducing unit of maltosc are substituted by sulfur [20]. The glycosidic C-1-S and S-C-4' bonds are 1.83 A length and this results in a larger separation of the two pyranose rings. The torsional angles at the thio-glycosidic linkage are 0-C-lS-C-4' = 89" and C-l-S-C4'-CS' = - 1 16.8", they correspond to a low energy conformation within a potential energy surface that is similar to that computed for maltose. In C-glycosides, a methylene group (C-g) replaces the glycosidic oxygen atom and the bonding sequence C-5-0-5-C- 1-C-g-Cx' replaces the original sequence present in the parent O-glycosidic molecules. It is therefore expected that in the crystalline conformation, the anomeric and exo-anomeric effects will no longer be driving forces in these molecules. The conformation of C-gentiobioside [21] in the solid state is characterized by torsion angles CD = 5S.9", Y = 175.1' and o = 63.9". There is indeed a striking difference between C-gentiobioside and its O-counterpart since there is a 120" difference between the magnitude of the CD angle in both molecule. The in uacuo potential energy calculations for C-gentiobioside indicates that the lack of influence of the exo-anomeric effect is not responsible for the occurrence

N-analog

RIWZIG S-analog CIBYAN

C-analog EABRAA

Code

a,@-N-trehalose octaacetate aGlc( 1-1)PGlc

aGlc( 1-4)aGlc aGlc( 1-4)aGlc PGlc(1-2)aGlc

Methyl a-thiomaltoside di-thio-maltose, CH30H 5 S-kijobiose

[19] PO01 [201

~ 3 1

aGlc(1-6)aGlc

C-isomaltose heptaacetate

PI

aGlc(l-2)PFru fiGlc(l-6)PGlc

Disaccharide

C-sucrose octaacetate Methyl-C-gentiobioside

Molecule

[22] PI1

Ref

61.6

-63.9

w

57.6 59.8

89.0 80.7 79.3

100.3 100.4 114.4 119.9 118.9

49.2

68.1 55.9

4

114.6

111.3 113.6

T

Table 4. References and selected structural information for the crystal structures of disaccharides analogs.

-78.8 -85.9

-116.8 -125.7 -140.8

64.4

-77.2 175.1

cp

H06.. .06’

H-bond

34.5 Packing Features

981

of such crystalline conformation; it may simply result from steric interaction of the substituents at the inter-residue linkage. Similarly, the crystalline conformation of C-sucrose, as dictated by torsion angles @ = -54" and Y = 44.1" is significantly different from that of its 0 analog [22]. Aminosugars are potent reversible inhibitors of glycosidases and can be used for many therapeutic agents providing that they exhibit sufficient stability. In diglycosylamines, where a N H group replaces the glycosidic oxygen atom, the basicity of the such glycosylamine nitrogen is expected to be decreased by the combined electron withdrawing effect of the two neighboring oxygen atoms. The crystalline and molecular features of the peracetylated a , 0-glucopyranosylamine, the N-analog of per-acetylated a, p trehalose, has been solved [23]. The hydrogen atom covalently linked to the nitrogen at the glycosidic linkage is well located, yielding to a R configuration for the nitrogen. The substitution of 0 by NH at the glycosidic linkage induces a significant perturbation. The glycosidic C-N-C bond angles in 6" larger than that of a, p trehalose, and a lengthening of the 0-5-C-1 bond is also observed. This feature was predicted to occur from calculations and is attributed to a stabilizing delocalization of the nitrogen lone pair into an antibonding C-0 orbital ( n N + ( ~ c - 0 '[24]. ) The torsional angles at the glycosidic linkages (57.6", -78.8") and (59.8", -85.9") are slightly different from the conformation found in a, p trehalose monohydrate [25] even though the potential energy surface computed for the N-analogue exhibits the same aspects than the one calculated for a, p trehalose, except that the accessible area is larger.

34.4 Hydrogen Bonding in Crystalline Oligosaccharides The carbohydrates offer rich examples to analyze the hydrogen bonding in crystals, from which rules can be extracted to be further used in molecular modeling situations. Most of the basic rules have been established throughout the analysis of high accuracy X-ray analysis and most evidently from those crystalline structures that have been derived from neutron diffraction investigations. Neutron diffraction determines nuclear coordinates whereas X-ray diffraction refers to maxima of the electron density distribution. A systematic investigation of the differences in bond lengths from X-ray and neutron diffraction analyses of the same crystal structure gave significant differencFs for d,-dN for C-H and 0 - H bonds, respectively -0.096(7) and -0.155(10) A [26]. This is the reason why the X-ray X-H bond lengths have to be normalized to standard, neutron diffraction, values for the interpretation of hydrogen bonding in molecular crystals. Essentially, the rules governing the establishment of hydrogen bonds obey to two basic concepts [27]: i) maximize the hydrogen-bond interactions throughout the participation of all hydroxyl groups and as many ring and, to a lesser extend, glycosidic oxygen atoms as possible. These features implies both two and three-center bonds.

988

34 Oligosaccharide Conformations by Diffraction Methods

ii) maximize cooperativity by forming as many finite and infinite chains of hydrogen bonds as possible. Several patterns have been identified by Jeffrey [28] and the crystal structure can be roughly divided equally between them. Class A: encompasses those structures where the cooperative effect is maximized by the formation of infinite chains or spirals of hydrogen-bonded hydroxyls. In class B, a more limited cooperative effect is obtained throughout the occurrence of finite chains with the acetal oxygen atoms acting as chain terminators. Class C, in which the maximum cooperative effect is achieved with all the hydroxyl groups, except one, which forms a separate two- or three-centered hydrogen bond to a ring or glycosidic oxygen. In class D, the infinite chains are retained and the ring and glycosidic oxygen atoms are included in the hydrogen bonding scheme throughout the occurrence of three-center bonds. In almost all cases, hydroxyl groups act as both hydrogen bond donors and acceptors. Compared with the other hydroxyl groups, the anomeric hydroxyls tend to be stronger donor and weaker acceptors. As for ring oxygen and glycosidic oxygen, stereochemical and packing reasons may prevent these atoms to be acceptors. Because there are more acceptors than donors in carbohydrates, it is not surprising to find a significant occurrence of three-centered hydrogen bonds. Hydration is a fairly common features of crystalline oligosaccharides, which may result from their molecular shapes that do not provide efficient packing. Therefore, voids from the water molecules are left in the crystalline structures. It should be clearly stated that the analysis of crystalline oligosaccharides cannot provide a limited insight into the preferred hydration effect. Whereas water molecules may influence the occurrence of a given conformation, these water molecules may be expelled upon formation of the crystalline arrangement. In the hydrate structures, the water molecules always donate two hydrogen bonds; they may accept one or two.

34.5 Packing Features As for their crystalline structure, more than 95% of the carbohydrates crystallize in non centro-symmetric structures. The rare examples of centro-symmetric space groups deal with some linear or cyclic polyalcohol along with some pyranoses and furanoses. In order to adopt such features, these molecules either have an internal symmetry or crystallize as racemic. Figure 3 displays the histogram of the occurrence of space groups in carbohydrate crystal structures, from which it is obvious that only four space groups, i.e. P212121, P21, PI and Ct account for more than 80% of the observed space groups. Despite the wealth of potential structural information, there is a lack of comprehensive investigation of the molecular principles underlying the formation of the oligosaccharide crystals. Only few reports have tried to analyze the anisotropy of intermolecular interactions in these crystals. A typical way of exploring the packing

34.5 Packing Features

989

space groups Figure 3. Histogram of the distribution of space groups among the crystal structures of oligosaccharides.

arrangement is to evaluate the intermolecular energy between one molecule, i.e. the reference molecule, and all its neighbors. This energy is evaluated by taking into account the intermolecular hydrogen bonds as well as the non-bonded interactions. The number of 'close' contacts corresponding to interactions distances less than 1.5 times the sum of the van der Waals interacting atoms may be evaluated. The estimation of electrostatic component is more straightforward as it may depend upon the way the atomic charges are evaluated along with the cut-off distance used in the calculations. Since most of the oligosaccharide structures solved up to now concern neutral carbohydrates, such a point has not be given much attention. The packing arrangements found in crystal structures of oligosaccharides are consistent with a high packing density, as each molecule is usually surrounded by 12 neighbors. Among these neighbors, which occur in pair, a certain anisotropy exist, which in many cases can be correlated to the dimensions of the unit cell, and/or the morphology of the crystals. From the differences in the magnitude of the intermolecular energies can be identity supramolecular elements such as molecular chains and/or molecular layers, that constitute the basis of the molecular layers. From the limited number of packing of oligosaccharide structures which have been investigated so far, the low energy molecular layers correspond to the few which underline the formation of the space groups cited above.

990

34 Oligosaccharide Conformations by Diflruction Methods

Figure 4. The molecular and crystal structure of Lewis X [29]. Graphical representation of the two trisaccharides constituting the asymmetric unit along with the water molecules hydrating them. Nine of the water molecules are part of the asymmetric unit content and the other ones are related by symmetry element.

34.6 Selected Examples Lewisx. Despite their well recognized biological role, the first crystal structure of a histo-blood group carbohydrate-dependent antigen, i.e. Lewis' [ PGall4[aFucl31PGlcNac was reported in 1996 [29] (Figure 4). Starting from chemically synthesized LeX trisaccharide methyl glycoside, single crystal could be grown from a slow evaporating solution of Lex, water/ethanol mixture. Over a period of 2 years, more than 20 crystals having dimensions suitable for X-ray investigations were obtained and investigated. One single crystal having dimensions of 0.5 x 0.25 x 0.05 mm was of a sufficient quality to diffract and yielded enough reflections to solve and refine the structure to a reliability index of 0.05. Le' crystallizes in the monoclinic space group P21, with unit cell parameters u = 12.147(6), b = 27.552(9), c = 8.8662(6) A and P = 91.71"(5). In such a unit cell, the asymmetric unit contains two independent LeXmolecules, and an unusually high number of water molecules. All hydrogen atoms of the 0-H groups and the water molecules could be located in this crystal structure, allowing a straightforward assignment of hydrogen bonding scheme.

34.7 Crystalline Conjovmations o j Oligosuccharides Complexed with Lectins

99 1

The two crystallographically independent LeX molecules differ in their overall conformations, and these differences are essentially located at the glycosidic torsion angles at the PGall4GlcNAc linkage for which angle (D differs by 10". Neither of the two trisaccharides exhibits any intramolecular hydrogen bonds. A strong interaction exists between the fucose and galactose residue, but only non polar van der Waals contacts are involved, each ring presenting its most hydrophobic face to the other one. Conformational studies using NMR and/or molecular modeling generally agree on a single conformation for LeXin solution, corresponding closely to the one found in the crystal structure. The nine water molecules present in the asymmetric unit are arranged in a cluster-like fashion. They establish hydrogen bonds to other water molecules within the cluster and to the surrounding carbohydrate molecule. They are arranged to fill up an empty space in the crystal packing. Six and seven water molecules, respectively, are involved in the hydration of the two trisaccharides. The crystal structure displays an extremely dense network of hydrogen bonds. Thirty six hydrogen bonds are observed in the asymmetric unit, 30 of them implying water molecules. Each trisaccharide is involved in 21 hydrogen bonds, the ration per glycosidic residue varying from 5 to 8. Such a high number of hydrogen bonds can be correlated to (i) the high hydration level and (ii) the peculiar folding of the oligosaccharides, burying the hydrophobic faces of the residue and presenting the hydrophilic faces to the external part. The analysis of the packing indicates a well defined hierarchy of intermolecular contacts, some of them suggesting how LeX-LeXinteractions may occur in biological conditions providing the molecular basis for cell-cell recognition event. Cellodextrins. The quest for the structural elucidation of the crystalline allomorphs of cellulose has stimulated interest in the resolution of the crystal structures of the low molecular weight oligomers. In the case of cellulose 11, it has been established that the cellodextrins having a degree of polymerization of four and higher, give X-ray powder diffractograms that are similar to those of the polysaccharides. Furthermore, powders of methyl P-cellotrioside, and higher members, give diffraction patterns having the cellulose I1 features. The crystal structure of methyl P-cellotrioside was solved as a monohydrate 0.25 ethanolate, using synchrotron data collected on crystals of dimensions 0.43 x 0.33 x 0.04 mm [6]. B-D-cellotetraose crystallizes in the triclinic space group PI with two independent molecules and one water molecule in the unit cell having dimensions a = 8.045( 12), b = 9.003(9), c : 22.51(2) A, a = 89.66(7), P = 94.83(13), y = 115.80(4)' [7]. Because of the very small dimensions of the crystals (0.40 x 0.15 x 0.015 mm) the X-ray diffraction data were collected at room temperature, using beam line at the European Synchrotron Radiation Facility in Grenoble. A monochromatic radiation with a wavelength of 0.925 A was used, and the data collection was achieved on image plate detectors. At about the same time were reported the results of an independent investigation that was performed using a combination of conventional Xray and synchrotron sources on crystals having dimensions 0.4 x 0. x 0.1 mm, [8]. In that case, the reported unit cell dimensions were a = 8.023(1), b = 8.951(2), c = 22.445(2) a = 89.26(2), P = 85.07(l ) , y = 63.93( l)', this corresponds to a different setting of the crystallographic axis.

A,

992

34 Oligosuccharide Conformations by Diflruction Methods

The least-squares superposition of the C and 0-atoms ofthese two structural investigations, has a root mean square deviation of 0.05 A for the non hydrogen atoms. Some several discrepancies are noticed in the positions of 0-H hydrogen atoms and consequently in hydrogen bonding scheme. The way H atom positions were located in the two investigations may explained such discrepancies as in one case they were located in ‘theoretical’ positions whereas in the other one, they were located from differences Fourier analyses. Among the many structural and conformational insights that such investigations are providing, one would like to emphasize the striking finding that the two crystallographically independent cellotetraose molecules are significantly different. These structural differences are not only observed with the ring puckering parameters, but also with the glycosidic @ and Y torsion angles. Whereas one molecule is in a ‘standard’ conformation; the other one displays a significant conformational strain. The influence of mode of packing of these two molecules in the crystal has been invoked to explain such a difference, even though no significant molecular mechanic calculations have yet been performed to elucidate the reasons underlying such significant deviations from normality (Figure 5). When a subcell is constructed of the two central D-glucopyranoses in the two independent molecule, the obtained cell dimensions are identical to those of the cellulose I1 allomorph. Based on this subcell, a new model for cellulose I1 has been proposed [81.

34.7 Crystalline Conformations of Oligosaccharides Complexed with Lectins Since the beginning of the 90’s an increasing number of crystal structures have been reported for glycoproteins and protein-carbohydrate complexes. The resolution of the first reported structures was rarely sufficient to provide reliable conformational information. Significant and rapid progresses arising from the use of synchrotron radiation are providing access to highly resolved structures. For example, the protein-carbohydrate literature provides structural data for 15 protein/cellobiose, 10 protein/lactose, 35 protein/maltose 4 protein/sucrose and 1 proteinla-a trehalose complexes crystal structures. Forces in protein-carbohydrate complexes should be more varied than in the fairly homogenous small molecule oligosaccharides crystals. The protein geometry reduces the influence of preferred packing modes, this is particularly true for hydrolytic enzymes that induces significant distortions both of the ring geometry and the conformations at the glycosidic linkages. Other bias may occur, resulting from the low-resolution of the structure, or from the fact that the crystallographic refinements are often guided with force fields. For these reasons, we are restricted the presentation to the class of lectin-oligosaccharide crystal structures, where there is no distortion occurring from hydrolysis and for which the crystallographic resolution is generally very high. Among proteins that interact non covalently with carbohydrates, lectins bind mono- and oligosaccharides reversibly

34.7 Crystalline Conformations of Oligosaccharides Complexed with Lectins

993

j I

Figure 5. The molecular structure of p-D-cellotetraose [S]. Two orthogonal views of the unit cell.

and specifically, while displaying no catalytic or immunological activity. More than 200 of crystal structures of lectins have been solved, most of them as complexes with carbohydrate ligands. From a data base of three-dimensional structures of lectins (http://www.cermav.cnrs/databank/lectins)Table 5 has been set, to illustrate the extraordinary wealth of information which is available about oligosaccharides. One can found information about simple disaccharides (maltose, sucrose,. . . .) and investigate how their ‘bio-active’ conformation may be similar/different from the one display in molecular crystals. Presumably more informative are those complexes involving biologically important oligosaccharides for which no structural information was available. For example these are: (i) the sialic acid containing oligosaccharides (sialyllactose, sialoglycopeptide, . . .), important antigens (T antigen, histo blood group antigen such as Lewis antigens and their sialylated or sulfated

Sialyllactose Sialogycopeptide Sialyllactose Galpl,3GalNAc Lactose Gal+ 1,3-GalNAc-a-O-benzyl Lactose T-antigen C-lactose N-acetyllactosamine aMan(lL3)uMan aMan( lL6)aMan

Agglutinin I (WGA)

Agglutinin II(WGA) Agglutinin Ricin (RCA) Agglutinin (ACA) Lectin (PNA)

Concanavalin A (ConA)

Lectin Lectin (DBL) Lectin (EcorL)

Soybean agglutinin (SBA)

Isolectin IV (GS4)

Isolectin I (LOL-I)

Isolectin I1 (LOL-11)

Lectine (LcL) Agglutinin (GNA)

Wheat germ

Wheat germ Maclura pomfera Ricinus communis Amaranthus caudatus Arachis hypogaea

Canavalia ensformis

Dioclea grandiflora Dolichos bijorus Erythrina corallodendron

Glycine max

Griffonia simplicfolia

Lathyrus ochrus

Lathyrus ochrus

Lens culinuris Galanthus niualis

Lewis B Lewis Y aMan(1-3)PMan( 1-4)GlcNAc Biantennary octasaccharide glycopeptide N2 fragment of lactotransferine Sucrose aMan(IL3)aMan Mannopentaose

N-ace tyllactosamine N-acetyllactosamine

Pentasaccharide from N-glycan aMan(lL6)[aMan(l-3)]aMan Forssman disaccharide Lactose

uMan(lL6)[aMan(l-3)]aMan

Glycan

Protein

Origin

1WGC [ 1011 2CWG [ 1021 2WGC [ 1011 lJOT [lo31 2AAI [ 1041 1JLX [ 1051 2PEL [ 1061 2TEP [ 1071 lBZW [I081 lCIW [lo91 IQDO [I101 lQDC /110] ICVN [ I l l ] IONA [ 1121 ITEI [I131 l D G L [114] l L U l [115] lLTE [116] IAXl [117] 1AX2 [117] 2SBA [ 11S] ISBD, ISBE, lSBF [I191 1LED [ 1201 lGSL [120] lLOG [121] 1LOF [ 301 lLGC [31] lLGB [31] lLES [ 1221 INIV [123] IJPC [I241

PDB code and ref

Table 5. References and selected structural information for the crystal structures of lectin-oligosaccharide crystal structures.

Galectin-1

Galectin-2 Galectin-7 Galectin-7 Congerin-I Sialoadhesin Pertussis toxin Cholera toxin

Heat-labile Enterotoxin Verotoxin- 1 Enterotoxin B

Coat protein

Hemagglutinin Tailspike protein

Bufo arenarum

Homo sapiens

Staphylococcus aureus

Murine Polyomavirus

Influenzae virus Phage P22

Escherichia coli

3’ Sialyllactose Salmonella octasaccharide Salmonella nonasaccharide

Lactose Receptor GB3 Lactose GM3 trisaccharide 3’ Sialyllactose Disialylated oligosaccharide

Mannose 6-P receptor Galectin-1

Bos taurus Bos taurus

Conger myriaster Mus Musculus Bordetella pertussis Vibrio cholerae

aMan(I -3)aMan Biantennary glycopeptide 3’sialyl Lewis X 3’sulfo Lewis X 4’sulfo Lewis X Mannopentaose N-acetyllactosamine Biantennary oligosaccharide N-acetyllactosamine Thio-digalactose Lactose Lactose N-acetyllactosamine Lactose 3’ Sialyllactose 3’ Sialylgalactose G M 1 pentasaccharide

Daffodil lectin Mannose binding protein A MBP-A mutant CL-K3

Narcissus pseudonarcissus Rattus rattus Rattus norvegicus

Q

5 lTYW [142]

i! !E

x

5

9 5 f?-:

00

.a

c*i

lNPL [ 1251 2MSB [ 1261 2KMB [ 1271 3KMB [I271 4KMB [127] 1C39 [ 1281 lSLT [ 1271 ISLA, ISLB, ISLC [I291 1CAN 1A78 lHLC [130] 4GAL [131] SGAL [131] l C l L [132] lQFO [I331 lPTO [134] lCHB [I351 2CHB, lCTl [32] 1LTT [ 1361 IBOS [137] 1SE4 [I381 1SE3 [138] lSID [139] lSIE [139] lVPS [ 1401 l H G G [141] lTYU, lTYX

996

34 Oligosaccharide Conformations by Difraction Methods

forms.), (iii) moieties of glycolipids such as G M l and GM3, (iv) oligosaccharides belonging to N-linked glycans, biantennary glycopetide, a N2 fragment of lactotransferine, (v) fragments of polysaccharides from the cell wall of pathogenic bacteria (Salmonella octa-and nona-saccharide). . . . Whereas the study of the conformations about the glycosidic torsion angles indicated a somehow limited flexibility in molecular crystal (vide infra) a different pictures emerges for the flexibility of oligosaccharides interacting with lectins. Figure 6 depicts the distributions of glycosidic torsion angles within three disaccharide segments: aMan(1-3)Man, PGlcNAc(1-2) Man and aNeuAc(2-3)Gal. In the case of the aMan(1-3)Man segment, the observed conformations are essentially located around a @ value of 80", with an excursion in the vicinity of 140". May be more interesting is the observation that a remote low energy area (located at @ = 90" and 3 ' ' = 310") can be occupied, as observed in the crystalline complex between Latyrus ochrus and a biantennary glycan [30]. The study of the dispersion of conformations observed for the disaccharide segment PGlcNAc( 1-2) Man, provides another illustration of the occurrence of conformations in remote energy well of the potential energy surfaces. The location of this well is 120" away from what would correspond to the stable conformation driven by the exo-anomeric effect in the case of an equatorial type of linkage. Such examples are observed in the crystalline complexes involving the isolectin I1 of Latyrus ochrus, complexed with high molecular oligosaccharides such as a biantennary octasaccharide [30], a glycopeptide or a N2 fragment of lactotransferin [31]. The aNeuAc(2-3)Gal offers an extreme case of conformational flexibility as it can be ten fold more flexible than the other disaccharides. Here again, the conformation corresponding to the establishment of the exo-anomeric effect CD = 60" is adopted in several case. Such a stabilizing influence can be easily overridden as exemplified by the occurrence of several low energy conformations having @ in the vicinity of -60"; this is observed for the GMl pentasaccharide interacting in the combining site a cholera toxin [32].

34.8 Concluding Remarks With this article, we have set out to provide enough information to initiate the general reader into the world of oligosaccharide structures as derived from diffraction methods. Oligosaccharide structures, numeric data and detailed reference lists have been presented, which may be useful to those directly involved in structural glycoscience. Some keys conformational features of oligosaccharides, have been dissected using the disaccharide moiety as a basic reference entity. This allowed us to provide a useful classification, both of the disaccharides and their analogs. Some of the general rules which dictate the formation of hydrogen bonding and packing patterns in crystalline oligosaccharides have been presented. They provide an efficient template to decipher those very complex structures. It must certainly be emphasized that despite the wealth of potential structural information available, there is a lack of comprehensive investigation of the molecular principles underlying the formation of oligosaccharide crystals.

References 9 N 0

03

0

N

::e 0

P 0

2

I

L

991

998

34 Oligosucchuride Conformutions by Dcjruction Methods

There has been a significant stagnation of the number of oligosaccharide structure analysis reported over the last decade. Usually, the minute amount of available pure oligosaccharides along with their natural reluctance to crystallize in a form suitable for X-ray diffraction studies may explain such a stagnation. It is nevertheless expected that progress in the synthesis or biosynthesis of carbohydrates as well as development in X-ray diffraction sources will be sufficient to overcome the paucity of structural and crystalline data on oligosaccharides. Interest is focusing on the more complex carbohydrate molecules, as diffraction methods extend to the field of protein-carbohydrate complex. Such a field benefits from the developments of methods for crystallizing these protein-based complexes as well as from the rapid progresses arising from the use of synchrotron radiation which is providing access to highly resolved structures. The conjunction of these new developments is likely to provide exciting new results and a greater insight into the biological mechanisms of this fascinating class of biomolecules.

References 1. F.J. Llewellyn, E.G. Cox, T.H. Goodwin, J. Chem. Soc. 1937, 883-894. 2. E.G. Cox, G.A. Jeffrey, Nature 1939, 143, 894-895. 3. C.A. Beevers, W. Cochran, Proc. R. Soc. London Ser. A. 1947, 190, 276-285. 4. G.M. Brown, H.A. Levy, Science 1963,141,921-924. 5. G.M. Brown, H.A. Levy, Science 1965, 147, 1038--1041. 6. S. Raymond, B. Hennissdt, D.T. Qui, A. Kvick, H. Chanzy, Carbohydr. Rex 19y5,277, 209229. 7. S. Raymond, A. Heyraud, D. Tran Qui, A. Kvick, H. Chanzy, Macromolecules 1995, 28, 2096--2100. 8. K. Gessler, N. Krauss, T. Steiner, C. Betzel, A. Sarko, W. Saenger, J. Am. Chem. Soc. 1995, 117, 11397-11406. 9. F.H. Allen, 0. Kennard, Chemical Design Automation News 1993, 8, 31-37. 10. K. Gessler, I. Uson, T. Takaha: N. Krauss, S.M. SMith, S. Okada, G.M. Sheldrick, W. Saenger, Proc. Nut. Acad. Sci. USA 1999, 96,4246-425 1. 11. W. Saenger, J. Jacob, K. Gessler, T. Steiner, D. Hoffman, H. Sanbe, K. Koizumi, S.M. Smith, T. Takaha, Chem. Rev. 1998, 98, 1787-1802. 12. W. Hinrichs, W. Saenger, J. Am. Chem. Soc. 1990. 112, 2789-2796. 13. A. Imberty, H. Chanzy, S. Perez, A. Buleon, V. Tran, J. Mol. Biol. 1988,201, 365-378. 14. G.A. Jeffrey, R. Taylor, J. Comput. Chem. 1980, 1, 99-109. 15. G.A. Jeffrey, R.A. Wood, P.E. Pfeffer, K.B. Hicks, J. Am. Chem. Soc. 1983,105, 2128-2123. 16. D. Cremer, J.A. Pople, J. Am. Chem. Soc. 1975,97, 1354-1358. 17. IUPAC-IUB, Arch. Biochem. Biophys. 1971, 145, 405-621. 18. R.U. Lemieux, S. Koto, D. Voisin In: Anomeric effect: origin and consequence, ACS symposium Series No. 87. Szarek, W. A. and Horton, D., Ed.; Amer. Chem. Soc.: Washington DC, 1979, pp 17-29. 19. S. Perez, C. Vergelati, Acta Cryst. 1984, B40, 294--299. 20. T. Weimar, U.C. Keis, J.S. Andrews, B.M. Pinto, Carbohydr. Res 1999, 315, 222-233. 21. A. Neuman, F. Longchambon, 0. Abbes, H. Gillier-Pandraud, S. Perez, D. Rouzaud, P. Sinay, Carbohydr. Rex 1990, 195, 187-197. 22. D.J. O’Leary, Y. Kishi, J. Org. Chem. 1993,58, 304-306. 23. A. Imberty, J. Gruza, N. Mouhous-Riou, B. Bachet, S. Perez, Carbohydr. Res. 1998, 311, 135-146. 24. I. Tvaroska, J.P. Carver, J. Chem. Phys. 1996, 100, 11305-11313. 25. T. Taga, Y . Miwa, Z. Min, Acta Crystallogr 1997, C53, 234-256.

References

999

26. F.H. Allen, Actu Cryst. 1986, B42, 515-522. 27. G.A. Jeffrey, Actu Cryst. 1990, B46, 89-103. 28. G.A. Jeffrey, W. Saenger Hydrogen bonding in biological structures; Springer-Verlag: Berlin, 1991. 29. S. Perez, N. Mouhous-Riou, N.E. Nifant'ev, Y.E. Tsvetkov, B. Bachet, A. Imberty, Clycobiology 1996, 6, 537-542. 30. Y. Bourne, P. Rouge, C. Cambillau, J. Biol. C h m . 1992, 267, 197-203. 31. Y. Bourne, J. Mazurier, D. Legrand, P. Rouge, J. Montreuil, G. Spik, C. Cambillau, Structure 1994, 2, 209-219. 32. E.A. Merritt, S. Sarfaty, M.G. Jobling, T. Chang, R.K. Holmes, T.R. Hirst, W.G. Hol, Protein Sci. 1997, 6, 1516-1528. 33. A. Otter, R.U. Lemieux, R.G. Ball, A.P. Venot, 0. Hindsgaul, D.R. Bundle, Eur. J. Biochem. 1999,259, 295-303. 34. R.A. Moran, C.F. Richards, Actu Cryst. 1973, B29, 2770-2783. 35. V. Warin, F. Baert, R. Fouret, C. Strecker, C. Spik, B. Fournet, J. Montreuil, Curbohydr. Res. 1979: 76, 11-22. 36. T. Taga, E. Inagaki, Y. Fujimori, S. Nakamura, Curbohydr. Res. 1993, 240, 39-45. 37. T. Taga, E. Inagaki, Y. Fujimori, S. Nakamura, Curbohydr. Res. 1994, 251, 203-212. 38. W. Pangborn, D. Langs, S. Perez, Int. J. Bid. Mucromol. 1985, 7, 363-369. 39. G.A. Jeffrey, D.-B. Huang, Carbohydr. Rrs. 1991,222, 47-55. 40. G.A. Jcffrey, D.-B. Huang, Carbohydr. Res. 1990, 206, 173-182. 41. W. Mackie, B. Sheldrick, D. Akrigg, S. Perez, Znt. J. Bid. Mucromol. 1986, 8, 43-51. 42. G.A. Jeffrey, Y.J. Park, Actu Cryst. 1972, B28, 257-267. 43. V. Ferretti, V. Bertolasi, G. Gilli, C.A. Accorsi, Acta Cryst. 1984, 40, 531-535. 44. D. Avenel, A. Neuman; H. Gillier-Pandraud, Acta Cryst. 1976, B32, 2598-2605. 45. J. Becquart, A. Newman, H. Gillier-Pandrdud, Curbohydr. Res 1982, 111, 9-21. 46. D.C. Rohrer, Actu Cryst. 1972, B28, 425-433. 47. S. Perez, F. Brisse, Actu Cryst. 1977, B33, 2578-2584. 48. G.A. Jeffrey, D.-B. Huang, Curbohydr. Res. 1991,210, 89-104. 49. G.A. Jeffrey, D.B. Huang, Curbohydr. Res. 1993,247, 37-50. 50. K. Noguchi, E. Kobayashi, K. Okuyama, S. Kitamurd, K. Takeo, S. Ohno, Curbohydr. Res. 1994,258, 35-47. 51. A. Linden, C.K. Lee, Acta Cryst. 1995, C51, 1007-1012. 52. J. Ollis, V.J. James, S.J. Angyal, P.M. Pojer. Curbohydr. Res. 1978, 60, 219-228. 53. C.-K. Lee, L.L. Koh, Curbohydr. Res. 1994, 254, 281-287. 54. G.M. Brown, D.C. Rohrer, B. Berking, C.A. Beevers, R.O. Gould, R. Simpson, Actu Cryst 1972, B28, 3145-3158. 55. G.A. Jeffrey, R. Nanni, Curbohydr. Res. 1985, 137, 21-30. 56. W.J. Cook, C.E. Bugg, Curbohydr. Res. 1973, 31, 265 ~275. 57. A. Linden, C.K. Lee, Actu Cryst. 1995, C51, 1012-1016. 58. C.-K. Lee, A. Linden, Curbohydr. Rex 1994, 264, 319-325. 59. T. Srikrishnan, M.S. Chowdhary, K.L. Matra, Curhohydr. Res. 1989, 186, 167-175. 60. J. Thiem, M. Kleeberg, K.-H. Klaska, Curbohydr. Res. 1989, 189, 65-77. 61. G . Svensson, J. Albertsson, C. Svensson, C. Magnusson, J. Dahmen, Curbohydr. Res. 1986, 146, 29-38. 62. D.K. Watt, D.J. Brasch, D.S. Larsen, L.D. Melton, J. Simpson, Carbohydr. Ras. 1996, 285, 1-15. 63. L. Eriksson, R. Stenutz, G. Widmalm, Actu Cryst. 1997, C53, 1105-1107. 64. A. Neuman, D. Avenel, F. Arene, H. Gillier-Fandraud, J.-R. Pougny, P. Sinay, Carbohydr. Res. 1980, 80, 15-24. 65. A. Neuman, D. Avenel, H. Gillier-Pandraud, Acta Cryst. 1978, B34, 242-248. 66. F. Takusagawa, R.A. Jacobson, Actu Cryst 1978, B34, 213-218. 67. M.E. Gress, G.A. Jeffrey, Actu Cryst. 1977, B33, 2490-2495. 68. I. Tanaka, N. Tanaka, T. Ashida, M. Kakudo, Actu Cryst. 1976, B32, 155-160. 69. S.S.C. Chu, G.A. Jeffrey, Actu Cryst. 1967, 23, 1038-1049. 70. M.E. Gress, G.A. Jeffrey, D.C. Rohrer, Actu Crystullogr. 1978, B34, 508-512.

1000

34 Oligosaccharide Conformations by Diffruction Methods

V. Mikol, P. Kosma, H. Brade, Carbohydr. Res. 1994,263, 35-42. C.-K. Lee, L.L. Koh, Actu. Cryst. 1993, C49, 621-624. J. Ohanessian, F. Longchambon, F. Arene, Actu Cryst. 1978, BM, 3666-3671. H. Takeda, N. Yasuoka, N. Kasai, Curbohydr. Res. 1977,53, 137-152. K. Noguchi, K. Okuyama, S. Kitamura, K. Takeo, Carbohydr. Res. 1992,237, 33-43. D. Lamba, A.L. Segre, S. Clover, W. Mackie, B. Sheldrick, S. Perez, Curbohydr. Res. 1990, 208,215-230. 77. M. Senma, T. Taga, K. Osaka, Chem. Lett. 1974, 1415-1418. 78. F. Longchambon, J. Ohanessian, H. Gillier-Pandraud, D. Duchet, J.-C. Jacquinet, P. Sinay, Actu Cryst. 1981, B37, 601-607. 79. C.E. Bugg, J. Am. Chern. SOC.1973, 95, 908-913. 80. J.H. Noordik, P.T. Beurskens, P. Bennema, R.A. Visser, R.O. Could, Z. Kristallogr. 1984, 168, 59-65. 81. K. Hirotsu, A. Shimada, Bull. Chem. SOC.Jpn. 1974, 47, 1872-1879. 82. M.M. Olmstead, M. Hu, M.J. Kurth, J.M. Krochta, Y.L. Hsieh, Acta Cryst. 1997, C53, 915916. 83. B. Sheldrick, W. Mackie, A. D., Curbohydr. Res. 1984, 132, 1-6. 84. S.S.C. Chu, G.A. Jeffrey, Actu Cryst., 1968, B24, 830-838. 85. P. Koll, M. Petrusova, L. Petrus, B. Zimmer, M. Morf, J. Kopf, Curbohydr. Res. 1993, 248, 37-43. 86. J.T. Ham, D.G. Williams, Acta Cryst. 1970, B26, 1373-1383. 87. F. Mo, L.H. Jensen, Acta Cryst. 1978, B34, 1562-1569. 88. F. Mo, Actu Ckem. Scand. 1979, A33, 207-218. 89. C. Burden, W. Mackie, B. Sheldrick, Acta Cryst. 1986, C42, 177-179. 90. F. Arene, A. Neuman, F. Longchambon, C. R. Acud. Sci. Paris 1979, (2288, 331-334. 91. J.A. Kanters, R.L. Scherrenberg, B.R. LeeAang, 3. Kroon, M. Mathlouthi, Cuvbohydr. Res. 1988,180, 175-182. 92. G.M. Brown, H.A. Levy, Acta Cryst. 1973, B29, 790-797. 93. C.A. Accorsi, F. Bellucci, V. Bertolasi, V. Ferretti, G. Gilli, Carbohydr. Rex, 1989, 191, 105116. 94. C.A. Accorsi, V. Bertolasi, V. Ferretti, G. Gilli, Curbohydr. Res. 1989, 191, 91-104. 95. R.V. Krishnakumar, S. Natarajan, Carbohydr. Res. 1996,287, 117-122. 96. T. Taga, E. Inagaki, Y. Fujimori, K. Fujita, K. Hara, Curbohydr. Res. 1992, 241, 63-69. 97. W. Dreissig, P. Luger, Actu Cryst. 1973, B29, 514-521. 98. G.A. Jeffrey, D.B. Huang, P.E. Pfeffer, R.L. Dudley, K.B. Hicks, E. Nitsch, Carbohydr. Res. 1992,226, 29-42. 99. T. Skrydstrup, D. MazCas, M. Elmouchir, G. Doisneau, C. Riche, A. Chiaroni, J.M. Beau, Chem. Eur. J. 1997,3, 1342-1356. 100. S. Mehta, K.L. Jordan, U.C. Kreis, T. Weimar, R.J. Batchelor, F.W.B. Einstein, B.M. Pinto, Tetrahedron Asym. 1994,5, 236772396, 101. C.S. Wright, J. Mol. Biol. 1990, 215, 635-651. 102. C.S. Wright, J. Jaeger, J. Mol. Biol. 1993,232, 620-638. 103. X. Lee, A. Thompson, Z. Zhang, H. Ton-that, J. Biesterfeldt, C. Ogata, Xu, L., R.A. Johnston, N.M. Young, J. Biol. Chem. 1998,273,6312-6318. 104. E. Rutenber, B.J. Katzin, S. Ernst, E.J. Collins, D. Mlsna, M.P. Ready, R. J.D., Proteins 1991,10,240-250. 105. T.R. Transue, A.K. Smith, H. Mo, I.J. Goldstein, M.A. Saper, Nut. Struct. Biol. 1997, 10, 779-783. 106. R. Banerjee, K. Das, R. Ravishankar, K. Suguna, A. Surolia, M. Vijayan, J. Mol. Biol. 1996, 259, 281-296. 107. R. Ravishankar, M. Ravindran, K. Suguna, A. Surolia, M. Vikayan, Cum. Sci. 1997, 72, 855861. 108. R. Ravishankar, A. Surolia, M.V.S. Lim, Y. Kishi, J. Amer. Chem. SOC.1998, 120, 1129711303. 109. R. Ravishankar, K. Suguna, A. Surolia, M. Vijayan, Acta. Crystullogr. 1999, D.55,13751382. 71. 72. 73. 74. 75. 76.

34.2 General Analysis

1001

110. J. Bouckaert, T. Hamelryck, L. Wyns, R. Loris, J. Bid. Chem. 1999,274, 29188-29195. 111. J.H. Naismith, R.A. Field, J. Biol. Chenz. 1996, 271, 972-976. 112. R. Loris, D. Maes, F. Poortmans, L. Wyns, J. Bouckaert, J. Biol. Chem. 1996, 271, 3061430618. 113. D.N. Moothoo, J.H. Naismith, Glycohiology 1998, 8, 173-181. 114. D.A. Rozwarski, B.M. Swami, C.F. Brewer, J.C. Sacchettini, J. Biol. Chem. 1998, 273, 328 18-32825. 115. T.W. Hamelryck, R. Loris, J. Bouckaert, M.-H. Dao-Thi, G. Strecker, A. Imberty, E. Fernandez, L. Wyns, M.E. Etzler, J. Mol. Bid. 1999,286, 1161-1177. 116. B. Shaanan, H. Lis, N. Sharon, Science 1991,254, 862-866. 117. S. Elgavish, B. Shaanan, J. Mol. Bid. 1998, 277, 917-932. 118. A. Dessen, D. Gupta, S. Sabesan, C.F. Brewer, J.C. Sacchetini, Biochemistry 1995, 34, 49334942. 119. L.R. Olsen, A. Dessen, D. Gupta, S. Sabesan, J.C. Sacchettini, C.F. Brewer, Biochemistry 1997, 36, 15073-1 5080. 120. L.T. Delbaere, M. Vandonselaar, L. Prasad, J.W. Quail, K.S. Wilson, Z. Dauter, J. Mol. Bid. 1993,230, 950-965. 121. Y. Bourne, P. Rouge, C. Cambillau, J. Bid. Chenz. 1990, 265, 18161-18165. 122. F. Casset, T. Hamelryck, R. Loris, J.R. Brisson, C. Tellier, M.H. Dao-Thi, L. Wyns, F. Poortmans, S. Perez, A. Imberty, J. Biol. Chem. 1995,270, 25619-25628. 123. G. Hester, C.S. Wright, J. Mol. Biol 1996, 262, 516-531. 124. C.S. Wright, G. Hester, Structure 1996, 4, 1339-1352. 125. M.K. Sauerborn, L.M. Wright, C.D. Reynolds, J.G. Grossmann, P.J. Rizkallah, J. Mol. Bid. 1999,290, 185-199. 126. W.I. Weis, K. Drickamer, W.A. Hendrickson, Nature 1992, 360, 127-134. 127. K.K. Ng, W.I. Weis, Biochemistry 1997, 36, 979-988. 128. L.J. Olson, J. Zhang, Y.C. Lee, N.M. Dahms, J.J.P. Kim, Journal of Biologica1 Chemistry 1999, 274, 29889-29896. 129. Y. Bourne, B. Bolgiano, D.I. Liao, G. Strecker, P. Cantau, 0. Herzberg, T. Feizi, C. Cambillau, Nature Struct. Biol. 1994, I , 863-870. 130. Y.D. Lobsanov, M.A. Gitt, H. Leffler, S.H. Barondes, J.M. Rini, J. Biol. Chem. 1993, 268, 27034-27038. 131. D.D. Leonidas, E.H. Vatzaki, H. Vorum, J.E. Celis, P. Madsen, K.R. Acharya, Biochemistry 1998, 37, 13930-13940. 132. T. Shirai, C. Mitsuyama, Y. Niwa, Y. Matsui, H. Hotta, T. Yamane, H. Kamiya, C. Ishii, T. Ogawa, K. Muramoto, Structure 1999, 7, 1223-1233. 133. A.P. May, R.C. Robinson, M. Vinson, P.R. Crocker, E.Y. Jones, Mol. Cell. 1998, I , 719-728. 134. P.E. Stein, A. Boodhoo, G.D. Armstrong, L. Heerze, S.A. Cockle, M.H. Klein, R.J. Read, Nut. Struct. Biol. 1994, 1, 591-596. 135. E.A. Merritt, S. Sarfaty, F. Van Den Akker, C. L’hoir, J.A. Martial, W.G.J. Hol, Protein Sci. 1994, 3, 166-175. 136. T.K. Sixma, S.E. Pronk, K.H. Kalk, B.A.M. Van Zanten, A.M. Berghuis, W.G.J. Hol, Nature 1992, 355, 561-564. 137. H. Ling, A. Boodhoo, B. Hazes, M.D. Cummings, G.D. Armstrong, J.L. Brunton, R.J. Read, Biochemistry 1998, 37, 1777-1788. 138. S. Swaminathan, W. Furey, J. Pletcher, M. Sax, Nature Struct. Bid. 1995, 2, 680-686. 139. T. Stehle, S.C. Harrison, Structure 1996, 4 , 183-194. 140. T. Stehle, S.C. Harrison, EMBO J. 1997, 16, 5139-5148. 141. N.K. Sauter, J.E. Hanson, G.D. Glick, J.H. Brown, R.L. Crowther, S.J. Park, J.J. Skehel, D.C. Wiley, Biochemistry 1992, 31, 9609-9621. 142. S. Steinbacher, U. Baxa, S. Miller, A. Weintraub, R. Seckler, R. Huber, Proc. Nut. Acad. Sci. USA 1996, 93, 10584-10588.

Carbohydrates in Chemistry and Biology Editor Beat Ernst, Gerald W. Hart, Pierre Sinay copyright OWILEY-VCH Verlag GmbH, 2000

35 Transfer NOE Experiments for the Study of Carbohydrate-Protein Interactions Thomas Peters

35.1 Introduction Many biological recognition events are based upon specific protein-carbohydrate interactions. For instance, self non-self discrimination in immune reactions, bacterial and viral adhesion to mammalian target cells as well as key steps in inflammatory processes involve protein-carbohydrate recognition as the heart of the matter at the molecular level [l-51. The broad interest in a detailed understanding of complexation reactions between carbohydrate ligands and corresponding receptor proteins certainly originates from considerations that go well beyond mere scientific questions. In most cases the biological recognition phenomenon is potentially coupled to pathophysiological conditions that could be cured by inhibition of the particular protein-carbohydrate interaction. Therefore, a strong driving force for research projects that address the molecular nature of protein-carbohydrate recognition reactions stems from potential pharmaceutical applications. A good example that has attracted much attention during recent years is the interaction between selectins and their carbohydrate ligands because these complexation reactions represent the first steps in the inflammatory cascade [6-lo]. In pathological situations such as myocardial infarction, transplantation, or rheumatoid arthritis, efficient suppression of the inflammatory cascade is very important. The design of potent inhibitors obviously requires precise knowledge about the underlying binding reaction, and more specifically about the bioactive conformation of the carbohydrate ligand [ 11-17]. In the case of the selectins, an X-ray structure of the carbohydraterecognition domain of E-selectin has been published [ 181. Attempts to co-crystallize E-selectin with its “natural” ligand, the sialyl Lewis’ tetrasaccharide (Figure 1), failed so far. Knowledge about the bioactive conformation of sialyl Lewis’ came from NMR spectroscopy utilizing experiments that are based upon the so called transfer NOE (trNOE) effect [ 19-23]. In general, the observation of transfer NOES allows an analysis of bound ligand conformations without requiring any knowledge about the complex protein NMR spectrum (for a comprehensive explanation of the trNOE effect see [24]).

1004

35 Transfer NOE Experiments

Neu5Ac

HO

Gal

~

GlcNAc

LOH

NHAc

OH

-OH HO

OH

Fuc

Figure 1. Chemical formula of sialyl Lewis' (sLeX),a-~-Neu5Ac-(2+3)-b-~-Gal-( 1 +4)-[a-~-Fuc(1+3)]-!3-D-GlcNAc.

35.2 The Transfer NOE Experiment Transferred NOEs were originally observed and their principles described more than twenty years ago [26, 271. The experiment is based on a chemical exchange between ligand molecules in solution and ligand molecules bound to large proteins. Various experimental implementations have been explored since then, ranging from one dimensional selective saturation transfer experiments [27, 281 to one and two dimensional transient NOE experiments [29, 301. The chemical exchange between free and bound ligand molecules leads to an averaging of physical properties characteristic for the free and for the bound state, respectively. One important physical parameter that distinguishes free and bound molecules is the so called motional correlation or tumbling time, zc. The motional correlation time zc is defined as the time that is required for a molecule to advance by one radian during translational and rotational diffusion. It is apparent that small molecules have short correlation times z, whereas large molecules such as high molecular weight proteins have long correlation times z,. As an approximation, Equation 1 relates the motional correlation time zc to the molecular weight of the molecule in question [24]:

with M , representing the molecular weight in kDa. This equation assumes isotropic overall molecular motion and neglects internal molecular motions but it is usually sufficient for a first estimate of z,. It is of fundamental importance to realize that the correlation time zc determines sign and size of the NOE [24]. Low or medium molecular weight molecules (MW < 1-2 kDa) have short tumbling times zc, and as a consequence such molecules exhibit positive NOEs, no NOEs, or very small negative NOEs depending on their molecular weight, shape and the field strength

35.2 The Transfer NOE Experiment

I

I

I

1005

I

small molecules

-0.5

-

-1.0

-

large molecules

0.1

1

10

100

Figure 2. Dependence of the NOE enhancement (q) on the motional correlation zc under steady state conditions [24]. The spectrometer frequency w was set at 500 MHz. The size of tetra- and pentasacchdrides usually leads to molecular tumbling times z, that cause the NOE to become zero or very close to zero.

applied. Large molecules such as proteins, polysaccharides or nucleic acids have long tumbling times T ~ ,and therefore exhibit strong negative NOEs reaching -100% at the extreme. The dependence of the NOE on the tumbling time and the field strength is depicted in Figure 2. It is obvious that the motion of a small molecule that binds to a large molecule is determined by the long correlation time of the large molecule during the life time of the complex. Upon dissociation of the complex the small ligand tumbles again with its own characteristic correlation time. During a normal NOESY experiment the large negative NOEs of the bound ligand will be transferred to the sharp signals of the free ligand. The NOEs that are observed under such circumstances are called transferred NOES (trNOEs). It is obvious that trNOEs contain information that is necessary to determine the conformation of the bound ligand, the so called bioactive conformation. One advantage of trNOE experiments is that they usually work at optimum with a ten to twenty fold excess of ligand, making it very easy to observe signals of free ligand molecules against the background of broad protein signals. The reason for the fact that the optimum is reached at excess ligand concentrations is inequality 2 [24]: Nb X CJb

>> Nf

X Of

(2)

and Nf are the number of ligand molecules in the bound (b) and the free (f) state, respectively. Cq, and q designate the cross relaxation rates of the ligand in the

Nb

1006

35 Transfer NOE Experiments

bound and free state. Cross relaxation rates are mainly a function of r p 6 , with r being the distance between two protons, and of the motional correlation time T~ (Equation 1). They determine sign and size of the NOE (cf. Figure 2). Since cross relaxation rates of large biomolecules such as proteins are much larger than cross relaxation rates of small molecules, inequality 2 is usually satisfied for a 10-20-fold excess of ligand. In fact, ofsometimes is so small that it does not play any significant role. This is the case for the turning point in Figure 2, where no NOEs can be observed. Usually, tetra- and pentasaccharides are of such a molecular weight (Equation 1) that NOEs are very close to zero. It is important to realize that the observation of trNOEs is only possible under certain conditions. In general, it can be said that trNOEs are only observable for to complexes with &-values approximately in the range between It follows that trNOE-experiments are ideally suited for the investigation of proteincarbohydrate complexes which are usually characterized by low binding affinities [31]. The reason for the limitations of the observation of trNOE lies in the nature of the NOE itself. Dipole-dipole relaxation is the source of NOEs, and therefore the observation of trNOEs depends critically on the kinetics of the associationdissociation process of a protein-carbohydrate complex [ 32-35]. In general, it is required that the dissociation of the complex is fast on the relaxation time scale. Whether the exchange process is fast or slow on the chemical-shift time scale plays no role for the observation of trNOEs, because it is always the signal of the free ligand that carries the information about the bound state. In other words, it is critical that a large amount of ligand molecules samples the binding pocket of the protein frequently enough during the mixing time of the NOESY experiment. This is visualized in Figure 3 . In practice, exchange frequencies of 50-100 Hz are sufficient to permit the observation of trNOEs.

35.3 Measurement of trNOEs The measurement of trNOEs is performed with exactly the same pulse sequences that are used for the measurement of NOEs. Therefore, the NOESY experiment is well suited to record trNOEs. As pointed out above, the right conditions for the measurement of trNOEs are found by varying the carbohydrate-protein ratio. This is usually done in a titration starting at ligand-protein ratios of approximately 4: 1 (for an example see [36])and than increasing the ratio until maximum trNOEs are observed. During titration it is usually sufficient to perform one-dimensional NOE experiments such as 1D-NOESY in order to save time. The mixing times used for the observation of trNOEs are much smaller than the mixing times required to obtain optimum NOEs for di-, tri-, or tetrasaccharides. This is illustrated in Figure 4 that displays typical NOE-curves in comparison to corresponding trNOE curves for the disaccharide a-~-Fuc-( 1-i6)-fL~-GlcNAc-OMefree and bound to Aleuria aurantia agglutinin (AAA) [37].

35.3 Measurement of trNOEs

1007

\

Figure 3. Schematic representation of the transfer NOESY (trNOESY) experiment. Ligands have to sample the binding site of the protein (shown in red) often enough during the mixing time in order to generate a transfer of NOES. It follows that dissociation of the protein-carbohydrate complex must be fast on the relaxation time scale. With mixing times ,z usually being 50-300 ms for trNOESY experiments, dissociation rates of ca. 50-100 Hz are sufficient.

Although trNOE-experiments are performed with a large excess of ligand the protein background obstructs the resulting NOESY spectra. Especially if a quantitative evaluation of trNOEs is desired, techniques should be applied to remove unwanted protein signals. The optimum solution is the use of a spin-lock filter that is applied after the first n/2 pulse in the NOESY sequence [38].During this time the protein signals undergo fast transverse relaxation whereas signals of the ligand survive because their transverse relaxation rates are negligible. If a quantitative analysis of trNOEs is planned trNOEs should be measured at different mixing times leading to trNOE curves (for an example see [39]). The trNOE step can also be combined with other types of NMR-experiments. For instance, the use of 3DTOCSY-trNOESY experiments and one-dimensional versions thereof has been demonstrated to relieve signal overlap [20]. Especially powerful is the combination of heteronuclear correlation experiments with trNOESY, such as the 3D-HSQCtrNOESY experiment [23]. Unfortunately, this usually requires the availability of 13C-labelledcarbohydrate ligands (see discussion below) that are time consuming and expensive to synthesize [23, 40-421. In general, any multi-dimensional NMRexperiment that contains a NOESY step can be used to detect trNOEs during this step. Once experimental trNOEs have been measured, a bioactive conformation may be derived.

1008

35 Transfer NOE Experiments

NOE in

t

n

eu

8Z

o -in

-20

-30

Figure 4. NOEs for free u-~-Fuc-( 1+6)-fi-~-GlcNAc-OMe (upper half of the graphics), and trNOEs for c~-L-Fuc-( 1+6)-p-~-GlcNAc-OMein the presence of Aleuriu uuruntiu agglutinin (lower half of the graphics), measured at 600 MHz as a function of the mixing time z,. Circles refer to proton pairs H6proRG'cNAc-H6pr~SGlcNAc and diamonds to H 1Fuc-H6pr~SG'cNAc , respectively. Precise experimental conditions have been described elsewhere [37]. Maximum NOEs for the free disaccharides are found around mixing times z, of 1000 ms, whereas maximum trNOEs are reached at mixing times lower than 300 ms.

35.4 Bioactive Conformations of Carbohydrate Ligands From trNOE Experiments As explained above, trNOEs carry the information about the geometry of the bound state of a ligand molecule binding to a receptor protein. Therefore, the bioactive conformation may be back-calculated from observed trNOEs. This analysis can be performed on different levels ranging from qualitative approaches to highly sophisticated protocols that take into consideration e.g. the binding kinetics and all protons of the receptor-protein binding site and the carbohydrate ligand (for recent reviews on these techniques see: [43, 441). The most simplistic approach would be to decide on the basis of the absence and/or presence of inter-glycosidic trNOEs which conformation is bound. From the examples presented in the following, it will become clear that this may easily lead to wrong conclusions mainly because of the probability of so called spin-diffusion processes where NOEs or trNOEs occur between protons that are not necessarily close in space. Only if spin-diffusion has been excluded by specific NMR experiments, this simple procedure is allowed. A semi-

35.5 Spin Difusion may Generate Misleading Distance Constraints

1009

quantitative method requires the acquisition of trNOE build-up curves that have been corrected for spin-diffusion. Since the initial slope of such buildup-curves is easily translated into inter-proton distances by the so called initial slope approximation (ISPA) [45] bound conformations can be deduced. Two circumstances limit the straightforward translation of initial slopes of trNOE build-up curves into threedimensional structures. First, inter-glycosidic trNOEs are sparse, and therefore several conformations may fulfill the observed trNOEs. Second, if more than one conformation is bound by the receptor protein “virtual” bioactive conformations may be obtained. In order to interpret data sets that are incomplete in terms of unambiguous structural information experimental restraints may be combined with MD or MMC simulations (see for instance [19, 22, 23, 461. Following such protocols a range of possible, or “allowed” conformations will be available. If a model for the carbohydrate-protein complex is accessible from X-ray data, or from e.g. homology modelling, extended protocols may be applied. One such approach is implemented in the program CORCEMA ([44] and references cited therein) that is based on full-relaxation matrix calculations taking into account the binding kinetics, and all protein protons that are located in the binding pocket of the protein. Unfortunately, often the structural information required for performing such complex calculations is not available.

35.5 Spin Diffusion may Generate Misleading Distance Constraints For large molecules such as proteins magnetization transfer in NOESY experiments not only takes place between protons that are close in space. A transfer of magnetization may also occur between protons that are separated by far larger distances than usually required for the observation of NOES. The reason is a phenomenon called spin diffusion which allows magnetization between two protons to be transferred via relay protons [24]. Spin diffusion becomes of increasing importance with increasing molecular weight and with increasing mixing times. Therefore, spin diffusion also plays an important role for the observation of trNOEs. It is clear that trNOEs that are due to spin diffusion cannot be used as distance constraints in the generation of bioactive conformations. To avoid spin diffusion short mixing times of less than 50 ms are required if no other precautions are taken. But usually at such low mixing times the intensity of trNOEs is so low that no meaningful analysis may be performed. An example will illustrate the misleading effects of spin diffusion artefacts in trNOESY spectra. The bioactive conformation of a (1 +6)-linked disaccharide, methyl 6-O-p-~galactopyranosyl-4-deoxy-2-deutero-4-fluoro-~-~-galactopyranoside bound to the Fab fragment of an anti-(1+6)-P-~-galactopyrananantibody had originally been analyzed on the basis of trNOEs that were not corrected for spin diffusion [47]. It was concluded that a significant conformational change about the (1+6)-glycosidic linkage occurs upon binding. The resulting bound conformation was not found in aqueous solution without the protein present. Later, the authors identified the key

1010

35 Transfer NOE Experiments

trNOEs that lead to the unusual bioactive conformation as spin diffusion artefacts [48]. They reached this conclusion by applying trROESY experiments and comparing the results to the trNOESY spectra. In trNOESY spectra spin diffusion leads to cross peaks that are of the same sign as cross peaks that are due to direct dipolar contacts, and therefore direct dipolar interactions cannot be distinguished from indirect (spin diffusion) interactions. In trROESY spectra cross peaks from indirect magnetization transfer are of opposite sign compared to direct effects, assuming that one “relay” proton mediates the indirect transport of magnetization. If more than one proton mediates spin-diffusion the sign of the corresponding cross peak changes in an alternating manner, but at the same time significantly losing intensity. Therefore, in such cases usually no cross peaks are observed. It follows that the detection of trROEs of opposite sign or of very weak intensity at positions where trNOEs were observed indicates spin diffusion. This is shown schematically in Figure 5a. In the case described above, trROE experiments showed that the critical cross peaks were due to spin diffusion, and therefore the model for the bound conformation had to be revised. Other groups found similar results [49, 501, and in general it can be stated that testing for spin-diffusion contributions is essential for the eluci-

ROESY

NOESY

-,*--A A

hA [

<3A

<3A

C

jc

AB

L

-v

,,_ _

AB

i C

-C

Figure 5. Schematic representation of three spins, A , B, and C interacting via direct dipolar contacts (black arrows) and/or spin diffusion (dashed arrows). a) Trace of a 2D NOESY spectrum (right) and a corresponding trace from a 2D ROESY spectrum (left). Spins A and B are close in space, and spin diffusion mediates magnetization transfer between protons A and C, that are not close in space. The 2D ROESY experiment allows unambiguous discrimination between direct and spin diffusion effects. b) Protons A and C are close in space, and in addition magnetization is transferred between the two protons via spin diffusion. The effect is a cancellation of the 2D ROESY signal. A cancellation can also occur if indirect magnetization transfer involves more than one relay proton (proton B in this case). Therefore, a discrimination between spin diffusion and direct dipolar interaction is not possible in this case.

35.6 The Conformation of Sialyl Lewis" Bound to E-selectin

1011

dation of bioactive conformations of carbohydrates (and other ligands). TrROESY experiments are a suitable tool to perform this test, but special constellations may exist that require additional experiments. In such cases direct and indirect (spin diffusion) dipolar interactions interfere and lead to a cancellation of trROEs (see Figure 5b) [39]. This observation would mimic the presence of spin diffusion and corresponding distance constraints would be removed, again leading to a false bioactive conformation. The problem here is to separate direct and indirect (spin diffusion) magnetization transfer. It has been shown, that so called QUIET-trNOESY experiments are well suitable for this task [51]. In these type of experiments spins of interest are inverted during the mixing time leaving only their mutual direct dipolar interaction and cancelling all other magnetization transfers [521. For small ligands such as carbohydrates binding to large receptor proteins such as antibodies the main source of spin diffusion are protons that are attached to the binding site of the protein [35, 39, 48, 51, 531. Therefore, band-selective inversion of the ring-proton region usually leads to spectra without spin-diffusion contributions from aliphatic or aromatic protein side chain protons [39]. Figure 5 displays a scheme that summarizes the occurrence of cross-peak patterns in the different type of experiments discussed above. Examples will illustrate the capabilities of the methodology described above. No attempt was made to review the field in total since several very good review articles have appeared during the recent years [54-56].

35.6 The Conformation of Sialyl LewisX Bound to E-selectin Several studies have been performed to elucidate the conformation of the sialyl Lewis' (sLe') epitope bound to E-selectin [ 19-23]. The selectins are membrane bound glycoproteins that are involved in the initial steps of the inflammatory cascade [6-lo]. It has been shown that these proteins mediate rolling of leukocytes by binding to the sLeXepitope on the surface of leukocytes. Clearly, this provides a key point of therapeutic interaction in the case of inflammatory diseases. In order to design potent mimics of sLeXthe bioactive conformation of the molecule must be known. In solution, the tetrasaccharide sLeXadopts several conformational states as this has been shown by many conformational analysis studies [ 19-23, 57-59]. Two main conformational families may be distinguished due to two different major orilinkage as this is depicted in Figure 6. entations of the a-~-NeuSAc-(2+3)-P-~-Gal The two conformational families are characterized by two inter-glycosidic NOES H3-Gal/H3ax-NeuSAc and H3-Gal/H8-NeuSAc, that are mutually exclusive. The precise distribution of the two orientations and the presence of other local minima in aqueous solution is still a matter of dispute. Upon binding to E-selectin only one of the two major conformational families generated by different orientations of the a-~-Neu5Ac-(2+3)-P-~.Gallinkage is recognized. The bound conformation is characterized by the observation of a trNOE between H3-Gal and HS-NeuSAc, whereas no interaction is observed between H3-Gal and H3ax-

1012

35 Transfer NOE Experiments

NeuSAc (Figure 6). This global result has been found by all conformational analyses of sLex bound to E-selectin performed so far [ 19-23]. Different protocols have been applied for the analysis of trNOEs, and therefore one would predict that different values are reported for the precise orientation of the a-~-NeuSAc-(2+3)-p-~Gal linkage in the bound state of sLe". Indeed, different values were found but nevertheless the gross orientation is similar in all cases. The latest study reported for the bioactive conformation of sLeX utilized fully I3C-labeled sLeX [23]. The 13C-enrichment made it possible to perform 3DNOESY-HSQC spectra. From these spectra additional inter-glycosidic trNOEs were extracted that cannot be obtained from homonuclear spectra because of severe spectral overlap. These additional trNOEs represent extra restraints for the con1-+4)-p-~-GlcNAc linkage and the a-~-Fuc-( 1+~)-P-Dformations of the p-~-Gal-( GlcNAc linkage but not for the a-~-Neu5Ac-(2-.3)-P-~-Gallinkage. Therefore, this study provides more integral data for the bioactive conformation of the Le" fragment than previous studies. The glycosidic torsion angles of the different bioactive conformations of sLeXbound to E-selectin reported in the studies mentioned above are compiled in Table 1. Stereo pictures of the different bioactive conformations of sLeXfrom Table I are shown in Figure 7. It has been shown that for biological activity the relative orientation of the fucose and the neuraminic acid residue are relevant [ 11, 121. Following this guide line, a potent mimic of sLeXhad been synthesized [ 11-13]. A conformational analysis of this sLeXmimic in the free and bound state [60] clearly showed that the bioactive conformation is very similar to the one derived originally for sLeX bound to E-selectin [19]. In none of the studies published so far the protein protons in the binding pocket were taken into account. For a more precise analysis this would be mandatory but it would also require knowledge about the orientation of the tetrasaccharide in the binding pocket. This information is currently not available, although several models for the binding of sLeXto E-selectin have been proposed. For a complete understanding of the carbohydrate recognition by E-selectin and consequently as a prerequisite for designing new drugs that block the sLeX/E-selectininteraction it would clearly be desirable to have a better knowledge about the orientation of sLeXin the binding pocket of E-selectin. It is interesting to compare the bioactive conformation of sLeXwhen binding to different receptor proteins. For instance, a different orientation of the a-~-NeuSAc(2+3)-p-~-Gallinkage was postulated for sLeXbound to L-selectin [22]. The biological implications of this finding remain to be elucidated, but certainly it would be interesting to collect more experimental data that demonstrate the presence of distinct bioactive conformations of the sLex ligand when binding to different receptor proteins. For instance, what will happen if totally different (from E-selectin) "carbohydrate recognition" proteins are employed? We therefore studied the binding of sLeXto the lectin Aleuria aurantia agglutinin (AAA) [61], a lectin from orange peel mushroom, that specifically recognizes fucose residues. Since this lectin is known to have binding specificity for fucose one could well assume that the Neu5Ac part of sLeXretains its flexibility even in the bound state, as this had been reported for other saccharide-protein complexes [50, 371. Surprisingly, this is not the case. Probably

I \

HO OH

1

OH

v,

H8-Neu5Ac

OH

NHAc

Figure 6. Two conformations of sLeXrepresenting the two major conformational families resulting from different orientations around the a-~-NeuSAc-(2+3)-P-~-Gal glycosidic linkage. Mutually exclusive inter glycosidic NOES occur.

~ 3 4 - 3 ~ 1 H3ax-Neu5Ac

/

HO

F

rl

3

3c

r:

$F'

Y

5

1014

35 Transfer NOE Experiments

Table 1. Torsion angles at the glycosidic linkages of bioactive conformations of sLeX bound to E-selectin 119, 22, 231 and to Aleuriu uuruntiu agglutinin (AAA) [61]. Reference

N(2+3)G

G( 1+4)GN

~~~

~ 9 1 [221 ~ 3 1 [611

-76"/+@ -58"-22" -43"/- 12" -61"/-4"

F(1+3)GN ~~

+39"/+12" +24"/+34" +45"/$19" +67"/+17"

+38"/+2@ +71"/+ 14" +29"/+4l" -25"/-28"

Figure 7. Different bioactive conformations of sLeX bound to E-selectin as reported in the literature [19, 22, 231. The bioactive conformation of sLeX bound to Aleuriu uuruntiu agglutinin (AAA) is also shown [61]. It is obvious that the latter conformation differs from the other conformations in the orientation of the fucose residue (compare Table 1).

because of unfavourable steric interactions with protein side chains, the Neu5Ac residue adopts a conformation that belongs to the same conformational family that also embraces the conformation of sLeXbound to E-selectin. As pointed out above, the inter-glycosidic trNOE H3-Gal/H8-NeuSAc is characteristic for this conformation [see Figure 61. Inspecting the data that are available for the bioactive conformation of sLeXbound to E-, L- or P-selectin [ 19-23] it is clear that in all of these conformations the galactose and the fucose pyranose rings have a stacked orientation giving rise to certain characteristic inter-glycosidic NOES H2-Gal/HS-Fuc, H2Gal/H6-Fuc, and H1-Fuc/NAc-GlcNAc. This orientation also corresponds to the

35.6 The Conformation of Sialyl Lewisx Bound to E-selectin

1.5

I

H3axN-

2.0 2.5

I

H2"

/

B

' f NHAc/

A

NHAc

1015

-' H3axN+ ,

B

3.0 3.5 4.0

ppm

5.0 4.5 4.0 3.5 3.0 2.5 2.0

1.5

5.0 4.5 4.0 3.5 3.0 2.5 2.0

1.5

Figure 8. 500 MHz NOESY spectrum of sLeX(left, 900 ms mixing time) compared to a 500 MHz trNOESY spectrum of sLe" bound to AAA (right, 150 ms mixing time). It is obvious that a number of inter glycosidic trNOEs are not observed (gray circles) where NOEs were observed (left). Important to notice is the absence of signal intensity for the contact between H2-Gal (H2G)and H5Fuc (HsF), a contact that would indicate stacking of fucose and galactose. An inter glycosidic trNOE is observed between protons HI-Fuc and H2-Gal. This trNOE is difficult to locate at the present expansion of the spectrum but the effect is clearly visible when inspecting the corresponding part of the trNOESY spectrum [61].

global minimum energy region of this glycosidic linkage, and is prevalent in aqueous solution. It follows that the hydrophobic side of the fucose residue is shielded from potential interactions with a protein, and it can be hypothesized that a fucosebinding lectin such as AAA will not recognize a fucose residue that is so well shielded. From trNOE experiments it is evident that AAA recognizes the fucose residue in sLex, but in a conformation that is considerably different from the global minimum and that does not obey the exo-anomeric effect. From the arguments given above this is understandable because recognition of the fucose would be hampered by the presence of the galactose-fucose stacking in the global energy minimum. The experimental evidence that led to this conclusion is clear cut. Upon binding to AAA the NOEs characteristic for the global minimum disappear. Instead, a trNOE is observed between H1-Fuc and H2-Gal (a NOESY and a trNOESY spectrum displaying the major features discussed are shown in Figure 8). This in1+3)-p-~-GlcNAc dicates that the above mentioned local minimum at the a-~-Fuc-( linkage is selected upon binding to AAA. The torsion angles at the (1+3)-glycosidic linkage are around $I = -25" and \1, = -28" (compare Table 1 and Figure 7). Since usually conformations that are recognized by proteins are "preformed" in aqueous solution we reinvestigated the solution conformation of sLeXand LeX to find out whether this "non-exo-anomeric" conformation is also present in aqueous solution. The presence of this minimum had been predicted recently on the basis of

1016

35 Transfer NOE Experiments

theoretical M D simulations and has also been discussed in a study that compared the conformational features of the disaccharide a-~-Fuc-( 1+3)-p-~-GlcNAc to its thio-analog [46]. But so far, no direct experimental evidence was available for the presence of this local minimum. The minimum is also predicted by MMC simulations that we performed for sLeXand Le". An accurate conformational analysis of LeXin aqueous solution reveals that the inter-glycosidic NOE between H1-Fuc and H2-Gal is indeed present although with a rather low intensity. For sLeXwe did not detect this NOE, probably because of the larger molecular weight and the charge of sLeXthat renders very weak negative NOES, as this was reported before [20, 57, 581. Nevertheless, our experiments allow the conclusion that a small portion of Le" and most likely of sLeXin aqueous solution is "preformed" in the local minimum conformation with dihedral angles close to = -25" and \Ir = -28". These findings clearly question the concept of rigidity for the LeXtrisaccharide core structure, and suggest that carbohydrate chains in general may carry different biological information that is encoded by different potential bioactive conformations. Depending on the cognate receptor protein different "conformational information" may be read out.

+

35.7 Interaction of Bacterial Lipopolysaccharide Fragments with Monoclonal Antibodies Much attention has been paid during the past ten years to the specific recognition of carbohydrate epitopes by monoclonal antibodies (mAbs) [62-661. Understanding these reactions at a molecular level will certainly help in generating new perspectives in diagnosis and therapy of related diseases. In our laboratory, we studied the interaction of synthetic lipopolysaccharide fragments with corresponding mAbs with NMR [39, 671. Several mAbs were generated against synthetic and isolated fragments of lipopolysaccharides (LPS) that are present on the surface of chlamydial bacteria [68].These parasites are responsible for a variety of diseases in humans and animals. During infection, antibodies are expressed against components in the outer membrane, with LPS as one of the major surface antigens. Here, we will focus on the disaccharide element a-Kdo-(2-+4)-a-Kdo that constitutes a common structural element of the core region of Gram-negative bacterial LPS in general [69].The conformational features of the binding of a-Kdo-(2+4)-a-Kdo-(2+O)-allyl (Figure 9) to two mAbs S25-2 and S23-24 will be described [39]. Compared to S25-2, S23-24 binds to disaccharide a-Kdo-(2+4)-a-Kdo-(2+O)-allyl with approximately 50-fold increased affinity, and the question arises if this also is reflected by different bioactive conformations that this disaccharide adopts binding to the two mAbs. It turned out that the acquisition of QUIET-trNOESY spectra was mandatory because interference of direct and spin-diffusion mediated magnetization transfer was observed. This is demonstrated in Figure 10 that depicts parts of trNOESY, trROESY and QUIET-trNOESY spectra. The critical cross peaks indicate short distances between protons attached to the carbon atom C8 of ring b and protons

35.7 Interaction of Bacterial Lipopolysaccharide Fragments

s/ 1

OH

HO

.

H

0 .

O

T

coo-

o

1017

H

coo-

..

1.

Figure 9. Chemical formula of the disaccharide a-Kdo-(2+4)-a-Kdo-(2+0)-allyl.

attached to C3 of ring c. The trROESY experiment suggests that these interglycosidic trNOEs are due to spin-diffusion because no intensitiy is observed at the critical positions (Figure lob). In contrast, QUIET-trNOESY experiments clearly show that only part of the interaction is due to spin diffusion (Figure 1Oc). The experiments show that direct dipolar interactions and spin diffusion occur at the same time, giving misleading results for the trROESY experiments. From a variety of NMR experiments it was concluded that protein protons are the major source of spin diffusion, and therefore doubly band selective experiments were performed where the regions of ring protons of 4.10-3.60 ppm and 2.17-1.67 ppm were inverted simultaneously during mixing. Buildup curves from such QUIETtrNOESY experiments were measured and translated into bioactive conformations using restrained MMC simulations. For a-Kdo-(2+4)-a-Kdo-(2+0)-allyl binding to S25-2 the analysis results in a rather limited part of conformational space that contains the bioactive conformation as this is shown in Figure 11. For the other antibody, mAb S23-24, it is impossible to deduce a single bound conformation. Obviously, this antibody recognizes two different bioactive conformations, with one being similar to the global minimum A (Figure l l ) of disaccharide a-Kdo-(2+4)-a-Kdo-(2+O)-allyl and the other being rather close to the conformation C (Figure 11) that is bound by the other antibody S25-2. The bioactive conformations found are in good agreement with the binding data. The antibody with higher affinity binds to a conformation that is highly populated in aqueous solution. This study underlines that one carbohydrate epitope may be recognized in different bioactive conformations by different receptor proteins (Figure 12). These results emphasize that the recognition of different bioactive conformations is an essential component of protein-carbohydrate recognition reactions. Indeed, other examples for this phenomenon have recently been published [66, 70, 711. As one example, the reader's attention should be drawn to studies that targeted the binding of C-glycosides to glycosidases. Using e.g. C-lactose as a non-hydrolizable substrate for Escherichia coli-p-galactosidase, it was shown that the enzyme binds to a high energy local minimum that represents the so called anti-conformation around the pseudo-glycosidic linkage [70]. From crystallographic studies it followed

1018

35 Transfer NOE Experiments

2.0

3axc 3eq"

c?

2.0

c?

C?

3

3ax" 3eq"

b I

8

2.0

Bax" t3eq"

c

0

4.2

4.0

3.8

3.6

Figure 10. Parts of 2D trNOESY (a), 2D trROESY (b), and 2D QUIET-trNOESY (c) of a-Kdo(2+4)-a-Kdo-(2+O)-allyl bound to mAb S25-2. The QUIET-trNOESY experiment was recorded with a 15 ms double-band selective 4 3 inversion pulse (inversion of regions 4.10-3.60 ppm and 2.17-1.67 ppm). Peaks within the inverted regions show an opposite sign (bold lines, c ) relative to the other cross peaks outside these regions. The mixing time was 250 ms for all experiments. A comparison of the spectra allows identification of spin-diffusion effects. Cross-peaks that are cancelled in the trROESY spectrum because spin diffusion and direct dipolar interactions take place at the same time (see discussion in the text) are marked with circles in the 2D trROESY spectrum (b). Reprinted with permission from Biochemistry [39].

that C-lactose binds to peanut lectin in a conformation that corresponds to the minimum energy conformation of 0-lactose [72]. Therefore, the selection of a high energy conformation of C-lactose by E. coltp-galactosidase may promote the cleavage of the glycosidic linkage by the enzyme [70]. At the present, there are too few examples to draw meaningful general conclusions about the biological significance of the recognition of different carbohydrate conformations by different receptor proteins. Certainly, in the future more such data will be collected for other biological relevant cases, and it will be a challenge to link the different bioactive conformations to different biological functions.

35.8 Conclusions and Future Directions -1 80 -

w

0

7

0

-180 180

180

-180

7

l

180

180

r ----I

I30-50% ---1 I50-70% 180

180 -180 180

0

1019

0

Figure 11. Contour plots showing the relative population of conformational space around the (2+4)-glycosidic linkage in a-Kdo-(2+4)-a-Kdo-(2iO)-allyl. The +/$-maps were divided into bins of lo" in and +-direction, and the number of conformations in each bin was counted. Then, contour levels were calculated relative to the highest populated bin (global minimum). The contour levels are color coded. Magenta encodes 1-10%, dark blue 10-30%, light blue 30-50%0, green 5070%, yellow 70-90%0, and red more than 90% of the number of conformations in the highest populated bin. (a) MMC simulation at 600 K. (b) MMC simulation at 2,000 K. (c) and (d) represent possible conformations of a-Kdo-(2+4)-u-Kdo-(2~O)-allyl bound to mAb S25-2. (c) all conformations from the MMC simulation at 2,000 K that satisfy only positive distance constraints (0.15% of the total number of conformations from the MMC simulation). (d) all conformations that satisfy positive and negative distance constraints (0.075% of the total number of conformations). Reprinted with permission from Biochemistry [39].

+-

35.8 Conclusions and Future Directions A major consequence from the studies reported above, and from other examples from the literature, it can be stated that the inherent flexibility of oligosaccharides as compared to e.g. globular proteins allows to encode different biological information for recognition reactions in a parallel manner. So far it appears that binding affinities for the recognition of a certain conformation parallel with the potential

1020

35 Transfer NOE Experiments

B

D

Figure 12. Stereo images of a-Kdo-(2+4)-a-Kdo-(2-+O)-allyl (relaxed view); (A) global minimum A; (B) local minimum B; (C) conformation C that is recognized by mAb S25-2; (D) local minimum D. Reprinted with permission from Biochemistry 1391.

energy of the conformation in aqueous solution, not bound to a protein. The way how nature takes advantage of this phenomenon is not known yet. In order to establish links between different bioactive conformations and biological functions, it will be important to study the thermodynamics and kinetics of the binding reactions in parallel. Microcalorimetry [31, 651 and surface plasmon resonance [e.g. 731 provide powerful tools to collect such data. For more detailed structural analyses it will be important to make use of new techniques that have emerged in protein NMR spectroscopy. The main breakthroughs are probably the introduction of TROSY spectroscopy [74], extending the size limit of biological macromolecules that can be subjected to NMR analyses, the creation of so called cross correlation experiments [75] that allow the direct measurement of dihedral angles, and the measurement of residual dipolar couplings [76] that complement data from NOESY spectroscopy. Unfortunately, all of these techniques require, or at least benefit from isotope enrichment which remains a difficult and expensive task for carbohydrates. Consequently, only few examples utilizing e.g. 13C-labelledcarbohydrates have been published so far [23, 40-421. It is

References

1021

obvious that the new tools that have been discovered for NMR spectroscopy will also greatly promote the analysis of protein-carbohydrate interactions. Interesting developments originate from experiments that use the trNOE effect to screen substance libraries for binding activity against receptor proteins. The first attempts to perform such bio-affinity protocols were rather successful [77, 781, and recently a new experimental strategy was introduced that allows a very robust and powerful screening of libraries, called STD-NMR [79, SO]. The new method is based on the principle of saturation transfer and can be combined with virtually any other NMR technique. Whereas trNOE experiments are well established tools to study bioactive conformations, especially of carbohydrate ligands, the use of bio-affinity NMR methods is still in its infants and promises more surprises in the near future.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 1I . 12. 13. 14. 15. 16. 17.

18. 19. 20. 21. 22. 23. 24. 25. 26.

T. A. Springer, Nature, 1990, 346, 425-434. T. Feizi, Curr. Opin. Struct. Biol., 1993, 3, 701-710. A. Varki, Glycobiology, 1993, 3, 97-130. R. A. Dwek, Chem. Rev., 1996, 96, 683-720. P. Sears, C.-H. Wong, Cell. Mol. Lifl. Sci. 1998, 54, 223-252. M. P. Bevilacqua, S. Stengelin, M. A. Gimbrone, Jr., B. Seed, Science, 1989, 243, 1160-1165. L. A. Lasky, Science, 1992: 258, 964-969. A. Levinovitz, J. Muhloff, S. Isenmann, D. Vestweber, J. Cell. Biol., 1993, 121, 449-459. M. Lenter, A. Levinovitz, A,, S. Isenmann, D. Vestweber, J. Cell. Biol., 1994, 125, 471-481. K. L. Moore, N. L. Stults, S. Diaz, D. F. Smith, R. D. Cummings, A. Varki, R. P. McEver; J. Cell. Biol., 1992, 118, 445-456. H. C. Kolb, B. Ernst, Pure Appl. Chem., 1997, 69, 1879-1884. H. C. Kolb, B. Ernst, Chem. Eur. J., 1997, 3, 1571-1578. K. E. Norman, G. P. Anderson, H. C. Kolb, K. Ley, B. Ernst, Blood, 1998, 91,475-483. T. F. Lampe, G. Weitz-Schmidt, C.-H. Wong, Angew. Chem., 1998,110, 1761-1764. T. Uchiyama, V. P. Vassilev, T. Kajimoto, W. Wong, H. Huang, C.-H. Wong, J. Am. Chem. Soc., 1995, 117, 5395-5396. E. E. Simanek, G. J. McGarvey, J. A. Jablonowski, C.-H. Wong, Chem. Rev., 1998, 98: 833862. P. Sears, C.-H. Wong, Angew. Chem. Int. Ed,, 1999,38, 2300-2324; Anyew. Chem., 1999, 111, 2446-247 1. B. J. Graves, R. L. Crowther, C. Chandran, J. M. Rumberger, S. Li, K.-S. Huang, D. H. Presky, P. C. Familletti, B. A. Wolitzky, D. K. Burns, Nature, 1994, 367, 532-538. K. Scheffler, B. Emst, A. Katopodis, J. L. Magnani, W. T. Wang, R. Weisemann, T. Peters, Angew. Chem., 1995, 107, 2034 -2037; Angew. Chem. Int. Ed. Engl., 1995, 34, 1841-1844. K. Scheffler, J.-R. Brisson, R. Weisemann, J. L. Magnani, W. T. Wang, B. Ernst, T. Peters, J. Biomol. N MR , 1997, 9, 423-436. R. M. Cooke, R. S. Hale, S. G. Lister, G. Shah, M. P. Weir, Biochemistry, 1994, 33, 1059110596. L. Poppe, G. S. Brown, J. S. Philo, P. V. Nikrad, B. H. Shah, J. Am. Chem. Soc., 1997; 119, 1727-1736. R. Harris, G. R. Kiddle, R. A. Field, M. J. Milton, B. Ernst, J. L. Magnani, S. W. Homans, J. Am. Chem. Soc., 1999,121, 2546-2551. D. Neuhaus, M. Williamson, The Nuclear Overhauser Efect in Strucutval and Conformational Analysis, VCH Wiley, New York, 1989. P. Balaram, A. A. Bothner-By, J. Dadok, J. Am. Chem. Soc., 1972, 94, 4015-4017. P. Balaram, A. A. Bothner-By, E. Breslow, J. Am. Chem. Soc., 1972, 94, 4017-4018.

1022 27. 28. 29. 30. 31. 32. 33. 34. 35. 36.

35 Transfer NOE Experiments

G. M. Clore and A. M. Gronenborn, J. Mugn. Reson., 1982,48,402-417. G. M. Clore and A. M. Gronenborn, J. Mugn. Reson., 1983,53,423-442. N. H. Andersen, K. T. Nguyen, H. L. Eaton, J. Magn. Reson., 1985, 63, 365-375. N. H. Andersen, H. L. Eaton, K. T. Nguyen, Mugn. Reson. Chem., 1987,25, 1025-1034. E. Toone, Curr. Opin. Struct. Biol., 1994, 4 , 719-728. R. E. London, M. E. Perlman, D. G. Davis, J. Mugn. Reson., 1992, 97, 79-98. F. Ni, J. M a p . Reson., 1992, 96, 651-656. W. Lee, N. R. Krishna, J. Mugn. Reson., 1992, 98, 36-48. H. N. B. Moseley, E. V. Curto, N. R. Krishna, J. Mugn. Reson. Ser. B, 1995, 108, 243-261. F. Casset, A. Imberty, S. Perez, M. Etzler, H. Paulsen, T. Peters, Eur. J. Biochem., 1997, 244, 242-250. 37. T. Weimar, T. Peters, Angew. Chem. Znt. Ed. Engl., 1994, 33, 88-91; Angew. Chem., 1994, 106, 79-82. 38. T. Scherf, J. Anglister, Biophys. J., 1993, 64, 754-761. 39. T. Haselhorst, J.-F. Espinosa, J. Jimenez-Barbero, T. Sokolowski, P. Kosma, H. Brade, L. Brade, T. Peters, Biochemistry, 1999, 38, 6449-6459. 40. M. A. Probert, M. J. Milton, R. Harris, S. Schenkman, J. M. Brown, S. W. Homans, R. A. Field, Tetrahedron Lett., 1997, 38, 5861-5864. 41. D. G. Low, M. A. Probert, G. Embleton, K. Sheshadri, R. A. Field, S. W. Homans, J. Windust, P. J. Davis, Glycobiology, 1997, 7, 373-381. 42. H. Shimizu, J. M. Brown, S. W. Homans, R. A. Field, Tetrahedron, 1998, 54, 9489-9506. 43. F. Ni, Prog. N M R Spectr., 1994, 26, 517-606. 44. N. R. Krishna, H. N. B. Moseley, in Biological Magnetic Resonance, 17: Structure and Dynamics in Protein NMR, Eds. Krishna and Berliner. Kluwer Academic/Plenum Publishers, New York, 1999, 223-307. 45. A. M. Gronenborn, G. M. Clore, Progr. Nucl. Mugn. Reson. Spectrosc., 1985, 17, 1-32. 46. B. Aguilera, J. Jimenez-Barbero, A. Ferandez-Mayoralas, Carbohydr. Res., 1998, 308, 1927. 47. C. P. J. Glaudemans, L. Lerner, G. D. Daves Jr., P. Kovac, R. Venable, A. Bax, Biochemistry, 1990,29, 10906-10911. 48. S. R. Arepalli, C. P. J. Glaudemans, G. D. Daves Jr., P. Kovac, A. Bax, J. Mugn. Reson., 1995, 106, 195-198. 49. T. Weimar, S. L. Harris, J. B. Pitner, K. Bock, B. M. Pinto, Biochemistry, 1995, 34, 1367213680. 50. J. L. Asensio, F. J. Caiiada, J. Jimenez-Barbero, Eur. J. Biochem., 1995, 233, 618-630. 51. S. J. F. Vincent, C. Zwahlen, C. B. Post, J. W. Burgner, G . Bodenhausen Proc. Nut1 Acud. Sci. USA, 1997,94,4383-4388. 52. C. Zwahlen, S. J. F. Vincent, L. Di Bari, M. H. Levitt, G. Bodenhausen, J. Am. Chem. Soc., 1994,116,362-368. 53. F. Ni, Y. Zhu, J. Mugn. Reson. Ser. B, 1994, 102, 180-184. 54. T. Peters, B. M. Pinto, Curr. Opin. Struct. Biol., 1996, 6, 710-720. 55. A. Poveda, J. Jimenez-Barbero, Chem. SOC.Rev., 1998,27, 133-143. 56. J. Jimenez-Barbero, J. L. Asensio, F. J. Caiiada, A. Poveda, Curr. Opin. Struct. BioL, 1999, 9, 549-555. 57. J. Breg, L. M. J. Kroon-Batenburg, G. Strecker, J. Montreuil, J. F. G. Vliegenthart, Eur. J. Biochem. 1989, 178, 727-739. 58. Y. Ichikawa, Y.-C. Lin, D. P. Dumas, G-J. Shen, E. Garcia-Junceda, M. A. Williams, R. Bayer, C. Ketcham, L. E. Walker, J. C. Paulson, C.-H. Wong, J. Am. Chem. Soc. 1992, 114, 9283-9298. 59. C. Mukhopadhyay, K. E. Miller, C. A. Bush, Biopolymers 1994,34,21-29. 60. W. Jahnke, H. C. Kolb, M. J. J. Blommers, J. L. Magnani, B. Ernst, Angew. Chem., 1997,109, 27 15-271 9. 61, T. Haselhorst, T. Peters, unpublished results. 62 D. R. Bundle, N. M. Young, Curr. Opin. Struct. Biol., 1992,2, 666-673. 63 D. R. Bundle, E. Eichler, M. A. J. Gidney, M. Meldal, A. Ragauskas, B. W. Sigurskjold, B. Sinnott, C. D. Watson, M. Ydguchi, N. M. Young, Biochemistry, 1994, 33, 5172-5182.

References

1023

64. D. R. Bundle, H. Baumann, J.-R. Brisson, S. M. Gagne, A. Zdanov, M. Cygler, Biochemistry, 1994,33, 5183-5192. 65. D. R. Bundle, B. W. Sigurskjold, Meth0d.r Enzymol., 1994, 247, 288-305. 66. M. J. Milton, D. R. Bundle, J. Am. Chem. Soc., 1998,120, 10547-10548. 67. T. Sokolowski, T. Haselhorst, K. Scheffler, R. Weisemann, P. Kosma, H. Brade, L. Brade, T. Peters, J. Biomol. N M R , 1998, 12, 123-133. 68. Y. Fu, M. Baumann, P. Kosma, L. Brade, H. Brade, Infect. Immun., 1992, 60, 1314-1321. 69. L. Brade, P. Kosma, B. J. Appelmelk, H. Paulsen, H. Brade, Zrzfect. Immun., 1987, 55, 462466. 70. J. F. Espinosa, E. Montero, A. Vian, J. L. Garcia, H. Dietrich, R. R. Schmidt, M. MartinLomas, A. Imberty, J. Cafiada, J. Jimenez-Barbero, J. Am. Chem. Soc., 1998, 120, 1309-1318. 71. M. Gilleron, H. C. Siebert, H. Kaltner, C. W. von der Lieth, T. Kozar, K. M. Halkes, E. Y. Korchagina, N. V. Bovin, H. J. Gabius, J. F. G. Vliegenthart, Eur. J. Biochem., 1998, 252, 41 6- 427. 72. R. Ravishankar, A. Surolia, M. Vijayan, S. Lim, Y. Kishi, J. Am. Chem. Svc., 1998, 120, 11297-11303. 73. C. R. MacKenzie, T. Hirama, J. T. Buckley, J. Biol. Chem., 1999,274, 22604-22609. 74. K. Pervushin, R. Riek, G. Wider, K. Wiithrich, Proc. Natl Acad. Sci. USA, 1997, 94, 1236612371. 75. S. J. Glaser, T. Schulte-Herbriiggen, M. Sieveking, 0. Schedletzky, N. C. Nielsen, 0. W. S0rensen, C. Griesinger, Science, 1998, 280, 421-424. 76. N. Tjandra, A. Bax, Science, 1997, 278, 11 11-1 114. 77. B. Meyer, T. Weimar, T. Peters, Eur. J. Biochem., 1997, 246, 705-709. 78. D. Henrichsen, B. Ernst, J. L. Magnani, W.-T. Wang, B. Meyer, T. Peters, Angerv. Chem. h t . Ed., 1999, 38, 98-102. 79. M. Mayer, B. Meyer, Angrw. Chem. Int. Ed., 1999, 38, 1784--1788. 80. J. Klein, R . Meinecke, M. Mayer, B. Meyer, J. Am. Chem. Soc., 1999, 121, 5336-5337.

Carbohydrates in Chemistry and Biology Editor Beat Ernst, Gerald W. Hart, Pierre Sinay copyright OWILEY-VCH Verlag GmbH, 2000

36 Carbohydrate-Protein Interactions: Use of the Laser Photo Chemically Induced Dynamic Nuclear Polarization (C1DNP)-NMR Technique Hans-Christian Siebert and Johannes F. G. Vliegenthart

36.1 Introduction The laser photo chemically induced dynamic nuclear polarization (CIDNP) method is a sophisticated NMR technique [ 11, which can be used for the detection of surface exposed Tyr-, His- and Trp-residues of a protein. It is possible to correlate the intensity of a CIDNP signal with the degree of accessibility of the corresponding CIDNP-sensitive amino acid residue in a protein-structure [2-71. To obtain CIDNP signals it is necessary that a laser-light excited dye undergoes radical reactions with one of the three CIDNP-sensitive amino acids. Two spectra are recorded: one with and one without laser-light irradiation at each CIDNP experiment. The resulting light spectrum is subtracted from the dark spectrum, thereby establishing the CIDNP difference spectrum, containing the signals of polarized residues only. Trpand His-signals occur in positive, Tyr-signals in negative direction in the aromatic part of a CIDNP-difference spectrum [3, 81. Trp- and His-signals can be discriminated by the strong pH-dependence of the His-signals resulting in alterations of chemical shift and signal intensity. The pH-dependence of the intensities of Tyr-, His- and Trp-CIDNP-signals have been described [ 3 , 9, lo]. CIDNP-measurements are usually carried out in D20 solution, therefore one has to consider the relation: pD = pHmeterreadins0.4. In general, it is possible to distinguish highly, partly and non-accessible Tyr-, His- and Trp-residues. In case a CIDNP-reactive amino acid is located in or in the vicinity of the binding pocket of a protein, the corresponding CIDNP signals can in many cases be suppressed by the addition of a specific ligand. In the bound form this ligand hinders the excited flavin dye to undergo radical reactions with the CIDNP-reactive residues. Provided that the alternative interpretation can be excluded that ligand binding alters the general conformation of a receptor thereby changing the surface accessibilities of various amino acid residues. For this reason, CIDNP results have always to be evaluated in close correlation to a computer supported molecular modelling procedure. The CIDNP-technique is a rather quick method, but restricted to special problems, namely the detection and

+

1026

36 Carbohydrate-Protein Interactions

analysis of Tyr-, His and Trp-residues on the surface of a protein. CIDNP data provide their optimal benefit when they are combined with molecular modelling data based on X-ray and NMR structures of similar molecules [5-71. The CIDNP method sometimes also allows to detect an influence on the conformation caused by ligand interaction or by site-directed mutagenesis. The validity of CIDNP data can be checked when protein-structures are completely examined by multi-dimensional NMR experiments. In contrast to multidimensional NMR experiments the CIDNP technique is not limited by the size of the molecule [ 1 11.

36.2 The CIDNP Method The CIDNP radical reaction is initiated by the flavin I mononucleotide (N3 of the isoalloxazine ring substituted with -CHzCOOH, and N10 with -CH3) as laserresponsive dye. The laser-light used is generated by a continuous-wave argon ion laser which operates in the multi-line mode with principal emission wavelengths of 488.0 and 514.5 nm, close to the edge of the 450 nm absorption band of the dye. By an optical fibre the laser light is directed to the sample and chopped by means of a mechanical shutter controlled by the spectrometer to prevent heating of the proteincontaining solution. The irradiation leads to the generation of protein-dye radical pairs by surface-exposed Tyr-, Trp-, and His-side chains. Nuclear spin-polarization is obtained from back-reactions of the radical pairs. The irradiated dye converts from an excited singlet state to an excited triplet state where it can undergo radical reactions with the corresponding groups of the CIDNP-sensitive amino acids. Recombination and escape reactions are possible. The recombination, which only occurs in the singlet state, depends on the spin of the nuclei leading to a nuclear spin polarization which can be detected by NMRspectroscopy [3, 12-14]. The Tyr-dependent CIDNP effect corresponds to a spin-density distribution of the intermediate phenoxy radical with strong negative signals of the protons H3, H5 and less intense positive signals of the protons H2, H6. The CIDNP signals of Trp are generated by an intermediate radical with strong spin densities at the positions H2, H4, H6 of the indole unit which all yield positive CIDNP signals. The CIDNP responses of H2, H4 of His occur as positive singlets [3]. A typical experimental setting consists of a presaturation pulse of 1 s, a light pulse of 0.5 s (5 W), a resonance frequency pulse (90" flip angle) of 5 ps, an acquisition time of 1 s and a delay of 5 s. In general, an adequate signal-to-noise ratio is obtained after 16, 32 or 64 light scans [3]. The individual mixtures consisted of 0.1 mM protein, 0.1 to 2 mM ligand and 0.4 mM flavin derivative. Chemical shifts were assessed relative to acetone (2.225 ppm) and HDO (4.76 ppm at a defined temperature and pH/pD-value).

36.4 Applications

1027

36.3 CIDNP-related Molecular Modelling As mentioned before CIDNP-data should whenever possible be correlated with model structures. In case no further NMR-data are available, X-ray coordinates from the studied molecule or even from homologous structures can be used. The values of the surface accessibility of CIDNP-reactive side chains have shown to be suited parameters for the combination of experimentally and theoretically derived results. Those values can be calculated with the help of the Connolly program in Insight11 following an established method [5-7, 15-19]. The calculated values correspond to the accessible exterior part of the molecule obtained by smoothening the van der Waals surface with a test sphere that has the average radius of a water molecule (1.5 A). Additionally, the average radius of flavin (4 A) can be used without yielding different conclusions concerning the degree of accessibility of the studied residues. The average radii of water (r = 1.5A) and flavin (r = 4 A) are calculated according to a published formula [20]. The dot density of the van der Waals-like spheres representing the Connolly surfaces is generally set to a value of 1 corresponding to a distribution of calculated surface coordinate values.

36.4 Applications Beside successful studies of protein-nucleic acid interactions [19, 21, 221 and protein-folding processes [23] the method has shown to be a proper tool in glycosciences since Tyr-, His- and Trp-residues are often constituents of the carbohydrate binding domain [5-7, 15, 24, 251. Many plant and animal lectins harbour CIDNPreactive amino acid residues as an essential part of their binding pocket architecture [ 26-28]. The presence of aromatic amino acids in the binding site of various lectins raises the question why such hydrophophic amino acids are important for the binding of the hydrophilic carbohydrates. An answer may be found in Figure l a and b which documents the hydrophilic and the hydrophobic parts of a monosaccharide and its position between polar and non-polar amino acids. The CIDNP-reactive amino acid residues can form hydrophobic contacts with the unpolar parts of the carbohydrate in the binding pocket. Lectins are interesting objects to study this aspect of binding in detail. These molecules are carbohydrate binding proteins from nonimmunological origin and without enzymatic activity. They play important roles in various regulation processes in the plant as well as in the animal organism and are therefore of great interest in the field of glycosciences [25, 26, 28-30]. Since Tyr-, His- and Trp-residues are CIDNP-responsive and often involved in carbohydrate binding, alterations of the respective signal intensities of lectins can be detected after addition of a specific saccharide. It is possible to correlate such findings with the theoretical results derived by molecular modelling techniques. CIDNP

1028

36 Carbohydrate-Protein Interactions

Figure la. Hydrophobic an hydrophilic site of a monosaccharide.

36.5 Hevein-like Lectins

1029

Figure lb. Schematic representation of a monosaccharide in a binding pocket with a hydrophobic and a hydrophilic site.

data provide furthermore complementary information to structural data derived by other biochemical and biophysical methods [3I]. Molecular dynamics simulations in combination with CIDNP experiments make it possible to refine the models obtained by interpretation of the rigid X-ray coordinates alone. The success of such a study can be investigated by isoforms and/or mutants of carbohydrate-binding proteins, when available. In case X-ray data from one isoform or the wild type are known, information about important structural features can already be obtained by CIDNP-experiments.

36.5 Hevein-like Lectins The role of Tyr-, His- and Trp-residues can be exemplarily studied in the small lectins hevein, pseudo-hevein (two isolectins from the rubber tree) and the B-domain of wheat germ agglutinin (WGA) which consist of 43 amino acid residues [6, 32, 331. All hevein-like lectins have a binding specificity for D-N-acetylglucosamine oligomers (GlcNAc,). The CIDNP signals of the corresponding amino acids in the binding pocket show a significantly weakened intensity after complexation with a ligand, as demonstrated for hevein at a low pD-value of 4.4 (Figure 2). The lectin Urtica dioica agglutinin (UDA) from the stinging nettle and wheat germ agglutinin (WGA) have also an affinity to GlcNAc,. UDA consists of two and WGA of four hevein-like domains. Beside its affinity to GlcNAc, WGA also binds oligosaccharides containing sialic acids. In all mentioned examples the corresponding signals of Tyr- and Trp-residues of the binding pocket are affected by the addition of GlcNAc, indicated by CIDNP signals with lower intensity after ligand complexation. These results were confirmed by two-dimensional NMR methods

1030

36 Curbohydrute-Protein Interactions

a

b

I Tyr 30 H3,5

flavin

9.00

8.00

flavin

7.00

6.00

I

9.00

8.00

7.00

, '

6.00

Figure 2. Laser photo CIDNP difference spectra (aromatic part) of hevein and hevein-GlcNAc4 complexes at pD = 3.5. a. ligand-free hevein; b. 0.5 mmol hevein + 1 mmol GlcNAc4.

[13, 34, 351. The degree of suppression for WGA and the hevein-like monomers was significantly stronger than for UDA. Interestingly, complexation of UDA with a small amount of GlcNAc3 leads to an altered CIDNP spectrum in which a new small Trp-signal has occurred. This finding argues in favour for a small conformational change of the binding pocket during ligand binding (Figure 3 ) [6]. In order to obtain further information about the architecture of the heveindomains the CIDNP data were correlated with X-ray crystallographic coordinates from WGA [36] and completely NMR-obtained conformations from hevein and hevein-ligand complexes [ 13, 32, 33, 371. Comparison of CIDNP-derived surface accessibilities with computationally obtained values on the basis of refined X-ray or NMR-structures leads to the following results: Buried Tyr-residues have a surface accessibility below 30 A2, partly buried ones have accessibilities of 30-80 A2, whereas a value above 80A2 is calculated for Tyr-residues which are considered to be the surface exposed ones causing intense CIDNP signals. Corresponding values

36.5 Hevein-like Lectins

b

a

Trp 21 H4

H4

v 7.80

1031

7.60

7 -80

3 7.60

Trp 21 H6 Trp 23 H6

Tm 23 H4

flavin

,I I

Trp 23 H4

II

r i

II

flavin

Tyr 30 H3,5 Tyr 76 H3,5 B .'oO

7 .bO

Tyr 76 H3,5 1-

8.00

7.00

Figure 3. Laser photo CIDNP difference spectra (aromatic part) of UDA and UDA-GlcNAc4 complexes. a. ligand-free UDA; b. 0.5 mmol UDA + 1 mmol GlcNAc4.

1032

36 Carbohydrate- Protein Interactions

are estimated for Trp- and His-residues. The calculations are carried out with a dot density of 1 and a test sphere radius of 1.5A [5-7, 15-19]. These data make it possible to use CIDNP-derived results for quality control of modelled structural data [5-71. The surface accessibility values of CIDNP-reactive amino acid residues of hevein-like lectins can be compared with the intensities of the corresponding CIDNP signals. CIDNP and modelling data indicate that the architecture of the binding pockets of hevein, pseudo-hevein and the B-domain of WGAl are similar [6]. In detail: different ways in designing the models have been used. In one case the B-domain of WGAl was directly taken from the X-ray structure of WGAl . In the other case the B-domain of WGAl has been constructed by amino acid replacement in the hevein-NMR-structure. The initial structures have been energetically minimized before starting the MD (molecular dynamics) simulations. Values of the surface accessibility of CIDNP-reactive amino acid residues are obtained during the MD-simulation and listed in a table [6]. Modelled structures of pseudo-hevein and the B-domain of WGAl which are in optimal agreement with the CIDNP-results are shown in Figure 4a, b.

36.6 Galactoside-binding Lectins from Plant and Animal Origin NMR-spectra of Erythrina corallodendron lectin (EcorL) are not resolved (Figure 5) without the use of special techniques, like CIDNP. Resolved CIDNP signals from EcorL, however, can be assigned by comparison of wild-type- and mutant-EcorLCIDNP-spectra, as published [ 71. For completion, galectins, which are galactosidebinding lectins from animal origin which do not need any Ca2+-ionsfor saccharide binding and which share structural homologies with EcorL, have been included in this study. From X-ray data as well as from the results of the CIDNP experiments one can conclude that the highly accessible amino acid residues Tyrl06 in EcorL and Trp68 in galectins are involved in the binding process. However, X-ray data of EcorL and bovine galectin also indicate that the extent of surface accessibility of other CIDNPreactive moieties besides Tyrl06 and Trp68 can be examined. Since the crystallographic data sets provide the starting point for knowledge-based homology modelling, MD (molecular dynamics)-derived conformational parameters for the related proteins can likewise be correlated to the results of the CIDNP spectroscopic measurements. This comparison is also facilitated by the availability of mutants for the legume lectin, in which Tyr-residues at position 106, 108 or 229, respectively, were replaced by Thr-, Gly- or Ala-residues [38, 391. Furthermore, a mutant was studied in which Trp is replaced by Ala at position 135. The introduction of the single-site mutations in the legume protein can cause non-uniform impacts on the calculated conformational parameters beyond the immediate vicinity of the site of mutation [7, 401. These results deserve attention for the interpretation of comparative data sets of wild-type and single-site mutant proteins. The structural predictions of molecular modelling are in agreement with the CIDNP-spectra. Regarding the

36.6 Galactoside-binding Lectins,from Plant and Animal Origin

1033

a

b Figure 4a. Energy minimum conformation of pseudo-hevein, emphasising the surface exposition of Tyr21 and Trp23 in the upper part of the picture; b. energy minimum conformation of domB of WGAl, emphasising the surface exposition of Tyr64 and Tyr66 in the lower part of the picture.

1034

36 Curbohydrute-Protein Inteructions

8

v;

8

d

8 G

8 od

8

0;

36.6 Galactoside-binding Lectins from Plant and Animal Origin

1035

intensity of the Tyr-signal, the CIDNP-spectrum of the mutant Trpl35Ala (Figure 6a) does not alter remarkably from that of the wild type [7]. A single site mutation of the highly accessible Tyr 106 leads to the complete disappearance of the Tyrsignal in the CIDNP difference spectrum, Figure 6b. However, also the CIDNPspectrum of an EcorL mutant wherein Tyrl08 is exchanged against a CIDNP-inert residue does not show any Tyr-CIDNP-signal, Figure 6c. This can only be the case when Tyrl06 loses its surface accessibility due to the replacement of Tyrl08. Our experimental finding is exactly reproduced by molecular modelling calculations which are based on the X-ray structure data of EcorL [41].The derived MD data [7, 421 confirm the experimental results, as demonstrated in Table 1. Furthermore, as can also be seen from Table 1, the single site mutations of Tyrl06 or Tyrl08 have an impact on the surface accessibilities of Tyr192 and Tyr229 which are lowered significantly. On the other hand a replacement of Tyr by Ala at position 229 influences the surface presentation of Tyrl06 and Tyrl08 [7] but no impact could be detected on the binding specificity 138, 391. The conformational rearrangement in this mutant is shown in Figure 7a, b. The CIDNP method was also applied to five members of the galectin family, namely bovine heart galectin, human lung galectin, CG- 14-the galectin from chicken intestine, CG- 16-the galectin from chicken liver and the recombinant murine galectin-3. The experimental data show a significant covering of Trp68 after addition of lactose whereas the Tyr-residues of the studied galectins are unaffected by the ligand. The X-ray structure from bovine heart galectin and the CIDNP data of all studied galectins have been used as valuable information for the design of four model structures [7, 431. CIDNP-responsive amino acids are therefore important constituents in the binding site of the P-galactoside specific galectins. One galectin-domain with a bound ligand and the involvement of Trp is shown in Figure 8. The availability of well-defined X-ray structures [44-461 and sequence homology between various galectins from different species allow the prediction of structures by use of the knowledge-based protein modelling approach [47, 481. The results obtained from CIDNP experiments provide valuable information concerning the reliability of structures obtained by different modelling techniques. The occurrence of an additional Trp-CIDNP signal in a spectrum obtained from a saccharidehuman galectinl complex argues in favour of a change in orientation of the Trp-ring. Furthermore, effects on the binding strength of galectins caused by site-directed mutagenesis [49] can be analysed by a combination of CIDNP and modelling techniques. Such a protein-carbohydrate interaction study can be completed when the results from transferred nuclear Overhauser experiments (trNOE) are integrated (50). TrNOE-experiments are a proper tool in glycosciences [50-581. These kinds of experiments are used when the conformation of a small ligand in the presence of a much bigger receptor is studied. Since the mentioned biophysical methods are complementary to each other, a combination of X-ray-data, NMR methods (multidimensional experiments, trNOE, CIDNP) and molecular modelling techniques is in general the best way to analyse protein-carbohydrate complexes on an atomic scale [48, 58, 591.

I

7.00

6.00

Tyr 106 8.00

!

7.00

"

"

,

-

6.00

9.00

7.00

6.00

Figure 6. CIDNP-difference spectra (aromatic part) of three single-site mutants of recombinant EcorL. a. [WI 35AJEcorL (ligand-free); b. [Y106GIEcorL (ligand-free); c. [Y108TIEcorL (ligand-free).

8.00

1

C

36.7 Sialidase from Clostridium Perfringens (Wild Type and Mutants)

1037

Table 1. Average surface accessibilities ( 2 ) and the differences of the surface accessibility areas of CIDNP-sensitive residues derived from molecular dynamics calculations between the wild-type of EcorL ( 2 ) and the three single-site mutants [7]. All area values are given in A2. Dot density: I ; test sphere radius: 1.5 A. **** denotes the site of mutation substituting the respective tyrosine residue with CIDNP-inert alanine. X

AY106

AY108

AY229

~

Tyr 53 Tyr 82 Tyr 106 Tyr 108 Tyr 121 Tyr 172 Tyr 185 Tyr 192 Tyr 229 Trp 45 Trp 60 Trp 135 Trp 207 Trp 231 His 58 His 142 His 180 His 226

10.0 69.3 67.7 75.3 49.4 7.9 6.1 65.8 81.0 100.9 30.9 69.1 113.3 3.0 28.7 0.0 45.0 0.0

+1.2 -10.7

****

10.0 +7.0 - 14.4

-30.7 +12.2 -7.4 +3.6 -18.6 -11.9 +5.2 $2.8 -20.6 -5.5 -1.0 -10.0 0.0 +13.0 f2.1

+15.7 +7.4 -1.6 9.0 -25.4 -29.0 -17.6 123.6 -42.2 -3.0 -3.6 0.0 +9.4 0.0

-

****

+15.5 +4.4 +14.1 -32.6 $11.6 -7.9 +16.3 -30.4

****

-15.5 -26.3 -2.8 -20.6 -3.0 - 14.9 0.0 + 17.2 0.0

36.7 Sialidase from Clostridium Perfringens (Wild Type and Mutants) The presence of a number of CIDNP-sensitive amino acids on the protein surface can enable the detection of conformational alterations resulting from single site mutations. This has been demonstrated for wild-type forms and various mutants of EcorL and of the sialidase of Clostridium perfringens [5, 71. Sialidase of C. perfringens is larded with CIDNP-responsive amino acid residues. Therefore, these residues have been used as valuable sensors in NMR and modelling studies in which CIDNP-spectra of the sialidase wild type and various mutants are compared and correlated with modelled structures. Starting with the crystal structure of the bacterial sialidase of Salmonella typhimurium [60] which is used as a framework, the knowledge based homology modelling produces a set of energy-minimised conformations for the small sialidase of C.

1038

36 Carbohydrate-Protein Interactions

Figure 7. MD-derived energy minimum structures of wild-type EcorL and single-site mutants. a. wild-type EcorL; b. mutant [Y229A]EcorL.

perfringens. Although the number of CIDNP-reactive amino acids is to large to unequivocally assign signals to defined residues, as has been feasible for hevein [6], the signal intensity can be set into correlation with the model-derived expectations. In addition to the wild-type enzyme the impact of introducing amino acid substitutions by site-directed mutagenesis has been theoretically and experimentally delineated as a further test of the model structure. Changes in surface accessibilities of widely separated residues affected by a Tyr/Phe- or a Cys/Ser-substitution could be measured and calculated. This leads to the conclusion that conformational changes of mutant enzymes relative to the wild-type form should not be underestimated. However, not only CIDNP-spectra and modelling data argue in favour of an influence of single site mutations on the overall conformation. The CIDNP- and modelling-results obtained for the sialidase of C. perfringens and several of its mutants are in perfect agreement with observations concerning the enzymatic activities of these molecules determined by biochemical methods. The differences in enzymatic activities between the wild type and various mutants confirm the conclusions drawn from CIDNP experiments and molecular modelling [5].

36.8 CIDNP Analysis of Glycoproteins

1039

Figure 8. Galectin-1 monomer with a specific disaccharide in its binding pocket.

The success of knowledge-based homology modelling is critically dependent on the predictive potency of the program of structure-based calculations, which attempts to translate homologous sequences into three-dimensional structures. In order to evaluate the actual relevance of molecular-modelling-supported crystal structures for the protein topology in solution, CIDNP data providing selected parameters of the protein’s conformation can be used for quality ranking.

36.8 CIDNP Analysis of Glycoproteins When CIDNP-reactive amino acids are accessible on the surface of a glycoprotein, a possible impact of the covalently bound oligosaccharide chain on the protein part can be scrutinized [14, 15, 611 by this method. Therefore, CIDNP-experiments are also suitable to improve structural models of glycoproteins. On the basis of a X-ray structure of a glycoprotein, very often only the coordinates for the protein part are available since the flexibility of the oligosaccharide chain hampers a complete crys-

1040

36 Carbohydrate-Protein Interactions

tallographic structure. NMR measurements of the oligosaccharide chain [62-661 in combination with molecular mechanic and molecular dynamic calculations make it possible to gain the needed structural data. CIDNP experiments were applied on the glycoprotein SAP (serum amyloid P-component) from human serum [67, 681. In one series of experiments the oligosaccharide chains were intact and in another series they were desialylated [14, 151. The respective spectra of SAP and enzymatically desialylated SAP were determined. Six Trp/His signals and one Tyr signal are present in the aromatic part of the CIDNP difference spectrum of SAP. The corresponding spectrum of desialylated SAP shows remarkable alterations. The chemical shift of one Trp/His-characteristic signal is decreased by 0.1 ppm. One Trp/His-signal disappeared and a new one was formed in the CIDNP difference spectrum of desialylated SAP, while the other signals were unaffected. The Tyr signal has a clearly enhanced intensity in desialylated SAP. Therefore, the removal of sialic acid residues from the single N-glycan of each monomer apparently affects the surface presentation of distinct CIDNP-reactive amino acids of SAP [ 14, 151. A conformational change of the protein part of SAP in relation with a different orientation of the desialylated oligosaccharide chain in comparison to the sialylated chains or a covering of CIDNP-responsive amino acids by the oligosaccharide chain itself arc possible explanations of the CIDNP results. In order to prove these experimental data molecular dynamics simulations of SAP with an intact and an asialo-saccharide chain were carried out. The structural data of the protein part of SAP were obtained by using X-ray crystallographic coordinates of SAP [69, 701 as a start structure for the molecular dynamics simulations. The computationally constructed oligosaccharide chains were energetically minimised with the SWEET program [71]and linked to the protein part (Figure 9) before the MD-run. The MD-simulations address the question in which way an interaction between the oligosaccharide and protein part is possible. Small long range interactions within oligosaccharide chains of 0-acetylated and non-0-acetylated gangliosides (90Ac-GDla and GDla) have already been identified with help of NMRsupported MD-simulations [72].

36.9 Conclusions In summary, the CIDNP method is a potent addition to the arsenal of biophysical methods which arc successfully used in the field of structural glycosciences. The combination of X-ray data, molecular modelling, trNOE-methods and laser-photo CIDNP provides valuable structural information about protein-carbohydrate complexes. This is of particular importance, when multidimensional NMR-data can not be recorded due to the size of the molecule. Furthermore, the analysis of similar carbohydrate-binding proteins sharing a sufficient sequence homology is possible when complete structural data are available only for a few of them and CIDNPand molecular modelling data are present for all of them. The successful role, which CIDNP-experiments can play for the solution of important structural glyco-

36.9 Conclusions

1041

Figure 9. X-ray structure of human serum amyloid P component with modelled oligosaccharide chains.

biochemical problems, is based on the important function of Tyr-, His- and Trpresidues in constituting the hydrophobic interactions between the corresponding combining sites of a carbohydrate and a protein. Acknowledgments This work was supported by the Human Capital and Mobility Program of the European Community, the Netherlands Foundation for Chemical Research (SON) and the Netherlands Organisation for Scientific Research (NWO).

1042

36 Carbohydrate-Protein Interactions

References 1. D. Neuhaus, M. P. Williamsen, The Nuclear Overhauser EfSect, 1989, VCH, Weinheim-New York. 2. R. Kaptein, K. Dijkstra, K. Nicolay, Nature, 1978, 274, 293-294. 3. R. Kaptein in Biological Magnetic Resonance, 1982,4, (L. J. Berliner, ed.) pp. 145-191, Plenum Press, New York. 4. P. J. Hore, R. W. Broadhurst, Progr. N M R Spectrosc., 1993,25, 345-402. 5. H.-C. Siebert, E. Tajkhorshid, C.-W. von der Lieth, R. G. Kleineidam, S. Kruse, R. Schauer, R. Kaptein, J. F. G. Vliegenthart, H.-J. Gabius, J. Mol. Model., 1996, 2, 446-455. 6. H.-C. Siebert, C.-W. von der Lieth, R. Kaptein, J. J. Beintema, K. Dijkstra, N. van Nuland, U.M.S. Soedjanaatmadja, A. Rice, J.F.G. Vliegenthart, C.S. Wright, H.-J. Gabius, Proteins, 1997,28, 268-284. 7. H.-C. Siebert, R. Adar, R. Arango, M. Burchert, H. Kaltner, G. Kayser, E. Tajkhorshid, C.-W. von der Lieth, R. Kaptein, N. Sharon, J. F. G. Vliegenthart, H.-J. Gabius, Eur. J. Biochem., 1997,249, 27-38. 8. S. Stob, R. Kaptein, Photochem. Photobiol., 1989, 49, 565-577. 9. P. F. Heelis, B. J. Parsons, G. 0. Phillips, Biochim. Biophys. Acta, 1979, 587, 455-462. 10. K. A. Muszkat, T. Wismontski-Knittel, Biochemistry, 1985, 24, 5416-5421. 11. R. M. Scheek, R. Kaptein, J. W. Verhoven, FEBS Lett., 1979, 107, 288-290. 12. Y. N . Molin (ed.), Spin Polarization and Magnetic Effects in Radical Reactions, 1984, Elsevier, Amsterdam. 13. H.-C. Siebert, R. Kaptein, J. J. Beintema, U. M. S. Soedjanaatmadja, C. S. Wright, A. Rice, R. G. Kleineidam, S. Kruse, R. Schauer, P. J. W. Pouwels, J. P. Kamerling, H.-J. Gabius, J. F. G. Vliegenthart, Glycoconj. J., 1997, 14, 53 1-534. 14. H.-C. Siebert, S. Andre, G. Reuter, R. Kaptein, J. F. G. Vliegenthart, H.-J. Gabius, Glycoconj. J., 1997, 14, 945-949. 15. H.-C. Siebert, S. Andre, G. Reuter, H.-J. Gabius, FEBS Lett., 1995, 371, 13 -16. 16. B. Lee, F. M. Richards, J. Mol. Biol., 1971, 55, 379-400. 17. M. L. Connolly, J. Appl. Cryst., 1983, 16, 548-558. 18. M. L. Connolly, Science, 1983, 221, 709-713. 19. E. Kellenbach, T. Hiird, R. Boelens, K. Dahlman, J. Carlstedt-Duke, J.-A. Gustafsson, G. A. van der Marel, J. H. van Boom, B. Maler, K. R. Yamamoto, R. Kaptein, J. Biomol. N M R , 1991, I , 105-110. 20. J. T. Edward, J. Chem. Educ., 1970, 47, 261-270. 21. F. Buck, H. Riiterjans, R. Kaptein, K. Beyreuter, Proc. Natl Acad. Sci. USA, 1980, 77, 51455148. 22. S. Stob, R. M. Scheek, R. Boelens, R. Kaptein, FEBS Lett., 1988, 239, 99-104. 23. R. W. Broadhurst, C. M. Dobson, P. J. Hore, S. E. Radford, M. L. Rees, Biochemistry, 1991, 30, 405-412. 24. F. Quiocho, Pure Appl. Chem.; 1989, 61: 1293-1306. 25. N. Sharon, H. Lis, H., Science, 1989,246, 227-246. 26. N. Sharon, Trends Biochem. Sci., 1993, 18, 221-226. 27. J. M. Rini, Annu. Rev. Biophys. Biomol. Struct., 1995, 24, 551-577. 28. H.-J. Gabius, S. Gabius, Glycosciences: Status und Perspectives, 1997, Chapman & Hall, Weinheim-London. 29. R. A. Dwek, Chem. Rev., 1996, 96, 683-720. 30. H.-J. Gabius, Eur. J. Biochem., 1997, 243, 543-576. 31. T. Diaz-Mauriiio, D. Solis, J. JimCnez-Barbero, M. Martin-Lomas, H.-C. Siebert, J. F. G. Vliegenthart, Carbohydr. Eur., 1998, in press. 32. J. L. Asensio, F. J. Caiiada, M. Bruix, A. Rodriguez-Romero, J. Jimenez-Barbero, Eur. J. Biochem., 1994,230, 621-633. 33. J. L. Asensio, F. J. Caiiada, M. Bruix, C . Gonzales, N. Khiar, A. Rodriguez-Romero, J. Jimenez-Barbero, Glycobiology, 1998,8, 569-577. 34. K. Hom, M. Gochin, W. J. Peumans, N. Shine, FEBS Lett., 1995,361, 157-161.

References

1043

35. H.-C. Siebert, R. Kaptein, J. F. G. Vliegenthart, in Lectins und Glycobiology (H.-J. Gabius, S. Gabius, eds), 1993, pp. 105-1 16, Springer Verlag, Heidelberg, New York. 36. C. S. Wright, J. Mol. B i d , 1989,20Y, 475-487. 37. N. H. Andersen, B. Cao, A. Rodriguez-Romero, B. Arreguin, Biochemistry, 1993, 32, 14071422. 38. R . Arango, E. Rodriguez-Arango, R. Adar, D. Belenky, F. G. Loontiens, S. Rozenblatt, N. Sharon, FEBS Lett., 1993, 330, 133-136. 39. R. Adar, N. Sharon, N., Eur. J. Biochem., 1996,239, 668-674. 40. T. P. Kogan, B. M. Revelle, S. Tapp, D. Scott, P. J. Beck, J. Biol. Chem., 1995, 270, 1404714055. 41. B. Shaanan, H. Lis, N. Sharon, Science, 1991,254, 862-866. 42. E. Moreno, S. Teneberg, R. Adar, N. Sharon, K.-A. Karlsson, J. Angstrom, Biochemistry, 1997,36,4429-4437. 43. E. Tajkhorshid, H.-C. Siebert, M. Burchert, H. Kaltner, G . Kayser, C.-W. von der Lieth, R . Kaptein, J. F. G. Vliegenthart, H.-J. Gabius, J. Mol. Model., 1997, 3, 325-331. 44. Y. D. Lobsanov, M . A. Gitt, H. Leffler, S. H. Barondes, J. M. Rini, J. Biol. Chen?.,1993, 26, 27034-27038. 45. Y. Bourne, B. Bolgiano, D. Liao, G. Strecker, P. Cantau, 0. Herzberg, T. Feizi, C. Cambillau, Nature Struct. Biol., 1994, I , 863-870. 46. D. Liao, G. Kapadia, H. Ahmed, G. R. Vasta, 0. Herzberg, Proc. Natl Acad. Sci. U S A , 1994, 91, 1428-1432. 41. M. S. Johnson, N. Srinivasan, R. Sowdhamini, T. L. Blundell, Crit. Rev. Biochem. Mol. B i d , 1994,29, 1-68. 48. M. W. MacArthur, R. A. Laskowski, J . M. Thornton, J. M., Curr. Opin. Struct. Biol., 1994, 4, 731-737. 49. J. Hirabayashi, in Glycosciences: Status und Perspectives (H.-J. Gabius, S. Gabius, eds.), 1997, pp. 355-368, Chapman & Hall, London-Weinheim. 50. H.-C. Siebert, M. Gilleron, H. Kaltner, C.-W. von der Lieth, T. Kozar, N. V. Bovin, E. Y. Korchagina, J. F. G. Vliegenthart, H.-J. Gabius, Biochem. Biophys. Rex Commun., 1996, 219, 205-212. 51. V. L. Bevilacqua, D. S. Thomson, J. H. Prestegard, Biochemistry, 1990, 29, 5529-5537. 52. V. L. Bevilacqua, Y. Kim, J. H. Prestegard, Bioeliemistry, 1992, 31, 9339-9349. 53. J. L. Asensio, F. J. Cafiada, J. Jimenez-Barbero, Eur. J. Biochem., 1995, 233, 618-630. 54. T. Weimar, T. Peters, Anyew,. Chem. Int. Ed Enyl., 1994, 33, 88-91. 5 5 . L. Poppe, G . S. Brown, J. S. Philo, P. V. Nikrad, B. H. Shah, J. Am. Chem. Soc., 1997, I I Y , 1727- 1736. 56. M. Gilleron, H.-C. Siebert, H. Kaltner, C.-W. von der Lieth, T. Kozar, N . V. Bovin, E. Y. Korchagina, H.-J. Gabius, J. F. G. Vliegenthart, Eur. J. Biochem., 1998, 252, 416-427. 57. H.-C. Siebert, C.-W. von der Lieth, M. Gilleron, G. Reuter, J. Wittmann, J. F. G. Vliegenthart, H.-J. Gabius, in Glycosciences: Status und Perspectives (H.-J. Gabius, S . Gabius, eds.), 1997, pp. 291-310, Chapman & Hall, London-Weinheim. 58. A. Poveda, J. Jimenez-Barbero, <'hem. Soc. Rev., 1998, 27, 133-143. 59. G. Wagner, S. G. Hyberts, T. F. Havel, Annu. Rec. Biophys. Biomol. Struct., 1992,21, 167-198. 60. S. J. Crennel, E. F. Garman, W. Graeme Laver, E. R. Vimr, G. L. Taylor, Proc. Nut/ Acud. Sci. U S A , 1993, YO, 9852-9856. 61. K. Hird, J. P. Kamerling, J. F. G . Vliegenthart, Carbohydr. Rex, 1992, 236, 315-320. 62. K. HBrd, J. F. G . Vliegenthart, in Glycobiology. A Prcicticul Approach (M. Fukuda, A. Kobata, eds.), 1993, pp. 223-~242,Oxford University Press, Oxford, New York, Tokyo. 63. T. de Beer, C. W. E. M. van Zuylen, K. HBrd, R . Boelens, R. Kaptein, J. P. Kamerling, J. F. G . Vliegenthart, FEBS Lett., 1994, 348, 1-6. 64. T. de Beer, C. W. E. M. van Zuylen, B. R. Leeflang, K. HBrd, R. Boelens, R. Kaptein, J. P. Kamerling, J. F. G. Vliegenthart, Eur. J. Biochem., 1996, 241, 229-242. 65. J. P. M. Lommerse, L. M. J. Kroon-Batenburg, J. Kroon, J. P. Kamerling, J. F. G. Vliegenthart, J. Biomol. N M R , 1995, 5, 79-94. 66. J. P. M. Lommerse, L. M. J. Kroon-Batenburg, J. P. Kamerling, J. F. G. Vliegenthart, Biochemistry, 1995, 34, 8196-8206.

1044

36 Carbohydrate-Protein Interactions

67. G. A. Tennent, M. B. Pepys, Biochem. Soc. Trans., 1994,22, 74-79 68. M. B. Pepys, T. W. Rademacher, S. Amatayakul-Chantler, P. Williams, G. E. Noble, W. L. Hutchinson, P. N. Hawkins, S. R. Nelson, J. R. Gallimore, J. Herbert, T. Hutton. R. A. Dwek, Proc. Nut1 Acad. Sci. USA, 1994, 91, 5602-5606. 69. J. Emsley, H. E. White, B.P. O’Hara, G. Oliva, N. Srinivasan, I.J. Tickle, T. L. Blundell, M. B. Pepys, S. P. Wood, Nature, 1994,367, 338-345. 70. A. K. Shrive, G. M. T. Cheetham, D. Holden, D. A. A. Myles, W. G. Turnell, J. E. Volanakis, M. B. Pepys, A. C. Bloomer, T. J. Greenhough, Nuture Struct. Biol.,1996, 3, 346-354. 71. A. Bohne, E. Lang, C.-W. von der Lieth, J. Mol. Model., 1997, 3, 1-5. 72. H.-C. Siebert, C.-W. von der Lieth, X. Dong, G. Reuter, R. Schauer, H.-J. Gabius, J. F. G. Vliegenthart, Glycohiology, 1996, 6, 561-572.

Carbohydrates in Chemistry and Biology Editor Beat Ernst, Gerald W. Hart, Pierre Sinay copyright OWILEY-VCH Verlag GmbH, 2000

37 Biacore Wolfgang Jager

37.1 Introduction Since the first commercially available Biacore system (Biacore AB, Uppsala, Sweden) was introduced back in 1990, the technology has advanced to become a standard method in studying biomolecular interactions. This Chapter describes the function and practical utilization of this technology, discusses typical user questions and offers suggestions for problem solving. Studies of oligosaccharide interactions are presented, while the whole application range includes all kinds of biomolecules. Since applications are continually optimized and new reagents and consumables are being developed, detailed protocols will not be given. The reader is referred to updated protocols provided by the supplier, to the literature cited in the text, and to the current reference list on the Internet under http://www.biacore.com. 37.1.1 Real-time Analysis by Surface Plasmon Resonance The detection principal of Biacore relies on the physical phenomenon of surface plasmon resonance (SPR: Figure 1). Specifically, it is based on an optical unit which measures the refractive index located on the matrix side of the flat sensor chips (Figure 2). One of the interacting partners is immobilized on the chip matrix. Substance accumulation by molecule binding to the immobilized partner directly influences the refractive index, which is detected in real-time and indicated immediately [ 11. Inside the instrument, the chip matrix forms a roof of small flow cells. In current instrument configurations, each sensor chip serves two or four of these flow channels, giving two or four separate measuring surfaces. The integrated flow system provides a continuous flow of buffer or samples over the chip surface. Sample injections and measurements run automatically: only specimen loading to a sample loop can be

1046

37 Biacore

A) srslscrr chip masrix

B) Mctlecule immobilisation

C ) Interaction analysis

D) Regeneration and reuse

Figure 1. Different steps for applying the biosensor are displayed. A) A suitable sensor chip is chosen. B) Immobilization of biomolecules in the chip matrix generates a specific measuring surface. C) Interaction is analyzed by injecting binding partners along the surface. D) After the regeneration the surface is used for the next analysis.

optical

Figure 2. The sensor chip is composed of a glass carrier coated with a thin gold layer on which a biocompatible matrix is attached. An evanescent field is generated when light under the condition of total internal reflection is directed to the glass side. This field extends about 300 nm into the solution on the other side (the matrix side) and interacts with the refractive index of the solution close to the gold surface. The electromagnetic phenomenon of surface plasmon resonance (SPR) arises in the gold film, resulting in extinction of the reflected light at a specific angle. The angle of minimum reflected intensity (the resonance angle) varies with the refractive index on the matrix side. Changes in the resonance angle are directly proportional to changes in mass concentration due to binding or dissociation of biomolecules. The signal is presented in resonance units (RU), where 1 RU corresponds approximately to 0.8 pg carbohydrate or 1 pg protein bound per mm2.

37.1 Introduction

1047

either automatic or manual, depending on the system used. Signal detection occurs simultaneously in all channels and a serial flow along different surfaces allows monitoring of separate interactions with just one sample injection, or measuring the in-line control signal at a reference surface simultaneously. Because detection directly reacts to substance accumulation, the signal depends both on the molecular weight and concentration of the injected partner, and on the number of binding sites present on the matrix. However, detectable limits are picomolar concentrations or a minimal mass of about 180 Da [2]. 37.1.2 Information in a Sensorgram In the sensorgram the actual signal corresponding to mass concentration at the sensor surface is plotted against time (Figure 3 ) . Interaction with the immobilized partner during sample injection causes an increase in signal. At the end of the injection, the system automatically switches to buffer flow and a decrease in signal now reflects the dissociation of interactants. The matrix is regenerated and reused.

Signal (kKli)

P

16

14

12

J

AAMAAAA

I

I

I

I

I

100

20

300

400

5 0

llime (sec)

Figure 3. Interaction progress is monitored in a sensorgram. Binding of molecules during sample injection and dissociation during buffer flow are indicated. The height or the initial slope of the signal are directly correlated to sample concentration, while the shape of the curves reflect association and dissociation kinetics. After each measurement, remaining material is washed away by injecting an appropriate regeneration solution leaving only the immobilized partner at the surface, and a new analysis cycle can be done.

1048

37 Biacore

Multiple usage of one surface greatly enhances reproducibility of results and, therefore, is especially valuable for comparative studies. Analysis can be purely qualitative (yes or no), in order to determine the presence or absence of binding activity in a particular sample, or under selected conditions. Whole complexes can be built up on the matrix, step by step, and simultaneously viewed on the screen. The concentration of biologically active molecules can be determined by use of a standard curve. Association and dissociation kinetics as well as the affinity of an interaction are calculated by the fitting routines of the evaluation software [3].

37.2 Experimental Procedures This biosensor allows experiments to be organized in different ways. The choice can be made as to which interacting partner to immobilize on the chip matrix and which to inject in solution. Different immobilization strategies may be applied, and both purified samples or extracts can be used for the measurements. Interaction analyses are either performed by direct binding to the immobilized molecule, or as an inhibition assay with both partners free in solution. With this flexibility, the easy handling of the instruments, and a wide diversity of standardized protocols, each user has the opportunity to immediately start their analysis. On the other hand, many assays can be carried out in different ways. Therefore, it is recommended to obtain a general view of the technology and to clearly define the goal for each measurement, in order to structure an analysis in the best way.

37.2.1 Immobilization of Biomolecules at the Sensor Surface Before the actual measurement, a sensor chip is specifically prepared by immobilizing the biomolecule of interest in the matrix. There is a variety of sensor chips available with different surface properties, and there are standardized methods for coupling just about every biomolecule. A summary of the currently available sensor surfaces and the corresponding immobilization strategies is presented in Table 1. The immobilization level can be controlled on the screen and varied to fit any particular application. It is recommended to use purified samples for the coupling in order to create a homogeneous measuring surface, while the total amount of material consumed is usually only 1-5 pg. Covalent coupling via amine groups is the standard immobilization procedure [21], which includes three steps. First, the carboxymethyl (CM)-groups on the sensor chip matrix are activated by a mixture of EDC (N-ethyl-N’-(3-dimethylaminopropy1)-carbodiimide) and NHS (N-hydroxysuccinimide), then covalently bound to free amine groups of the biomolecules, and finally all remaining activated

37.2 Experimental Procedures

1049

Table 1. Sensor surfaces. Sensor chip matrix

Coupling via*

Immobilized molecules/comment

Standard CM-dextran

Proteins or small compounds

Streptavidin-dextran

NH2-, SH-, CHO- or COOH-groups Biotin

NTA-dextran** Flat hydrophobic surface**

Histidin-6mer Hydrophobic adsorption

Biotinylated molecules like Oligosaccharides or nucleic acids His-tag fusion proteins Lipid-monolayers

New sensor chips (continually developed, surfaces might change) Low carboxylated dextran Shortened dextran Flat CM-surface Lipid-anchor-dextran** Pure gold surface

As standard CMdextran As standard CMdextran As standard CMdextran Liposome capture User-defined

Reduced non-specific binding of culture medium or cell homogenate Reduced non-specific binding of serum, injection of phages or cells When dextran is not required, injection of phages or cells Liposomes (lipid bilayer) Customized surfaces, spin-coating, selfassembled monolayers

* for detailed protocols please refer to information supplied by the manufacturer ** surface can be completely regenerated

groups are deactivated by an excess of ethanolamine. In the same manner, CMsurfaces can be chemically modified, in order to bind proteins or other molecules via thiol-, aldehyde-, or carboxyl-functions [4]. In some cases, covalent coupling can result in loss of biological activity. Then, the immobilization strategy has to be changed, for instance from amine to thiol coupling or to indirect binding by high affinity capture. Covalent immobilization ensures a stable baseline, which is important whenever low signals are expected in the measurements. Molecule capture is an indirect immobilization, carried out for instance by covalent coupling of an antibody for capturing the corresponding antigen. This method allows removal of the captured interaction partner after measurement, and the surface can be re-loaded with the same or a new molecule. Moreover, specifically captured molecules need not be purified, but can be present in extracts, so that the danger of protein inactivation by a purification protocol is eliminated [ S ] . Carbohydrates are usually biotinylated and bound to a streptavidin matrix. A simple procedure has been published for the preparation of 4-(biotinamid0)phenylacetylhydrazine and the employment of this substance to biotinylate oligosaccharides at the reducing end [6]. An additional incubation step at 4°C at pH 3.5 greatly reduces the mixture of stereoisomers resulting in an excess of the cyclic P-glycoside form, which also exists in natural N-linked oligosaccharides. This aspect is important in that the type of modification can influence the activity of bio-

1050

37 Biucore

Signal (mu)

Lipsome capture

Toxin binding

Conditioning

1 1

Regeneration

18 16

12 14: 10

.Li . ........... i..

I.........I

0.5 mM DMPC with 2% GMl

.........l

1................I

I.........I

5omM l0mM NaOH glycine pH 1.5

I

2.5@Id cholera

l0mM glycine pH 1.5

toxin

20 mM chaps I

I

I

1

I

molecules. Heparin, for instance, showed slightly changed binding characteristics depending on whether amino or oxidized cis-diol groups were biotinylated [7]. Glycolipids can be incorporated in liposomes, directly captured on the matrix of the sensor chip (Figure 4),and completely removed after a measurement [22]. In a similar approach using immobilized antibodies for liposome capture, bacterial toxin affinity and binding specificity for glycolipid receptors were determined [ 81. 37.2.2 Surface Regeneration

After immobilization, surface activity and regeneration conditions are tested. Repeated injections with the same sample and regeneration of the surface must lead to identical binding signals and to a stable baseline. Regeneration is needed, whenever the injected partner does not dissociate in buffer flow within an acceptable time. Care should be taken while testing the conditions, in order not to inactivate the surface-bound molecules. Injections of 1 min pulses are recommended, using successively harsher solutions, until all complexes have dissociated and the signal is back on the original baseline level. Acidic solutions down to pH 1 (100 mM HCl), basic solutions up to pH 13 (100 mM NaOH), hydrophobic solutions (up to 50% ethylene glycol), chaotropic ions, detergents, nonpolar solvents, and also chelating agents may be applied. If the inactivation of the matrix-bound molecule by the regeneration procedure cannot be avoided, then a capturing strategy should be applied, by completely regenerating the surface after each analysis and re-loading it with an active interaction partner.

1051

37.2 Experimental Procedures

37.2.3 Interaction Analysis and Controls Sample injections are typically performed at flow rates of 5-30 yl min-' using sample concentrations corresponding to the affinity of the interaction (K D ,affinity dissociation constant). A measurement will immediately give an answer as to whether binding occurs or not. Nevertheless, each response should be verified through a surface and a sample control. A second trace on the same sensor chip, chemically modified in the same manner as the active surface (e.g. activated/ deactivated), usually serves as a control channel. However, an optimized reference surface carries an inactive molecule, which is similar to the active molecule and immobilized at the same amount. Negative samples like non-specific proteins or carbohydrates, or inactive extracts, serve as a control in solution. The interaction signal during sample injection is divided into two main parts, the so-called bulk response at the beginning and the end of the injection and the actual binding signal (Figure 5). Since the detector reacts to changes in the refractive index, every solution which is optically denser or thinner than the running buffer causes this signal jump. Using the sample buffer as running buffer greatly minimizes this effect and improves the data quality. Possible background binding to the chip surface can be minimized by choosing a different sensor chip (see Table 1). If background is based on ionic interaction with the CM-groups, then the negative charge at the matrix should be reduced by expanding activation and deactivation times to about 15 min each. Increasing the ionic strength in the injected sample up to 500 mM sodium chloride can also help.

Si,

I

I

I

I

100

0 1

2oo

I

I

i*

I

Time (sec)

1052

37 Biucove

For the analysis of crude samples, like extracts or serum, a 1 : 10 dilution with running buffer is often sufficient to decrease any background to an acceptable level. Non-specifically bound material or aggregated proteins can usually be removed from the surface by using detergent or 5-50 mM sodium hydroxide.

37.2.4 Determination of Kinetic Rate Constants The dynamics of an interaction are characterized by the speed of complex building and decay and described by the rate constants of association and dissociation. Quantitative kinetic analysis is performed by using a set of curves obtained by injecting samples in concentrations of about 0.1-100 times KD. Based on the mathematical rate equation for the assumed interaction model, a global fitting routine directly evaluates the measured curves. 1 : 1 binding and reactions, like heterogeneity, conformational changes, or bivalence, are predefined in the software, and customized models can be created by the model editor. Published dissociation rate constants range from 6 x lop7 sec-' [9] to 10 sec-' [lo]. Association rate constants, which can be determined by global fitting, are shown in Figure 6. The quality of a fit is assessed by means of the corresponding residual plot. Differences between measured and calculated curves should be near the noise level and regularly dispersed. Any trend in the residuals might indicate use of a wrong interaction model or inaccuracy during the analysis. The choice of the correct model is important as well as proof that the binding partners react in the way the model assumes. For example, a monovalent molecule will bind multivalently if it ag-

-

.---

Very little kinetic information

_.--

Quantitative kinetic information

lo'

-

Flow rate = 30 p l h h I

I

I

I

__*.--.

-/--

/--

37.2 Experimental Procedures

1053

gregates in solution. This was in fact found during the kinetic analysis of a single chain Fv antibody raised against Salmonella 0-polysaccharide [ 111. Deviations from a 1 : 1 interaction are indicated by altered dissociation curves following injections with different contact times [3]. Mass transport to the surface can affect binding signals of interactions with fast association rates, since the last distance to the immobilized partner must be overcome by diffusion. Nevertheless, due to the defined laminar flow mass transport can be mathematically described and, if necessary, taken into account during the data evaluation [ 121. Transport has no influence on the interaction, when the initial slopes of binding curves obtained at flow rates of 15 and 75 pl min-' do not deviate more than 10% from each other. In general, kinetic analyses should be carried out at flow rates of 2 3 0 p1 min-' on a surface with a low immobilization level corresponding to a maximum surface binding capacity (Rmax)of 10-100 RU. The experimental arrangement should be as simple as possible, in order to ensure robust and secure data evaluation. It can be achieved, for instance, by coupling a bivalent molecule to the surface and injecting the monovalent partner. This results in a 1 : 1 binding scheme, whereas in an opposite construction a bivalent interaction would occur. 37.2.5 Affinity Determination

Affinity can be directly determined from the equilibrium response, which is reached during an injection, when as many complexes are formed as decay and the signal no longer shows any change. The sample concentration leading to 50% surface saturation directly corresponds to the affinity dissociation constant ( K D ) The . data can be evaluated by the existing software program or a standard Scatchard plot [13]. Affinity determination with both partners free in solution can be performed by an inhibition assay [ 14, 151. Interaction partners are mixed and injected after equilibrium binding has been reached in solution. Biacore specifically detects free molecules of one of the interactants. This concentration is inversely proportional to the amount of formed complexes. Low molecular weight carbohydrates often show weak and rapid interactions, reaching equilibrium within seconds after injection and dissociating rapidly during buffer flow. The corresponding response looks like a square pulse, similar to bulk signals. Nevertheless, even transient binding can be quantitatively analyzed by realtime detection (Figure 7). When working with small sugars, active and reference surfaces should be immobilized at the same density, in order to avoid differences in the bulk signal [ 161, or a correction factor should be included [ 2 3 ] . A comparison of affinity and kinetic data ensures consistency of the analysis. For a 1 : 1 interaction, affinity is equal to the quotient of dissociation and association rate constants ( K D = kdiss/k,,,). However, some reactions do not follow a 1 : 1 interaction, and it must be noted that the apparent affinity of bi- or multivalent reactions can be directly affected by the density of immobilized interaction partners. Nevertheless, the chip surface resembles the situation on a cell surface and possibly reflects the in vivo environment better than an assay free in solution [ 171.

1054

37 Biacore

Req vs Conc plot ,

1500rn

t7 I/

Figure 7. Maltose binding to immobilized anti-maltose antibody is shown. The concentration of maltose ranges over 5.8-1500 1 M with duplicate injections of each concentration. Depicted are the binding signals after subtraction of the background response measured on a surface, which was coated with an unspecific antibody. Affinity determination was done by fitting data from the Plot of equilibrium response (Req)against concentration to the Langmuir 1 : 1 binding isotherm using BIA evaluation software 3.0.

37.3 Application Areas The application range of Biacore for studying biomolecular binding is quite diverse. Some of the references describing analyses of oligosaccharide interactions have already been cited in the text, while three publications are presented below in more detail. 37.3.1 Selectin Binding to a Glycoprotein Ligand Selectins are transmembrane proteins with membrane-distal Ca2+-dependent lectin domains. This family of cell adhesion molecules is involved in the tethering and rolling of leukocytes on the blood vessel endothelium. Nicholson and colleagues [ 101 published the analysis of leukocyte selectin, CD62L, binding to the mucin-like glycoprotein, Gly-CAM-1. The extracellular portion of CD62L was expressed as a fusion protein with two domains of CD4. The resulting soluble monomeric form of CD62L was used to measure the monovalent affinity and kinetics of its interaction with the native GlyCAM-1. Injection of CD62L-CD4 along a matrix coated with GlyCAM-1 resulted in an immediate increase in signal, attaining equilibrium within seconds. Different concentrations of CD62L-CD4 were applied and subtracting the signals on the control surface carrying no GlyCAM-1 from the signals on the active surface resulted in the actual binding response. Fitting the graph of equilibrium binding level against concentration to a 1 : 1 binding model showed a low affinity of K D NN 108 pM. No binding occurred when CD62L-CD4 was applied in the presence of EDTA, or when the control chimera sCD48-CD4 was injected. The binding response dropped with a half time of 0.07 sec after switching to buffer, corresponding to a dissociation rate constant of kdlss NN 10 sec-'. But, since it

37.3 Application Areas

1055

also takes this time to wash the sample out of the flow cell, dissociation of the complex may even be faster and was therefore given as kdi,, 2 10 sec-'. Direct measurement of the association rate constant was not possible because the reaction reached equilibrium within one sec. However, with the values for affinity, 108 pM, and dissociation rate, 210 sec-', the association rate could be calculated as k,,, 2 lo6 M-' sec-'. A higher affinity was found when the analysis arrangement was changed by immobilizing CD62L-CD4 and injecting GlyCAM-1 along the surface. Binding was apparently detectable at concentrations as low as 30 nM GlyCAM-1, and the dissociation rate constant was reduced approximately by a factor of 10,000 when compared to the data obtained in the above 1 : 1 binding assay. This strongly suggested that GlyCAM-1 interacts in a multivalent manner to immobilized CD62L-CD4. The data supported the hypothesis that the highly dynamic cell adhesion process is based on the low affinity and fast kinetics of selectin interaction. GlyCAM-1 was shown to bind multivalently to immobilized CD62L at concentrations just above its mean serum level, indicating a physiological role by interacting with CD62L embedded in leukocyte membranes. 37.3.2 Oligosaccharide Characterization

An analytical method was described by Blikstad et al. [ 181 to detect unlabelled oligosaccharides or glycopeptides in column effluents. Sumbucus nigru agglutinin (SNA) specific for sialic acid, and Ricinus comrnunis agglutinin (RCA) specific for terminal galactose were covalently immobilized and used to identify different biand triantennary N-linked oligosaccharides. After each measurement, the lectin surfaces were regenerated with a 3 min pulse of 100 mM HCl. Surface stability was found for at least 300 measurements over a period of 2 weeks. For 25 consecutive analysis cycles the coefficient of variation was calculated to be as low as 1.21%. Detection was highly specific, since Sialylated oligosaccharide bound to SNA but not to RCA, whereas desialylated oligosaccharide with a terminal galactose interacted only with RCA. The sensitivity of the analysis was determined by injecting a concentration series of trisialyated triantennary N-linked oligosaccharide (A3) and disialylated biantennary N-linked oligosaccharide (A2) over the SNA surface. Both oligosaccharides could be detected at concentrations of less than 1 nM, while above 200 nM no further increase in signal occurred, indicating surface saturation. Similarly, asialo-galacto biantennary N-linked oligosaccharide (NA2) was detectable at a concentration of 10 nM, using immobilized RCA as detector. During the measurements only 35 pl of sample were consumed for each injection. The feasibility of this assay was demonstrated by using a pronase digest of transferrin. After a chromatographic separation, individual fractions from the column were diluted and injected over the SNA surface. While measuring binding activity to SNA, one major peak showed the presence of responding glycopeptides. These fractions showed practically no UV absorbance. By using other lectins, the method can be extended to detect oligosaccharides containing other structures.

1056

37 Biucore

Combining gel filtration, different glycosidases and different lectins for an analysis, the overall structures of oligosaccharides could be deduced.

37.3.3 In situ Modification of Immobilized Carbohydrates Beside qualitative detection of oligosaccharides and quantitative affinity or kinetic analyses, the in situ modification of carbohydrates immobilized at a sensor chip surface has been published [6, 191. Van der Menve et al. [20] used this approach to analyze the binding of B-lymphocyte antigen, CD22, to the highly glycosylated leukocyte surface protein CD45. Native CD45 (CD45-thy) purified from rat thymus was covalently coupled to sensor chip CM5. CD22Fc (domains 1-3 of mouse CD22 fused to the Fc portion of human IgG) interacted with immobilized CD45-thy. The in situ modification of the sialoglycoconjugates present on CD45-thy by desialylation and resialylation provided the basis for a detailed characterization of this interaction. Muuckiu umurensis agglutinin (MAA) and Sumbucus nigru agglutinin (SNA) lectins, which are specific for a2-3- and 1x2-6-linkedsialic acids, respectively, bound to CD45-thy, indicating the presence of the corresponding sialic acids. Treatment of immobilized CD45-thy by a 30 min injection of Vibrio cholerae sialidase abolished CD22Fc interaction and substantially decreased both MAA and SNA binding. Resialylation of CD45 with GalPl-4GlcNAc a2-6-sialyltransferase using NeuAc as substrate increased SNA binding, while CD22Fc did not bind. CD22Fc interaction was fully restored after 1x2-6-resialylatingCD45-thy using NeuGc as a substrate. Binding of SNA and the lack of MAA binding confirmed the specificity of the a2-6resialylation. The results showed that CD22Fc binds to NeuGca2-6Gal~1-4GlcNAc, which are carried on CD45-thy N-glycans. Moreover, the experiments demonstrated the potential for analyzing binding specificities to selectively modified carbohydrates immobilized at a sensor chip.

References 1. Jonsson U. and Malmqvist M. (1992) Real time biospecific interaction analysis. The integration of surface plasmon resonance. Detection, general biospecific interface chemistry and microfluidics into one analytical system, Adv. Biosensors 2:29 1-336. 2. Karlsson R. and Stihlberg R. (1995) Surface plasmon resonance detection and multi-sensing for direct monitoring of interactions involving low molecular weight analytes and for determination of low affinities, Anal. Biochem. 228:274-280. 3. Karlsson R. and Falt A. (1997) Experimental design for kinetic analysis of protein protein interactions with surface plasmon resonance biosensors, J. Immunol. Methods 200:121-133. 4. Johnsson B., L o f h S., Lindquist G., Edstrom A,, Miiller-Hilgren R.-M. and Hansson A. (1995) Comparison of methods for immobilization to carboxymethyl dextran sensor surfaces by analysis of the specific activity of monoclonal antibodies, J. Mol. Recognit. 8: 125-131. 5. Nath D., van der Menve P. A,, Kelm S., Bradfield P. and Crocker P. R. (1995) The aminoterminal immunoglobulin-like domain of sialoadhesin contains the sialic acid binding site. Comparison with CD22, J. Biol. Chem. 270:26184-26191.

Rejerences

1057

6. Shinohara Y., Sota H., Gotoh M., Hasebe M., Tosu M., Nakao J. and Hasegawa Y. (1996) Bifunctional labeling reagent for oligosaccharides to incorporate both chromophore and biotin groups, Anal. Chem. 68:2573-2579. 7. Mach H., Volkin D. B., Burke C. J., Middaugh C. R., Lindhardt R. J., Fromm J. R., Loganathan D. and Mattsson L. (1993) Nature of the interaction of heparin with acidic fibroblast growth factor, Biochemistry 325480-5489. 8. MacKenzie C. R., Hirama T., Deng S., Bundle D. R., Narang S. A. and Young N. M. (1997) Quantitative analysis of bacterial toxin affinity and specificity for glycolipid receptors by surface plasmon resonance, J. Biol. Chem. 2725533-5538. 9. Deka J., Kuhlmann J. and Muller 0. (1998) A domain within the tumor suppressor protein APC shows very similar biochemical properties as the microtubule-associated protein tau, Eur. J. Biochem. 253591-597. 10. Nicholson M. W., Barclay A. N., Singer M. S., Rosen S. D. and van der Merwe P. A. (1998) Affinity and kinetic analysis of L-selectin (CD62L) binding to glycosylation-dependent celladhesion molecule-1, J. Biol. Chem. 273:763-770. 11. MacKenzie C. R., Hirama T., Lee K. K., Altmann E. and Young N. M. (1996) Analysis by surface plasmon resonance of the influence of valence on the ligand binding affinity and kinetics of an anti-carbohydrate antibody, J. Biol. Chem. 271:1527-1533. 12. Myszka D. G., He X., Dembo M., Morton T. A. and Goldstein B. (1998) Extending tha range of rate constants available from BIACORE: Interpreting mass transport-influenced binding data, Biophys. J. 75583-594. 13. Karlsson R., Fagerstam L., Nilshans H. and Persson B. (1993) Analysis of active antibody concentration. Separation of affinity and concentration parameters, J. Immunol. Methods 166:75-84. 14. Karlsson R. (1994) Real-time competitive kinetic analysis of interactions between low-molecular-weight ligands in solution and surface-immobilized receptors, Anal. Biochem. 221:142-1 5 1 . 15. Nieba L., Krebber A. and Pluckthun A. (1996) Competition BIAcore for measuring true affinities: Large differences from values determined from binding kinetics, Anal. Biochem. 234: 155-165. 16. Ohlson S., Strandh M. and Nilshans H. (1997) Detection and characterization of weak affinity antibody antigen recognition with biomolecular interaction analysis, J. Mol. Recognit. 10:135138. 17. Shinohara Y., Hasegawa Y., Kaku H. and Shibuya N. (1997) Elucidation of the mechanism enhancing the avidity of lectin with oligosaccharides on the solid phase surface, Glycobiology 7: 1201-1208. 18. Blikstad I., Fagerstam L. G., Bhikhabhai R. and Lindblom H. (1996) Detection and characterization of oligosaccharides in column effluents using surface plasmon resonance, Anal. Biochem. 233:42-49. 19. Hutchinson A. M. (1994) Characterization of glycoprotein oligosaccharides using surface plasmon resonance, Anal. Biochem. 220:303-307. 20. van der Merwe P. A,, Crocker P. R., Vinson M., Barclay A. N., Schauer R. and Kelm S. (1996) Localization of the putative sialic acid-binding site on the immunoglobulin superfamily cellsurface molecule CD22, J. Biol. Chem. 271:9273-9280. 21. Johnson B. and Lofis S. (1991) Immobilization of proteins to a carboxymethyldextran modified gold surface for biospecific interaction analysis in surface plasmon resonance, Anal. Biochem. 198:268-277. 22. Cooper, M. A,, Hansson, A,, Lofas, S. and Williams, D. H. (2000) A vesicle capture sensor chip for kinetic analysis of interactions with membrane-bound receptors, Anal. Biochem. 277r196-205. 23. Karlsson, R., Kullmann-Magnusson, M., Hamalainen, M., Remaeus, A,, Anderson, K., Borg, P., Gyzander, E. and Deinum, J. (2000) Biosensor analysis of drug target interactions, Anal. Biochem. 278:l-13.

Part I Volume 2

V Carbohydrate-Carbohydrate Interactions

Carbohydrates in Chemistry and Biology Editor Beat Ernst, Gerald W. Hart, Pierre Sinay copyright OWILEY-VCH Verlag GmbH, 2000

38 Carbohydrate-Carbohydrate Interactions Dorothe Spillmann and Max M. Burger

38.1 Introduction Cell recognition and adhesion are important events in the formation and maintenance of functional tissues. These processes must be selective and flexible in order to guarantee proper development as in embryogenesis, during the development of the nervous system, for continuous activity of the lymphoid system, but can also be misused by parasitic intruders or aberrant cells like tumor cells. How are glycans involved in these recognition and adhesion phenomena so important for all living organisms? Carbohydrates are the most ubiquitous and prominently exposed molecules on the surface of living cells. The combination of a few building blocks in a seemingly countless array of structures is a principle characteristic of carbohydrate chains that distinguishes carbohydrate sequences from all other biomolecules whether nucleic acids or peptides. Such, these chains offer simultaneously flexible, ordered and easily modulatable motives for cellular interactions. The specific arrangement of individual carbohydrate sequences on surfaces and within matrices creates yet another dimension of diversity: Well defined structures or patches of carbohydrate motives can be created within linear stretches of glycans and glycosaminoglycans (GAG), attached to core protein backbones or located on freely movable lipid anchors for interaction with other molecules [ 1, 21. Glycans are therefore predestined to serve as crucial molecules for recognition and attachment at the moment of cell encounter, Carbohydrates have been recognized as interaction sites in many different instances. Many of these processes are mediated by lectin- or lectin-like molecules that recognise specific carbohydrate motifs. Lymphocyte homing is a multistep process with increasing interaction strengths. A first rolling stage of circulating lymphocytes is dependent on recognition of glycan ligands by protein receptors, the selectins, on the endothelium as well as on the lymphocytes [2, 31. Neuronal development depends on reversible cell extensions and connections followed by disruption which requires the proper interaction with cells, extracellular matrix and solu-

1062

38 Curbohydrute-Carbohydrute Interactions

ble factors like growth factors. Many of these interactions have glycans implicated [4-61. Plant and mammalian lectins have defined roles in tissue maintenance and defense. Glycan ligands are also used as primary receptors by microbial pathogens [7, 81. Many of these interactions are characterized by fairly weak forces which, however, are easily potentiated by orders of magnitude through multimerization of ligand and receptor [2, 9, lo]. Further along this conceptual line is the idea of carbohydrate-carbohydrate interactions mediated by specific carbohydrate sequences in an ordered polyvalent array [ l l ] . Instead of the protein receptor for a glycan ligand, a second glycan is receptor for the first. Such interactions could be both specific and easily controlled by external influences like ionic conditions, shear forces, availability of substrates and biosynthetic activity of genes involved [ 121 and would therefore be ideal for early steps in cell interactions where links should have modest affinities in order to guarantee a quick and flexible reversion. The property of potentially weak interactions, on the other hand, raises one major difficulty, namely the question how to measure such interactions. In cases of structural components as found for instance in plants where direct carbohydrate-carbohydrate interactions have been postulated first, the number of binding sites is extremely large and therefore the avidity reasonably high to be measurable by classic methods of biophysics and biochemistry. In more dynamic systems, like in mammalian tissues that undergo continuous restructuring or underlay shear forces from body fluids the possibilities of carbohydrate-carbohydrate interactions are obvious, but hard to measure. The number of publications dealing with this special aspect of glycobiology is therefore limited. However, the advent and continuous refinement of modern methods like atomic force microscopy, nuclear magnetic resonance, surface plasmon resonance or the increasing availability of tools like anti-

Figure 1. Schematic overview of carbohydrate-carbohydrate interactions. Binding of carbohydrate sequences to one another can be envisaged both in cis and trans mode at the cell surface between different or identical glycoconjugates and between identical or different types of sequences. Also cell-matrix interactions might be stabilized in this way. I protein core, 1 lipid core, I carbohydrate chains

38.2 From Structural Components to Cell Recognition

1063

bodies and homogeneous, synthetic ligands should allow it to study such phenomena equally well as other types of interactions where carbohydrates are involved. This Chapter is divided into a first general part with an overview of the biological background for the different models and is followed by a discussion of the molecular details assumed to play a role in such interactions. Finally, methods in use for assessing carbohydrate-carbohydrate interactions are described. An extensive description of “cook-book’’ recipes is omitted as the use of most methods is discussed in detail in other parts of this series and does not differ for different applications whether carbohydrate-carbohydrate or carbohydrate-protein interactions are analyzed.

38.2 From Structural Components to Cell Recognition 38.2.1 Carbohydrate-Carbohydrate Interactions as Part of Structural Components Extracellular Matrix of Seaweeds-Agarose, Carrageenan and Alginate

The classical view on carbohydrates is the one of the energy store and the structural component, whether extracellular space filler and cushion or as part of a cell wall. In this latter form, carbohydrate-carbohydrate interactions have been proposed for the first time. Agarose, carrageenan and alginate, are examples of carbohydrate networks which are produced by different species of algae as intercellular matrix. They are examples for structural networks created from regular and random stretches of carbohydrate chains which form hydrated, elastic gels, that are stabilized by interchain hydrogen-bonding and ionic complexing thereby providing a very basic form of carbohydrate-carbohydrate interaction. Agarose and carrageenan are each built of two alternating sugar residues that form a repeating unit ([3,6-anhydrofor agarose ~-Galal-3o-Galpl-4]and [3,6 anhydro-~-Galwl-3~-Ga1(4-OS0~)/31-4] and K-carrageenan, respectively). The chains can form double helices which are interrupted at sites where the chains kink due to further modification of the basic units by sulfation or dehydration. The extent of modification, which is under biosynthetic control, determines the structural behaviour of the chains and thereby affects the functional properties of the gelling polysaccharides [ 131. Alginates are another group of glycan networks. They are composed of homo- and heteropolymers of p-D-mannuronic acid and its C5 epimer a-L-guluronic acid, and are produced by brown algae and certain strains of Pseudomonas and Azotobacter bacteria [ 141. The special characteristics of alginates are determined by the sequence of building blocks in the single chain. Blocks of homogeneous mannuronate or guluronate polymers alternate with heteropolymers containing both units in variable order of sequence. Due to the conformation of the two uronic acids, the different types of sequences adopt different three-dimensional conformations that in turn determine the extent of chain-interaction. While the mannuronic acid is in a preferred 4C1 chair conformation and the mannuronan-homopolymer therefore adopts

1064

38 Carbohydrate-Carbohydrate Interactions

an extended structure of either a ribbon in its acid form or a threefold helix in its salt form, the guluronic acid conformation of Cq drives the guluronan-homopolymer to adopt a twofold buckled ribbon, that is more condensed [ 141. In presence of Ca2+ ions these so called G-blocks complex to egg-box like structures, that lead to intra- and interchain cross-linking, for the first time suggested by Rees 1131. A high content of guluronic acid is consequently resulting in much stiffer gels due to carbohydrate-carbohydrate binding while high mannuronic acid contents provide more loosely, flexible gels that have a low degree of chain contacts. The extent of interaction can be modulated not only by the ratio and distribution of guluronic and mannuronic acid along the chain, but also by post-polymerization modification. Some of the bacterial alginates for instance, though identical in basic structure to the algae structures, are 0-acetylated on mannuronate which changes their Ca2+uptake and complexing characteristics [ 151. Cell Walls Many cell walls from bacteria to plants are predominantly composed of carbohydrates. Especially capsular polysaccharides of bacteria have attracted the interest because of their properties being similar to other polysaccharide structures found in mammalians and as virulence factors. In the case of Bucteroides fragilis the identity of a virulent capsular polysaccharide has been identified and reveal an unusual dimeric carbohydrate complex of a polysaccharide chains A and B. The two polysaccharides are linked by non-covalent electrostatic interactions between negatively and positively charged groups within both polysaccharide chains providing a carbohydrate-carbohydrate interaction [ 161. Considering findings with isolated polysaccharides as bacterial colominic acid 1171 or the behavior of GAGS (18, 191, one might anticipate similar phenomena to occur within capsular structures in more than the Bucteroides example. Another form of very simple carbohydrate-carbohydrate interactions can be found in plant cell walls. These walls contain a high content of carbohydrates of which approximately 20% are cellulose microfibrils, some 70-80% are non-cellulose polysaccharides, hemicelluloses and pectins, and up to 10% of the mass is glycoproteins. While some of these structures are very conserved over a wide range of organisms (like cellulose), others are not, as detected by specific antibodies 1201. Cellulose, the homogeneous polymer of P-4-linked D-glucose and the polymer with the largest biomass on earth, forms the basic frame of plant cell walls with rigid microfibrils. Hydrogen-bonds and hydrophobic interaction between the individual, parallel chains stabilize the assembly to a near crystalline form [21, 221. These homogeneous fibrillar structures are cross-linked by an uncounted number of different heteropolysaccharides. Hemicelluloses (xylans, gluco- and galactoglucomannans, xyloglucans and branched glucans), on the one hand, and pectins (galacturonans, rhamnogalacturonans, arabinans, galactans and arabinogalactans), on the other hand, create a fine network between different cellulose fibres. The non-cellulose chains are all more heterogeneous and linear and branched elements create even more versatile structural properties. In this way a three-dimensional network is created that guarantees the physical properties of the plant cell wall. Covalent ester

38.2 From Structural Components to Cell Recognition

b

0

1065

1066

38 Carbohydrute-Carbohydrate Interactions

Figure 3. Segment of a plant cell wall (schematic drawing). Bundles of cellulose microfibers in near crystalline packing (thick bundles) are crosslinked by hemicelluloses (drawn as double-lined intercalated network) and pectins (fine lines with egg-box interaction sites). Carbohydrate-carbohydrate interaction occurs between individual cellulose chains, between cellulose and xyloglucan (a hemicellulose) and between different hemicelluloses and pectins. Besides different forms of non-covalent interactions (Ca2+ mediated “egg-box” structures as well as ion-independent forms of molecular interactions) also covalent ester linkages between carbohydrate chains are found (adapted from ~31).

linkages between individual polysaccharides are found as well as ion-complexed “egg-box” structures and non-covalent carbohydrate-carbohydrate hydrogenbonding [21, 23, 241. Xyloglucan for instance associates in carbohydrate--carbohydrate manner to cellulose, both within the cell wall as well as extracted in in vitro assays [21, 2.51. Still there are plenty of questions open concerning the threedimensional structure of different cell walls especially considering the needs of the plant during development and growth, when the cells should overcome the rather rigid network and be able to extend themselves [20]. Mammalian Extracellular Matrix Components

Although the number of different molecules present in matrices of higher organisms is clearly larger than in more simple organisms similar concepts of structural organization of carbohydrate networks might still serve as a basic fundament. The extracellular matrix and cell surfaces of higher eukaryotic organisms consist of a rich assembly of highly glycosylated molecules [26-281. Among those, proteoglycans with their long GAG chains provide motifs that could serve not only for binding proteins [ l , 291 but might also show a higher order of structural organisation. Matrix constituents are glycoproteins, proteoglycans and the only type of free carbohydrate chains, hyaluronan (HA). This last one is not only an important component of extracellular matrices, but also essential component of the vitreous body of the eye and of synovial fluid. HA chains are the most simple type of GAG chains with the basic structure of an alternating glucuronic acid and glucosamine unit to HA chains in form long, uniform chains of the structure [GlcA~l-3GlcNAc~l-4].. solution are highly hydrated randomly kinked coils, which are entangled at concentrations of less than 1 mg/ml as seen by sedimentation analysis and viscosity

38.2 From Structurul Components to Cell Recognition

1067

measurements, function as lubricator and as controller of liquid and macromolecular flow (for review see [30]). Although they interact with proteins in the matrix and on the cell surface their gel forming properties are also occurring in protein-free preparations [311, by direct interaction of chains with one another. Fibrillarization of HA chains with one another has been observed by electron microscopy [ 19, 321 as well as other techniques like spectroscopic methods [33] and viscosity measurements [34]. This purely carbohydrate mediated network might well serve as basic cushion into which other molecules are integrated. In the crystal an antiparallel helix has been described [35, 361. Driving force for the complexing could be the exposure of hydrophobic stretches stabilized by intramolecular hydrogen bonding [37] which would lead to hydrophobic association of the chains. The protein-bound forms of GAG chains, namely the heparan sulfates (HS) and chondroitin sulfates (CS) that are parts of the ubiquitous proteoglycans have been tested for direct carbohydrate-carbohydrate interaction in vitro. Isolated GAG chains free from their core proteins were found to form double-helical structures with other GAGS [38]. GAG chains immobilised on beads to mimic their natural, polyvalent appearance on protein cores, showed GAG specific interactions [ 39, 401. HA modified beads aggregate with chondroitin-6-sulfate (CSC) beads and less so with chondroitin-4-sulfate (CSA) but only very modestly with heparin or dermatan sulfate (DS) modified beads. Neither of the beads showed homotypic aggregation. These interactions were independent of Ca2+ ions and required intact, full-length chains, as neither oxidized nor fragmented chains were mediating bead aggregation, thus indicating a strong need for polyvalent interaction [39]. Similarly, rotary shadowing and electron microscopy showed aggregation to meshes of CSC with itself or HS, but not with CSA, and CSA did not form networks with itself [ 181. A potential explanation could be the interaction of hydrophobic sites within CSC to aline with hydrophobic stretches within HA, or in another molecule of CSC, whereas the 4-sulfation of N-acetyl-galactosamine in CSA would disturb these hydrophobic patches explaining the reduced self-complexing of CSA compared to CSC [18]. HS and DS chains, on the other hand, were proposed to bind in a homotypical way to the respective type of GAG either as free chains or multimerized on the core protein [41,42]when tested by affinity chromatography on HS-modified matrices. This affinity interaction was dependent on intact sulfation and carboxylation of the chains, desulfation or reduction of the carboxyl groups did abolish the binding [43], and on the cooperative interactions of several sites within an intact chain [44]. Such GAG-GAG interactions as measured in vitro could be of value at the interface of cells and matrix in tissues with a high content of GAGS (i.e. cartilage) while they are probably of rather negligible importance in others considering the large number of potentially GAG-binding proteins present in the tissue. Of relevance they might be in particular pathogenic situations where GAG-chains become accumulated [45] as for instance in the deposition of plaques in amyloidoses. Glycolipids have been postulated to assume structural purposes via carbohydratecarbohydrate mediated interactions for instance by stabilizing myelin layers around axons. Galactosylceramide (GalCer) and cerebroside sulfate (CBS) are the predominant glycolipids found in myelin in higher vertebrates and it has been suggested that the

1068

38 Curbohy~rute-Carbohydrate Interactions

Table 1. Different classes of glycans suggested in carbohydrate-carbohydrate interactions. Structure”

Type of interactionb

Organism/System’

K-Carrageenan Agarose Alginate Cellulose Cellulose/Hemicellulose Neutral glycans Glycosaminoglycans Acidic glycans Zwitterionic glycans Glycolipids

homo homo homo homo hetero hetero homo/hetero homo/hetero? hetero homo/hetero

Algae Algae Algae, bacteria Plants, bacteria Plants In vitro In vitro Marine sponge, In vitro Bucteroides frugilis Different cell types

Reference

a Classes of glycan structures associated with carbohydrate-carbohydrate binding hIndicates whether identical or different sequences of the same class of glycans are interacting with one another “Orgdnism/model in which these interactions have been described

amount of these ceramides correlates with the stability and compactness of the myelin sheet [46]. In vitro liposome assays have demonstrated a preferred heterogeneous interaction of GalCer-liposomes with CBS-liposomes in presence of Ca2+ over a homogeneous interaction of either glycolipid type with itself [46, 471. 38.2.2 Carbohydrate-Carbohydrate Interactions as Part of Recognition Keys? In all of the examples described so far, often homogeneous types of carbohydrate sequences interact with neighboring chains. Stable structures are created by cooperative action of a multitude of low affinity binding sites that accumulate to an avidity keeping for instance cell walls together. In none of these examples, however, is there flexibility tolerated unless on the expense of stability. A hyaluronan-pad can be compacted or extended by exchange of H 2 0 and ions, but it is stable unless it becomes degraded by enzymatic or chemical means. Cell walls, once created, are rigid and flexibility must be “bought” by the action of reconstructive enzymes, sofisticated hydrolysis processes and biosynthetic adaptation to new requirements [23].A totally different situation arises, if carbohydrate-carbohydrate are considered in recognition and adhesion process of tissues that are dependent on flexibility. Here, advantage could be taken from the fact that carbohydrate-carbohydrate interactions are rather low affinity interactions and be used as primary, reversible contact glues that allow the organism to quickly release or reinforce adhesion between two cells. Mainly two lines of research have been followed testing these possibilities: The properties of an extracellular aggregation molecule that mediates aggregation of marine sponges, on the one hand, and cell surface glycoconjugates that participate in the adhesion of embryonic and tumor cells, on the other hand.

38.2 From Structural Components to Cell Recognition

1069

Carbohydrate Interactions in Invertebrates-The Marine Sponge Microciona prolifera as a Model System Sponges are the most simple multicellular organisms. They do not have any defined tissues or organs instead their pluripotent cells remain motile within the animal and maintain different functions. For this simplicity and for the fact that mechanically isolated cells have the property to reaggregate and reform animals [48, 491 the marine sponges have been model systems for studying recognition and adhesion in multicellular organisms since the beginning of the century. Early it was recognized that cell suspensions from two different sponges do sort out when mixed and reaggregate in a species-specific fashion. As the sponge is feeding on filtration basis, the animal is continuously in contact with foreign particles a fact which probably has helped to let the animal evolve a modus for recognising self u s foreign. Equally well has this sponge model been one of the first examples where carbohydratecarbohydrate interactions have been postulated in context of a recognition phenomenon [50].The extracellular matrix of the sponge is similarly composed as higher eukaryotic tissues, containing both proteoglycan-like complexes as well as collagens and other glycoproteins. A major adhesion mediating factor of the sponge tissue has been isolated in form of a proteoglycan-like complex from Microciona prolifera [51].This large complex, the Microciona aggregation factor (MAF), with a carbohydrate content of more than half the molecular mass of the macromolecule has been shown to mediate species-specific aggregation of dissociated sponge cells [52-54] in a Ca2+-rich environment (-10 mM, as in sea water). In the electron microscope, the macromolecule appears as a sunburst [55]with a ring of -200 nm diameter to which up to 20 arms are attached, a feature confirmed by more physiological assessments of the structure by atomic force microscopy (AFM) in its hydrated form [56]. EDTA-treatment of the complex disrupts the arms from the central ring core, indicating that the cation is important for the structural integrity of the complex [54]. Several hundred Ca2+-binding sites are also important for the Ca2+-dependent self-interaction [ 571. Two functionally distinct binding sites have been identified for the complex, one of which is a Ca2+-independent binding site to cellular receptors and the second, the Ca2+-dependent self-association site which is providing the intercellular adhesion force [57]. Cellular receptors may include membrane-associated glycoproteins of 210 kDa and 68 kDa [58] and have a high affinity to both, the cells as well as the aggregation factor [59, 601. The carbohydrates are essential for the adhesion and recognition process as either glycosidase treatment [61, 621 or periodate oxidation abolish the aggregation of isolated sponge cells [57].A carbohydrate-carbohydrate interaction of the glycans by themselves was suggested by experiments with protein-free glycan preparations from MAF, where a Ca2+-dependent aggregation of beads could be demonstrated with multimerized glycans [I I]. Species-specific binding could be achieved by either the intact aggregation factor or the polymerized glycans to create de novo polyvalence [50, 631. Ca2+ is essential for aggregation of the glycans and cannot bereplaced by other divalent cations [64]. At much lower charge concentrations, polycations as polybrene or polylysine are able to replace Ca2+ indicating a cooperative effect of the polycations with the polyanionic carbohydrate chains. The quantitative

1070

38 Carbohydrate- Carbohydrate Interactions

Figure 4. Model of polyvalent interaction between two Microciona prolifera cells. The high molecular weight proteoglycan-like complex of MAF is composed of -20 arms attached to a central ring each composed both of protein and carbohydrate components. The small, N-linked g6 glycans are predominantly localized in the arms of the molecule [55] and supposed to interact with the cell surface [67]. The large glycans associated with the self-association property of the complex [I 11have not been unambiguously localized but might correspond to the larger glycans identified in the central ring ([55] and J. Jarchow personal communication) and are recognised by anti-MAF-antibodies blocking the self-aggregation [ 11, 68, 691, adapted from [ 121.

differences in interactive strength of various polycations could reflect the variations around an optimal fit of the polycations to the polyanionic glycan chains of MAF as demonstrated by phase-partition assays [65]. Whereas many different polycations are able to precipitate the intact factor glycans, the self-interaction of glycans under physiological conditions, i.e. in the presence of Ca2+ ions, is restricted to glycan chains of the same species [50, 631. These findings indicate that the self-interaction

38.2 From Structuvul Components to Cell Recognition

1071

of the glycan chains is not due to unspecific charge interaction of polyionic chains. Rather, the glycans contain a species-specific arrangement of residues in a proper spacing of interactive sites, whether these are single-charged residues in a defined distance [65] or sequences of several residues 1661. Monoclonal antibodies which inhibit the glycan mediated self-aggregation of factor molecules [ 11, 671 are directed against distinct carbohydrate motives [68, 691 favoring the sequence model. Observation of cross-reactivity of some of the monoclonal antibodies with glycans from different species [ I I ] and the observation that different sponge type cells may first aggregate randomly before sorting out [88] could easily be explained by the different arrangement of otherwise identical or practically identical motives. Carbohydrate Interactions in Vertebrates-Embryonal and Tumor Cells Eukaryotic cells express, besides glycoproteins, high amounts of different glycosphingolipids (GSL) on their plasma membranes and many of them are subjected to developmental regulation. Not only embryonic cells but also tumor cells undergo changes in the expression pattern of GSL and therefore these glycoconjugates became classified as onco-developmental antigens [47, 701. Early mouse embryos at the 8- to 16-cell stage when expression of the stage specific embryonic antigen 1 (SSEA-1) is highest, are susceptible to decompaction by the analog free hapten, namely the blood group antigen LewisX (Le": Gal01 +4[Fucul-+3]GlcNAc~l+ 3Gal/31+4Glcp) [71]. The embryonal carcinoma cell line F9, which shows a selfaggregation capacity, expresses also a large number of LeX epitopes, whereas the more differentiated embryonal cell line PYS-2 does not self-aggregate and does not express LeXepitopes [72]. The correlation of self-aggregation capacity with Le" expression level was corroborated by binding properties of these cells to Le" coated culture dishes indicating that F9 cells depend on a Le" based binding mechanism. F9 cells were therefore used to analyze the mechanism behind the LeX mediated decompaction of either morula stage embryos or aggregated tumor cells. Indeed, also F9 cell aggregation turned out to be sensitive for competition by Le" containing oligosaccharides. In isolates from F9 cell membranes both Le"-containing glycoproteins [72, 731 as well as GSL having this hapten were found to bind in a Ca2+-dependent fashion to Le"-modified matrices [73]. The glycan nature of the aggregation mediator was further supported by the observation that protein-free embryoglycan chains from the F9 cells precipitated in presence of Ca2+, but not EDTA and were sensitive towards removal of fucose residues, suggesting that the LeXepitopes can mediate a Ca2+-dependent carbohydrate-carbohydrate interaction [72]. Not only Le" epitopes can mediate cell interaction but also diverse other glycans. Undifferentiated human embryonal carcinoma 2 102 cells express high amounts of the LeXprecursor nLc4 and SSEA-3 (with the major epitope GalGb4) and Gb4 and these cells bind well to Gg3 (GalNAc~1+4Gal~1+4Glc~l+lCer) (GalNAc~1+3Galu1+4Gal~l+4Glc~l+lCer) coated dishes while they do not adhere to several other tested surfaces. The binding is dependent on the differentiation state of the cells as retinoic acid or bromodeoxyuridine induced differentiation alters both, the expression level of these GSL on the cell surface and the adhesion phenotype of the cells [74]. Furthermore, binding to Gb4 was reported to induce

1072

38 Curbohydrute~CurhohydrateInteractions

activation of the transcription factors AP1 and CREB, while binding to Gg3 did not result in any cell activation [74] indicating that there is also some qualitative difference in binding to different GSL layers. A similar observation was made for B16 melanoma cells either plated on non-coated, G M (NeuAca2+3Galpl+ ~ 4Glcpl+lCer) or Gg3 coated surfaces. Only on the preferential coating, the Gg3 surface, adhesion went along with enhanced tyrosine phosphorylation of FAK 1751. That direct binding between G& on the cell surface to Gg3 on the culture dish is the activating mechanism, is suggested by the ability to activate FAK also with ~ antibodies against G M1751. When cells interact with other cells or with the extracellular matrix, not only glycolipids may be involved in the interaction with the surroundings but equally well glycoproteins. In what context do these different molecules act? What is the role of other adhesion molecules that are well known mediators of cell interaction and what forces could be tolerated by the different combinations? Work with the mouse melanoma cell line B16 addressed these questions. As seen before with the embryonal carcinoma cell lines, also the melanoma cells could adhere in a GSLdependent manner to either other cells or glycolipid coated culture dishes [76]. The adhesion and spreading on coated culture dishes was most obvious at early stages of cell plating indicating that GSL mediated interactions are very early phenomena in cell interactions and will be overtaken by protein-mediated binding 1771. This fact was also confirmed by plating cells on dishes coated both with GSL and extracellular matrix proteins. There, melanoma cells could adhere and spread faster on glycolipid coated plates while the adhesion and spreading on laminin or fibronectin coated plates were slower [78]. Different clones of the melanoma line B16 with different expression levels of the predominant GSL were tested for their binding behavior to non-activated endothelial cells. Binding of melanoma cells was de~ level in the static system and faster, but weaker pendent on their G Mexpression than binding to laminin or fibronectin. Under shear forces, binding strength of the , also stronmutant B16 cells was still correlated to the expression level of G M ~but ger than the binding via the proteins. From these findings the hypothesis was raised that metastatic tumor cells make use of the high expression rates of certain glycolipids to attach to the unstimulated endothelium, before the next steps in cell activation and transmigration are mediated by protein-protein interactions [79]. Repulsive Carbohydrate-Carbohydrate Interactions

Where there are adhesive carbohydrate-carbohydrate interactions, there must also exist anti-adhesive or repulsive ones. Aggregation factors and glycans from different sponge species show only little or no cross-reactivity [ I l l or similarity in glycan composition [80], and beads coated with aggregation factors from different species sort out [SO, 811 as cells have been known to do since long ago. These properties might suggest that there are sequence compositions that are not promoting adhesion as explained by the zipper model [ 121 and therefore participate in the self us non-self recognition phenomenon of sponge cells. Examples of non-adhesive carbohydrate-carbohydrate interactions are also seen between different GSLs. GSL-GSL interactions have been found to promote

38.2 From Structural Components to Cell Recognition

1073

Table 2. Carbohydrate motifs suggested to interact in carbohydrate-carbohydrate interactions. MotiP -

-

Organism/System"

Reference

Algae, bacteria In vitro Early mouse embryo, teratocarincoma cells Melanoma/lymphoma cells Melanoma/endothelial cells Teratocarcinoma cells Teratocarcinoma cells In vitro liposome assay In vitro liposome assay In vitro liposome assay

~ 4 1 [171 [71, 1221 [761 [791

~

[GulA],-(GulA], PSA-PSA Lex-LeX G~3-Gg3 GM3-lacCer Gb4-nLc4 Gb4-GalGb4 GalCer-Gal(S03)Cer H-H H-Ley

PI

(741 [461 ~901 [901

Identified structure or epitope suggested to interact in a carbohydrate-carbohydrate interaction bDescribes in what context these observations were made

a

binding, to be neutral or to be anti-adhesive [82]. An example of an anti-adhesive GSL interaction has been observed for B 16 melanoma cells coated on GM3-plates. While cells coated on Gg3 were nicely spreading, less than 1% of all cells did so on G M ~Sialidase . treatment of the dish could revert the effect, so that cells did spread to control levels, indicating that a G~3-Gh.13interaction is of repulsive nature [77]. Control of GSL expression could therefore easily allow the switch from a nonadhesive to an adhesive phenotype or vice versa by the activation/inactivation of a sialyltransferase. A fascinating oligosaccharide modulating neuronal plasticity by repulsive carbohydrate-carbohydrate interactions is polysialic acid (PSA). PSA is a homogeneous polymer of a2-8 linked sialic acid residues [83], which in neuronal tissue is expressed almost exclusively on the neuronal cell adhesion molecule (N-CAM) attached to a N-linked core glycan [84]. The role of PSA on embryonal forms of N-CAM, but also on adult forms as on axons that undergo plasticity or synaptic remodelling, is to prevent axon fasciculation and thereby guaranteeing easier axon migration towards the target, and higher accessibility of the axon for signalling molecules from the target. The injection of endo-neuraminidase during embryonal development leads for instance to serious pathfinding problems for the motorneurons as they should brake away from their neighbors in the plexus once they have emerged from the neural tube. In adult tissue, where the amount of PSA on N-CAM is drastically reduced this repulsive effect is lost. The mechanism behind this anti-adhesive carbohydrate-carbohydrate interaction could be pure ionic repulsion or physical hindrance due to the large hydration volume of the polymer. In either way, cell contacts can be affected in trans or cis fashion, hindering either receptor-receptor interaction between two cells or affecting receptor mobility and interaction with other molecules within the cell membrane which ultimately also leads to altered cell-cell contacts [84]. An unexpected finding with isolated PSA from either bacterial or mammalian source, however, demonstrates an adhesive capacity of the

1074

38 Carhohydrate-Carbohydrate Inteructions

Table 3. Carbohydrate motifs suggested to function in anti-adhesive mode. Motif

Organism/System

Reference ~-

PSA>
LeY>
Embryonal brain Melanoma cells In vitro liposome assay Marine sponges

[61 [771 [901

P O , 811 ~~

"A carbohydrate-carbohydrate interaction is only indirectly suggested as the sponge glycans are postulated to bind in a homophilic way [ 11, 671. By bead-aggregation and AFM, however, sorting out of MAF and HAF modified beads could be demonstrated, whereas a direct proof of MAFglycan-HAF-glycan interaction has not yet been established.

chains as observed by atomic force microscopy. Similar to HA, PSA forms bundles in the presence of Ca2+ beyond a chain length of 12 units and longer fragments have a tendency to form branched bundles [17]. These findings refresh the discussion about the mechanism and functioning of PSA in the developing brain and in bacterial cell walls [ 171.

38.3 Molecular Aspects of Carbohydrate Interactions 38.3.1 Polyvalence to Inforce Weak Interactions Molecular interactions where carbohydrates are involved are usually weak interactions. Proteins usually recognize and bind to just a few carbohydrate residues within an oligosaccharide, whether it is an antibody or a lectin. Nature adopts to the needs of higher affinity requirements by the multimerization of either protein or carbohydrate or both units in order to reach avidities that can hold molecules and cells under physiological conditions [2, 661. Even more true is the need for polyvalence in the case of carbohydratecarbohydrate interactions. Association of glycoconjugates as small as a mono- or disaccharide with another sugar molecule can be observed by methods as mass spectrometry [85, 861 or crystallography [87]. The affinities are, however, so low that the stability of the complexes would not even survive the measuring time in NMR [72] not to think of a situation in vivo. Polyvalence, i.e. the repetition of a binding motif in either of different modes, allows these molecules to bind together and gives the scientist the chance to measure them by different methods. Sponge cell aggregation mediated by isolated glycans needs the de nouo polymerization of the isolated chains into longer, multivalent complexes to achieve approximately the size of the native aggregation factor [50, 631. Polyvalence could also be achieved by attachment of glycans to beads which could aggregate in a species-specific manner [ 11, 501 confirming earlier proposals and data [65].

38.3 Molecular Aspects of Carbohydrate Interactions

1075

The same needs hold true in the GSL-mediated interactions of tumour cell lines. The adhesion capacity of cells as well as liposome aggregation and binding are dependent on the concentration of glycolipids on the receptor side. Correlation of binding strength was both seen for the amount of a specific GSL expressed on a certain cell line as well as to the concentration of the glycolipid on the substrate side 174, 771. Also the length and degree of hydroxylation of the fatty acid portion has an influence on the interaction of glycolipids with one another, suggesting that developmental control or pathological changes can affect carbohydrate-carbohydrate mediated cell interactions [46]. In the carbohydrate-carbohydrate model of cell interaction advantage is taken from single low affinity sites. Control of the polyvalence by various means (surface density of presented structures, ionic strength to modulate attractive vs repulsive forces, subtle changes in biosynthesis of the carbohydrate sequences, etc.) to change the affinity of the interactive molecules provides an adaptable system beyond a structural scaffold. Such recognition systems are created which allow different cells to test surrounding surfaces and release or reinforce interactions [SS]. Speciesspecific sorting is one example of the value for moderate interaction strengths which achieve biological relevance through polyvalence. Moderate strength is also required within one species, e.g. within an individual sponge, otherwise the cells could not migrate past each other. In an experiment where factor-mediated aggregation was induced by polybrene instead of Ca2+, secondary migration of cells was inhibited. This observation demonstrates the drastic effect of “locking” the molecules in a permanent tight interaction mode 1651. 38.3.2 Arrangement of Motifs and the Possibility to Control Specificity

A Velcro pad or a zipper can be used as simple models to highlight the value and simplicity by which nature may create specific and lasting binding sites by such carbohydrate-carbohydrate interactions from compositionally rather similar structures [66]. The creation of repetitive, interacting glycan sequences is feasible in different modes. Oligosaccharide motives can be repeated along the primary glycan sequence as seen in the examples of plant cell wall carbohydrates. Glycans can also be arranged in a repetitive pattern along a backbone structure which is not directly participating in binding, as illustrated by mucin structures on protein backbones [2] or by branching carbohydrates on a glycan scaffold as for instance blood group antigens on poly-N-acetyllactosamine type glycan backbones [70, 891. Finally, proteoglycans, glycoproteins or glycolipids can be presented in clusters or surface superstructures partly due to mobile anchors. Only in this last version advantage is taken of the fluidity of the biomembrane for control of the surface density of these glycans and therefore their avidity [2, 66, 901. There are indications that GSL form clusters within membranes and are not randomly distributed 1751. Indeed, GSL accumulate on apical surfaces of epithelial cells [91, 921, are observed as patches in erythrocyte membranes [93] and peripheral lymphocytes 1941 or as morphological clusters in vitro in liposomes as detected by electron microscopy of freeze fractured samples 1951.

1076

38 Carbohydrate-Carbohydrate Interactions

Primary glycan sequences determine whether there is the possibility of interaction, but how these sequences will be arranged in a three-dimensional context will define the specificity of interaction. Even though sequences could be similar, different spacing in the three-dimensional architecture could create a “non-binding” phenotype as illustrated with two half-chains of a zipper being of different dimensions. These architectural differences could even be gradual, allowing for instance two different sponges to first slightly complex before sorting out again [88]. Or a tumor cell with a high density of a certain GSL be an effective metastatic colonizer, whereas the one with a lower concentration of the same GSL be less efficient due to reduced primary binding capacity [96]. The next step in binding via carbohydrates is the question how for instance a certain combination of GSL would trigger the cells to activate signalling molecules while another combination, although still mediating binding would not activate the cell [75].

38.3.3 Molecular Basis of Carbohydrate-Carbohydrate Interactions The forces playing between carbohydrates are no different from those acting between other biomolecules. Most carbohydrates are neutral or negatively charged due to carboxy- and sulfate groups, although positively charged glycans also occur [16, 971 and ionic interactions can be anticipated. Except for rare ionic interactions between oppositely charged glycans [ 161, direct carbohydrate-carbohydrate interaction between two anionic carbohydrate rather leads to repulsive forces as nicely illustrated for PSA [84].With the aid of divalent cations, this effect can be overcome as demonstrated for sponge cell aggregation, where both carboxyl and sulfate groups as well as Ca2+-ionsare important for proper species-specific aggregation of cells [57, 981. However, not just any kind of acidic sponge glycan is interacting in the presence of Ca2+ indicating that the effect of Ca2+ is not merely a charge effect. Another possibility of the action of Ca2+ is the property to complex carbohydrates via a suitably arranged combination of sugar hydroxyl groups. Single hydroxyl groups in sugars are too weak to coordinate cations in the presence of water molecules. However, in the combination of two to three well positioned hydroxyls on one sugar residue or over two adjacent residues, cations can become coordinated to the carbohydrate which can force the chain into a specific conformation or lock it there as seen for instance with guluronic acid sequences in alginate [ 141 or pectins [23].The preference of Ca2$ over other cations lies in the molecular dimensions and complexing properties: Complexing strength raises from mono- to trivalent cations, while the ionic radii determine how well the ions fit into molecular dimensions of the molecules which are optimal around 100-1 10 pm as seen for Na+, Ca2+,La3+ [99].Some, but not all, of the GSL interactions have been shown to depend on Ca2+ and it is not clear what is the difference between those. For the interactions of G M ~ with Gg3 and Le”-Le” an association of the molecules via their hydrophobic sides that complement each other followed by a locking through Ca2+ has been proposed based on molecular modelling [ 72, 761. However, a similar molecular modelling approach has led to the suggestion that Ca2+ is the interlinking force between the

38.3 Molecular Aspects of Carbohydrate Interactions

1017

two Le" molecules [85].No Ca2+ was present in the Le" crystal [87], why further studies are required to resolve the exact role of Ca2+ in this interaction. Carbohydrates offer a rich source for hydrogen-bonds due to hydroxyl-, amineand carboxy-groups. Hydrogen bonding can be seen intramolecular as well as intermolecular or in combination with the solvent. An extremely high number of hydrogen bonds could be seen in the crystal of Le" [87] with bonds between the trisaccharide and water and between the carbohydrates themselves. An indirect consequence of intramolecular hydrogen bonding has been suggested for HA. As with the locking properties of Ca2+ a similar fixing the conformation has been proposed for neighbouring residues in HA, which would lead to exposure of a larger hydrophobic patch in the chain, which in turn would favor hydrophobic interactions between different chains explaining chain interactions [ 371. An opposite interpretation of similar data from molecular dynamics models for short saccharide sequences would suggest the rapid exchange of different intramolecular H-bonds in favor of a prolonged solubility of even high concentrations of HA [ 1001. Improved measuring methods must be awaited for deciding which ones are the driving forces in vivo. Also hydrophobic interactions and Van der Waals forces are likely to occur as not only HA-chains present hydrophobic patches [ 18, 1011. Hydrophobicity of the COO'

Figure 5. Schematic drawing of forces that might act between two carbohydrate chains in sponge glycans. Possible forces are sketched as ionic interaction via Ca2+ ions between two carboxyl groups (Ca2+ between arrows), hydrogen-bonds (arrows with dashed lines) might occur within the chain, between chains and between a chain and the solvent, and hydrophobic interactions (broad interaction band) between apolar surfaces of the sugar rings (adapted from [ 121).

1078

38 Curbohydrate- Carbohydrate Inteructions

carbohydrates depends on the conformation of the rings, the epimer conformation and the glycosidic linkages and therefore not all carbohydrate chains are equally amphiphilic [ 1021.

38.4 Experimental Approaches 38.4.1 General Considerations Low affinity interactions are difficult to measure and the intrinsic problem is to distinguish specific from non-specific interaction. In this context, it is important to realise that not all biological interactions are equally “specific”, in other words, to understand specificity not merely as a term of high affinity, unique binding site, but rather as an adaptation to the physiological possibilities and needs. The coexistence in space and time of potential partners in an interaction scheme is therefore a more important requirement than the existence of unique structural motifs recognized by individual molecules. A measure for the strength of two molecules interacting with one another is the affinity between them. This can be expressed in terms of an association constant which can differ for many orders of magnitude depending on the kind of interaction. Association strengths between two sugar molecules can be below lo2 M-l, whereas low affinity interactions have a range of approximately 102-104 M-’ and high affinity interactions more than lo5 M-’, already indicating that the possibilities of detection and choices to measure such interactions can be limited by the inherent technical possibilities of different approaches. As measurements of interactions often happen in dissected systems where molecules are tested in other than the original context, it should be crucial to test a model in conceptually different assay set-ups to validate findings and try to eliminate experimental artefacts. These requirements can be difficult to achieve for the carbohydrate-carbohydrate interaction; even more difficult than to find the natural components of protein-carbohydrate interaction which is already a difficult task as demonstrated by the hunt for the “real” selectin ligands [103]. Much of older literature on carbohydrate-carbohydrate interaction does not necessarily hold the requirements for unambiguous testing in different systems, but due to lack of possible alternatives at the time they were performed they have been included in this summary. Improvement of methods and technology should give the chance to test these concepts in future. The collection of methods presented below is chosen on the basis of experiments described in the literature used for studies of carbohydrate-carbohydrate interactions. Rather than being a collection of ready to use recipes the different methods are discussed considering important points as most of the techniques are discussed in full extent in other parts of these volumes and do not differ conceptually whether measuring carbohydrate-carbohydrate interactions or any other type of interaction.

38.4 Experimental Approaches

1079

38.4.2 Affinity Interactions

In principle all studies to measure interaction between two molecules are affinity interactions. What is described under this heading could also be described as traditional cell biological or biochemical approaches to measure binding between two systems. It was historically started with the observation of whole cells interacting, before more sophisticated methods of cell fractionation and artificial reconstruction have progressed to the point that almost any molecule can be isolated and packed into a de nozio assembly to test for its role in the intact system. Cell Binding Studies The advantage to use intact cells for binding studies is that the molecular assembly is in its original form and effects are less likely due to altered exposure of molecules, artificial combinations, lack or changed arrangement of components. Obviously that creates also a disadvantage in itself, as designing which of the components does what is less ambiguous in an intact system. Cell binding studies can be performed in several different approaches. Single-cell suspensions are allowed to aggregate with themselves or another cell type, to bind to a supporting cell layer or to an artificial substrate. Cell aggregation

Cell aggregation is a classical type of experimental set-up and has been used extensively, hemagglutination just being the most prominent one. Mechanically dissociated and well washed cells are allowed to aggregate in the presence of a potential adhesion mediator or inhibitor. Cells can be left untouched or exposed to shear forces for instance on a gyratory shaker. To exemplify this approach, sponge cells can be mechanically dissociated from the sponge matrix, washed extensively and allowed to reaggregate in the presence of Ca2+ and an aggregation factor [51]. This approach can be further extended to test the role of different components of the adhesion molecule. Chemical treatment of the factor or enzymatic removal of certain sugars [57, 61, 621, fragmentation into smaller subunits [63] or chemical cross-linking of factor glycans to a multimeric complex [ I l l aid to substantiate the mode of interaction between two cells. The reversed scenario can also be applied: A cell complex is broken up by adding competitors from which one assumes that they have the capacity to inhibit the interaction. This can be antibody molecules against different surface molecules or glycan motifs to be tested for their effect as done with early embryos [71]. Cell adhesion to substrate

The binding capacity of cells to different substrates is a first step in dissecting the system and testing individual components for binding. While the cell is still an intact entity and serves as a tool to screen with, the complementary adhesion site is modified and theoretically can be built from almost any kind of combination of

1080

38 Carbohydrate-Carbohydrate Interactions

Table 4. Cell lines applied to GSL-mediated binding studies.

Cell line"

Expressionb

Binding"

Substrated

B 16 melanoma

( 3 ~ 3high

Strong

B16 melanoma

( 3 ~ 3high

Strong

B 16 melanoma

G M high ~

Weak

B16-FI0 clone

G Mhigh ~

Strong

B16-F1 clone

G M medium ~

Medium

B16-WA4 clone

G M lOW ~

Weak

F9 teratocarcinoma PYS-2 teratocarcinoma 2102 embryonal carcinoma

LeXhigh Le" low

Yes No

LacCer-coated plate, non-stimulated endothelial cells (LacCer expression) Gg3 coated plate, Tcell lymphoma L5178 AA12 (high Gg3 expression) T-cell lymphoma L5178 AV27 (low Gg3 expression) Non-stimulated endothelial cells (LacCer expression) Non-stimulated endothelial cells (LacCer expression) Non-stimulated endothelial cells (LacCer expression) Le" coated plate Le" coated plate

High nLc4 and GalGb4, moderate Gb4, no Gg3 High nLc4 and GalGb4, moderate Gb4, no Gg3

Yes

Gg3 and Gb4-coated plate

No

Gb3, Lc3, Lea, Le", G M ~GM3-coated , plate

2 102 embryonal carcinoma

Reference

Cell lines and clones used for studying GSL-mediated binding Expression levels are described with low, moderate or high according to the original description 'The binding strength is graded as described in the original publication a

molecules. The advantage lies in the simplicity of the test and the better control of components, whether concerning amounts or combinations. The caveat arises from the same point as mentioned earlier, namely the artificial combination of components and the uncertainty as to the steric properties of the molecules when immobilized on a surface. Nevertheless this type of approach has been used extensively in the study of GSL mediated cell binding. Individual GSL alone or in combination with different glycoproteins have been immobilized to culture dishes and testing occurred with different cell lines under static or flow conditions. Conclusions drawn from this kind of assays are that different GSL have different adhesion promoting/ preventing potentials for a given cell line, an effect which is clearly dependent on

38.4 Experimental Approaches

1081

both the concentration of GSL expressed on the cell surface and that coated on the culture dish [74, 75, 771. Combination of GSL with proteins and testing cell adhesion under both static and flow conditions indicate that GSL-GSL interactions are primarily mediating cell adhesion in the early phases of cell coating and are replaced by protein-carbohydrate and protein-protein interactions after that, but are more prominent under shear forces than under static conditions [78, 791. Aggregation of de novo Complexes

Once a cell is dissected into single components and these should be tested individually, different options can be tested. Aggregation or complexing are easily observed and do not need too much of technical equipment. Macromolecules that allow such observations have been favored due to simplicity and easiness to create such polyvalent assemblies. Modification of inert beads and integration of molecules into liposomes have been two different approaches used extensively in the study of carbohydrate-carbohydrate interaction. Aggregation of beads can be followed macroscopically or with a normal light microscope, or, when using fluorescent beads, by fluorescent microscopy. Liposome aggregation can be followed simply by measuring absorption with a spectrometer. ModiJication of inert beads The choice for beads is merely empirical. Two important points must be considered before an experiment: First, how shall the molecules be coupled to the beads in order to exert a minimal effect of the immobilization method on the assay. Is noncovalent adsorption possible and stable enough or does it need a covalent-linking scheme. The chemistry of linking molecules to beads depends on molecular characteristics (which chemical groups are available for linking, which ones should not be used to conserve the biological activity) and can affect the properties of the linked molecule in terms of its binding properties, i.e. its functional integrity and its way of presentation (steric factor, architecture on the bead). Integrity and access of the molecule on the bead should therefore be tested with a suitable probe as could be an antibody, a lectin or another molecule for which the binding behavior is known. Second, the chemical groups on the beads used for covalent coupling of the sample must not affect the binding properties of the beads. It must be possible to neutralize them without changing the binding characteristics of the bead. This problem is usually rather easily circumvented as control beads can be modified and blocked in parallel without any ligand or an alternative ligand and be used as negative (or positive) control. Beads have been used to demonstrate that either intact aggregation factor or immobilized factor glycans can mediate the sponge cell aggregation [57, 671, that GAGScan bind to one another [39,44]and to demonstrate GSL-GSL interaction [72]. Integration of glycans into liposomes Glycolipids are highly suited to be integrated into liposomes for testing glycandependent liposome aggregation and liposome binding to surfaces. No chemical

1082

38 Carbohydrate- Carbohydrate Interactions

modification is needed, as the quality of liposomes be controlled by the choice of lipids used to create the liposomes. The integration of radioactively marked glycolipids can help to control the level of incorporation. Different glycan motifs can be integrated and tested in parallel. The possibility to create neoglycolipids from short glycans also opens the possibility to test glycans that originally are not isolated as glycolipids but may originate from a glycoprotein and could be tested independent of the protein core [ 1041. Hypotheses of GSL-mediated cell interactions have been tested by demonstration of the aggregability of GSL-liposomes and adhesion of such liposomes to glycolipid coated surfaces [46, 82, 901. Affinity Chromatography Alternative to using immobilized glycans in aggregation assays, cell or liposome binding and similar experiments, they can also be used in chromatography. Whereas high affinity interactions are easily seen by affinity chromatography, low affinity interactions might be overlooked due to insufficient retention. For larger glycans that are inherently polyvalent, retention on a glycan modified column could be manifested on the basis of carbohydrate-carbohydrate [42].As holds true for the bead coating, the inclusion of an identically modified column matrix without the ligate of choice attached should be considered as a control or precolumn to avoid looking at non-specific matrix-ligand interactions. At least the theoretical possibilities of weak affinity chromatography are impressive if one succeeds to immobilize the ligate at high enough densities to cope with the low affinity of carbohydratecarbohydrate interactions [ 1051. Distribution between Compartments Boyden chamber

The distribution of molecules between two phases or two liquid chambers separated by a membrane has been occasionally used for the demonstration of carbohydratecarbohydrate interaction. In the Boyden chamber, two compartments with identical buffer content are separated by a membrane with suitable molecular weight cut-off. To one chamber the molecules to be tested, for instance a soluble oligosaccharide and a potential macromolecular ligand are added, and the diffusion of the oligosaccharide into the other chamber is followed. If it does not interact with the macromolecule, the distribution of the oligosaccharide should be under free diffusion and an equilibrium must be reached. Similarly, two freely diffusible species can be observed, if their interaction creates complexes that become too big for free diffusion (by correct choice of the membrane pore size), and will remain in the chamber they have originally been placed in. Binding of radiolabelled oligosaccharides to stationary liposomes containing different glycolipid species was observed in the Boyden chamber. Oligosaccharides that bound to the GSL in the liposomes did not pass the membrane, whereas oligosaccharides that could not bind to the GSL were recovered also from the other compartment [73].

38.4 Experimental Approaches

1083

Phase partitioning In a similar type of approach, phase partitioning, the distribution of molecules between two different liquid phases, is analyzed without any separating membrane. In this assay, the solubility and interaction with the solvent in the two different phases is driving force for the separation behavior of the analytes. The distribution of 1251-MAFbetween a lower, dextran-rich and an upper, polyethylene glycol phase had been analyzed as a function of cations added. While the non-complexed MAF distributed to the dextran-rich phase due to its polyanionic character, aggregation of MAF in the presence of cations reduced the net charge of the aggregate and resulted in a shift to the polyethylene glycol phase [65]. The limitation of this assay system is, however, that the effects of pure charge neutralization by the addition of cations to the polyanion cannot be distinguished from the effect of cation-mediated complexing of the polyanion to a less charged macrocomplex. The change of distribution properties of liposomes modified with neutral glycans like dextran, pullulan or mannan between a carbohydrate-rich and a carbohydrate-poor phase, might rather been taken as a sign of direct carbohydrate-carbohydrate interaction between the carbohydrate on the liposome and the bulk carbohydrate [ 1061.

38.4.3 Microscopy Electron Microscopy Tertiary structures beyond the cellular level can be observed by electron microscopy (EM). Both, macromolecular complexes of proteoglycan-like aggregation factors [ 551 and isolated glycosaminoglycan chains forming fibrillar structures have been observed by EM [19, 321. Many of the observations have been made on replicas of rotary shadowed structures that had been dried on the carrier. These dehydrated samples have most obviously lost some of their physiological properties beside the fact that the preparation as such could be deleterious to the sample and result in artefacts. The development of atomic force microscopy (AFM) in the 1980s is therefore a valuable improvement of technology towards more physiological sample application possibilities. Atomic Force Microscopy AFM is an attractive development which has resulted from scanning tunneling microscope technology. In AFM a sensor is raster scanned in nanometer distance over a surface with the immobilized sample giving a picture of the specimen. In this imaging mode, measurements can be achieved for single molecules as well as whole cells allowing the observation of biological phenomena in a more physiological context. In contrast to EM, samples do not need to be fixed or shadowed and they can be observed in hydrated form. What was anticipated a few years ago to be a general method for imaging from molecules to cells [ 1071 has already been successfully used for imaging cells [108], proteins [109], ribonucleic acid [lo81 and carbo-

1084

38 Carhohydrute-Carbohydrate Interactions

hydrates [56, 1101 but even more to measure forces acting between molecules [56, 1101. In the force-distance approach, the tip with an immobilized molecule is moved in an approach-and-retract cycle towards the sample on the substrate surface. The forces acting on the sensor are recorded under all the approach-and-retract cycle and give a picture of singular binding events, the distance at which the binding ruptures, the adhesion probability and, at least theoretically, also the energy dissipated during the process [ 11 11. In order to really be able to measure forces acting between the two samples and not those between the samples and their respective immobilization surface, it is important that the molecules are immobilized so well as to resist forces larger than the ones acting between the molecules. As discussed with other immobilization techniques, it is also important with AFM, that the immobilization procedure has to conserve the right orientation and binding activity of the molecules to be analyzed. And equally well is it important to discriminate against any unspecific interaction with the surface and sensor [ 1111. Applying AFM technology, imaging demonstrated the same contours of the aggregation complex from sponges as had earlier been observed by EM [56]. In the force mode, an average molecular contour length of -220 nm and arm lengths of -150 nm for MAF could be determined [ 11 11. Furthermore, the average forces of individual interaction sites between two MAF molecules were estimated to -40-50 pN [56, 1111. As multiple jump-off events of -40 pN f 15 pN were seen in a singular approach-and-retract cycle, but the average forces measured between the molecules amounted to -125 pN with maxima up to -400 pN, this result seems to favor multivalent binding of 3-10 binding pairs [56]. As it is, however, still unresolved how the single glycan species are integrated in the whole aggregation factor complex a conclusive picture of this interaction awaits further experimentation. An unexpected finding by AFM is the observation of PSA filaments in the presence of Ca2+-ions [ 171. 38.4.4 Crystallography Crystallography is one method of choice to receive detailed conformational information about molecules. The flexibility and heterogeneity of oligosaccharides, however, sets serious limitations to the analysis of either glycoconjugates or carbohydrate ligands bound to other macromolecules [ 1 121. Crystallization of carbohydrates might be more difficult and even impossible for certain structures, but still has a considerable value in cases where the crystallization succeeds. Not surprisingly HA, which is very homogeneous in sequence, has been described as a crystal and assumes a double helical form [35, 361. More recently, crystallization of a blood group antigen succeeded for the first time ever [87]. This Le" crystal is characterized by a high degree of hydration with water molecules filling all spaces in the crystal and participating in the intermolecular hydrogen bonding pattern. Hydrophobic interaction within a single trisaccharide can be seen between fucose and galactose rings, but no intramolecular hydrogen bonding. The conformation of the trisaccharides in the crystal fit to the conformations predicted from NMR and molecular modelling data [ 1131, and the order of the trisaccharide in a head-to-head arrange-

38.4 Experimental Approaches

1085

ment in the crystal would suggest that a tight packing of glycolipids could occur in a similar fashion [87].

38.4.5 Mass Spectrometry Mass spectrometry (MS) has been used since its beginnings for assisting the characterization of carbohydrate structures. More recently it has gained much of value due to improved ionization tools that provide milder treatment of the sample without fragmentation risk. Whereas the advent of matrix assisted laser desorption ionization (MALDI) MS has improved the characterization of even larger compounds, electrospray ionization (ESI) MS has opened the field to studies of interactions between molecules. In ESI-MS highly charged liquid droplets in a strong electric field are dispersed and evaporated into the mass analyzer [ 1141. Due to this soft ionization technique even non-covalent interactions can therefore be observed. By ESI-MS homodimers could be observed for both LeX and Le"-LacCer in presence of divalent, but not monovalent cations. A heterospecific interaction in presence of divalent cations could also be observed for Lex-LacCer in interaction with LacCer, GalCer and even Cer with a decreasing affinity in this order [85] confirming findings realized by other methods. Another example for the value of ESI-MS is provided by the study of GalCer with CBS, that shows a Ca2+ dependent oligomerization and confirms results from liposome assays. A qualitative picture of the affinity between different oligomers can be gained by measuring the stability at different declustering potentials. Under increasing declustering potentials GalCerCa2+ was the most stabile complex which contrasts to findings with the liposome assay. The divergence might, however, be explained by the choice of the solvent used for the ESI-MS measurements [86]. Collision induced decomposition of the complexes, on the other hand, showed that the non-covalent interactions between the carbohydrate portions and Ca2+ are more stable than covalent linkages, as the ceramide portion could be split off before the complex broke. Combining the results from changing declustering potentials and performing collision induced decomposition suggests therefore, that the GalCer-CBS-Ca2+ complex is the most stabile form [ 861 which corresponds to the liposome assays. Similar observations were also made with the Lewis antigens where ceramide and fucose were split off before the Ca2+-stabilized carbohydrate-carbohydrate interaction was disintegrated [85].

38.4.6 Nuclear Magnetic Resonance The advantage of nuclear magnetic resonance (NMR) lies in the fact that measurements occur in aqueous solutions, i.e. the most natural form if one considers biological materials. NMR is widely used for structural characterizations of carbohydrates even though the requirements for homogeneous probes and relative insensitivity can pose limitations. The flexibility of carbohydrates sets another factor of limitation and therefore structural and dynamic properties must be determined in a combination of different methods [ 1121. Conformations of bound carbohydrates

1086

38 Carbohydrate-Curbohydrute Intermtiom

us non-bound carbohydrates can be measured by transfer nuclear Overhauser effect

(trNOE) experiments. These trNOEs can only be determined for a carbohydrate complexed to a large molecular weight molecule, for instance a protein, when the relaxation is governed by the tumbling time of the protein. The only information gained in this way is the difference between non-bound and bound carbohydrate. To elucidate the structure of the entire complex needs also the parallel analysis of the protein structure by either crystallography or NMR and the combination of the data acquired in the different experiments with molecular modelling techniques [ 112, 1131. As carbohydrate-carbohydrate interactions are usually weak and require polymerization of both binding partners resulting in larger complexes, trNOEs are not feasible for measuring carbohydrate-carbohydrate interactions. The only possibility for analyzing characteristics of two carbohydrate molecules against one another is by modelling interactions with data obtained for the individual components from either NMR or crystallographic approaches. 38.4.7 Molecular Modelling

Due to the nature of carbohydrates, appropriate force-fields must be used to describe the conformation and dynamic properties of these structures. A number of different programs have been adapted or created especially for modelling carbohydrates (for review see [ 113, 1151). Due to the limited number of data for larger carbohydrates, models only exist for a few carbohydrate-carbohydrate interactions mediated by small oligosaccharides. The interactions of the GSL headgroups Gg3 ~ been modelled by creating minimum energy conformation models of and G Mhave the two head groups based on hard sphere exoanomeric calculations, suggesting the interaction via an exposed hydrophobic patch in the two molecules [76]. Similarly, Le"-LeX interaction was modelled and proposed to assume a head-to-head alignment of the molecules [72] as also suggested by crystal data [87]. The increasing potential of calculation power and development of enhanced measuring capacity with NMR equipment will probably add information to the field of carbohydratecarbohydrate interactions.

38.4.8 Tools Important tools for the study of carbohydrate interactions have been antibodies directed against carbohydrate epitopes. Even more important is the synthesis of carbohydrate sequences to create homogeneous, well defined sequences which allow the study of interactions without problems created by heterogeneous populations of glycans in functional and structural tests. Synthetic Oligosaccharides

The synthesis of well defined and homogeneous structures does allow the study of interaction phenomena in different ways. For obvious reasons oligosaccharide syn-

1087

38.4 Experimental Approaches

thesis involves more effort than peptide- or oligonucleotide-synthesis, but nevertheless the number of structures synthesized for different purposes is increasing. Sponge epitopes [ 1161, GSL-epitopes [ 1171 and, the most prominent examples, Lewis blood group antigens have been synthesized and applied for direct measurement of interaction by NMR, crystallography 1871, ESI-MS [85] and other methods [118]. Antibodies against Carbohydrate Motifs Antibodies against carbohydrate determinants have been widely used to study the functional aspects of carbohydrate interactions. They are applied to control the expression level of a given carbohydrate epitope as GSLs 1771, to block binding sites [ 111 or to affinity purify and detect oligosaccharide structures containing the epitope [68]. One important aspect with antibodies is the fact that their reactivity can depend both on the conformation of the epitope due to different surroundings and on the degree of polyvalence due to clustering [ 1 191. The recognition of the same antenna structure on glycopeptides and glycolipids has been shown to differ for various antibodies as a consequence of different conformation of these epitopes depending on the respective anchoring sequences [ 1 191. Two monoclonal antibodies inhibiting Microciona aggregation are directed against the carbohydrate portion of the aggregation factor (Block 1 and Block 2), while a third one does not block the aggregation although recognizing the carbohydrate portion [ 111. All three have been characterized by a combination of chemical degradation and characterization of the recovered structures by enzyme susceptibility, NMR and MS [68, 691. Several antibodies directed against glycolipid structures have been used in the characterization of GSL-GSL interaction to Table 5. Monoclonal antibodies applied for studying carbohydrate-carbohydrate mediated interaction.

Antibody

Epitope recognized

Assay systema

Reference

Block 1

Microciona aggregation

[681

Block 2 DH2

Pyr-4,6Galp 14GlcNAcp1-3F~c GlcNAc(30S03)p1-3Fuc Anti-GM3

~ 9 1 ~761

2D4

Anti-Gg3

T5A7 9G7 1B2 MC63 1

Anti-LacCer Anti-Gb4 Anti-nLc4 Anti-SSEA-3 (GalGb4 main epitope) Anti-SLex Anti-Lex Anti-H

Microciona aggregation Melanoma adhesion to T-cell lymphoma Melanoma adhesion to T-cell lymphoma B16 melanoma adhesion 2 102 lymphoma cell adhesion 2 102 lymphoma cell adhesion 2 102 lymphoma cell adhesion B16 melanoma adhesion GSL-GSL in vitro B16 melanoma adhesion

[791

SNH3 SH I BE2

“Assay system indicates in which context the antibodies were used for binding studies

[761 1791 [ 741

P I P I

PI

[791

1088

38 Curbohydrute-Curbohydrute Interactions

quantitate the GSL-expression on cell lines and to interfere with the binding of cells and liposomes to one another [76, 77, 79, 82, 117, 1201. Cells

What started in the beginning of the century with the observation of cells taken from different sponges to allow the study of recognition processes [48, 491 has developed into a major research tool in biology with the use of both, primary and immortalized cells from many different sources. The selection of different clones from the same parent line has further stimulated the application and direct comparison of cells with similar but not identical surface composition. Examples of such cell lines that have been used for the study of carbohydrate-carbohydrate mediated phenomena are variants of the mouse melanoma cell line B16 expressing different levels of GM3-ceramides [79]or T-cell lymphoma clones with different levels of Gg3ceramides on the cell surface [76] (see also Table 4).

References D. Spillmann and U. Lindahl, Curr. Opin. Struct. B i d , 1994, 4, 677-682 A. Varki, Proc. Natl Acad. Sci. USA, 1994, 91, 7390-7397 T. A. Springer, Cell, 1994, 76, 301-314 H. Rauvala and H. B. Peng, Proy. Neurobiol., 1997, 52, 127-144 M. Schachner, Curr. Opin. Cell Biol., 1997, 9, 627-634 U. Rutishauser, J. Cell. Biochem., 1998, 70, 304-312 K.-A. Karlsson, Curr. Opin. Struct. Biol., 1995, 5, 622-635 K. S. Rostand and J. D. Esko, Infect. Immun., 1997, 65, 1-8 N. Sharon and H. Lis, Science, 1989,246, 227-234 10. T. Feizi, Trends Biochem. Sci., 1994, 19, 233-234 11. G. N. Misevic, J. Finne and M. M. Burger, J. Bid. Chem., 1987,262, 5870-5877 12. D. Spillmann and M. M. Burger, J. Cell. Biochem., 1996,61, 562-568 13. D. A. Rees, Biochem. J., 1972, 126, 257-273 14. P. Gacesa, Carbohydr. Pol., 1988, 8, 161-182 15. F. Clementi, Crit. Rev. Biotech., 1997, 17, 327-361 16. A. 0. Tzianabos, A. Pantosti, H. Baumann, J.-R. Brisson, H. J. Jennings and D. L. Kasper, J. Biol. Chem., 1992,267, 18230-18235 17. J. Toikka, J. Aalto, J. Hayrinen, L. J. Pelliniemi and J. Finne, J. Biol. Chem., 1998, 273, 28557-28559 18. J. E. Scott, Y. Chen and A. Brass, Eur. J. Biochem., 1992,209, 675-680 19. J. E. Scott, C. Cummings, A. Brass and Y. Chen, Biochem. J . , 1991,274, 699-705 20. P. Albersheim, J. An, G. Frshour, M. S. Fuller, R. Guillen, K.-S. Ham, M. G. Hahn, J . Huang, M. O’Neill, A. Withcombe, M. V. Williams, W. S. York and A. Darvill, Biochem. Soc. Trans., 1994,22, 374-378 21. M. McNeil, A. G. Darvill, S. C. Fry and P. Albersheim, Annu. Rev. Biochem., 1984,53, 625663 22. L. M. J. Kroon-Batenburg and J. Kroon, Glycoconj. J . , 1997, 14, 677-690 23. N. C. Carpita and D. M. Gibeaut, Plant Cell, 1993,3, 1-30 24. N. Carpita, M. McCann and L. R. Griffing, Plant Cell, 1996,8, 1451-1463 25. G. 0. Aspinall In The Biochemistry of Plants; Academic Press: New York, 1980; pp 473-500. 26. T. N. Wight, M. G. Kinsella and E. E. Qwarnstrom, Curr. Opin. Cell Biol., 1992, 4, 793-801 27. H. Lis and N. Sharon, Eur. J. Biochem., 1993,218, 1-27 28. A. Varki, Glycobiology, 1993, 3, 97-130 1. 2. 3. 4. 5. 6. 7. 8. 9.

References

1089

29. L. KjellCn and U. Lindahl, Annu. Rev. Biochem., 1991, 60, 443-475 30. T. C. Laurent, U. B. G. Laurent and J. R. E. Fraser, Imm. Cell. Biol., 1996, 74, A3-A7 31. L. A. Sellers and A. Allen In Mucus and Related Topics; University Press Corporation: 1989; pp 65-71. 32. M. Morgelin, M. Paulsson, T. E. Hardingham, D. Heinegdrd and J. Engel, Biochem. J., 1988, 253, 175-185 33. R. E. Turner, P. Lin and M. K. Cowman, Arch. Biochem. Biophys., 1988,265, 484-495 34. E. R. Morris, R. D. A. and E. J. Welsh, J. Mol. Biol., 1980, 138, 383-400 35. J. K. Sheehan, K. H. Gardner and E. D. T. Atkins, J. Mol. Biol., 1977, 117, 113-135 36. S. Arnott, A. K. Mitra and S. Raghunathan, J. Mol. Biol., 1983, 169, 861-872 37. J. Ed Scott, Ciba Found. Symp., 1989, 143, 6-20 38. L.-A. Fransson and L. Coster, Biochim. Biophys. Acta, 1979, 582, 132-144 39. E. 4. Turley and S. Roth, Nature, 1980,283, 268-271 40. L.-& Fransson, I. Sjoberg and V. P. Chiarugi, J. Biol. Chem., 1981,256, 13044-13047 41. L.-A. Fransson, L. Coster, A. Malmstrom and J. K. Sheehan, J. Biol. Chem., 1982, 257, 6333-6338 42. L.-A. Fransson, I. Carlstedt, L. Coster and A. Malmstrom, J. Biol. Chem., 1983, 258, 1434214345 43. L-4.Fransson and B. Havsmark, Carbohyd. Res., 1982, 110, 135-144 44. L.-A. Fransson, B. Havsmark and J. K. Sheehan, J. Biol. Chem., 1981,256, 13039-43 45. A. D. Snow, H. Mar, D. Nochlin, R. T. Sekiguchi, K. Kimata, Y . Koike and T. N. Wight, Am. J. Pathol., 1990, 137, 1253-1270 46. R. J. Stewart and J. M. Boggs, Biochemistry, 1993, 32, 10666-10674 47. S.4. Hakomori, Pure Appl. Chem., 1991, 63, 473-482 48. H. V. Wilson, J. Exp. Zool., 1907, 5, 245-258 49. P. S. Galtsoff, J. Exp. Zool., 1925, 42, 223-251 50. G. N. Misevic and M. M. Burger, J. Biol. Chem., 1986,261, 2853-2859 51. T. Humphreys, Dev. Biol., 1963, 8, 27-47 52. T. Humphreys In The Speczjicity of Cell Surfaces; Englewood Cliffs: Prentice Hall, NJ, 1967; pp 195-210. 53. P. Henkart, S. Humphreys and T. Humphreys, Biochemistry, 1973. 12, 3045-3050 54. C. B. Cauldwell, P. Henkart and T. Humphreys, Biochemistry, 1973, 12, 3051-3055 55. S. Humphreys, T. Humphreys and J. Sano, J. Supramol. Struct., 1977, 7, 339-351 56. U. Dammer, 0. Popescu, P. Wagner, D. Anselmetti, H.-J. Giintherodt and G. N. Misevic, Science, 1995,267, 1173-1 175 57. J. E. Jumblatt, V. Schlup and M. M. Burger, Biochemistry, 1980, 19, 1038-1042 58. J. A. Varner, M. M. Burger and J. F. Kaufman, J. Biol. Chem., 1988,263, 8498-8508 59. J. A. Varner, J. Cell Sci., 1995, 108, 3119-3126 60. J. A. Varner, J. Biol. Chem., 1996,271, 16119-16125 61. S. R. Turner and M. M. Burger, Nature, 1973,244, 509-510 62. G. N. Misevic and M. M. Burger, J. Cell. Biochem., 1990, 43, 1-8 63. G. N. Misevic, J. E. Jumblatt and M. M. Burger, J. Biol. Chem., 1982,257, 6931L6936 64. D. J. Rice and T. Humphreys, J. Biol. Chem., 1983,258, 6394-6399 65. W. Burkart and M. M. Burger, J. Supramol. Struct. Cell. Biochem., 1981, 16, 179-192 66. D. Spillmann, G/ycoconj. J., 1994, I I , 169- 171 67. G. N. Misevic and M. M. Burger, J. Biol. Chem., 1993,268, 4922-4929 68. D. Spillmann, K. Hgrd, J. Thomas-Oates, J. F. G. Vliegenthart, G. Misevic, M. M. Burger and J. Finne, J. Biol. Chem., 1993,268, 13378-13387 69. D. Spillmann, J. E. Thomas-Oates, J. A. van Kuik, J. F. G. Vliegenthart, G. Misevic, M. M. Burger and J. Finne, J. Biol. Chem., 1995,270, 5089-5097 70. T. Feizi, Nature, 1985, 314, 53-57 71. B. A. Fenderson, U. Zehavi and S.4. Hakomori, J. Exp. Med., 1984,160, 1591-1596 72. N. Kojima, B. A. Fenderson, M. R. Stroud, R. I. Goldberg, R. Habermann, T. Toyokuni and S.4. Hakomori, Glycoconj. J., 1994, 11, 238-248 73. I. Eggens, B. Fenderson, T. Toyokuni, B. Dean, M. Stroud and S.4. Hakomori, J. Biol. Chem., 1989, 264, 9476-9484

1090

38 Carbohydrate- Carbohydrute Interactions

74. Y. Song, D. A. Withers and S.4. Hakomori, J. Biol. Chem., 1998, 273, 2517-2525 75. K. Iwabuchi, S. Yamamura, A. Prinetti, K. Handa and S.-i. Hakomori, J. Bid. Chem., 1998, 273, 9130-9138 76. N. Kojima and S.4. Hakomori, J. Biol. Chem., 1989,264,20159-20162 77. N. Kojima and S.4. Hakomori, J. Biol. Chem., 1991,266, 17552-17558 78. N. Kojima and S.4. Hakomori, Glycobiology, 1991, 1, 623-630 79. N. Kojima, M. Shiota, Y. Sadahira and S.-i. Hakomori, J. Biol. Chem., 1992, 267, 1726217270 80. J. Jarchow and M. M. Burger, Cell Adhes. Comm., 1998. 81. 0. Popescu and G. N. Misevic, Nuture, 1997, 386, 231-232 82. S.4. Hakomori, Biochem. Soc. Trans., 1993, 21, 583-595 83. J. Finne, J. Bid. Chem., 1982,257, 11966-11970 84. U. Rutishauser and L. Landmesser, Trends Neurosci., 1996, 19, 422-427 85. G. Siuzdak, Y. Ichikawa, T. J. Caulfiedl, B. Munoz, C.-H. Wong and K. C. Nicolaou, J. Am. Chem. Soc., 1993,115, 2877-2881 86. K. M. Koshy and J. M. Boggs, J. Biol. Chem., 1996,271, 3496-~3499 87. S. PCrez, N. Mouhous-Riou, N. E. Nifant’ev, Y. E. Tsvetkov, B. Bachet and A. Imberty, Glycobiology, 1996, 6, 537-542 88. M. M. Burger In The Role of Intercellulur Signals: Nuvigution, encounter, outcome; Verlag Chemie: Berlin, 1979; pp 119-134. 89. M. Fukuda, Biochim. Biophys. Actu, 1985, 780, 119-150 90. N. Kojima, Trends Glycosci. Glycotechnol., 1992, 4 , 491-503 91. K. Simons and G. van Meer, Biochemistry, 1988, 27, 6197-6702 92. D. A. Brown and J. K. Rose, Cell, 1992, 68, 533-544 93. T. W. Tillack, M. Allietta, R. E. Moran and W. W. J. Young, Biochim. Biophys. Actu, 1983, 733, 15-24 94. M. Sorice, I. Parolini, T. Sansolini, T. Garofalo, V. Dolo, M. Sargiacomo, T. Tai, C. Peschle, M. R. Torrisi and A. Pavan, J. Lipid Res., 1997, 38, 969-980 95. P. Rock, M. Allietta, W. W. j. Young, T. E. Thompson and T. W. Tillack, Biochemistry, 1990,29, 8484- 8490 96. S.-i. Hakomori, Cancer Rex, 1996, 56, 5309-5318 97. M. Salmivirta, K. Lidholt and U. Lindahl, FASEB J., 1996, 10, 1270-1279 98. W. J. Kuhns, G. Misevic and M. M. Burger, Biol. Bull., 1990, 179, 358-365 99. S. J. Angydl, Adv. Curhohydr. Chem. Biochem., 1989, 47, 1-43 100. A. Almond, A. Brass and J. K. Sheehan, Glycobiology, 1998, 8, 973-980 101. R. U. Lemieux, Chem. Soc. Rev., 1989,119, 347-374 102. D. Balasubramanian, B. Raman and C. S. Sundari, J. Am. Chem. Soc., 1993,115,74-77 103. A. Varki, J. Clin. Invest., 1997, 99, 158-162 104. T. Feizi, M. S. Stoll, C. T. Yuen, W. Chai and A. M. Lawson, Methods Enzymol., 1994, 230, 484-519 105. D. Zopf and S. Ohlson, Nature, 1990, 346, 87-88 106. E.-C. Kang, K. Akiyoshi and J. Sunamoto, Int. J. Biol. Mucromol., 1994, 16, 348-353 107. M. Radmacher, R. W. Tillmann, M. Fritz and H. E. Gaub, Science, 1992,257, 1900-1905 108. H. G. Hansma, K. J. Kim, D. E. Laney, R. A. Garcia, M. Argaman, M. J. Allen and S. M. Parsons, J. Struct. Biol., 1997, 119, 99-108 109. D. J. Muller, C.-A. Schoenenberger, F. Schabert and A. Engel, J. Struct. Biol., 1997, 119, 149-157 110. M. Rief, F. Oesterhelt, B. Heymann and H. E. Gaub, Science, 1997, 275, 1295-1297 111. J. Fritz, D. Anselmetti, J. Jarchow and X. Fernandez-Busquets, J. Struct. Biol., 1997, 119, 165-1 71 112. T. Peters and B. M. Pinto, Curr. Opin. Struct. Biol., 1996, 6, 710-720 113. A. Imberty, Curr. Opin. Struct. Biol., 1997, 7, 617~-623 114. G. Siuzdak In Muss Spectrometry f o r Biotechnology; Academic Press, Inc.: San Diego, New York, Boston, London, Sydney, Tokyo, Toronto, 1996; pp 1-161. 115. R. J. Woods, Glycoconj. J., 1998, 15, 209-216 116. T. Ziegler, Curbohyd. Rex, 1994, 262, 195-212

References

1091

1 1 7. M. R. Stroud, S. B. Levery, S. Mgrtensson, M. E. K. Salyan, H. Clausen and S.4. Hakomori,

Biochemistry, 1994, 33, 10672- 10680 118. T. Le Bouar, F. Pincet, J.-M. Mallet, J. Esnault, Y. M. Zhang, E. Perez and P. Sinay, Glycoconj. J., 1997, 14, Sl2-73 119. S. Saito, S. B. Levery, M. E. K. Salyan, R. I. Goldberg and S.4. Hakomori. J. Biol. Chem., 1994,269, 5644-5652 120. M. R. Stroud, S. B. Levery, E. D. Nudelman, M. E. K. Salyan, J. A. Towell, C. E. Roberts, M. Watanabe and S.4. Hakomori, J. Biol. Chem., 1991,266, 8439-8446 121. D. Gupta, R. Arango, N. Sharon and C. F. Brewer, Biochemistry, 1994,33, 2503-2508 122. I. Eggens, B. A. Fenderson, T. Toyokuni and S.4. Hakomori, Biochem. Biophys. Rex Commun., 1989, 158, 913-920

Part I

Volume 2

VI Carb0hydrate-Nucleic Acid Interactions

Carbohydrates in Chemistry and Biology Editor Beat Ernst, Gerald W. Hart, Pierre Sinay copyright OWILEY-VCH Verlag GmbH, 2000

39 Carbohydrate-Nucleic Acid Interactions Heinz E. Moser

39.1 Introduction Among the various classes of biomolecules, nucleic acids play a central role not only as carrier of genetic information but also in the form of RNA as important link for the production of proteins. Specific interactions both on DNA and RNA are essential to retrieve the information required at various stages of life cycles. In general this is largely achieved by a variety of DNA or RNA binding proteins to regulate accessibility and transcription of DNA as well as modification, transport and translational processes of messenger RNA. Molecular interference with these processes can result in more or less specific disturbance of normal cell functions and evolution produced a variety of relatively small ligands that bind with high affinity to DNA or RNA, thereby disrupting crucial functions for life. This often translates in cytostatic or cytotoxic activity and depending on specificity, quite a few of these molecules found a therapeutic application as antiinfective or antitumor agents. Only a limited number of these compounds can be considered as pure carbohydrates. The binding to highly charged nucleic acids required structural properties that are usually not common for carbohydrates. Features like reduction of hydrophilicity and incorporation of positive charges are often found in carbohydrate containing molecules that tightly bind nucleic acids and they indicate how nature managed to optimally benefit from the richness of structural information in carbohydrates by adopting required physicochemical properties. Only a few structures of pure carbohydrate nucleic acid complexes are known thereby limiting our understanding of these interactions. However, many of the DNA ligands known to date carry one or more carbohydrate residues (recognition elements [ 11). The increasing availability of x-ray and NMR structures helps to understand their influence on interaction with either DNA or RNA at the molecular level. A small number of review articles have previously addressed carbohydrate nucleic acid interactions [2-51. The focus of this overview is the discussion of

1096

39 Carbohydrate-Nucleic Acid Interactions

selected examples from recently published structures with the aim to better understand common rules that govern interactions between nucleic acids and modified carbohydrates.

39.2 Carbohydrates Binding to DNA 39.2.1 Ene-Diyne Antibiotics and Antitumor Agents A family of extremely potent antitumor antibiotics were discovered over the past decades with many common features [ 3 , 61. These molecules all bind double stranded DNA in the minor groove and consist of a bicyclic core containing a strained 9- or 10-membered ring with an ene-diyne functionality and at least one sp2-like bridgehead carbon atom. Either an intra- or intermolecular nucleophilic attack at this central ring triggers a conformational change allowing a spontaneous cyclization to take place. The resulting, highly reactive aryl diradical was demonstrated to initiate single or double strand DNA cleavage based on H-abstraction from the deoxyribose backbone. In addition, most compounds carry one or more carbohydrate side chains and/or an intercalating group, adding to structural complexity and tuning their biological function. Selected members of this family include neocarzinostatin (l),esperamicin A1 (2), calicheamicin y1 (3), and dynemicin A (4) (Scheme 1). Recent publications of high resolution NMR structures of such ligands to short DNA duplexes [7, 81 allow a much better understanding of DNA interaction and modification by this complex class of molecules, including the important function contributed by carbohydrate moieties. The incredibly potent family of calicheamicins were discovered independently almost 20 years ago at Lederle Laboatories and Brystol Myers [9-121 based on a focused effort to identify natural products with the ability to inhibit tumour growth. Esperamicin A1 (2) and calicheamicin y1 (3) were among the most potent members isolated (approximately 1000-fold more potent than adriamycin (8b) against murine tumors, Scheme 3), and they both share the same aglycon but vary in their side chains influencing their specificity to interact with and damage DNA. Their biological activity triggered extensive research efforts to understand structure and function of aglycon and side chains to gain better insight at the molecular level of this fascinating and complex class of compounds. A key component to unravel this puzzle was the synthetic accessibility of the core ene-diyne structure that was successfully tackled by various groups [ 13-20]. Mastering the synthetic accessibility to these complex molecules paved the way to explore structure activity relationship of analogs. Esperamicins

Esperamicin A1 (2) and related members of this family, originating mainly from chemical hydrolysis or nuclephilic attack inducing the cycloaromatization, were

39.2 Cuvbohydrutes Binding to D N A

1097

h

3 Q

.-S .-0

fc

0”

P

1098

39 Curbohydrate-Nucleic Acid Interactions

studied extensively to understand their mechanism of action in relation to biological activity [21-241. Esperamicin A1 (2) was shown to be highly cytotoxic in various eukaryotic cell lines with ICSOSof 0.3-4.5 ng/ml. Removal of parts D and E (esperamicin C) reduced the cytotoxic effect by roughly 200-fold. Additional removal of carbohydrate moiety B (esperamicin D) further reduced the potency by approximately five-fold. Interestingly, this trend was also found for their ability to induce DNA breaks in HCT116 cells; however, the increase of concentration needed to trigger a similar amount of strand breaks varied 5000-fold for removal of D and E and additional 30-fold for esperamicin D, indicating that additional functions might be involved in the observed cytotoxicity [21]. Surprisingly examination of DNA scission in vitro showed equipotent behaviour of both esperamicin A1 and C, indicating a role of the deoxyfucose-anthranilate moiety D and E for drug accumulation in cells. However, esperamicin A1 predominantly produced single strand breaks at low concentration, pointing out the importance of the deoxyfucoseanthranilate moiety to position the ene-diyne aglycon and ultimately the aryl diradical in the minor groove of DNA, facilitating H-abstraction mainly from one of the two sugar phosphate backbones. These findings as well as sequence preferences and the proposed intercalation of the methoxyacrylyl-anthranilate E [24, 251 were experimentally supported by the NMR solution structure of esperamicin A1 (2) bound to the self complementary DNA duplex d(CGGATCCG) [S] (Figure la). Upon closer examination of these published structures some of the previously observed properties of the various esperamicins can be much better understood and indicate the role of the carbohydrate side chains. The ene-diyne aglycon R and the methoxyacrylyl-anthranilate E are linked by the deoxyfucose D which serves as ideal spacer accommodating the necessary 90" turn and distance between these residues. In addition, a hydrogen bond between the axial hydroxyl group and the carbonyl group of cytosine (C7', see Figure lb) contributes to the overall stability of the complex. Remarkably, this hydrogen bond could hypothetically be formed also with the three other bases replacing cytosine as they all contain a hydrogen bond acceptor at this position (thymine: 0-2; adenine: N-3; guanosine: N-3), thereby supporting the lack of a clear sequence preference for this particular interaction. The trisaccharide A-B-C on the opposite side of aglycon R has a key role to help fixing the ene-diyne moiety deep in the minor groove (Figure lb). All sugar residues are deoxygenated and carry rather unusual groups optimizing favorable interactions (hydrophobic and/or opposite charges) with the DNA duplex. Carbohydrate moiety A serves as connection point between the aglycon R and residues B and C, facing the relatively wide minor groove with the 6-membered ring and interacting with the phosphodiester backbone via its equatorial hydroxyl group. Residue B is placed through the unusual NH-0 linkage deep into the minor groove with the ring parallel to the rims of the groove, establishing extensive hydrophobic contacts in addition to a hydrogen bond with N-3 of adenine via its only axial hydroxyl group. As mentioned above, this specific interaction could in principle take place with the other three nucleic acid bases as well. An additional interaction was pointed out between the large polarizable sulfur atom on ring B and the exposed exocyclic amino proton of the proximal guanine base, detectable by the large upfield shift for this proton. This observation would indicate a sequence pref-

39.2 Carbohydrates Binding to DNA

U

1099

U

b Figure 1. Stereo representation of esperamicin A1 bound to the self complementary oligodeoxyribonucleotide duplex d(CGGATCCG)Z ([8], structures were retrieved from the Brookhaven database: lpik; to simplify the discussion only structure #2 is represented, with the hydrogen atoms omitted). a) Different parts of esperamicin A1 are colored as follows: aglycon R and methoxyacrylyl-anthranilate E (light grey), carbohydrate residues A , B, C, and D (dark grey). b) Same structure with one base pair omitted at each end. Specific heteroatoms intermolecularly interacting between esperamicin A1 and the DNA via hydrogen bonds are highlighted as small balls. Oxygen atoms are represented in light grey and both, nitrogen and sulfur, are shown in dark grey; all the other elements are black.

1100

39 Carbohydrate-Nucleic Acid Interactions

erence for a guanine at this position; even though such sites are found upon close examination of esperamicin A1 induced cleavage patterns, overall this interaction does not seem to play a dominant role. Residue C is placed over the sugar phosphate backbone of the DNA duplex forming a hydrogen bond between its charged isopropyl ammonium group and one of the phosphodiester oxygens. All the interactions discussed above are in excellent agreement with the observed, relatively weak sequence preferences for DNA cleavage by esperamicin AJ. The reported structure provides a plausible explanation for the preferred break of only one strand. The ene-diyne aglycon R is tightly placed deep in the minor groove fixed on both sides by the anthranilate-deoxyfucose D-E and the trisaccharide A-B-C sidechains, respectively, allowing only minimal conformational changes relative to the duplex DNA. Whereas one of the potential radical positions will be ideally placed to abstract HI’ from the bottom strand (Figure lb), the other position on the opposite site is not only further away from Hs’ of the other strand but also not ideally directed at this position. It is important to note that these differences have to be read with caution as in particular the position of Hs’ is not as well defined and the interpretation of this structure in terms of cleavage prediction is limited. Nevertheless, the hypothesis that esperamicin A J predominantly cleaves one strand based on conformational fixation is supported by the cleavage behavior of esperamicin C: removal of the intercalating part E and linking sugar D yields an esperamicin analog that exclusively produces double strand scissions [211. Calichearnicins

Calicheamicin y, (3) shares the same ene-diyne aglycon R and highly similar trisaccharide A-B-C (for 2) or A-B-E (for 3) with esperamicin Al (2) but lacks the intercalating methoxyacrylyl-anthranilate deoxyfucose part and contains an additional extension on the trisaccharide [6].Two major differences were found between these antibiotics: calicheamicin y1 (3) shows a clear sequence preference for TCCT (found as well for the oligosaccharide moiety alone [26-281) and produces mainly double strand cleavages even at low concentrations with a typical 3 base 3‘-shift of the cleavage sites on opposite strands, commonly observed for DNA cleavers bound to the minor groove of DNA [6, 291. Further insight was gained from the cleavage behaviour of calicheamicins either lacking rings D or E:Similar to esperamicin A1 , omission of the 4-ethylamino sugar E did not cause a change in the observed specificity patterns but dropped the efficiency by 2-3 orders of magnitude indicating a function for binding affinity but not specificity. The calicheamicin lacking the terminal rhamnose D exhibited a similar cleavage specificity with a roughly 100-fold lower efficacy [30]. Many groups initiated various studies to better understand the molecular interaction between DNA and calicheamicin y1 (3); in particular NMR proved to be extremely valuable [31-341. Finally, a high resolution structure of calicheamicin y1 (3) bound to the 23-mer hairpin duplex d(CACTCCTGGTTTTTCCAGJgG19AGTG) was obtained and published that is used here as basis for the discussion of specific DNA carbohydrate interactions [7]. Calicheamicin y1 (3) binds in an extended conformation to the minor groove of DNA, placing the terminal ene-diyne aglycon deep into the groove (Figure 2a).

39.2 Carbohydrates Binding to D N A

1101

h

Figure 2. Stereo representation of calicheamicin y, (3) bound to the 23-mer hairpin duplex d(CACTCCTCGTTTTTCCAGGAGTG) ([7], the structure was retrieved from the Brookhaven database as 2pik and due to simplicity only the first structure is represented omitting hydrogen atoms and the T5 hairpin loop. a) Different parts of calicheamicin y, (3) are coloured as follows: aglycon R and thiobenzoate ring C (light grey), carbohydrate residues A , B, D, and E (dark grey). b) Same structure with one base-pair omitted at the top end. Specific heteroatoms intermolecularly interacting between calicheamicin y, (3) and the DNA via hydrogen bonds are highlighted as small balls. Oxygen and iodine atoms are represented in light grey and both nitrogen and sulfur, are shown in dark grey; all the other elements are black.

1102

39 CarbohydrateeNucleic Acid Interactions

Determination of the structure without target DNA under various conditions indicated a high degree of conformational preorganization [ 351, allowing this rather big molecule to bind its target site in an extended conformation with high affinity. The interactions for sugar rings A , B, and E closely resemble those discussed for the similar counterparts in esperamicin Al (2) discussed previously (see Figure 2b). It is worthwhile to point out that the required orientation between rings A and B is maintained with the unusual N - 0 bond, ideally adopting an eclipsed conformation with a torsion angle of close to 120" [36, 371 (-128.4 2.6" [7]) providing a relatively rigid, gentle curvature in the middle of the molecule to optimally achieve the overall shape to bind the minor groove. A change of the natural configuration at the anomeric center on ring A completely disrupts the binding affinity of the corresponding aryltetrasaccharide to the target DNA site [28], further supporting the importance of structural preorganization. As mentioned for esperamicin A*, the sulfur atom is in proximity with the exocyclic amino group of guanine 19 causing a large difference in the chemical shift of the corresponding proton. However, replacing this guanine by inosine lacking the exocyclic amino group caused only a modest change in the binding affinity of the aryltetrasaccharide and therefore indicates this interaction to be of minor importance [ 381. The thiobenzoate ring C is buried similar to sugar B deep in the minor groove with its methyl and iodine substituents facing the bottom of the groove and establishing stacking interactions with the deoxyribose rings of T7 and A20 on opposite strands. The large polarizable iodine atom is in proximity to the exocyclic amino group of guanine 18 favorably contributing to the sequence preference of calicheamicin y1 (3) [ 11. Substitution of the iodine atom by bromine, chlorine, fluorine, methyl, or hydrogen [38] indicated the importance of iodine in this position for binding of the corresponding aryltetrasaccharide as these substitutions progressively reduced the binding affinity as experimentally determined by a competition cleavage experiment. In agreement with these findings, substitution of the critical guanine 18 by inosine greatly diminishes calicheamicin y1 (3) binding and subsequent DNA cleavage [38]. Rings B and D both showed the highest resolution with the least standard deviations of the four structures published [7] giving a high degree of confidence in this central part of the molecule. The terminal rhamnose sugar D is twisted relative to the rims of the minor groove and interacts via its 2-hydroxyl group with the proximal phosphodiester group on the purine strand of the recognition site. In addition, the second hydroxyl group is within distance to interact with a phosphodiester oxygen on the opposite strand. Finally, the ene-diyne aglycon R lacking further substitution on the opposite side of its aryltetrasaccharide substituent, interacts via its unsubstituted hydroxy group with the proximal oxygen atom on the terminal phosphodiester of the complex. The calicheamicin oligosaccharide 5 (Scheme 2) was shown to bind with low micromolar affinity to its target site with similar selectivity as compared to calicheamicin y1 (3) [38] (see also references in [39]). This unique property could be used to compete off specific transcripton factors from their target sequence at the expected concentrations [39]. The limited selectivity and potency, however, were not sufficient for a meaningful biological application and the recent progress in the synthesis of complex oligosaccharides allowed this approach to be explored further.

39.2 Carbohydrates Binding to D N A

1103

1 104

39 CurbohydruteeNucleic Acid Interactions

Based on structural knowledge of the calicheamicin oligosaccharide bound to its target DNA tetramer site, head to head and head to tail dimers 6 and 7 were designed and synthesized [40-421. Examination of the inhibitory activity on calicheamicin y, (3) induced DNA cleavage revealed an estimated 100-fold higher affinity of the head to head dimer 6 and more than 1000-fold improved binding of dimer 7 for the target sites as compared to monomer 5 [43]. Both dimers bound with >100-fold selectivity to the corresponding target site but dimer 6 interacted differently with two distinct target sites indicating the influence of neighboring sequences on the overall activity. This dimer was subjected to structure elucidation by NMR as complex with the self complementary oligonucleotide duplex d(CGTAGGATATCCTACG)2 [34]. Close examination of the head to head dimer 6 interaction with the DNA duplex essentially revealed a highly similar binding mode of each carbohydrate unit as compared to the monomer 5, reproducing all the individual interactions discussed previously (see above). The recently published structure of the head to tail dimer 7 confirmed the previous findings and added an interesting insight on sequence selectivity [44]: the structure was resolved with the target duplex d(GCACCTTCCTGC) x d(GCAGGAAGGTGC) containing two slightly different affinity sites. As discussed with calicheamicin y, (3) itself, the 3-OH group of ring B hydrogen bonds with N-3 of A20 [7] (see Figure 2b). Flipping this base pair in the duplex positions the 0 - 2 carbonyl function in almost exactly the same place allowing the hydrogen bond to be formed as evident from the NMR structure. The calicheamicin y1 (3) and calicheamicin oligosaccharide structures 6 and 7 discussed above [7, 34, 441 revealed an almost perfect fit between duplex DNA and bound carbohydrate residues. Structural analysis of the DNA binding sites revealed no major conformational changes at the sites of carbohydrate interactions whereas the accommodation of the en-diyne moiety required a widening of the minor groove. These findings are in contrast to an earlier hypothesis [31, 331 that assumed the conformational flexibility of the DNA binding site to be of crucial importance for the selective binding event. The head to head dimer 6 was also tested for its ability to interfere with transcriptional events [45]. At low micromolar concentrations oligosaccharide 6 was able to disrupt binding of transcription factors (AP-1, STAT-3, NFI, and PU.l) to the corresponding target sites containing a overlapping binding site as shown by gel mobility assays and in vitro transcription by polymerase 11. These experiments demonstrate the validity of modified carbohydrate units as minor groove recognition elements, not limiting their function as originally believed to tune the pharmacokinetic behavior. The required degree of conformational preorganization with an evolutionary adaption of the individual units by decreasing hydrophobicity and adding positive charges make the functionally rich carbohydrates a truly unique scaffold for minor groove recognition of DNA. Whereas the work described above was mainly based on further elaboration of oligosaccharides from natural sources, the increasing structural understanding on molecular level paired with powerful technologies like combinatorial (carbohydrate) chemistry, improved synthetic methodologies, and experimental binding optimization by NMR [46-481 might lead to new horizons for DNA-carbohydrate interactions. In particular, probing preformed DNA-carbohydrate complexes by NMR for adjacent

&,

39.2 Cuvhohydrates Binding to DNA

\

g

0

t

-

t

o

O \ /

= / \ 0

-

0

0

-

0 3 v

h

r

r

v

e

G6

P p

1105

1106

39 Curhohydrute-Nucleic Acid Interactions

binding events of structurally diverse carbohydrate residues (or other low molecular weight compounds) could lead after appropriate linkage to new classes of high affinity DNA ligands that will go beyond the scope of currently available molecules from natural sources.

39.2.2 Anthracyclins Anthracycline antibiotics or antitumor agents are a class of biologically active compounds isolated from various Streptornyces strains. They have indicator-like properties and consist of a core tetracyclic anthraquinone chromophore carrying one or multiple carbohydrate substituents. The first members of this family were isolated roughly half a century ago and shown to display potent antibacterial activities (for recent reviews see 149-521). The high toxicity in mice, however, precluded their further application in man and this changed in the sixties with the discovery of the structurally closely related daunorubicin (or daunomycin, 8a) 153, 541 and doxorubicin (or adriamycin, 8b) [55], both currently used in the clinical treatment of leukaemias and solid tumors, respectively. They interfere with DNA replication and transcription by tightly binding to double stranded DNA via intercalation, preferably at alternating pyrimidine-purine tracts, placing the carbohydrate moiety in the minor groove. In addition, they interfere with topoisomerase I1 activity 1561. Due to an intense interest in their biological behaviour and shortcomings in the clinical application (cardiotoxic side effects and development of resistance) efforts were initiated to understand the interaction with DNA on the molecular level. To date well over 40 high resolution structures are available from the Brookhaven database, providing insight in the role of the carbohydrate-DNA interaction that might lead to the design of derivatives with improved properties. Daunorubicin (8a) was the first anthracycline antibiotic for which structural information at high resolution became ayailable of its DNA-bound state 1571. This structure was refined to remarkable 1.2 A 1581 and, later on, was followed by additional structures using various DNA templates 159-611. In the highly refined 2: 1 complex with d(CGTACG)2 [58],the tetracyclic aglycon is intercalated between the terminal CG base pairs, oriented almost at a right angles to the long axis between the base pairs (Figure 3a). In this complexed form the conformation of ring A varies from the x-ray structure obtained from daunorubicin (8a) alone [62]: only C-9 lies outside the plane defined by the fused aromatic rings, positioning 0-9 almost perfectly perpendicular to this plane and wtthin hydrogen bonding distance to N-3 and N-2 of G2 (O-N distances: 2.6 and 2.9 A, respectively; see Figure 3b). Interestingly, this hydroxy group at C-9 is essential for the biological activity of daunorubicin, supporting the importance of this intermolecular interaction. Daunorubicin (Sa) was reported to preferentially bind to 5’-A/T-CG-3’ or 5’-A/T-GC-3’ triplets, albeit with only modest preference over other sequences 163, 641. This is in agreement with the hydrogen bonding interaction discussed above, as the more important interaction with N-3 of GZcan take place with the other three bases as well (N-3 of adenosine or 0 - 2 of either thymine or cytosine). The second main difference between the two x-ray structures is the torsion angle of the glycosyl linkage (C7-07-C1’-C2’)

1107

39.2 Curbohydrutes Binding to D N A

a

Figure 3. Stereo pictures of daunorubicin (8a) bound to d(CGTACG)* [58]. The co-ordinates have been retrieved from the Brookhaven database (1 l d l ) and for clarity, hydrogens are omitted. a) Representation of the DNA duplex with two daunorubicin molecules intercalated; the tetracyclic anthraquinone aglycon is represented as CPK model with carbon atoms in dark grey and oxygen/ nitrogen atoms in light grey. b) Part of the same structure including selected interacting water molecules (small balls in light grey) and one sodium atom (ball in dark grey). Heteroatoms involved in hydrogen bonding are drawn as small balls. Carbon and phosphorous atoms are shown in black, oxygen in light grey and nitrogen in dark grey.

1108

39 Carbohydrute-Nucleic Acid Interactions

which is smaller in the DNA-bound molecule. This small change allows the aminosugar (daunosamine) to adopt the correct orientation for an optimal fit in the minor groove of the right-handed DNA duplex. In order to optimally accommodate the carbohydrate in the minor groove the proximal G2-Cll base pair needs to be dislocated by roughly 1.3 A towards the major groove. Opposite to interactions seen in other complexes, the amino group does not directly bind to the negatively charged phosphodiester backbone but hydrogen bonds to 0-2 of C-5 and two water molecules. The axial hydroxy group at C-4’ points away from the DNA duplex into the solvent region where it might play an important role for the interaction with topoisomerase 11. As shown previously by footprinting studies of various anthracyclines with DNA, the attached carbohydrate residues do not largely influence the sequence selectivity for DNA binding [65], even though they greatly influence both the binding affinity [66] and biological activity [49]. These observations are supported by the high resolution structure of the complex: the binding affinity is increased by co-operative van der Waals contacts of the carbohydrate moiety in the minor groove. However, specific hydrogen bond interactions between carbohydrate and bases are lacking. The wealth of detailed information of the complex discussed above served as starting point for structure-based design of new bis-intercalating anthracycline antibiotics [67]. In all structures reported to date, a 2: 1 ratio of daunorubicin (8a) to target DNA was observed. Both drug molecules intercalate at the end of the duplex positioning their carbohydrate groups in the minor groove pointing at each other (see Figure 3a). In this bound form, the two amino groups of both daunorubicins are roughly 7 apart and therefore could easily be bridged with an appropriate linker to give the bis-intercalator 9. Binding studies confirmed the bis-intercalating binding mode and revealed a picomolar binding affinity (27 pM) to herring sperm DNA at 20”C, roughly four orders of magnitude higher than daunorubicin @a). The hypothesized binding mode of dimer 9 was later confirmed by structure determination using x-ray crystallography [68] and high resolution NMR [69]. The structural information on daunorubicin (8a) and calicheamicin y1 (3) bound to duplex DNA guided the design of a hybrid molecule between the daunorubicin aglycon and the calicheamicin y1 tetrasaccharide [70]. The main goal of this approach was the creation of an intercalating daunorubicin analog with improved sequence selectivity by replacing the carbohydrate moiety. Examination of models indicated the requirement of a P-glycosidic linkage and a five-atom spacer (CH2CH20CH2CH2) between aglycon and tetrasaccharide. The corresponding “calichearubicin” was indeed shown to intercalate double stranded DNA, whereas an analog lacking the five atom spacer failed to intercalate. Footprinting studies on a 155 base pair fragment of pBR322 with methidium propyl [71] indicated a gain in binding specificity compared to daunorubicin. Due to the difficult interpretation of these results and the lack of discrete binding affinities for this hybrid molecule, a clear picture on sequence preference could not be obtained. During the late sixties the structurally more complex nogalamycin (10) [72] was isolated and characterized. This anthracycline is unique as two carbohydrate groups (nogalose and aminoglucose) are linked to the opposite ends of the tetracyclic aglycon, yielding a dumbbell shaped molecule. As a consequence, one of the bulky

A

39.2 Curbohydrutes Binding to D N A

1109

sugar residues needs to move through the helix to reach the opposite side, most likely requiring a local melt of base pairs. This special feature is reflected in the slow on- and off-rates for binding. The publication of various high resolution structures of nogalamycin (9) to short DNA duplexes [73-811 confirmed the proposed binding mode (Figure 4a) which is discussed in more detail below. In all the structures reported to date nogalamycin 10 is intercalating DNA, preferentially between purine-pyrimidine tracts, with the doubly connected aminoglucose on ring D positioned in the major groove. The conformation of the DNA bound ligand [77] closely resembles the ligand itself [82], indicating a well preorganized structure. A subtle difference is the slightly closer proximity between nogalose and aminoglucose in the bound state that allows nogalamycin (10) to optimize its interactions with the DNA in the complex (see below). The aglycon intercalates almost perpendicular to both flanking GC base pairs, similar to the orientation reported for daunorubicip (8a). The main difference, however, is the shift towards the minor groove by -2 A, caused by interactions of the aminoglucose in the major groove and the large nogalose in the minor groove. The aminoglucose faces the GC base pair with its flat surface of the six-membered ring, positioning both hydroxy groups (OzG and 04G) in hydrogen bonding distance to N-7 of GZ and N-4 of m5Cll, respectively (Figure 4b). The dimethylamino group is not directly interacting with the DNA or its negatively charged phosphodiester backbone but is bridged through a water molecule to N-6 of A,". Ring A of the aglycon adopts a similar conformction as reported for daunorubicin with CS dislocated from the mean plane by 0.53 A, placing the hydroxy group in an equatorial position pointing away from the minor groove. The axial methyl ester at C-10 points in direction of the minor groove and interacts with N-2 of GI2 through a hydrogen bond. The large hydrophobic nogalose inserts sideways in the minor groove and primarily interacts by Van der Waals contacts, thereby widening the groove by roughly 3 A. Only one direct interaction was proposed between the glycosidic 0-7 between nogalose and ring A and the hydrogen donating N-2 of G2. The distance between these two and the lower electron density of the involved oxygen lone pair due to atoms (3.5 the stereoelectronic conjugation with the neighboring C-0 bond, however, suggest that this interaction is only weak. As seen previously for the calicheamicin antibiotics, the anthracyclins represent another family of DNA binding ligands carrying modified carbohydrates. To gain energetic benefits from binding to DNA, carbohydrates have to be less hydrophilic and/or contribute to the overall charge neutralization. This allows optimal interaction at the bottom and sidewalls of the lipophilic minor groove allowing excellent shape recognition through van der Waals contacts. In addition, the conformationally rigid six-membered ring is an excellent scaffold to place hydroxy or amino groups such that they can specifically interact with the functional groups of nucleotide bases on the floor of both grooves. Nature chose a variety of approaches to make sugar residues successful DNA or RNA ligands: often the carbohydrate is deoxygenated and some or all of the remaining hydroxy groups are methylated as seen in nogalose. The aminoglucose found in nogalamycin (10) is deoxygenated as well and, in addition, contains a dimethylamino functionality that energetically contributes to the interaction with DNA by charge neutralization. Both the remain-

A)

1110

39 Carbohydrate-Nucleic Acid Interactions

a

b Figure 4. Stereo representation of the x-ray structure of nogalamycin (9) (Brookhaven: ld21; [77]) bound to d[m5CGT(pS)Am5CG]2.The hydrogen atoms are omitted to simplify the representation of the complex. a) 2 nogalamycin molecules are shown in space filling models intercalated in the modified duplex with water molecules removed, carbon atoms are shown in dark grey and oxygen/ nitrogen atoms in light grey. b) Part of the same structure including selected interacting water molecules (small balls in light grey). Heteroatoms involved in hydrogen bonding are drawn as small balls. Carbon and phosphorous atoms are shown in black, oxygen in light grey and nitrogen in dark grey.

39.2 Carbohydrates Binding to DNA

1 11 1

ing hydroxyl groups are perfectly positioned for direct interaction with hydrogen bond donor or acceptor sites at the binding site. 39.2.3 Pluramycins and Aureolic Acids

Pluramycins are closely related to nogalamycin (10): they consist of a central, tetracyclic chromophore with up to three substituents (altromycins) at different positions. Besides carbohydrates they contain epoxy functionalities that provide them with alkylating properties in the bound state. This class of antibiotics was recently reviewed [83] and will not be discussed further in this review. Chromomycin A3 (11) is an antitumor antibiotic belonging to the group of aureolic acids along with related analogs like mithramycin and olivomycin. Chromomycin is believed to bind DNA, thereby inhibiting RNA transcription and regulation [84, 851. It was demonstrated, that it binds in the presence of Mg2+ to double-stranded DNA [86] with a sequence specificity to GC rich sites [87-901. Even though the central, tricyclic aglycon indicated an intercalative binding mode, NMR structure elucidation revealed a unique binding mode of the magnesium bound dimer in the minor groove of DNA [91, 921. The complex is bound symmetrically around the central (GC)2 sequence of the target duplex d(AAGGCCTT)Z (Figure 5, [91]) with its unligated hydroxy group of the aglycon directly interacting with the exocyclic amino group of the central guanosines. The binding of the dimer requires a widening of the minor groove to accommodate both chromophores and

Figure 5. Stereo picture of a Mg2+ chelated chromomycin A3 (11) dimer bound to d(AAGGCCTT)z [91] as retrieved from the Brookhaven database (ld83). The hydrogen atoms are omitted to simplify the representation of the complex. The DNA duplex is represented as a CPK model (grey) and both the chromomycin dimer and magnesium ion are shown in black.

1112

39 CurbohydruteeNucleic Acid Interactions

sugar moieties C and D side by side in a parallel orientation to the walls of the groove. Carbohydrates E extend on both sides in the minor groove, facing the bottom of the groove. The trisaccharide C-D-E along with mainly with the sugar moiety A , primarily stabilize the complex by van der Waals intermolecular interactions within the minor groove of the target duplex. As seen with the previous antibiotics, the hydrophilic nature of all carbohydrate moieties has been altered by deoxygenation (ABCDE), methylation ( B E ) and acetylation ( A E ) to improve the overall binding characteristics. This complex reveals a different molecular recognition approach for DNA binding molecules that is related to the recent finding of alternate binding modes for distamycin A or netropsin related compounds [93, 941. However, the metal ion plays a crucial role indicating new possibilities in the design of specific ligands that achieve higher molecular weight by chelating to physiologically present metal ions. A different but nevertheless related approach was described for protease inhibitors that bind the target enzyme as dimers chelated to Zn2+ [95]. In summary, the properly functionalized, six-membered carbohydrates can fit into the minor groove of DNA, depending on the groove width either facing the floor or squeezed between the phosphodiester backbone facing the walls. Besides changing the overall properties like solubility and pharmacokinetic behavior, these carbohydrates usually improve the overall binding affinity either in a selective or non-selective manner. As discussed previously, these carbohydrates are modified to increase their lipophilic properties and in some cases carry cationic functionalities to contribute to the overall charge neutralization. Such scaffolds seem to be ideally suited as DNA recognition elements due to size, conformational preorganization, and the high degree of potential functionalization. The synthetic difficulties, however, to fully explore this potential explain why there is currently only a limited understanding on energetic contributions of individual groups to overall DNA binding. The combination of combinatorial techniques with improvements in carbohydrate chemistry and biosynthesis might not only contribute to our understanding of interactions at the molecular level but could also provide compounds with improved therapeutic potential.

39.3 Carbohydrates Binding to RNA Quite a few antibiotics containing modified carbohydrates interfere with protein biosynthesis rather than transcriptional processes by binding to DNA. Examples are erythromycin, streptomycin (12), tetracycline, spectinomycin, hygromycin, edeine and the neomycin family of aminoglycosides. Even though the observed biological effects have been correlated with perturbation of particular ribosomal events, the understanding of specific interactions on molecular level between such antibiotics and ribosomal components remained unknown for a long time. In analogy to enzyme inhibitors, it was assumed for these molecules to directly and specifically interact with functional sites of ribosomal components but until quite recently, no convincing evidence was found. The hypothesis that rRNA is the

39.3 Carbohydrates Binding to RNA

11 13

essential determinant of ribosome function lead to close examination of the interaction between a variety of antibiotics and rRNA by chemical footprinting [96, 971. Over the past decade increasing information has been provided to support specific and direct interaction with rRNA but size and complexity of this target [98] rendered structural elucidation extremely difficult. Two high resolution structures were reported only recently [99, 1001. They provide for the first time structural insight in molecular recognition between aminoglycosides and RNA. Further information was obtained by attempts to optimize either the low molecular weight aminoglycosides or the target RNA using chemical synthesis or affinity selection and amplification approaches, respectively. In addition to ribosomal RNA, aminoglycosides have been shown to interact with various, biologically relevant RNA structures. Among those are catalytic group I introns [ 101, 1021, hammerhead ribozymes [ 1031061, the rev-binding element of HIV [ 107-1091, and Hepatitis delta virus ribozyme [ 1101. 39.3.1. Aminoglycosides

Aminoglycosides are a class of aminosugar and aminocyclitol containing antibiotics that inhibit prokaryotic cell growth by interfering with ribosomal protein biosynthesis [ 111-1 141 (Scheme 4). They have been in clinical use for several decades and lack of oral bioavailability, toxicity and frequent occurrence of resistance limit their current use in humans. They are known to perturb translational events, causing a marked decrease in the fidelity of translation and/or inhibiting translocation. A variety of experimental findings led to an increased understanding of aminoglycoside function. Antibiotic resistance was not only found by enzymatic modifications of the aminoglycosides or mutations in ribosomal proteins but also by direct changes of the rRNA sequence [ 1151, indicating points of direct contact. The small 30s ribosomal subunit of procaryotes consists of 16s ribosomal RNA (1540 nucleotides in length) and 21 proteins. Sequence analysis of this RNA from aminoglycoside resistant strains indicated functionally important areas for binding. Streptomycin resistance in Escherichia coli was originally shown to be governed by ribosomal protein S12. More recently, however, mutants have been found with a C to U transition at position 912 of 16s rRNA that was directly correlated with the observed aminoglycoside resistance [ 116, 1171. This finding indicated that streptomycin (12) directly binds to RNA rather than proteins. Experimental support for this assumption was reported from RNA footprinting studies of intact 70s ribosomes and sequence analysis of the 16s rRNA: chemical modification by dimethylsulfate was strongly suppressed in the presence of streptomycin (12) at adenosines 913-915 and weak but significant protection was noted for both, uracil 91 1 and cytosine 912. This protection or shielding from chemical modification surrounds and includes the same nucleotide that was mutated in a resistant strain thereby indicating the importance of streptomycin binding at this RNA site for antibiotic function. The neomycin family of aminoglycosides displays a different pattern of RNA protection and mutations associated with resistance. For example paromomycin resistance has been shown to result from single mutations in both

1114

39 Carbohydrate-Nucleic Acid Interactions N

-r

h

8 e0 FII 2 A oc

II

I

0

Y

I"

z

I"

z \

39.3 Carbohydrates Binding to RNA

1115

yeast mitochondria1 [ 1 181 and Tetrahymena rRNA [ 119, 1201. In accordance to the numbering of E. coli 16s ribosomal RNA, both identified mutations aim at the disruption of the base pair between positions 1409 and 1491 (Scheme 5) that is part of the A-site involved in the decoding by binding aminoacyl-tRNAs. As described for streptomycin (12), an excellent correlation was found with chemical footprinting pointing at the A-site as binding site for paromomycin (16): adenosine 1408 and guanosine 1494 are strongly protected from alkylation at N-1 and N-7, respectively. Similar behavior was detected for neomycin and gentamycin, whereas the protection for the smaller kanamycin is less complete at both nucleotides. An additional clue is provided by antibiotic producing microorganisms in which the resistance is achieved by methylation of specific nucleotides on rRNA. The cloned methylases can be expressed in heterologous organisms like E. coli which they render resistant towards selected aminoglycosides by transforming either guanosine 1405 or adenosine 1408 in the A-site of 16s rRNA [121]. This site was mimicked by a self-structured oligoribonucleotide that bound both antibiotic and RNA ligands of the 30s subunit in a similar manner as compared to the normal subunit [ 1221. This oligoribonucleotide was further reduced in size to the 27-mer 21 (Scheme 5) that was shown to specifically bind paromomycin (16) with similar binding affinity and protection pattern as found for E. coli 16s rRNA in the 30s subunit [ 123, 1241. This complex turned out to be suitable for structure elucidation by NMR spectroscopy, providing insight in molecular interactions between aminoglycoside and target RNA [99, 125, 1261 (Figure 6a). Paromomycin (16) binds in the major groove of the A-site within a pocket that is formed by the asymmetric internal loop, thereby changing the local structure of the bound RNA [126]. Not bound to RNA, paromomycin (16) has in solution rather rigid ring conformations with certain flexibility at the glycosidic linkages connecting the four rings. In the bound form, however, paromomycin adopts a L-shaped conformation with rings 11, I11 and IV linearly arranged. Both, rings I and I1 are conformationally highly conserved; they adopt chair conformations placing all the substituents at the 6-membered ring in equatorial positions with a number of direct contacts to the RNA (see Figure 6b). Ring I stacks on the purine ring of guanosine 1491 and its substituents interact with the phosphodiesters of guanosine 1491 and adenosines 1492 and 1493. Both hydroxy groups, 3j-OH and 4/-OH, are not essential for aminoglycoside function [ 1271 but the replacement of the pro-R oxygens of the phosphates which point into the minor groove at positions 1492 and 1493 with sulfur interfere with paromomycin binding [ 1281. The 6/-OH group might possibly not only interact with the proximal phosphodiester of adenosine 1493 but also with its N-7 of the purine base. The only amino group at C-2’ strongly forms an intramolecular hydrogen bond with ring I11 and might additionally interact with the phosphodiester of guanosine 1491. Both amino groups of ring I1 (2-deoxystreptamine) directly interact with N-7 of guanosine 1494 and 0 - 4 of uracil 1495 and possibly with the phosphodiester between adenosines 1492 and 1493. The CG base pair between 1407 and 1494 is essential for aminoglycoside binding [ 1231, and carboxyethylation at N-7 of guanosine 1494 interferes with paromomycin binding. These crucial interactions between paromomycin (16) and the A-site model oligoribonucleotide 21 have been largely confirmed through an affinity chromatography

1116

1400

39 Carbohydrate-Nucleic Acid Interactions

c

G-C C :A’

..

C U C-G 1405

GVc 1495 U G C A A A C-G 1410 A- u 1490 C-G C-G

I t

3

5’

A

AG Ci5 U U lo U-A

u

G-C G-C C-G 1405 G-C U

1495

G

C

A

A

A

c -G 1410 A- u 1490 C-G C-G

U

G

u c 21

F C A G-C *O A-U

G-C 5C-G A-U C-G25 G-C G-C s

3’

22

20 Scheme 5. Secondary structure on Escherichia coli 16s RNA in the region of decoding A-site 20 and its shortened hairpin version (27-mer model oligoribonucleotide 21) used for the structure determination with bound paromomycin (16) [ 1251. Nucleotides which are protected from chemical modification by bound aminoglycosides, methylated in resistant strains or essential for aminoglycoside binding are shown in bold. For more detailed discussion refer to text. The sequence of the 27-mer RNA aptamer 22 with its secondary structure is shown. The consensus sequence identified for this family of aptamers is highlighted in bold [141].

based assay looking at binding interference between immobilized paromomycin (16) and various, base or backbone modified RNAs [ 1281. Rings I11 and IV of paromomycin (16) extend into the minor groove and in particular ring IV contributes through positive charges to the overall binding affinity without adding much to the specificity. This is supported by the fact that both, ribostamycin (14) and neamine (13), lacking ring IV and rings 111 and IV, respectively, bind to 16s rRNA and cause miscoding, albeit at higher concentration. Measured dissociation constants by surface plasmon resonance [I081 are 19 nM, 25 pM, and 7.8 pM for neomycin B (15), ribostamycin (14), and neamine (13), respectively (paromomycin (16): 0.20 pM) [129]. In addition, ring IV is partially disordered in the ensemble of 20 NMR structures providing the least confidence for conformational determination. Exchanging ring IV (2,6-dideoxy-2,6-diamino-P-~-

39.3 Carbohydrates Binding to R N A

11 17

a

,

Figure 6. Minimized average stereo representation of paromomycin (16) binding to the model 27mer oligoribonucleotide 21 (Scheme 5 ) [99, 1251 as retrieved from the Brookhaven database (Ipbr). The hydrogen atoms are omitted to simplify the representation of the complex. a) Complex with DNA in black and paromomycin (16) as a CPK model with carbon atoms shown in dark grey and oxygenlnitrogen atoms in light grey. b) Part of the same structure with the non-interacting base pairs and the loop removed. Carbon and phosphorous atoms are shown in black, oxygen in light grey and nitrogen in dark grey. Heteroatoms potentially involved in hydrogen bonding are drawn as small balls.

11 18

39 Carbohydrate-Nucleic Acid Interactions

idopyranose) by either simple amines or deaminated idoses revealed the importance of charges for high affinity binding even though flexible amines with identical charge were not fully able to substitute for the hydroxy group bearing and preorganized ring IV [130]. A library approach was chosen by Wong and co-workers to identify new A-site binding molecules based on the modification of ring I of neomycin B (15), containing a core 1,3-hydroxyamine motif [ 131, 1321. Despite various binding affinities (between 10 pM and >500 pM), no molecules were identified containing either a high binding affinity or selectivity for the target A-site model RNA, presumably due to lack of sufficient conformational preorganization. These aminoglycoside antibiotics of the neomycin family act preferentially on prokaryotic organisms. Comparison between ribosomes of either eukaryotic or prokaryotic origin reveals a 10-15-fold higher sensitivity of the latter. The main difference is the substitution of adenosine 1408 in prokaryotes for a guanosine at this position in eukaryotes. The adenosine pair 1408-1491 was shown to be essential for aminoglycoside binding, causing a specific binding pocket that accommodates ring I. As mentioned before, the base pair between positions 1409 and 1491 is essential for aminoglycoside binding as it forms the floor of the antibiotic binding pocket. Disruption of this base pair in prokaryotes leads to aminoglycoside resistance. Higher eukaryotic organisms, including humans, have both disruptions in their cytoplasmic rRNA sequences and as a consequence these aminoglycosides cannot bind the ribosomal target with high affinity. The perfect optimization of a RNA binding ligand is tedious and as experience demonstrates is not that straightforward. Independently, the groups of Gold [ 1331, Joyce [134],and Szostak 113.51 described a new technique that allowed optimization of the RNA (or DNA) binding partner rather than the small molecule ligand. This elegant and powerful approach, often referred to as ‘SELEX’, allows the optimization of certain properties from either a diverse RNA or DNA pool by repeating cycles of affinity selection and amplification of selected nucleic acid pools. Within weeks it became possible to identify optimized DNA or RNA sequences from 1014-1016of initial members with optimized properties not only to bind a variety of different molecules but also for catalyzing specific reactions (for reviews see [ 136-1381). Under conditions of high selection stringency, usually conserved RNA sequences are identified with low nanomolar binding affinities for their target molecule. This in vitro selection process has been applied to optimize RNA sequences for binding neomycin [ 1391, kanamycin [ 1401, and tobramycin (19) [ 1411. For the latter target aminoglycoside, on particular clone (named 56) was isolated coding for a 60-mer oligoribonucleotide insert with a dissociation constant for tobramycin (19) of 0.77 nM [142]. Key structural component of this RNA was shown to be a small stem loop that was incorporated in a 27-mer oligoribonucleotide 22 subjected to structure elucidation by NMR in the tobramycin (19) bound form (Figure 7a, [loo]). This structure is of particular interest as the affinity of tobramycin (19) to this RNA aptamer is roughly three orders of magnitude greater as compared to the ribosomal RNA target site and structural insight should provide the answer of how tighter binding to RNA can be achieved by aminoglycosides. Unfortunately, the resolution of the reported structure is not sufficient to locate specific intermolecular interactions at atomic level with possibly the exception of a

39.3 Carbohydrates Binding to RNA

1119

a

b Figure 7. Structure of tobramycin (19) bound to RNA apatmer 22 [loo]. From the co-ordinates of the 7 structures deposited in the Brookhaven database ( I tob), only the last was chosen to visualize interactions. a) Representation of the whole complex with the RNA displayed in surface view. b) Stick model of the actual binding sites with direct interactions by potential hydrogen bonds highlighted. Carbon and phosphorous atoms are shown in black, oxygen in light grey and nitrogen in dark grey.

1 120

39 Carbohydrate-Nucleic Acid Interactions

hydrogen bond between one of the amino groups at ring I and N-7 of guanosine 9 (Figure 7b). However, tobramycin (19) is well resolved with all three rings adopting a chair conformation connected by well-defined torsion angles. Some of the key features responsible for this specific and tight interaction can be summarized as follows: i) the increased width of the major groove at the binding site is governed by the mismatch between uracils 11 and 16 as part of a well defined and conserved loop that is critical for tobramycin binding; ii) cytosine 15 that is flapping over ring I11 of the bound tobramycin (19) and contributes to the large surface area (52%) that is covered by the RNA ligand in the complex; iii) the defined floor of the major groove that is formed by the edges of base pairs interacting with the ligand; and finally iv) the overall shape complementarity between tobramycin (19) and RNA aptamer 22. The work discussed above clearly indicates the importance of RNA as molecular target for small molecules to specifically (or non-specifically) interfere with crucial biological functions, an aspect that was recently reviewed by Michael and Tor [ 1431. A standard RNA duplex or stem with its wide, shallow minor groove and deep, narrow major groove is not ideally shaped to accommodate a low molecular weight compound. Distortions like bulges or internal loops often create the necessary binding pockets required for the binding event. The modified carbohydrates discussed here greatly helped to understand interactions between this class of molecules and structured RNA at the molecular level. Based on limitations like hydrophilicity and multiple positive charges, however, general designing rules for RNA binding molecules that would be of therapeutic use are difficult to make and the question if neutral molecules might specifically bind a specified target RNA with high affinity remains largely unanswered. First experimental evidence for such an approach are provided by Mei and Czarnik with the identification of neutral molecules disrupting either tat protein-TAR RNA interaction [ 144, 1451 or inhibiting a self-splicing group I intron ribozyme [ 1441. Interestingly, they demonstrated by footprinting techniques that binding is not necessarily competitive and can disrupt protein binding by induction of conformational changes on the RNA level. Additional evidence for the feasibility of neutral molecules to bind RNA with high affinity was provided by the identification of aptamers designed to bind a variety of different molecules. This approach, however, is extremely powerful as ‘perfect’ sequences can be selected from large pools containing roughly l O I 5 individual members and it might prove to be more difficult designing small molecules with drug-like properties for a given RNA target of biological relevance. References 1. R.C. Hawley, L.L. Kiessling, S.L. Schreiber. Proc. Nut1 Acud. Sci. U.S.A. 1989, 86, 11051109.

References 2. 3. 4. 5. 6. 7. 8. 9.

1121

D. Kahne. Chem. Biol. 1995,2, 7-12. J. Hunziker. Chimiu 1996, 50, 248-256. P. Sears, C.H. Wong. Proc. Nutl Acad. Sci. U.S.A. 1996, 93, 12086-12093. P. Sears, C.H. Wong. Chem. Commun. 1998, 1161-1170. M.D. Lee, G.A. Ellestad, D.B. Borders. Arc. Chem. Res. 1991, 24, 235-243. R.A. Kumar, N. Ikemoto, D.J. Patel. J. Mol. Biol. 1997,265, 187-201. R.A. Kumar, N. Ikemoto, D.J. Patel. J. Mol. Biol. 1998,265, 173-186. J. Golik, J. Clardy, G. Dubay, G. Groenewold, H. Kawaguchi, M. Konishi, B. Krishnan, H. 1987, 109, 3461-3462. Ohkuma, K. Saitoh, T.W. Doyle. J. Am. Chem. SOC. 10. J. Golik, G. Dubay, G. Groenewold, H. Kawaguchi, M. Konishi, B. Krishnan, H. Ohkuma, K. Saitoh, T.W. Doyle. J. Am. Chem. Soc. 1987,109, 3462-3464. 11. M.D. Lee, T.S. Dunne, M.M. Siegel, C.C. Chang, G.O. Morton, D.B. Borders. J. Am. Chem. SOC. 1987,109, 3464-3466. 12. M.D. Lee, T.S. Dunne, C.C. Chang, G.A. Ellestad, M.M. Siegel, G.O. Morton, W.J. McGahren, D.B. Borders. J. Am. Chem. Soc. 1987, 109, 3466-3468. 13. S.L. Schreiber, L.L. Kiessling. J. Am. Chern. SOC.1988, 110, 631--633. 14. S.J. Danishefsky, N.B. Mantlo, D.S. Yamashita, G. Schulte. J. Am. Chem. Soc. 1988, 110, 6890-6891. 15. W.L. Jorgensen. Chemtracts: Org. Chem. 1988, I, 485-487. 16. A.S. Kende, C.A. Smith. Tetrahedron Lett. 1988,29,4217-4220. 17. K.C. Nicolaou, Y. Ogawa, G. Zuccarello, E.J. Schweiger, T. Kumazawa. J. Am. Clzem. Soc. 1988, 110, 4866-4868. 18. P. Magnus, P.A. Carter. J. Am. Chem. Soc. 1988, 110, 1626-1628. 19. K. Tomioka, H. Fujita, K. Koga. Tetruhedron Lett. 1989, 30, 851-854. 20. S.M. Kerwin, C.H. Heathcock. Chemtracts: Org. Chem. 1989,2, 21-23. 21. B.H. Long, J. Golik, S. Forenza, B. Ward, R. Rehfuss, J.C. Dabrowiak, J.J. Catino, S.T. Musial, K.W. Brookshire, T.W. Doyle. Proc. Nutl Acud. Sci. U.S.A. 1989, 86, 2-6. 22. Y. Sugiura, Y. Uesawa, Y . Takahashi, J. Kuwahara, J. Golik, T.W. Doyle. Proc. Nut/ Acud. Sci, U.S.A . 1989, 86, 7612-7676. 23. M. Uesugi, Y . Sugiura. Biochemistry 1993, 32, 4622-4627. 24. L. Yu, J. Golik, R. Harrison, P. Dedon. J. Am. Chem. Soc. 1994, 116, 9733-9738. 25. N . Ikemoto, R.A. Kumar, P.C. Dedon, S.J. Danishefsky, D.J. Patel. J. Am. Chem. Soc. 1994, 116, 938779388, 26. J. Drak, N. Iwasawa, S. Danishefsky, D.M. Crothers. Proc. NatlAcad Sci. U S A . 1991, 88, 7464-7468. 27. J. Aiyar, S.J. Danishefsky, D.M. Crothers. J. Am. Chem. Soc. 1992, 114, 7552-1554. 1992, 114, 7555-7557. 28. K.C. Nicolaou, S.C. Tsay, T. Suzuki, G.F. Joyce. J. Am. Chem. SOC. 29. N. Zein, A.M. Sinha, W.J. McGahren, G.A. Ellestad. Science (Washington, D. C.) 1988,240, 1198-1201. 30. N . Zein, M. Poncin, R. Nilakantan, G.A. Ellestad. Science (Washington, D.C.) 1989, 244, 697-699. 31. S. Walker, J. Murnick, D. Kahne. J. Am. Chem. SOC. 1993, 115, 7954-7961. 32. S.L. Walker, A.H. Andreotti, D.E. Kahne. Tetrahedron 1994, 50, 1351-1360. 33. L.G. Paloma, J.A. Smith, W.J. Chazin, K.C. Nicolaou. J. Am. Chem. Soc. 1994, 116, 36973708. 34. G. Bifulco, A. Galeone, L. Gomez-Paloma, K.C. Nicolaou, W.J. Chazin. J. Am. Chem. Soc. 1996, 118, 8817-8824. 35. S. Walker, K.G. Valentine, D. Kahne. J. Am. Chem. SOC. 1990, 112, 6428-6429. 36. S. Walker, D. Yang, D. Kahnc, D. Gange. J. Am. Chenz. Soc. 1991,113,4716-4717. 37. S. Walker, V. Gupta, D. Kahne, D. Gange. J. Am. Chem. Soc. 1994, 116, 3197-3206. 38. T. Li, Z. Zeng, V.A. Estevez, K.U. Baldenius, K.C. Nicolaou, G.F. Joyce. J. Am. Chem. Soc. 1994, 116, 3709-3715. 39. S.N. Ho, S.H. Boyer, S.L. Schreiber, S.J. Danishefsky, G.R. Crabtree. Proc. Nut1 Acad. Sci. U.S.A. 1994, 91, 9203-9207. 40. K.C. Nicolaou, K. Ajito, H. Komatsu, B.M. Smith, T. Li, M.G. Egan, L. Gomez-Paloma. Angew. Chem., Int. Ed. Engl. 1995, 34, 576-518.

1122

39 Carbohydrate-Nucleic Acid Interactions

41. K.C. Nicolaou, K. Ajito, H. Komatsu, B.M. Smith, P. Bertinato, L. Gomez-Paloma. Chem. Commun. 1996, 1495-1496. 42. K.C. Nicolaou, B.M. Smith, J. Pastor, Y. Watanabe, D.S. Weinstein. Synletters 1997, 401410. 43. K.C. Nicolaou, B.M. Smith, K. Ajito, H. Komatsu, L. Gomez-Paloma, Y. Tor. J. Am. Chem. Soc. 1996, 118,2303-2304. 44. G. Bifulco, A. Galeone, K.C. Nicolaou, W.J. Chazin, L. Gomezpaloma. J. Am. Chem. SOC. 1998,120, 7183-7191. 45. C. Liu, B.M. Smith, K. Ajito, H. Komatsu, L. Gomez-Paloma, T. Li, E.A. Theodorakis, K.C. Nicolaou, P.K. Vogt. Proc. Nut1 Acad. Sci. U.S.A. 1996, 93, 940-944. 46. S.B. Shuker, P.J. Hajduk, R.P. Meadows, S.W. Fesik. Science (Washington, D.C.) 1996,274, 1531-1 534. 47. P.J. Hajduk, et al. J. Am. Chem. Soc. 1997,119, 5818-5827. 48. P.J. Hajduk, R.P. Meadows, S.W. Fesik. Science (Washington, D.C.) 1997, 278, 497--499. 49. J.W. Lown. Chem. Soc. Rev. 1993,22, 165-176. 50. W.R. Strohl, M.L. Dickens, V.B. Rajgarhia, A.J. Woo, N.D. Priestley. Drugs Pharm. Sci. 1997, 82, 577-657. 51. F.M. Arcamone. Biochimie 1998, 80, 201-206. 52. A.H. Wang. Adv. DNA Sequence Specific Agents 1996,2, 59-100. 53. F. Arcamone, G. Franceschi, P. Orezzi, S. Penco, R. Mondelli. Tetrahedron Lett. 1968, 33493352. 54. R.H. Iwamoto, P. Lim, N.S. Bhacca. Tetrahedron Lett. 1968, 3891-3894. 55. F. Arcamone, G. Franceschi, S. Penco, A. Selva. Tetrahedron Lett. 1969, 1007-1010. 56. P.B. Jensen, B.S. Soerensen, M. Sephested, E.J. Demant, E. Kjeldsen, E. Friche, H.H. Hansen. Biochem. Pharmacol. 1993, 45, 2025-2035. 57. G.J. Quigley, A.H. Wang, G. Ughetto, G. van der Marel, J.H. van Boom, A. Rich. Proc. Natl Acad. Sci. U.S.A. 1980, 77, 7204-7208. 58. A.H. Wang, G. Ughetto, G.J. Quigley, A. Rich. Biochemistry 1987, 26, 1152-1163. 59. M.H. Moore, W.N. Hunter, D. Langlois, 0. Kennard. J. Mol. Biol. 1989,206, 693-705. 60. G.A. Leonard, T.W. Hambley, K. McAuley-Hecht, T. Brown, W.N. Hunter. Acta Crystallogr., Sect. D: Biol. Crystallogr. 1993, 049, 458-467. 61. Y. Gao, H. Robinson, E.R. Wijsman, G. van der Marel, J.H. van Boom, A.H. Wang. J. Am. Chem. SOC.1997,119, 1496-1497. 62. S. Neidle, G. Taylor. Biochim. Biophys. Acta 1977, 479, 450-459. 63. J.B. Chaires, J.E. Herrera, M.J. Waring. Biochemistry 1990, 29, 6145-6153. 64. C.J. Shelton, M.M. Harding, A S . Prakash. Biochemistry 1996, 35, 7974-7982. 65. K.R. Fox. Anti-Cancer Drug Des. 1988, 3, 157-168. 66. S. Kunimoto, Y. Takahashi, T. Uchidsa, T. Takeuchi, H. Umezawa. J. Antibiot. 1988, 41, 655-659. 67. J.B. Chaires, F. Leng, T. Przewloka, I. Fokt, Y.H. Ling, R. Perez-Soler, W. Priebe. J. Med. Chem. 1997, 40, 261-266. 68. G.G. Hu, X. Shui, F. Leng, W. Priebe, J.B. Chaires, L.D. Williams. Biochemistry 1997, 36, 5940-5946. 69. H. Robinson, W. Priebe, J.B. Chaires, A.H. Wang. Biochemistry 1997, 36, 8663-8670. 70. K.M. Depew, S.M. Zeman, S.H. Boyer, D.J. Denhart, N. Ikemoto, S.J. Danishefsky, D.M. Crothers. Angew. Chem., Int. Ed. Engl. 1997: 35, 2797--2800. 71. R.P. Hertzberg, P.B. Dervan. Biochemistry 1984, 23, 3934-3945. 72. P.F. Wiley, F.A. MacKellar, E.L. Caron, R.B. Kelly. Tetrahedron Lett. 1968, 663-668. 73. M.S. Searle, J.G. Hall, W.A. Denny, L.P. Wakelin. Biochemistry 1988, 27, 4340-4349. 74. Y.C. Liaw, Y.G. Gao, H. Robinson, G. van der Marel, J.H. van Boom, A.H. Wang. Biochemistry 1989, 28, 9913-9918. 75. X. Zhang, D.J. Patel. Biochemistry 1990, 29, 9451-9466. 76. L.D. Williams, M. Egli, Q. Gao, P. Bash, G. van der Marel, J.H. van Boom, A. Rich, C.A. Frederick. Proc. Nut1 Acad. Sci. U.S.A. 1990, 87, 2225-2229. 77. Y.G. Gao, Y.C. Liaw, H. Robinson, A.H. Wang. Biochemistry 1990,29, 10307-10316. 78. M. Egli, L.D. Williams, C.A. Frederick, A. Rich. Biochemistry 1991, 30, 1364-1372.

References

1123

H. Robinson, D. Yang, A.H. Wang. Gene 1994, 149, 179-188. C.K. Smith, G.J. Davies, E.J. Dodson, M.H. Moore. Biochemistry 1995, 34, 415-425. C.K. Smith, J.A. Brannigan, M.H. Moore. J. Mol. Biol. 1996,263, 237-258. S.K. Arora. J. Am. Chem. Soc. 1983, 105, 1328-1332. M.R. Hansen, L.H. Hurley. Acc. Chem. Res. 1996,29,249-258. W. Kersten, H. Kersten, F.E. Steiner, B. Emmerich. Z. Physiol. Chem. 1967,348, 1415-1423. W.E. Mueller. Pharmacol. Ther., Part A 1977, I , 451-474. Ward, D. C., Reich, E., and Goldberg, Irving H. Science (Washington, D.C.) 1965, 12591263. 87. M.W. Van Dyke, P.B. Dervan. Biochemistry 1983,22, 2373-2377. 88. K.R. Fox, N.R. Howarth. Nucleic Acid3 Res. 1985, 13, 8695-8714. 89. B.M. Cons, K.R. Fox. Nucleic Acids Res. 1989, 17, 5441-5459. 90. A. Stankus, J. Goodisman, J.C. Dabrowiak. Biochemistry 1992, 31, 9310-9318. 91. X. Gao, P. Mirau, D.J. Patel. J. Mol. Bid. 1992, 223, 259-279. 92. D.L. Banville, M.A. Keniry, M. Kam, R.H. Shafer. Biochemistry 1990, 29, 6521-6534. 93. J.G. Pelton, D.E. Wemmer. Proc. Nut1 Acad. Sci. U.S.A. 1989, 86, 5723-5727. 94. C.L. Kielkopf, S. White, J.W. Sewczyk, J.M. Turner, E.E. Baird, P.B. Dervan, D.C. Rees. Science ( Washington, D.C.) 1998, 282, 111-1 15. 95. K.D. Rice, R.D. Tanaka, B.A. Katz, R.P. Numerof, W.R. Moore. Curr. Pharm. Des. 1998,4, 381-396. 96. D. Moazed, H.F. Noller. Nature (London) 1987, 327, 389-394. 97. J. Woodcock, D. Moazed, M. Cannon, J. Davies, H.F. Noller. EMBO J. 1991, 10, 30993103. 98. E.V. Puglisi, J.D. Puglisi. Mod. Cell Bid. 1997, 17, 1-21. 99. D. Fourmy, M.I. Recht, S.C. Blanchard, J.D. Puglisi. Science (Washington, D.C. ) 1996,274, 1367-1371. 100. L. Jiang, A.K. Suri, R. Fiala, D.J. Patel. Chem. Biol. 1997, 4, 35-50. 101. U. von Ahsen, J. Davies, R. Schroeder. J. Mol. Bid. 1992, 226, 935-941. 102. U. von Ahsen, H.F. Noller. Science (Washington, 0 . C . ) 1993,260, 1500-1503. 103. T.K. Stage, K.J. Hertel, O.C. Uhlenbeck. R N A 1995, I , 95-101. 104. H. Wang, Y. Tor. Angew. Chem., Znt. Ed. 1998,37, 109-111. 105. H. Wang, Y. Tor. J. Am. Chem. Soc. 1997, 119, 8734-8735. 106. T. Hermann, E. Westhof. J. Mol. Bid. 1998, 276, 903-912. 107. M.L. Zapp, S. Stern, M.R. Green. Cell (Cambridge, Muss.) 1993, 74, 969-978. 108. M. Hendrix, E.S. Priestley. G.F. Joyce, C.H. Wong. J. Am. Chem. Soc. 1997,119, 3641-3648. 109. S.H. Wang, P.W. Huber, M. Cui, A.W. Czarnik, H.Y. Mei. Biochemistry 1998, 37, 55495557. 110. J.S. Chia, H.L. Wu, H.W. Wang, D.S. Chen, P.J. Chen. J. Biomed. Sci. (Basel) 1997, 4, 208216. 11 1. Puglisi, Joseph D. ‘Structural basis foy aminoglycoside antibiotic action’ in Many Faces RNA, (8th SmithKline Beecham Pharm. Res. Symp.), ed. by D.S. Eggleston, Ed. Academic Press, San Diego, CA 1998, p. 97. 112. Schroeder, U. von Ahsen. Nucleic Acid,? Mol. Biol. 1996, 10, 53-74. 1 13. Wank, Herbert and Schroeder, Renee. ‘Antibiotics that interfere with R N A J R N A interactions’ in Ribosomal RNA Group I Introns, ed. by R. Green and R. Schroeder, Eds. Landes, Austin, TX 1996, p. 129. 114. C. Bochaton, S. Rochegude, R. Roubille. Lyon Pharm. 1997, 48, 226-239. 115. P. Chakrabarti. Proc. Natl Acad. Sci., India, Sect. B 1997, 67, 169-179. 116. P.E. Montandon, R. Wagner, E. Stutz. EMBO J. 1986,5, 3705-3708. 117. P.E. Montandon, P. Nicolas, P. Schuermann, E. Stutz. Nucleic Acid.y Res. 1985, 13, 42994310. 118. M. Li, A. Tzagoloff, K. Underbrink-Lyon, N.C. Martin. J. Biol. Chem. 1982,257, 5921-5928. 119. E.A. Spangler, E.H. Blackburn. J. Bid. Chem. 1985,260, 6334-6340. 120. M. Li, A. Tzagoloff, K. Underbrink-Lyon, N.C. Martin. J. Biol. Chem. 1982,257, 5921-5928. 121. A.A. Beauclerk, E. Cundliffe. J. Mol. Bid. 1987, 193, 661-671. 122. P. Purohit, S. Stern. Nature (London) 1994,370, 659-662. 79. 80. 81. 82. 83. 84. 85. 86.

1 124

39 Carbohydrate-Nucleic Acid Interactions

123. M.I. Recht, D. Fourmy, S.C. Blanchard, K.D. Dahlquist, J.D. Puglisi. J. Mol. Biol. 1996, 262, 421-436. 124. H. Miyaguchi, H. Narita, K. Sakamoto, S. Yokoyama. Nucleic Acids Res. 1996, 24, 37003706. 125. D. Fourmy, M.I. Recht, J.D. Puglisi. J. Mol. Biol. 1998,277, 347-362. 126. D. Fourmy, S. Yoshizawa, J.D. Puglisi. J. Mol. Bid. 1998, 277, 333-345. 127. R. Benveniste, J. Davies. Antimicrob. Agents Chemother. 1973, 4, 402-409. 128. S.C. Blanchard, D. Fourmy, R.G. Eason, J.D. Puglisi. Biochemistry 1998, 37, 7716-7724. 129. C.H. Wong, M. Hendrix, E.S. Priestley, W.A. Greenberg. Chem. Biol. 1998, 5, 397-406. 130. P.B. Alper, M. Hendrix, P. Sears, C.H. Wong. J. Am. Chem. Soc. 1998, 120, 1965-1978. 131. M. Hendrix, P.B. Alper, E.S. Priestley, C.H. Wong. Angew. Chem., Int. Ed. Engl. 1997, 36, 95-98. 132. C.H. Wong, M. Hendrix, D.D. Manning, C. Rosenbohm, W.A. Greenberg. J. Am. Chem. SOC. 1998, 120, 8319-8327. 133. C. Tuerk, L. Gold. Science (Washington, D.C.) 1990,249: 505-510. 134. D.L. Robertson, G.F. Joyce. Nature (London) 1990,344,467-468. 135. A.D. Ellington, J.W. Szostak. Nature (London) 1990,346, 818-822. 136. G.F. Joyce. Curr. Opin. Struct. Bid. 1994, 4, 331-336. 137. L. Gold, B. Polisky, 0. Uhlenbeck, M. Yarus. Annu. Rev. Biochem. 1995,64, 763-797. 138. J.R. Lorsch, J.W. Szostak. Ace. Chem. Res. 1996,29, 103-110. 139. M.G. Wallis, U. von Ahsen, R. Schroeder, M. Famulok. Chem. Biol. 1995,2, 543-552. 140. S.M. Lato, A.R. Boles, A.D. Ellington. Chem. Bid. 1995,2, 291-303. 141. Y. Wang, R.R. Rando. Chem. Biol. 1995,2, 281-290. 142. Y. Wang, J. Killian, K. Hamasaki, R.R. Rando. Biochemistry 1996, 35, 12338-12346. 143. K. Michael, Y. Tor. Chem. Eur. J. 1998, 4, 2091-2098. 144. H.Y. Mei, M. Cui, S.M. Lemrow, A.W. Czarnik. Bioorg. Med. Chem. 1997, 5, 1185-1195. 145. H.Y. Mei, M. Cui, A. Heldsinger, S.M. Lemrow, J.A. Loo, K.A. Sannes-Lowery, L. Sharmeen, A.W. Czarnik. Biochemistry 1998, 37, 14204-14212.

Part I1 Volume 3

i

Biosynthesis of Glycoconjugates

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

1 Metabolism of Sugars and Sugar Nucleotides Hudson H. Freeze

1.1 Introduction Glycoconjugate biosynthesis requires activation of monosaccharides to sugar nucleotides. Specific glycosyltransferases then donate the activated sugars to various acceptors. To achieve this, monosaccharides must be imported into the cell, salvaged from degraded glycoconjugates or derived from other monosaccharides within the cell. Most glycosylations occur in the lumen of the ER or Golgi, but monosaccharide metabolism occurs in the cytoplasm. Sugar nucleotide specific transporters carry the activated donors into the Golgi. This chapter explores various pathways leading to the synthesis and delivery of precursors for glycosylation. Most of the biosynthetic reactions are covered in standard biochemistry texts, but this chapter will point out some recent observations suggesting that salvage pathways make substantial contributions to glycoconjugate synthesis. This bedrock of biochemistry may have undiscovered layers.

1.2 Basic Principles In 1951 Nobel Laureate L.F. Leloir showed that a nucleotide triphosphate such as UTP reacts with a glycosyl-1-P forming a high energy donor for glycoconjugate synthesis [l]. There are a few variations on the theme, but all sugars must either be activated by a kinase (reaction 1 below) or generated from a previously synthesized activated sugar nucleotide (reactions 2 and 3 below):

-

NTP PPi Sugar + ATP +Sugar-P

L?

Sugar-NDP

4

1 Metabolism of Sugars and Sugar Nucleotides

Sugar(A)-NDP Sugar(A)-NDP

+

Sugar(B)-NDP Sugar(B)-l-P Sugar(B)NDP

+ Sugar(A)-l-P

(2) (3)

All the donors are dinucleotides, except for sialic acid which is activated as a mononucleotide, CMP-Sia. Iduronic acid does not have a sugar nucleotide parent since it is formed by epimerization of glucuronic acid after this sugar is incorporated into the glycosaminoglycan chains. Some reactions use uncommon nucleotides as donors which are formed from the usual ones by a nucleotide exchange reaction (reaction 3 above).

1.3 Transporters Deliver Monosaccharides to Cells Carbohydrate degrading enzymes (such as a-amylase, disaccharidases, etc.) are secreted into the digestive tract or localized to the brush border of intestinal epithelial cells. These enzymes generate monosaccharides for sugar transporters located at the plasma membrane. Glucose and fructose account for most of the monosaccharide traffic in mammals. Glucose is transported from the digestive tract using energydependent sodium-glucose transporters, (SGLT) found on the intestinal epithelial cell surface. This transporter is also found on kidney tubules where it reabsorbs glucose from the kidney filtrates. Their K,s for glucose are < 1 mM [2]. Hexoses enter other cells via a set of facilitated diffusion transporters that mostly transport glucose (hence they are called GLUT transporters) 13-51. K,s for glucose are in the 2-20 mM range. Fructose is transported by one of the GLUT family members, GLUT 5. Even though the GLUT transporters can recognize other hexoses such as mannose and galactose, only glucose or fructose are in sufficiently high concentrations to be physiologically relevant. Other sugar-specific transporters occur in higher organisms and they may have physiological importance. Two types of mannose transporters are known [6-111. One is an energy-dependent transporter analogous to the SGLT variety used for glucose and located on the apical surface of enterocytes and kidney tubule epithelial cells [7, 9, 111. It presumably transports mannose liberated from digested macromolecules and retrieves mannose from the kidney filtrates, analogous to SGLT used for glucose. The second type of transporter is a highly mannose-specific, energyindependent transporter found on the surface of many types of mammalian cells [6, 8, 101. It is practically insensitive to inhibition by glucose and its is about 50 pM, which is near the average concentration of mannose in the blood of mammals (average 75 pM, range 30-120 pM). Following intravenous injection in rats, mannose is primarily taken up by the liver and intestine [ 101. Hepatoma cells and fibroblasts preferentially use mannose for up to 80% of their glycoprotein synthesis when given physiological concentrations of both mannose (50 pM) and glucose (5 mM) [lo, 121. Direct conversion from glucose (see below) is a relatively minor pathway in these cells. This is quite striking considering the 100-fold molar excess of

1.4 Intracellulur Sources of Sugars

5

glucose in the blood. The relative contributions of mannose and glucose in different tissues and organs in intact animals is unknown. A fucose-specific transporter [ 131 has also been reported in mammalian cells. The Kuptakeis -250 pM, which is much higher than the fucose concentration in human blood (4pM). We do not know if this transporter is physiologically important, but much of what is taken up can be converted into GDP-Fuc and incorporated into glycoproteins. In cell culture, most GDP-Fuc is thought to be derived from GDP-Man [ 141 (see below). No cell surface transporters specific for galactose, amino sugars, or sialic acids are known in higher organisms, but about 25%) of radiolabeled glucosamine injected into the peritoneum of rats or rabbits is incorporated into serum glycoproteins as GlcNAc and sialic acid [15].

1.4 Intracellular Sources of Sugars 1.4.1 Salvage

Monosaccharides can also be salvaged from glycoconjugates degraded within the same cell. Salvage pathways have not been systematically explored, however, their contribution to glycosylation may be quite substantial (see Table 1). For instance, -80'%1 of radiolabeled GlcNAc derived from a glycoprotein degraded in liver lysosomes is converted into UDP-GlcNAc [ 161. At least a third of that amount is used to synthesize secreted glycoproteins. Also, glycoconjugates endocytosed by fibroblasts are degraded in the lysosomes and about 50% of the amino sugars are used for new glycoprotein synthesis [ 171. Thyroid glands blocked in de novo synthesis of UDP-GlcNAc from fructose-6-P degrade thyroglobulin and use the liberated GlcNAc to maintain protein glycosylation [ 181. Salvage is not limited to GlcNAc. Glycosylation-impaired mutant CHO cells (fdl-D)that require supplements of Gal and GalNAc for normal 0-glycosylation can endocytose and degrade glycoproteins from the 5-10% fetal bovine serum in the culture medium, and use the liberated Table 1. Evidence for the importance of monosaccharide salvage pathways. Amino Sugars * More than 50'1/0of label reused for new glycoconjugates (GalNAc, GlcNAc) * Approximately 70% of 3H-GlcNAc incorporated into sugar nucleotides in rat liver * 15-909'0 Sialic acids recycled in some cells * GlcNAc/ManNAc kinase present in many tissues * GalNAc-I-kinase distinct from Gal-1-kinase Hexose and Fucose * Mannose transporter uses physiological concentrations of Man * Cells lacking UDP-Gal/GalNAc-4-epimerase can use serum glycoproteins for Gal and GalNAc * Fucose transporter identified * Correction of GDP-Man-GDP-Fuc deficiency with exogenous Fuc

6

I Metabolism of’ Sugars and Sugar Nucleotides

monosaccharides for glycosylation [ 19, 201. It is likely that 30-900/0 of sialic acids released in the lysosome can be reutilized for new glycoconjugate synthesis [21, 221. Since glycoconjugate degradation occurs in the lysosome, and monosaccharide activation in the cytosol, reutilization of liberated monosaccharides requires their exit from the lysosome. To do this, the lysosome has separate carriers for neutral hexoses, N-acetylated amino sugars, and acidic hexoses [23]. The neutral hexose carrier has a K, of 50-75 mM and also transports fucose and xylose [23]. Although these sugars could diffuse through the lysosomal membrane, their efflux rate may not be high enough to keep up with synthetic demands. The N-acetylhexosamine carrier with a 4.4 mM K, cannot use non-acetylated amino sugars. Sialic and glucuronic acids will not diffuse out of the lysosome. Their carrier has a relatively low K, (-300-550 pM) to prevent accumulation of these sugars in the lysosome [23]. Once the sugars reach the cytoplasm, they can be activated to sugar nucleotide donors as described below. Is sugar transport from the lysosome associated with sugar activation? This has not been demonstrated, but a small amount of GlcNAc kinase is found in lysosomal preparations from rat liver [24]. The quantitative contributions of exogenous sugar, salvage pathways, and interconversions probably vary with the cell type and amount of glycoproteins they synthesize. This is clearly the situation seen in the case of CMP-sialic acid synthesis described later in this chapter. Most of the work on monosaccharide and sugar nucleotide metabolism centers on the direct activation or interconversions of various sugars, regardless of their source. Nearly all cell culture studies supply growth media that contains only glucose and 10-20% serum. It is usually assumed that glucose generates all of the other monosaccharides. Certainly, it can, but does it normally do so? Metabolic radiolabeling with glucose is inefficient, in part because most glucose enters the glycolytic pathway, but few experiments have directly assessed the amount of glucose converted to other monosaccharides and then into various glycans [12, 25, 261. The relative contributions of diet, salvage, and various interconversions is really not known [25-271. Recent studies suggest some key biosynthetic enzymes are quite restricted in their distribution in different organs [28]. Other studies show that the clinical conditions of patients with a human glycosylation disorder (Carbohydrate Deficient Glycoprotein Syndrome, CDGS) can be improved substantially by providing the missing monosaccharides as dietary supplements [29, 301. Clearly, these pathways can play an important role in maintaining health.

1.4.2 Activation and Interconversion of Monosaccharides The major pathways for sugar activation and monosaccharide interconversions are shown in Figure 1. Glycogen Glycogen is the major storage polysaccharide in animal cells. This mega-dalton-size molecule contains up to 100,000 glucose units, arranged in Glcal,4Glc repeating

7

1.4 Intracellukar Sources of Sugars

Tm- t v

Gal-1-P

GlycTn

$,

UTPI UDP-Glc

inir

G-;P il J,

Glycolysis

-

I

I

I

I 2,Fry-6-P I

-

-J

$131ACC~A

ATP

Man-6-P J,

ATP F :-1-P

Man-I-P

I GTP

GTP

GIcNAc-6-P

IUTP

bMaANAc

/

ManNAc-6-P

1

NADP

PEP

*-

Figure I . Biosynthesis and interconversion of monosaccharides. The relative contributions of each and (-donors; =-monosaccharides; under physiological conditions are unknown. control points (Used with the permission of the Consortium of Glycobiology Editors).

a

disaccharides with periodic ul,6Glc branches. Glycogen synthesis and degradation (glycogenolysis) are highly regulated for energy utilization. Some of the glucose used to synthesize other sugars may be derived from glycogen as well as from glucose transported into the cell. Glycogen is synthesized by addition of single glucose units from UDP-Glc, and it is degraded by glycogen phosphorylase. The non-ATP dependent reaction forms Glc-1-P which can be used directly to form UDP-Glc or epimerized to Glc-6-P for glycolysis. Glycogen is actually a glycoprotein. The core protein, called glycogenin, comprises < 1% of the mass of the molecule and has the unusual property of being selfglucosylating. Up to seven glucose residues are polymerized on a single Tyr (Tyr 194) of the protein. Glycogen synthase extends the primed glycogenin into huge glycogen molecules [31-33].

Glucose Glucose is convertible into all other monosaccharides (Figure 1). Hexokinase begins the glycolytic pathway generating Glc-6-P which can then either form Fru-6-P by phosphoglucose isomerase or Glc- 1-P by phosphoglucomutase. Reaction of Glc-1-P

8

I Metabolism of Sugars and Sugar Nucleotides

with UTP generates UDP-Glc. The large UDP-Glc pool is used for glycogen synthesis or glucose-containing molecules such as glucosylceramide and dolichol-P-Glc in N-linked oligosaccharide biosynthesis. Some proteins with EGF-like modules contain P-Glucose in 0-linkage to Ser/Thr [34]. Glucuronic acid Oxidation of UDP-Glc to UDP-glucuronic acid requires NAD. This acidic sugar is used primarily for glycosaminoglycan biosynthesis. Its importance is emphasized in developmental mutants of Drosophila (sugarless, sgl) that are defective in this enzyme [35-38]. UDP-GlcA is used to add GlcA to N - and 0-linked oligosaccharides and glycolipids, for the glucuronidation of bile acids [39-411 and detoxification of xenobiotic compounds in the ER [42]. Some evidence sugests that UDP-GlcA is derived from glycogenolysis rather than gluconeogenesis [43]. Iduronic acid This uronic acid (IdoA), the 5-epimer of GlcA, is found only in glycosaminoglycans. IdoA is not directly synthesized from a sugar nucleotide donor. Instead, it arises by epimerization of GlcA after its incorporation into the GAG chain. Xylose Decarboxylation of UDP-GlcA produces UDP-Xyl which is used primarily to initiate glycosaminoglycan synthesis, but it is also found as an extension to the 0-PGlucose residues found on some proteins with EGF-like modules [34]. The decarboxylation reaction that generates UDP-Xyl appears to occur in the lumen of the ER or early in the Golgi, as well as in the cytoplasm [44, 451. Mannose Mannose is a key sugar used for N-linked oligosaccharide and GPI-anchor synthesis. GDP-Man can be formed in two ways. First, directly from mannose beginning with hexokinase, or in some invertebrates, by a mannokinase [46].The second, and by far the best known way, is by conversion of Fru-6-P to Man-6-P through the enzyme phosphomannose isomerase (Figure 1). Genetic loss of a majority of this enzyme produces a potentially fatal human disease called CDGS (Type lb) [29, 30, 47, 481. However, the defect can be rescued by providing the patient with oral mannose supplements. In yeast, loss of PMI is lethal, but it can also be rescued with exogenous mannose [49]. Mannose is essential for glycosylation, but at least for bees, too much mannose is lethal [ 501. This curious phenomenon, called the “honeybee syndrome”, occurs when bees feed on mannose instead of sucrose or glucose. When mannose is the sole energy source for the bees, it serves as an ATP sink. Hexokinase converts mannose to Man-6-P, but further metabolism via glycolysis requires conversion to Fru-6-P. Honeybees have relatively low PMI activity compared to hexokinases which creates a bottleneck, and excess Man-6-P accumulates. It is quickly degraded by phospha-

1.4 Intracellular Sources of' Sugars

9

tase to free mannose, and again phosphorylated with the ever-diminishing supply of ATP. Very high concentrations of mannose are teratogenic in rats for a similar reason [51, 521. Early in development, the embryo relies more on glycolysis than on oxidative phosphorylation. At high concentrations, mannose enters cells via the typical glucose transporters. but once inside, the low PMI activity again creates a bottleneck in energy production, causing abnormal or arrested development. After Man-6-P is formed, it is converted to Man-1-P by phosphomannomutase. Two genes, P M M l and PMM2, produce two different isozymes in humans. Loss of PMM2 produces another form of CDGS (Type la) that results from underglycosylation of proteins [53, 541. Depletion of Man-6-P or Man-1-P limits the formation of GDP-Man and Dol-P-Man needed for lipid-linked oligosaccharide synthesis. Dol-P-Man is also the donor for synthesis of glycophospholipid anchors and the recently identified C-mannosylation reaction [55]. Another type of human Congenital disorder of glycosylation (CDG) results from defects in Dol-P-Man synthesis [56].

Fucose GDP-Fucose can be derived from GDP-Man by oxidation of the 6-OH of Mannose to CH3. This reaction is catalyzed by a three-step reaction that uses two enzymes. In the first step, the C-4 Man of GDP-Man is oxidized to a ketone, GDP-4-dehydro-6deoxy-mannose by the enzyme GDP-Man 4,6-dehydratase along with the reduction of NADP to NADPH. A single polypeptide called Fx catalyzes the next two reactions. GDP-4-keto-6-deoxy-mannose is first epimerized at C-3 and C-5 to form GDP-4-keto-6-deoxyglucose and then reduced with NADPH at C-4 to form GDPFuc [57-591. F x is well-conserved from bacteria and mammals. The first dehydration step is feedback inhibited by GDP-Fuc [60]. Free fucose can also be converted to GDP-Fuc after it is first converted to Fuc-I-P. Some mutant cell lines that cannot convert GDP-Man to GDP-Fuc form underfucosylated proteins [ 141, but the block can be bypassed with exogenous fucose. The human glycosylation disease called Leukocyte Adhesion Deficiency Type-I1 (LADII) may result from reduced GDPMan+GDP-Fuc conversion [61] or reduced utilization of GDP-Fuc. Regardless of the actual mechanism, underfucosylated glycoproteins result. These include sialyl Lewis' glycans that are required for selectin-mediated leukocyte rolling and extravasation into inflamed tissues [62]. Patients with LADII have chronic infections since they cannot effectively extravasate leukocytes to the sites of inflammation. They require constant antibiotic treatment. In addition, these patients show mental and psychomotor retardation and a severe failure to thrive. Fucose can be transported by mammalian cells, as mentioned above, and then converted to Fucose-I-P by a fucose kinase [63]. GDP-Fuc synthase generates the activated donor. The quantitative contribution of this fucose salvage pathway under normal conditions is not known, but one LADII patient was treated with oral fucose supplements and showed dramatic improvements in his clinical condition after only a few months of therapy [64]. Fucose was absorbed well. neutrophil levels returned to normal, infections ceased, antibiotics were discontinuted, and having a medically

10

I Metabolism of Sugars and Sugar Nucleotides

stable base, the patient showed significant psychomotor improvement. Clearly, there is much yet to learn about the importance of direct utilization of fucose by higher mammals. In pigs, fucose can also be fully oxidized by the liver, but this catabolic pathway is incomplete in rats [65]. Galactose

Galactose is activated to UDP-Gal in several ways. First, by direct phosphorylation to give Gal-1-P that reacts with UTP to form UDP-Gal. Alternatively, Gal-1-P is converted to UDP-Gal via a uridyl transferase exchange reaction with UDP-Glc displacing Glc-1-P. A deficiency in the uridyl transferase activity results in a severe human disease called galactosemia, leading to mental retardation, liver damage and eventual death [66]. Finally, UDP-Gal can be formed from UDP-Glc by the NADdependent reaction catalyzed by UDP-galactose 4-epimerase [67]. The 4-OH group is first converted to a 4-keto derivative with the formation of NADH from NAD. In the next step, NAD reforms and the hydroxyl group is regenerated producing one of the two 4-epimers. The same enzyme converts UDP-GalNAc to UDP-GlcNAc. N-Acetylglucosamine

Synthesis of UDP-GlcNAc begins with the formation of glucosamine-6-P from fru6-P by transamidation using glutamine as the -NH2 donor. Glucosamine-6-P is N acetylated via acetyl-CoA to form GlcNAc-6-P and then isomerized to GlcNAc-1-P via a 1,6 bis-phosphate intermediate. Similar to the other activation reactions, GlcNAc-1-P + UTP then yields UDP-GlcNAc and pyrophosphate. Alternatively, GlcNAc can be directly phosphorylated to form GlcNAc-6-P using a GlcNAc/ ManNAc kinase [24]. A mutase then converts GlcNAc-6-P to GlcNAc-1-P. This route may account for the efficient salvage of GlcNAc from lysosomal degradation. GlcN can be very efficiently used for liver glycoprotein synthesis in intact animals, showing that direct phosphorylation can be important [15]. Not all GlcN is necessarily used for glycoprotein synthesis, since it depends on the cell type. For instance, in high energy requiring tissues such as neurons and transporting epithelial cells in the kidney and intestine, glucosamine-6-P deaminase (GlcN-6-P- - - - - ->Fru-6P + NH4) (GNPDA) generates Fru-6-P for glycolysis [68]. This is not simply the reverse of the biosynthetic transamidation reaction which uses glutamine as a donor. The GNPDA reaction proceeds in the forward direction, since the concentration of NH4 required for GlcNH2 formation would be extremely high. This enzyme is essentially absent from liver where other sources of energy are available. Thus, UDP-GlcNAc can be formed from Fru-6-P or from GlcNAc via kinase. Separate pools of UDP-GlcNAc may arise for GalN and GlcN [69]. N-Acetylgalactosamine

UDP-GalNAc can arise from two routes. One is the direct reaction of GalNAc-1-P with UTP. GalNAc-1-P is formed by a specific kinase that probably uses ATP, but can also use ITP [70]. This enzyme is distinct from galactose-1-kinase. UDPGalNAc can also be formed by epimerization of UDP-GlcNAc using the same NAD-dependent epimerase that converts UDP-Glc to UDP-Gal [ 191.

1.5 Suyur Nucleotide Trunsportrrs

11

Sialic acids Sialic acid is the generic name given to a group of more than 30 different variations of the parent compound. The modifications include oxidation, single and multiple acetylations, sulfation, and methylation. Except for the formation of the glycolyl derivative as an activated sugar nucleotide, all the other modifications of sialic acid probably occur in the Golgi after transfer of the sialic acid to the acceptor glycan [71, 721. CMP-N-acetyl (or glycolyl) neuraminic acid is the immediate donor. It has a complicated de novo biosynthetic pathway that begins with UDP-GlcNAc which is converted to N-acetylmannosamine by a single enzyme with two catalytic activities. First, is the 2-epimerization and cleavage of the UDP to yield N-acetylmannosamine. Next, this same polypeptide uses ATP as a donor forming N-acetylmannosamine-6P. In the next step, this compound condenses with phosphoenolpyruvate forming Nacetylneuraminic acid 9-P. Phosphate is removed and reaction with CTP yields CMP-N-acetylneuraminic acid PPi. All the steps, except the last one, occur in the cytosol. CMP-Sialic acid is formed in the nucleus with subsequent export to the cytoplasm [71]. This is the only monosaccharide donor made in the nucleus. The reason for this unusual localization is not known. There is an alternate biosynthetic pathway leading to sialic acid. The epimerase/ kinase enzyme activity can only be detected in liver, salivary gland and intestinal mucosa [73]. The mRNA can only be detected in the liver [28]. It is clear that many other tissues contain sialylated glycoproteins and glycolipds, making it likely that other pathways exist. Salvage of sialic acids from other proteins is extremely efficient in some cells and may explain the absence of epimerase/kinase. Another possibility is that a widely distributed 2-epimerase converts GlcNAc to ManNAc. No ManNAc specific kinase has been described, but the GlcNAc kinase can also use ManNAc to form ManNAc-6-P although the K, for ManNAc (0.9 mM) is about 15-fold higher than GlcNAc (0.06 mM) [74].

+

1.5 Sugar Nucleotide Transporters Sugar nucleotides synthesized in the cytosol (or nucleus for CMP-sialic acid) are carried into the ER and Golgi using sugar nucleotide transporters, actually antiporters, that deliver sugar nucleotides into the lumen of these organelles. This occurs with simultaneous exit of nucleotide monophosphates which must first be derived from the nucleotide diphosphates. K , of the transporters range from 1-10 pM. Using in vitro systems, the transporters have been shown to increase the concentration of the sugar nucleotides within the Golgi lumen by 10-50 fold. This is usually sufficient to reach or exceed the calculated K, of sugar nucleotides for most glycosyltransferases [75, 761. Most of the antiporters are found in the Golgi, but some also occur in the ER [75, 761. They are organelle specific and their location usually corresponds to the location of the known glycosyltransferases. Their import into the Golgi is not energy

12

I Metabolism of Sugars and Sugar Nucleotides

dependent or affected by ionophores, however they are competitively inhibited by the corresponding nucleoside mono- and di-phosphates, but not by the monosaccharides. In addition to sugar nucleotide-specific transporters, separate transporters exist for ATP and PAPS, the donor for protein and carbohydrate sulfation. Glucuronidation of bile and xenobiotic compounds in the ER [41, 421 is consistent with the presence of the UDP-GlcA transporter in the ER, and the identification of reglycosylation of misfolded glycoproteins in the ER explains the need for an ER UDP-Glc transporter 1771. The ER UDP-Xyl transporter may be needed for initiating GAG chain synthesis there 144, 451. The existence of UDP-GlcNAc, and UDP-GalNAc transporters in the ER is more difficult to explain. For most glycosylation reactions, the sugar nucleotide donates the sugar residue resulting in the formation of nucleoside diphosphate, which must be converted into a monophosphate by the nucleoside diphosphatase that occurs in the Golgi lumen. In yeast a GDPase converts GDP to GMP. Disrupting the gene that codes for this Type I1 membrane protein reduces glycosylation of all the mannosylated glycoconjugates 1781. The antiporter system has the advantage of coupling the rate of sugar nucleotide utilization with its import. However, the Golgi can have a pool of transported, but unutilized, sugar nucleotides that can participate in glycosylation. This can be seen in in vitro Golgi preparations that are capable of glycosylating partially completed endogenous glycoproteins as well as freely diffusable glycoside acceptors 179, 801. The artificial acceptors enter the Golgi where they encounter glycosyl transferases and, in some cases, glycosylation proceeds even though no exogenous donors have been added [81]. Another advantage of using the antiporter system is that the precursor monophosphate is returned to the cytosol where it is available for another round of activation. This creates a highly efficient salvage system for the nucleotide precursors. Several transporters, including CMP-Sia, UDP-Gal and UDP-GlcA, have been reconstituted into proteoliposomes and used for functional assays and purification of the proteins. Several transporters have been cloned from mammals (CMP-Sia, UDP-Gal), yeast (UDP-GlcNAc) 175, 76, 821 and protozoa (GDP-Man) [83]. All are very hydrophobic and appear capable of spanning the membrane many times. They appear to function as homodimers. Several mutant mammalian cell lines lack specific sugar nucleotide transporters, e.g. UDP-Gal and CMP-Sia, and as a result they make incomplete sugar chains. However, they are “leaky” mutants. For instance, loss of the UDP-Gal in the Golgi of mutant MDCK cells, decreases the synthesis of galactosylated glycoproteins, glycolipids and keratan sulfate, but heparan and chondroitin sulfate synthesis is unaffected [84]. The galactosyl transferases that synthesize the core region tetrasaccharide common to GAG chains probably have low K,,, values for their sugar nucleotide donors. Mammalian cells do not have GDP-Man transporters but the trypanosomal parasite Leishmania donovani does [ 831 and it is required for the synthesis of lipophosphoglycan (LPG). Yeast also has a GDP-Man transporter which is essential for mannan synthesis 1821. In both microbes, specific inhibition of GDP-Man transporters might have potential therapeutic value. Theoretically, glycosylation could be controlled by regulating the presence or activity of sugar nucleotide transporters within the Golgi. At this time, the sub-

1.7 Possible Future Directions

13

compartmental location (cis, medial, trans) of the transporters in the Golgi is unknown as is their physical relationships to the various glycosyl transferases they service. Golgi localization has focused on glycosyltransferases, but there have been few revealing studies about how the actual glycosylation reactions occur within the Golgi. There is physical evidence for complexes containing several mannosyl transferase in yeast [ 851 and circumstantial evidence for association of selected glycosyltransferases in complexes in mammalian cells, but no studies have yet been able to determine whether the transporter is part of a complex.

1.6 Control of Sugar Nucleotide Levels This topic may be physiologically very important, but it is not well understood. A human genetic disorder called sialuria shows the importance of at least one prediction from in vitro studies. In this condition, large amounts of sialic acid (5-7 g/day) are secreted into the urine along with various intermediates in the UDPGlcNAc--->CMP-Sia pathway [86, 871. This was confirmed to be due to a defective feed-back inhibition of the 2-epimerase/ManNAc kinase, the first step in the pathway [88]. Most of the precursor pools turnover within a few minutes [12], and the size of various sugar nucleotide pools have been determined by methods with different reliability. But even with the best methods, it is hard to interpret the measured numbers and translate them into a clear picture because the relative distribution of the precursors in the cytosol and Golgi is not known. The average cellular concentration of sugar nucleotide precursors may not be very meaningful, since “cytosol” is an operational definition (100,000 g supernatant) that may not detect compartments of cytoplasmic organization or substrate channeling. Providing moderate amounts (< 1 mM) of glucosamine to muscle cells increases the size of their UDP-hexosamine pools and creates the biochemical phenotype of non-insulin dependent diabetes [89]. Glucosamine-incubated cells do not increase glucose uptake or form glycogen in response to insulin [90, 911. Glucosamine can induce a similar response when given to animals [91, 921. The pathway and mechanism for this effect is not known, but finding O-GlcNAc on many regulatory proteins may suggest a link [93, 941. Alternatively, since the UDP-GlcNAc concentration is second only to ATP (1-2 mM), some of the effects could simply be due to consumption of ATP in producing UDP-GlcNAc [95].

1.7 Possible Future Directions The production and utilization of glycoconjugate precursors is more complex than the metabolic chart indicates. Future studies may want to address various sources of monosaccharides, the organization of metabolic enzymes within the cells, and the

14

I Metabolism of Sugars and Sugar Nucleotides

influence of diet and environment on the metabolic control of complex sugar chain synthesis [26]. These studies will be especially important for patients with glycosylation disorders who are being treated with monosaccharide therapy [29, 30, 641.

References 1. Leloir, L.F. In: Phophorus Metabolism, 1st ed.; McElroy, W.D., Glass, B. (Eds.); Johns Hopkins Press: Baltimore, 1951; pp 67. 2. Thorens, B. Glucose transporters in the regulation of intestinal, renal, and liver glucose fluxes. Am. J. Physiol. 1996,270, G541LG553. 3. Could, G.W.; Bell, (3.1. Facilitative glucose transporters: an expanding family. Trends. Biochem. Sci. 1990, 15, 18-23. 4. Could, G.W.; Holman, G.D. The glucose transporter family: structure, function and tissuespecific expression. Biochem. J. 1993,295, 329-341. 5 . Walmsley, A.R.; Barrett, M.P.; Bringaud, F.; Could, G.W. Sugar Transporters from Bacteria, Parasites and Mammals: Structure-activity Relationships. TIBS 1998, 23, 476-48 1. 6. Panneerselvam, K.; Freeze, H.H. Mannose enters mammalian cells using a specific transporter that is insensitive to glucose. J. B i d Chem. 1996, 19t271, 9417-9421. 7. Ogier-Denis, E.; Blais, A,; Houri, J.J.; Voisin, T.; Trugnan, G.; Codogno, P. The emergence of a basolateral 1-deoxymannojirimycin-sensitivemannose carrier is a function of intestinal epithelial cell differentiation. Evidence for a new inhibitory effect of 1-deoxymannojirimycin on facilitative mannose transport. J. Bid. Chem. 1994, 269, 4285-4290. 8. Ogier-Denis, E.; Codogno, P.; Chantret, 1.; Trugnan, G. The processing of asparagine-linked oligosaccharides in HT-29 cells is a function of their state of enterocytic differentiation. An accumulation of Man9,8-GlcNAc2-Asn species is indicative of an impaired N-glycan trimming in undifferentiated cells. J. Biol. Chrm. 1988, 263, 6031-6037. 9. Ogier-Denis, E.; Trugnan, G.; Sapin, C.; Aubery, M.; Codogno, P. Dual effect of 1deoxymannojirimycin on the mannose uptake and on the N-glycan processing of the human colon cancer cell line HT-29. J. Bid. Chem. 1990,265, 5366-5369. 10. Alton, G.; Hasilik, M.; Niehues, R.; Panneerselvam, K.; Etchison, J.R.; Fana, F.; Freeze, H.H. Direct utilzation of mannose for mammalian glycoprotein biosynthesis. Glycohiologj~1998, 8, 285-295. 11 Silverman, M.; Aganon, M.A.; Chinard, F.P. Specificity of monosaccharide transport in dog kidney. Am. J. Physiol. 1970, 218, 743-750. 12 Panneerselvam, K.; Etchison, J.R.; Freeze, H.H. Human fibroblasts prefer mannose over glucose as a source of mannose for N-glycosylation. Evidence for the functional importance of transported mannose. J. Biol. Chem. 1997,272, 23123-23129. 13 Wiese, T.J.; Dunlap, J.A.; Yorek, M.A. L-fucose is accumulated via a specific transport system in eukaryotic cells. J. Biol. Chem. 1994, 269, 22705-2271 1. 14 Ripka, J.; Adamany, A.; Stanley, P. Two Chinese hamster ovary glycosylation mutants affected in the conversion of GDP-mannose to GDP-fucose. Arch. Biochem. Biophys. 1986, 249, 533545. 15 Shetlar, M.R.; Capps, J.C.; Hern, D.L. Incorporation of radioactive glucosamine into the serum proteins of intact rats and rabbits. Biochim. Biophys. Acta. 1964, 83, 93-101. 16 Aronson. N.N., Jr.; Docherty, P.A. Degradation of [6-3H]- and [ l-'4C]glucosaminelabeled asialo-alpha 1-acid glycoprotein by the perfused rat liver. J. Biol. Chem. 1983, 258, 4266-4271. 17 Rome, L.H.; Hill, D.F. Lysosomal degradation of glycoproteins and glycosaminoglycans. Efflux and recycling of sulphate and N-acetylhexosamines. Biochemical J. 1986, 235, 707713. 18. Trujillo, J.L.; Gan, J.C. Glycoprotein biosynthesis. VI. Regulation of uridine diphosphate N-acetyl-D-glucosamine metabolism in bovine thyroid gland slices. Biochim Biophys. Acta. 1973,304, 32-41.

References

15

19. Krieger, M.; Reddy, P.; Kozarsky, K.; Kingsley, D.; Hobbie, L.; Penman. M. Analysis of the synthesis, intracellular sorting, and function of glycoproteins using a mammalian cell mutant with reversible glycosylation defects. Met1zod.y Cell B i d . 1989, 32, 57-84. 20. Kingsley, D.M.; Kozarsky, K.F.; Hobbie, L.; Krieger, M. Reversible defects in 0-linked glycosylation and LDL receptor expression in a UDP-Gal/UDP-GalNAc 4-epimerase deficient mutant. Cell 1986, 44, 749 -759. 21, Schauer, R. Sialic acids as antigenic determinants of complex carbohydrates. Adc. Expeu. Med. B i d . 1988, 228, 47-72. 22. Chigorno, V.; Tettamanti, G.; Sonnino, S. Metabolic processing of gangliosides by normal and Salla human fibroblasts in culture. A study performed by administering radioactive GM3 ganglioside. J. Biol. Chem. 1996. 271, 21738-21744. 23. Lloyd, J.B. Metabolite efflux and influx across the lysosome membrane. Subcell. Biochem. 1996,277, 361-386. 24. Van Rinsum, J.; Van Dijk, W.; Hooghwinkel, G.J.; Ferwerda, W. Subcellular localization and tissue distribution of sialic acid precursor-forming enzymes. Biochemical Jouunul1983,211),2128. 25. Martin, A,; Rambal, C.; Berger, V.; Perier, S.; Louisot, P. Availability of specific sugars for glyconjugate biosynthesis --a need for further investigations in man. Biochimie 1998,8O, 75-86. 26. Berger, B.; Perier, S.; Pachiaudi, C.; Normand, S.; Louisot, P.; Martin, A. Dietary specific sugars for serum protein enzymatic glycosylation in man. Metabolism 1998, 47, 1499-1503. 27. Freeze, H.H. Human Glycosylation Disorders and Sugar Supplement Therapy. Bioclienz. Biophys. Res. Commun. 1999,255, 189- 193. 28. Hinderlich, S.; Stasche. R.; Zeitler, R.; Reutter, W. A bifunctional enzyme catalyzes the first two steps in N-acetylneuraminic acid biosynthesis of rat liver. Purification and characterization of UDP-N-acetylglucosamine 2-epimerase/N-acety~mannosdmine kinase. J. Bid. Cliem. 1997, 272, 2431 3-24318. 29. Niehues, R.; Hasilik, M.; Alton, G.; Korner, C.; Schiebe-Sukumar, M.; Koch, H.G.; Zimmer, K.P.; Wn, R.; Harms, E.; Reiter, K.; Von Figura, K.; Freeze, H.H.; Harms, H.K.; Marquardt, T. Carbohydrate-deficient glycoprotein syndrome type Ib. Phosphomannose isomerase deficiency and mannose therapy. J. Clin. Inwst. 1998, 101, 1414-1420. 30. Babovic-Vuksanovic, D.; Patterson, M.C.; Schwenk, W.F.; O’Brien, J.F.; Vockley, J., Freeze, H.H.; Mehta, D.P.; Michels, V.V. Severe hypoglycemia as a presenting symptom of carbohydrate-deficient glycoprotein syndrome. J Pediatr. 1999 Dec; 135(6): 775-81. 31. Baque, S.; Guinovart, J.J.; Ferrer, J.C. Glycogenin, the primer of glycogen synthesis, binds to actin. FEBS Lett. 1997, 41 7, 355-359. 32. Lomako, J.; Lomako, W.M.; Whelan. W.J. Glycogen metabolism in quail embryo muscle. The role of the glycogenin primer and the intermediate proglycogen. Euu. J. Biockem. 1995, 234, 343-349. 33. Lomako, J.; Lomako, W.M.; Whelan, W.J. A self-glucosylating protein is the primer for rabbit muscle glycogen biosynthesis. FASEB J. 1988, 2, 3097-3103. 34. Harris, R.J.; Spellman, M. W. 0-linked fucose and other post-translational modifications unique to EGF modules. Glycobiolmqj 1993, 3, 219- 224. 35. Lin, X.; Buff, E.M.; Perrimon, N.; Michelson, A.M. Heparan sulfate proteoglycans are essential for F G F receptor signaling during Drosophila embryonic development. Development. 1999 Sep; 126(17): 3715-23. 36. Benevolenskaya, E.V., Frolov, M.V.; Birchler, J.A. The sugarless mutation affects the expression of the white eye color gene in Drosophila melanogaster. Mol Gen Genet. 1998 Nov; 260(2-3): 131-43. 37. Hacker, U.; Lin, X.; Perrimon, N. The Drosophila sugarless gene modulates Wingless signaling and encodes an enzyme involved in polysaccharide biosynthesis. Development 1997, 124, 35653573. 38. Cumberledge, S.; Reichsman, F. Glycosaminoglycans and WNTs: just a spoonful of sugar helps the signal go down. Trends Genet. 1997, 13, 421-423. 39. Yuen, C.T.; Chai, W.; Loveless, R.W.; Lawson, A.M.; Margolis, R.U.; Feizi, T. Brain contains HNK-I immunoreactive 0-glycans of the sulfoglucuronyl lactosamine series that terminate in 2-linked or 2,6-linked hexose (mannose). J. Bid. Chem. 1997, 272, 8924-8931.

16

I Metabolism of Sugars and Sugar Nucleotides

40. Tereyama, K.; Seiki, T.; Nakamura, A,; Matsumori, K.; Ohta, S.; Oka, S.; Sugita, M.; Kawasaki, T. Purification and characterization of a glucuronyltransferase involved in the biosynthesis of the HNK-1 epitope on glycoproteins from rat brain. J. Biol. Chem. 1998,273, 30295-30300. 41. Kim, B.H.; Takahashi, M.; Tada, K.; Bosma, P.J.; Roy-Chowdbury, J.; Roy-Chowdbury, N. Cell and gene therapy for inherited deficiency of bilirubin glucuronidation. J. Perinatol. 1996, 16, S67-S72. 42. Burchell, B.; Brierley, C.H.; Rance, D. Specificity of human UDP-glucuronosyltransferases and xenobiotic glucuronidation. Life Sci. 1995, 57, 1819-1831. 43. Banhegyi, G.; Garzo, T.; Antoni, F.; Mandl, J. Glycogenolysis-and not gluconeogenesis-is the source of UDP-glucuronic acid for glucuronidation. Biochim. Biophys. Acta 1988, 967, 429-435. 44. Vertel, B.M.; Walters, L.M.; Flay, N.; Kearns, A.E.; Schwartz, N.B. Xylosylation is an endoplasmic reticulum to Golgi event. J. Biol. Chem. 1993,268, 11 105--11112. 45. Kearns, A.E.; Vertel, B.M.; Schwartz, N.B. Topography of glycosylation and UDP-xylose production. J. Biol. Chem. 1993,268, 11097-1 1104. 46. Stetten, M.R.; Goldsmith, P.K. Two hexokinases of Homarus americanus (lobster), one having great affinity for mannose and fructose and low affinity for glucose. Biochirn. Biophys. Acta 1981, 657,468-481. 47. Freeze, H.H. Disorders in protein glycosylation and potential therapy: Tip of an iceberg? J. Pediatr. 1998, 133, 593-600. 48. Jaeken, J.; Matthijs, G.; Saudubray, J.M.; Dionisivici, C.; Bertini, E.; Delonlay, P.; Henri, H.; Carchon, H.; Schollen, E.; Vanschaftingen, E. Phosphomannose isomerase deficiency-A carbohydrate-deficient glycoprotein syndrome with hepatic-intestinal presentation. Am. J. Hum. Genet. 1998, 62, 1535-1539. 49. Smith, D.J.; Proudfoot, A,; Friedli, L.; Klig, L.S.; Paravicini, G.; Payton, M.A. PMI40, an intron-containing gene required for early steps in yeast mannosylation. Mol. Cell. Biol. 1992, 12, 2924-2930. 50. Freinkel, N.; Lewis, N.J.; Akazawa, S.; Roth, S.1.; Gorman, L. The honeybee syndrome-implications of the teratogenicity of mannose in rat-embryo culture. N. Enr). J. Med. 1984, 310, 223-230. 51. Buchanan, T.; Freinkel, N.; Lewis, N.J.; Metzger, B.E.; Akazawa, S. Fuel-mediated teratogenesis. Use of D-mannose to modify organogenesis in the rat embryo in vivo. J. Clin. Invest. 1985, 75, 1927-1934. 52. Moore, D.C.; Stanisstreet, M.; Beck, F.; Clarke, C.A. The effects of mannose on rat embryos grown in vitro. Life Sci. 1987, 41, 1885-1893. 53. Krasnewich, D.; Gahl, W.A. Carbohydrate-deficient glycoprotein syndrome. Adu. Pediutr. 1997, 44, 109-140. 54. Matthijs, G.; Schollen, E.; Van Schaftingen, E.; Cassiman, J.J.; Jaeken, J. Lack of homozygotes for the most frequent disease allele in carbohydrate-deficient glycoprotein syndrome type 1A. Am. J. Hum. Genet. 1998, 62, 542-~550. 55. Doucey, M.A.; Hess, D.; Cacan, R.; Hofsteenge, J. Protein C-mannosylation is enzymecatalysed and uses dolichyl-phosphate-mannose as a precursor. Mol. Bid. Cell 1998, 9, 291300. 56. Kim, S.; Westphal, V.; Srikrishna, G.; Mehta, D.P.; Peterson, S.; Filiano, J.; Karnes, P.S.; Patterson, M.C.; Freeze, H.H. Dolichol phosphate mannose synthase (DPM1) mutations define congenital disorder of glycosylation Ie (CDG-Ie). J Clin Invest. 2000 Jan; 105(2): 191-8. 57. Sullivan, F.X.; Kumar, R.; Kriz, R.; Stahl. M.; Xu, G.Y.; Rouse, J.; Chang. X.J.; Boodhoo, A,; Potvin, B.; Cumming, D.A. Molecular cloning of human GDP-mannose 4,6-dehydratase and reconstitution of GDP-fucose biosynthesis in vitro. J. Biol. Chem. 1998,273, 8193-8202. 58. Chang, S.; Duerr, B.; Serif, G. An Epimerase-Reductase in L-Fucose Synthesis. J. Biol. Chem. 1988,263, 1693-1697. 59. Tonetti, M.; Sturla, L.; Bisso, A,; Benatti, U.; De Flora, A. Synthesis of GDP-L-fucose by the Human FX Protein. J. Biol. Chem. 1996, 271, 27274427279, 60. Kornfeld, R.H.; Givsburg, V. Control of Synthesis of Guanosine 5’-Diphosphate D-Mannose and Guanosine 5‘-Diphosphate L-Fucose in Bacteria. Biochim. Biophys. Actu 1966, 117, 7987.

References

17

61. Sturla, L.; Etzioni, A,; Bisso, A,; Zanardi, D.; Deflora, G.; Silengo, L.; Deflora, A.; Tonetti, M. Defective intracellular activity of GDP-D-Mannose-4,6-Dehydratase in leukocyte adhesion deficiency type I1 syndrome. FEBS Lett. 1998, 429, 274-278. 62. Etzioni, A,; Phillips, L.M.; Paulson, J.C.; Harlan, J.M. Leukocyte adhesion deficiency (LAD) TI. Ciba Found. Symp. 1995,189, 51-58. 63. Park, S.H.; Pastuszak, I.; Drake, R.; Elbein, A.D. Purification to apparent homogeneity and properties of pig kidney L-fucose kinase. J. Bid. Chem. 1998, 273, 5685-5691. 64. Marquardt, T.; Luhn, K.; Srikrishna, G.; Freeze, H.H.; Harms, E.; Vestweber. D. Fucose Therapy for Leukocyte Adhesion Deficiency Type I1 (LAD TI). Blood 1999, 94. 3976-3985. 65. Chan, J.Y.; Nwokoro, N.A.; Schachter. H. L-Fucose metabolism in mammals. The conversion of L-fucose to two moles of L-lactate, of L-galactose to L-lactate and glycerate, and of Darabinose to L-lactate and glycollate. J. Biol. Chenz. 1979, 2.54, 7060-7068. 66. Holton, J.B. Galactosaemia: pathogenesis and treatment. J. Inherit. Metab. Di,x 1996, 19, 3-7. 67. McDowell, G.; Gahl, W.A. Inherited disorders of glycoprotein synthesis: cell biological insights. Proc. Soc. Exp. Biol Med. 1997. 215, 145-157. 68. Wolosker, H.; Kline, D.; Bian, Y.; Blackshaw. S.; Cameron, A.M.; Fralich, T.J.; Schnaar, R.L.; Snyder, S.H. Molecularly coned mammalian glucosamine-6-phosphate deaminase localizes to transporting epithelium and lacks oscillin activity. FASEB J. 1998, 12, 91-99. 69. Pels Rijcken, W.R.; Ferwerda, W.; Van den Eijnden, D.H.; Overdijk, B. Influence of Dgalactosamine on the synthesis of sugar nucleotides and glycoconjugates in rat hepatocytes. Gll'cobiology 1995, 5, 495- 502. 70. Pastuszak, I.; Drake, R.; Elbein, A.D. Kidney N-acetylgalactosamine (GalNAc)-1-phosphate kinase, a new pathway of GalNAc activation. J. Bid. Chem. 1996, 271, 20776-20782. 71. Kean, E.L. Sialic acid activation. Glycabiologj>1991, 1, 441-447. 72. Varki, A. Diversity in the sialic acids [published erratum appears in Glycobiology 1992 Apr; 2(2): following 1681. GlycobiologjJ 1992, 2. 25-40. 73. Reuter, G.; Gabius, H.J. Sialic acids structure-analysis-metabolism-occurrence-recognition. Biol. Chern. Hoppe-Seyler 1996, 377, 325-342. 74. Allen, M.B.; Walker, D.G. Kinetic characterization of N-acetyl-D-glucosamine kinase from rat liver and kidney. Bioclwm. J. 1980, 185, 577 582. 75. Hirschberg, C.B.; Robbins, P.W.; Abeijon, C. Transporters of nucleotide sugars, ATP, and nucleotide sulfate in the endoplasmic reticulum and Golgi apparatus. Annu. Reu. Biochem. 1998,67,49-69. 76. Abeijon. C.; Mandon, E.C.; Hirschberg, C.B. Transporters of nucleotide sugars, nucleotide sulfate and ATP in the Golgi apparatus. 7 i m t l . r Biodzem. Sci. 1997, 22, 203-207. 77. Trombetta, E.S.; Helenius, A. Lectins as chaperones in glycoprotein folding. Curr. Opin. Stvuct. Bid. 1998, 8, 587-592. 78. Abeijon, C.; Yanagisawa, K.; Mandon, E.C.; Hausler, A,; Moremen, K.: Hirschberg, C.B.; Robbins, P.W. Guanosine siphosphatase is required for protein and shpingolipid glycosylation in the Golgi lumen of Saccharomyces cerevisiae. J. eel/ Biol. 1993, 122, 307-323. 79. Varki. A. Factors controlling the glycosylation potential of the Golgi apparatus. Trends Cell Biol. 1998, 8, 34-40. 80. Hayes; B.K.; Freeze, H.H.; Varki, A. Biosynthesis of oligosaccharides in intact Golgi preparations from rat liver. Analysis of N-linked glycans labeled by UDP-[6-3H]N-acetylglucosamine. J. Biol. Chenz. 1993, 268, 16 1 39- 16 154. 81. Etchison, J.R.; Freeze. H.H. A new approach to mapping co-localization of multiple glycosyl transferases in functional Golgi preparations. Glj~cohiology1996, 6 , 177-1 89. 82. Dean, N.; Zhang, Y.B.; Poster, J.B. The VRG4 Gene is Required for GDP-mannose Transport cereuisiue. J. Bid. Client 1997, 272, into the Lumen of the Golgi in the Yeast, Sacc~l~tivornyces 319O8--31914. 83. Ma, D.; Russell, D.G.; Beverley, S.M.; Turco, S.J. Golgi GDP-mannose uptake requires Leishmania LPG2. A member of a eukaryotic family of putative nucleotide-sugar transporters. J. Bid. Cliem. 1997, 272, 3799-3805. 84. Toma, L.; Pinhal, M.A.; Dietrich, C.P.; Nader, H.B.; Hirschberg. C.B. Transport of UDPgalactose into the Golgi lumen regulates the biosynthesis of proteoglycans. J. Bid. Che~z.1996, 271. 3897-3901.

18

1 Metabolism of Sugars and Sugar Nucleotides

85. Jungmann, J.; Munro, S. Multi-protein complexes in the cis golgi of Saccharomyces cerevisiae with alpha-l,6-mannosyltransferaseactivity. EMBO J. 1998, 17, 423-434. 86. Thomas, G.H.; Reynolds, L.W.; Miller, C.S. Overproduction of N-acetylneuraminic acid (sialic acid) by sialuria fibroblasts. Ped. Res. 1985, 19, 451-455. 87. Strecker, G. Genetic disorders of N-acetylneuraminic acid metabolism: sialurias and sialidoses. Comp. Rend. Seances Soc. Bid. Fil. 1999, 179, 567-576. 88. Weiss, P.; Tietze, F.; Gahl, W.A.; Seppala, R.; Ashwell, G. Identification of the metabolic defect in sialuria. J. Biol. Chem. 1989, 264, 17635-17636. 89. Hawkins, M.; Angelov, I.; Liu, R.; Barzilai, N.; Rossetti, L. The tissue concentration of UDPN-acetylglucosamine modulates the stimulatory effect of insulin on skeletal muscle glucose uptake. J. Biol. Chem. 1997, 272. 4889--4895. 90. Spiro, R.G. The effect of N-acetylglucosamine and glucosamine on carbohydrate metabolism in rat liver slices. J. Biol. Chem. 1958, 233, 546-550. 91. Hawkins, M.; Barzilai, N.; Liu, R.; Hu, M.; Chen, W.; Rossetti, L. Role of the glucosamine pathway in fat-induced insulin resistance. J. Clin. Invest. 1997, 99, 2173-2182. 92. Rossetti, L.; Hawkins, M.; Chen, W.; Gindi, J.; Barzilai, N. In uivo glucosamine infusion induces insulin resistance in normoglycemic but not in hyperglycemic conscious rats. J. Clin. Invest. 1995, 96, 132-140. 93. Snow, D.M.; Hart, G.W. Nuclear and cytoplasmic glycosylation. Intl Rev. Cyfol. 1998, 181, 43-74. 94. Yki-Jarvinen, H.; Vogt, C.; Lozzo, P.; Pipek, R.; Daniels, M.C.; Virkamaki, A,; Makimattila, S.; Mandarino, L.; DeFronzo, R.A.; McClain, D.; Gottschalk, W.K. UDP-N-acetylglucosamine transferase and glutamine: Fructose 6-phosphate amidotransferase activities in insulin-sensitive tissues. Diabetologia 1997, 40, 76-8 1. 95. Hresko, R.C.; Heimberg, H.; Chi, M.M.Y.; Mueckler, M. Glucosamine-induced insulin resistance in 3T3-Ll adipocytes is caused by depletion of intracellular ATP. J. Biol. Chem. 1998, 273, 20658-20668.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

2 Nucleotide Sugar Transporters Rita Gerardy-Schahn and Matthias Eckhardt

2.1 Introduction The cellular loci for the production of secreted and membrane bound glycoproteins and glycolipids are the endoplasmic reticulum (ER) and the Golgi apparatus. Both compartments are sealed vesicular structures that require active in- and outward transport of metabolites that are substrates or products of catalytic reactions taking place at the luminal side. Studies on the secretory pathway carried out in the 1960s, together with the delineation of the steps involved in N-glycosylation in the 1970s, made the need for specific transport systems apparent. Transport was demonstrated in vitro in isolated Golgi vesicles, and genetic proof for the existence of specific nucleotide sugar transporters was obtained when Chinese hamster ovary (CHO) cell mutants with asialo- or asialo-agalacto-surfaces were shown to be inactive in translocating CMP-sialic acid and UDP-galactose, respectively, from the cytoplasm into the Golgi lumen. Further biochemical analysis of the transport process demonstrated that nucleotide sugars are specifically recognized by individual translocators that work in a temperature-dependent, and saturatable manner. Nucleotide sugars can be concentrated in the organellar lumen up to 50-fold. The energy required for this active transport results from exchanging luminal nucleoside monophosphates with cytosolic nucleotide sugars. The nucleoside monophosphates thereby follow their concentration gradients. Attempts to purify nucleotide sugar transporters led to highly enriched and functionally active protein fractions, however, homogeneous purification is a very recent accomplishment and is discussed in this Chapter. Cloning of the first nucleotide sugar transporter genes in 1996 identified a family of highly hydrophobic multimembrane spanning proteins, with molecular masses of 34-42 kDa. Ten transmembrane domains have been demonstrated in the CMP-sialic acid transporter and due to similarities in sequence and hydrophobicity, it seems plausible to assume that the membrane topology is the same for most, if not all, nucleotide sugar transporters. Progress in studying nucleotide sugar transporters over the last few years stimu-

20

2 Nucleotide Suyur Transporters

lated a series of excellent reviews that are recommended to the interested reader 11-51. In this communication, the focus will be on advance made in identifying structure-function relationships in this important protein family. Furthermore, we discuss recent observations that suggest nucleotide sugar transporters may be involved in the regulation of glycosylation, and in the manifestation of a new form of carbohydrate deficiency glycoprotein syndrome (CDGS). Finally, we address the future perspectives of nucleotide sugar transporters in generating improved cell systems for use in the field of biotechnology.

2.2 General considerations The glycocalyx is a well known characteristic of the animal cell and has been recognised for more than 100 years. Its “outstanding” position per se implies that the sugar rim plays an important role in cellular functions associated with communication and contact. Nevertheless, deciphering the genetic code and the development of numerous powerful molecular biological techniques, concentrated research efforts on genes and on their linear translation products, the proteins. The subsequent observation that in many cases, recombinant proteins were also active in the absence of their physiological carbohydrate additives, substantiated the widely accepted view that carbohydrates provide a kind of luxury decoration to functionally active proteins. This view was reversed when sugars were found to be directly involved in a large variety of biological reactions 16, 71 including the stability and immunological properties of recombinant proteins [ 81.

2.3 The Requirement for Nucleotide Sugar Transporters and Their Mechanism of Function: A Comprehensive Overview of the Last 20 Years The addition of carbohydrates to secreted and membrane-bound proteins and to lipids starts in the ER and continues along the secretory pathway up to the transGolgi-network [9, 101. Maturation of N-glycans involves the interplay of two classes of enzymes: the trimming glycosidases and the elongating glycosyltransferases (for review, see 1111). In contrast, 0-glycosylation of proteins and the transfer of glycans to lipids, is the result of sequentially active glycosyltransferases, exclusively [ 121. The spatial organisation of the processing enzymes within the membranes of ER and Golgi corresponds to their position in the reaction sequence of oligosaccharide maturation [ 131. While the majority of ER glycosyltransferases use monosaccharides that are activated via dolichol phosphate, some ER glycosyltransferases and the glycosyltransferases of the Golgi apparatus use nucleotide sugars as donor substrates. Nu-

2.3 The Requirenicntjor Nuclrotidp Sugar Transporters

21

NDPase

UDP-Gal

I I"1" "Wlr ~

Galactose-I-phosphate uridylyltransferase Gal-1-P

JL

Glc-I-P

UDP-Glc

/

/

synthetase

Cytosol

Figure 1. Synthesis and transport of UDP-galactose and CMP-sialic acid. This scheme demonstrates three terminal steps in the synthesis of sialylated glycoproteins. The nucleotide sugar donors UDP-galactose (UDP-Gal) and CMP-sialic acid (CMP-Sia) are synthesized in the cytosol and in the nucleus, respectively. From the cytosol, the nucleotide sugars are transported into the lumen of the Golgi apparatus, the cellular locus where secreted and membrane bound glycoproteins and glycolipids are produced. The transport of nucleotide sugars across the Golgi membrane requires specific transporter proteins (black boxes). Nucleotide sugar transporters act as antiporters. The counter-transported substrates are the nucleoside monophosphates UMP and CMP. CMP is a product of the lumenal sialyltransferase (Sia-T), while UMP production requires an additional enzymatic step, the hydrolysis of UDP by the nucleoside diphosphatase (NDPase). The second product in the NDPase reaction is phosphate (Pi). The existence of a specific phosphate transporter has not been demonstrated yet.

cleotide sugars are generated in the cytosol, with the exception of CMP-sialic acid, which is synthesised in the nucleus [14, 151 (Figure 1). Translocation systems that bring the activated monosaccharides into the reaction compartments of the glycosyltransferases have therefore been postulated, and were demonstrated to exist, using in vitro studies [16, 171. Transport assays were carried out on purified ER and Golgi vesicles and on proteoliposomes (nucleotide sugar transporter enriched protein fractions reconstituted into artificial phospholipid vesicles) to investigate the kinetic and mechanistic properties of these proteins. The results of these studies demonstrate: 1) each nucleotide sugar is recognised by a specific protein; 2) nucleotide sugar transporters act in a temperature-dependent and saturatable manner with apparent K , values of 1-10 pM; 3 ) substrates can be actively transported against steep concentration gradients; 4) the driving force for the intra-organelle accumulation of nucleotide sugars is provided by the simultaneous outward transport of nucleoside monophosphates down their concentration gradients [ 161. Hence, nucleotide sugar transporters are believed to act as antiporters [2, 18, 191. Clear evidence for the antiport mechanism has been obtained in the case of GDP-fucose

22

2 Nucleotide Sugur Trunsporter.r

transport, where an equimolar exchange of [ ''C]GDP-fucose and [ 3H]GMP was measured in rat liver Golgi vesicles after preincubation with tritiated GDP-fucose, followed by exposure to the [ ''C]-labelled nucleotide sugar [ 191. In a later study, the exchange of UDP-GlcNAc against UMP was demonstrated to reach an optimum when preloading of proteoliposomes was carried out at pH 7.5, indicating that the dianionic form of UMP is counter-transported with UDP-GlcNAc [20]. Because uptake of nucleotide sugars into proteoliposomes was stimulated when vesicles were preloaded with the respective nucleoside monophosphates, and transport was inhibited after disruption of nucleoside monophosphate gradients, the antiport mechanism can be regarded as the general mechanism used by nucleotide sugar transporters [2, 191. Only in the case of CMP-sialic acid is the nucleoside monophosphate a direct product of the glycosyltransferase reaction. All other sugars are activated as nucleoside diphosphates, and these diphospho-nucleosides are the products released by glycosyltransferases. The Golgi-resident nucleoside diphosphatases have therefore been hypothesized to complete the cycle of nucleotide sugar import. Guanosine diphosphatase was purified from Succhuromyces cerevisiue [21] and the gene (gdul) was cloned [22]. Genetic disruption of the gdul locus interfered with N- and 0glycosylation, and with sphingolipid synthesis. Golgi vesicles isolated from gene targeted cells were found to have a five-fold reduced GDP-mannose uptake capacity [23]. A second nucleoside diphosphatase (YND1) with a broader substrate specificity has been cloned very recently from S. cerevisiur [24, 251. Over-expression of the YNDl protein in GDAl deficient strains was able to partially complement the defect, while the double mutantion Agdul, Ayndl generated a lethal phenotype [24, 251. Although precise data on the function and subcellular localisation of YNDl are presently not known the current data suggest the existence of two guanosine diphosphatases in S. cereuisiue which, under physiological conditions, may co-operate in producing GMP and so may be able to influence the velocity of GDP-fucose uptake. The speculation that nucleotide sugar transport into the Golgi lumen is metabolically controlled is in good correlation with earlier experiments demonstrating that the extra-vesicular concentration of nucleoside mono-, di-, and triphophates influences transport activity [ 1g]. Moreover, cross inhibition has been observed between nucleotide sugars. For example, UDP-GlcNAc translocation can be modulated by variations in the concentrations of UDP-glucose and UDPGalNAc [26], CMP-sialic acid transport is sensitive against elevated UDP-GlcNAc concentrations [ 18, 271 and GDP-fucose transport against GDP-mannose [ 181. It is, however, worth mentioning that inhibition is most efficient between sugar derivatives that are activated via the same nucleotide.

2.4 Molecular Cloning of Nucleotide Sugar Transporters Somatic mutants exhibiting defects in cell surface glycosylation have been isolated from mammalian cell lines as clones that withstand the cytotoxic activity of plant

2.4 Molecular Cloning of Nucleotide Sugur Trunsporters

23

lectins (for review see [28]) or acquired resistance towards Newcastle disease virus infection [29, 301. Cell fusion experiments allowed the subdivision of individual clones into genetic complementation groups [31]. Two of more than 40 complementation groups identified so far comprise mutants that are defective in Golgi nucleotide sugar transport systems. Lec2 cells are unable to translocate CMP-sialic acid [32] and cells of the complementation group Lec8/Had-1 are impaired in the Golgi-associated transport of UDP-galactose [33, 341. Additional mutants, lacking specific nucleotide sugar transport activities, have been identified in yeast [ 17, 35371, protozoa, [38] and in Caenorhahditis e1egun.r [39-411. Complementation cloning in these mutants identified the first nucleotide sugar transporter genes [36, 37, 42-48]. The cDNAs for both hamster and mouse CMPsialic acid transporter were isolated using Lec2 cells as recipients for wild type cDNA libraries [43, 441. Complementation to wild type cells was monitored via the cell surface marker polysialic acid, a post-translational modification of the neural cell adhesion molecule that is expressed in wild type CHO cells [49]. Detection of polysialic acid with the monoclonal antibody 735 [50] provides a useful strategy, because the antibody recognises polysialic acid chains with unusual high sensitivity. A helical segment of nine a2,8-linked sialic acid residues forms the epitope [51, 521. Since polymerization degrees of n > 50 have been determined, a single polysialic acid chain contains a multitude of 735 epitopes [52] and explains the sensitivity of the antibody reaction. Monitoring polysialic acid re-expression is a sufficient strategy to isolate any gene whose inactivation causes a severe reduction in sialic acid addition to cell surface glycans, such that polysialic acid is not synthesized [15, 43, 44, 491. Primary sequence analysis of nucleotide sugar transporters revealed that all are hydrophobic type I11 membrane proteins with molecular masses of 34-42 kDa (see [53] and literature cited therein). Database searches identified a large number of structurally related proteins of as yet unidentified specificity in human, Drosophilu, C. elegans and yeast. The dendrogram shown in Figure 2 summarizes all the sequences that were found to exhibit significant similarity to experimentally identified nucleotide sugar transporters. In good correlation with the nine different nucleotide sugars that need to be translocated, 7-9 clusters are discernible on the basis of primary sequence similarities. In the C. eleyuns genome, which has been completely sequenced, 14 genes encoding putative nucleotide sugar transporters have been identified. Of course, the functional activity of these genes awaits verification, but strong evidence has already been obtained that the gene U50135 (identical to sqv-7 [40]) encodes a nucleotide sugar transporter [41]. Another interesting observation that arises from primary sequence comparisons in the group of nucleotide sugar transporters, is that UDP-GlcNAc transporters isolated from Kluyveromyces luctis and MDCK cells are only 22% identical, while the mammalian transporters for CM P-sialic acid, UDP-galactose, and UDPGlcNAc are very similar, suggesting that these genes have developed from a common ancestor. Despite the low conservation between transporters of identical specificity in distant species, all the information required for the formation of a substrate specific transporter is contained within the primary sequence. Functional expression of the

Hsapiens

C.elegans (AFO16438)

S.pombe (AL021816)

hapiens (D87989)

R.norvegicus (D87991)

M.musculus (D87990)

rc

C.elegans (AF003383)

C.elegans (AF045639)

C.elegans (AFO245W)

C.elegans (AFO16438)

I

UDPGlcNAc

GDP-Man GDP-Man

C.elegans (270750)

C.elegans (U50135)

H.sapiens (D87449)

H.sapiens (AJOO5866)

L.donovani

S.cerevisiae

Scerevisiae (P40027)

C.elegans (268215)

C.elegans (AFO36696)

S.pombe (AL031856)

K.ladis

S.cerevisiae (P40004)

C.elegans (UOCC6-I)

Figure 2. Dendrogram identifying putative nucleotide sugar transporters. The dendrogram shows relationships among identified and putative nucleotide sugar transporters. Putative nucleotide sugar transporters were identified by BLAST searches against the sequences of known transporters. The evolutionary tree was generated using Megalign (DNAStar). The substrates of identified transporters are indicated. In all other cases, the data base accession numbers are included. The accession numbers of the identified transporters are (from top to bottom): AF057365, AL022598, P78381, 461420, 008520, P78382, U48413, P40107, U26175.

I

CMPSia

C.griseus C.elegans ((102334)

CMPSia CMPSia

M.musculus

C.elegans (281102)

L n

I ’

UDPGal

C.elegans (282288)

H.sapiens

Dmelanogaster (AL023874)

UDPGlcNAc UDPGal

Spornbe

C.elegans (AF016674) C.familiaris

ru

P

t3

2.5 The Structure of Nucleotide Sugar Tr~insporters

25

murine CMP-sialic acid [54] and the human UDP-galactose transporter [55] on the zero background of S. cerevisiae, and in transporter deficient CHO cells [64]demonstrated that the cloned cDNAs in fact, encode the transporters and not accessory proteins. In line with this, expression cloning in heterologeous systems identified additional family members, although the genes that have been isolated show very low conservation to their orthologues. The canine UDP-GlcNAc transporter was isolated in the UDP-GlcNAc transporter negative clone of K. Zactis [56] and a plant cDNA that encodes a putative UDP-galactose transporter was identified by phenotypic correction of CHO Lec8 cells [57]. These studies point the way to rapidly isolating a large panel of functionally related molecules that will improve the basis for studies aimed at enlightening structure-function relationships in the family of nucleotide sugar transporters.

2.5 The Structure of Nucleotide Sugar Transporters Hydrophobicity analyses and secondary structure predictions suggested that 6- 10 transmembrane domains exist in the different nucleotide sugar transporters [4]. As shown in Figure 3, ten transmembrane domains have been identified in the CMPsialic acid transporter by using an epitope insertion approach [58]. Complementary DNAs carrying epitope tags inserted into putative hydrophilic loop structures were transiently expressed in CHO cells, and the orientation in relation to the Golgi membrane was determined by indirect immunofluorescence in semi-permeabilized cells [58]. It will be interesting to see whether all nucleotide sugar transporters exhibit a common membrane topology. The high similarity observed in hydrophobicity plots and the conservation of structural elements between the different transporters argue for this possibility. At the theoretical level, ten hydrophobic domains can be clearly distinguished in the human UDP-galactose [46], canine UDP-GlcNAc [56], and GDP-mannose transporters of Leishmania donovani [47]. In addition the ten transmembrane domain model is compatible with the majority of the sequences shown in Figure 2. Only the genes designated as Q02334 and P40027 that have been identified in C. elegrxns and 5'. ceuevisiae, respectively, lack regions corresponding to the first two transmembrane domains. Because functional data on these putative transporters are not available so far, it remains unclear if these genes encode functional proteins, A mutant of the hamster CMP-sialic acid transporter that lacks the first two transmembrane domains was found to be functionally inactive [59]. Future experiments are needed to reveal whether ten transmembrane domains are the rule in nucleotide sugar transporters. The oligomeric state of functional nucleotide sugar transporters has been proposed to be the homodimer. This hypothesis was established on the basis of structural analogy existing between nucleotide sugar transporters and the Golgi transporter for activated sulfate, which is a dimer in its active state [60]. The identification of potential leucine zipper motifs in some of the cloned nucleotide sugar transporters supported this hypothesis [3]. However, the importance of the leucine zipper motifs

Figure 3. Membrane topology of the CMP-sialic acid transporter. Using an epitope insertion approach, ten transmembrane domains have been identified in the CMP-sialic acid transporter. Positions in the primary sequence where the HA-epitope was inserted are indicated by triangles. All constructs summarised in this figure were correctly transported to the Golgi apparatus. The orientation of the epitope in relation to the Golgi membrane was analyzed by indirect immunofluorescence. Filled triangles (HA14, HA7. HA6) indicate insertion mutants that were unable to translocate CMP-sialic acid. Transport active constructs are marked with open triangles. Polar and charged residues are indicated by filled circles and t,-, respectively.

H z6

2

‘5

2.6 The Subcellular Distrihution qf Nucleotide Sugur Transporters

27

is still questionable. Experimental inactivation of the leucine zipper in the CMPsialic acid transporter did not change the transport activity of the protein, demonstrating that the motif is not required for the assembly of an active CMP-sialic acid transporter [59]. Homogeneous purification has now been achieved for the rat UDP-N-acetylglactosamine transporter [ 531. The purified protein retained functional activity. In SDS-PAGE analysis, a protein band with an apparent molecular mass of 43 kDa was found, but in native glycerol gradients, the sedimentation behavior is identical to a protein of 80-90 kDa [ 531. These data demonstrate that the transporter is a dimer in this soluble state. The final decision on the aggregation state of the membrane bound nucleotide sugar transporters and on whether the transport active unit is a mono-, di-, or oligomer requires direct proof such as that provided recently for the yeast phosphate carrier [61].

2.6 The Subcellular Distribution of Nucleotide Sugar Transporters The distribution of nucleotide sugar transport activities over the membranes of ER and Golgi have been determined in vitro by measuring transport activities on purified ER and Golgi vesicles (for review, see [4]).The data summarized in Table 1 demonstrate that all transport activities can be found in the Golgi apparatus and only some are translocated into the ER. The transporter patterns show species specific variations. Measuring transport activity on vesicle fractions isolated from animal cells, the UDP-galactose transporter has been restricted to Golgi membranes (Table 1). A very recent study on the subcellular localization of UDP-galactose:

Table 1. Summary of nucleotide sugar transport activities identified in Golgi- and ER fractions and genes that have been isolated so far. Nucleotide sugar

Localization

Cloned genes

CMP-sialic acid

Golgi

UDP-galactose

Golgi

UDP-N-Acetylglucosamine

Golgi

UDP-N-Acetylgalactosamine UDP-glucose UDP-x ylose UDP-glucuronic acid GDP-fucose GDP-mannose

Golgi + ER (Golgi) + ER Golgi + ER Golgi + ER Golgi Golgi

mouse [43] human [45] hamster [44] human [46] Scl~i~o.succhoron~yee.~ ponqbe [ 371 Kluyuerorriyces Iuctis [42] dog [561

+ ER

Leishrnunia donovani [38] Succhuromyces cerevisiue [48]

28

2 Nucleotide Sugar Transporters

ceramide galactoyltransferase (CGalT) carried out by [62], however, suggests that UDP-galactose is also transported into the ER lumen. CGalT synthesises galactosylceramide, a major component of myelin, from ceramide and UDP-galactose. With the aid of immunological techniques, CGalT has been shown to be an ER resident [62]. The catalytic domain of the class I integral membrane protein is orientated towards the ER lumen [62]. In the UDP-galactose transporter negative Lec8 cell [33], CGalT activity is blocked [62]. These data strongly suggest the existence of the UDP-galactose transporter in ER membranes. Sensitive immunological techniques are required to conclusively identify the subcellular expression pattern also for other nucleotide sugar transporters. Furthermore, additional studies are required to find out whether nucleotide sugar transporters expressed in the ER and Golgi membranes represent identical proteins, alternative splice variants, or are encoded by different genes. Two splice variants that differ at the carboxyl terminus have been isolated in the case of the human UDP-galactose transporter. Both isoforms are active but only one contains the putative ER-retrieval sequence K(X)KXX [45]. Five of the sequences shown in the dendrogram (Figure 2) exhibit a K(X)KXX motif at the carboxyl terminus. The question on the subcellular distribution of nucleotide sugar transporters should soon be solved, since cDNAs are available that enable the production of specific immunoreagents [63].

2.7 Molecular Defects that Cause Inactive UDP-Galactose and CMP-Sialic Acid Transporters As mentioned above cloning of nucleotide sugar transporters was achieved by complementation of cells that previously had been shown to exhibit defects in the respective nucleotide sugar transport system. UDP-galactose transporter negative Had-1 clones have been isolated from the mouse mammary tumor cell line FM3A [29, 301. Clones belonging into the CMP-sialic acid transporter negative Lec2 group were isolated by lectin resistance [28] or immunoselection on the mAb 735 [49]. Individual clones of both genetic complementation groups have been analyzed at the molecular level. The majority of the clones contain mutations that lead to internal deleted or truncated translation products [59, 651. In two mutants, however, the inactivation of transport was found to be caused by single amino acid exchanges. Inactivation of transport in the FM3A subclone Had-lm can be explained by exchange of glycine-178 to aspartic acid [65]. The Lec2 subclone 9D3 carries a mutation that leads to replacement of glycine-189 with glutamic acid [59]. in both cases, the inactive proteins are correctly targeted to the Golgi-apparatus, demonstrating that the mutations do not interfere with protein folding and subcellular transport [59, 651, but concern recognition or transport of the nucleotide sugar. If mutated positions were replaced by alanine, transport activity was reconstituted [ 59, 651, while the introduction of large or charged amino acid residues in position 189 of the CMP-sialic acid transporter phenocopied the mutant phenotype [59]. These data together with the fact that glycine-178 and -189 are highly conserved residues

2.9 Inuolrmicnt o j Nuclwtide Sugar Trunxp0rter.r

29

allow us to speculate that these positions belong to structural elements that are required for the transport process.

2.8 Association Between Defects in Nucleotide Sugar Transporters and Diseases An observation made by three collaborating German groups [66] provides the first evidence for the association between a nucleotide sugar transport defect and a carbohydrate deficiency glycoprotein syndrome. Leukocytes and fibroblasts isolated from a patient suffering from a new type of leukocyte adhesion deficiency syndrome (LAD 11) express drastically reduced levels of fucosylated carbohydrates [67,68].The patient suffers from multiple infections and hence shows very high leukocyte counts [ 681. Immunological analyses demonstrated deficiencies of several fucosylated carbohydrates that are generated by different fucosyltransferases in activated leukocytes and endothelial cells [68]. The defect in fucose metabolism could be partially corrected by oral supplementation of fucose. After five months of fucose therapy, leukocyte counts turned to normal levels and psychomotor capabilities of the patient improved [69]. Biochemical studies carried out on leukocytes and primary fibroblasts from the LAD I1 patient demonstrated that the enzymes involved in GDP-fucose synthesis are active at normal levels [70]. In contrast, the transport of GDP-fucose into Golgi vesicles was found to be reduced by 80% and the cytoplasmic concentration of GDP-fucose was increased. Control studies demonstrated normal Golgi-update for UDP-Gal and normal GDPase activity [71]. The data obtained so far allow the conclusion that the defect in this patient concerns the Golgi GDP-fucose transporter or an element indirectly involved in GDP-fucose translocation. Attempts to isolate the cDNA that rescues the transport deficient cells are underway and will define the molecular defect that causes this new form of LAD 11 syndrome.

2.9 Involvement of Nucleotide Sugar Transporters in the Regulation of Glycosylation The pattern of glycan structures produced by a given cell depends on its glycosyltransferases. In accord with this, changes in carbohydrate structures that occur during onto- and onco-genetic development [74, 751 and in relation with different stages of cellular activity [6, 76, 771, are accompanied by variations in the expression of glycosyltransferases. Despite this situation, attempts to control N-glycan processing by overexpressing specific glycosyltransferases have met with variable success [78, 791 and suggest that prevalence of glycosyltransferases is not the sole determinator of glycosylation. Metabolic control also seems related to the availability and transport of nucleotide sugars into the respective compartment [22, 27,

30

2 Nucleotide Sugar Transporters

SO]. In a proteoglycan-producing MDCK cell mutant 1801, where availability of UDP-galactose in the Golgi lumen is limited, the synthesis of keratan sulfate drops down to a low level, while chondroitin sulfate and heparan sulfate are produced at normal concentrations [SO]. Possible explanations for this observation are: 1) Biosynthesis of the different glycoconjugates occurs in isolated compartments that are supplied with nucleotide sugar transporters of different kinetic properties. 2) Glycosyltransferases of different substrate specificity exhibit different K, values. 3) Glycosyltransferases form functional complexes with nucleotide sugar transporters and this quaternary organisation may cause a hierarchical order in substrate availability. Very interestingly in this context, is the observation that fucose therapy in the LAD I1 patient described above 1681 partially restores the functional form of Pselectin ligand PSGL-1 on neutrophils but not E-selectin binding to neutrophils 1691. Fucosylation of ESL-1 is mediated by FucT IV, while fucose transfer to PSGL-1 depends on FucT VII 1731. Cloning the GDP-fucose transporter and analyzing its subcellular and functional organisation will therefore be of major interest also in this aspect.

2.10 Future Perspectives Since cDNAs became available in 1996, research with nucleotide sugar transporters has progressed quickly and some of the questions that have been described as future aims in recent reviews [4,641, have already been answered [53, 58, 59, 641. Insights into structure-function relationships have been obtained by defining the membrane topology of the CMP-sialic acid transporter 1581 and by analyzing the sedimentation behaviour of homogeneously purified UDP-GlcNAc transporter from rat liver Golgi membranes 1531. The results of these studies suggest nucleotide sugar transporters to contain ten transmembrane spanning domains [58] and form dimers in the active state [53]. Analyzing loss-of-function mutants in the complementation groups Lec2 [59] and Had-1 1651 suggest two conserved glycine residues (position 189 of the CMP-sialic acid transporter, position 179 in the UDP-Gal transporter) take part in the transport cycle. Nevertheless, structure-function studies are only beginning, and much more information is required to define the molecular architecture of nucleotide sugar transporters. Protein domains involved in specific substrate recognition, in mediating the translocation process, and in regulating their activity need to be determined. Related studies have been carried out with a large variety of transporters that localize to plasma membranes and inner mitochondria1 membranes of mammalian, yeast, and bacterial cells [81-83]. Published experimental protocols can be used to design future experiments aimed at determining the secondary, tertiary and quaternary organization of nucleotide sugar transporters. As shown in Table 1, some transporters are found in Golgi membranes exclusively, others localise to both ER and Golgi. Poly- and monovalent immunoreagents against these proteins shall soon be available and can be used to reinvestigate and potentially complete the topography of nucleotide sugar transporters. It will be

2.10 Future Perspectives

31

interesting to see whether transporters that are targeted to different membranes are also different at protein level and whether ER and Golgi resident forms exhibit different kinetic properties, providing an additional level for the regulation of glycosylation patterns. Nucleotide sugar transporters are the first multiple membrane spanning residents of the Golgi apparatus identified so far and provide new model proteins to study Golgi-targeting, trafficking, and retention. Heterologous expression of mammalian transporters for CMP-sialic acid [54] and UDP-galactose [55] on the zero background of S. cereuisiae showed a substantial proportion of the recombinant proteins correctly targeted to the Golgi membranes, indicating that the information for subcellular destination is contained in the primary sequence. Moreover, these data provide initial evidence that the targeting machinery is, at least in part, conserved amongst different species. Identifying the physiological functions and the subcellular localisation of the many genes that have been identified as transporter-related in the sequence data bases (see Figure 2) should largely contribute to understanding structure-function relationships. There is increasing evidence that glycoconjugate synthesis is controlled at the metabolic level [3, 27, SO]. Recent support for this hypothesis has been obtained by analyzing the fucosylation patterns in the LAD I1 patient described above [69]. Oral supplementation therapy partially restored expression of the functional form of PSGL-I and core fucosylation of secreted proteins, while reconstitution of Eselectin binding was not obtained [69]. These observations suggest that increased expression of nucleotide sugar transporters in eukaryotic cells can be used to modulate and improve glycosylation pathways e.g. in biotechnological processes [27]. Last but not least, functional expression of mammalian nucleotide sugar transporters in yeast opens new perspectives to modulate glycosylation pathways in these organisms. The first example for successful neo-galactosylation has been described by [84]. Two heterologeous genes, ymuZ2+ encoding u 1,2-galactosyltransferase from Schizosuccharomyces pornbe and UGT2 encoding the human UDP-galactose transporter, were co-expressed in galactosylation negative Saccharomyces cerenisiue cells. An efficient transfer of galactose to N- and O-glycans was observed, if cells were supplied with the donor sugar and acceptor substrates. Again, the level of UDP-galactose transporter expression and not the level ul,2-galactosyltransferase was found to limit formation of galactosylated structures. Obviously, we are approaching the time when cellular systems expressing desired glycoforms can be designed for use in biotechnological processes.

Acknowledgments We are most grateful to our colleagues for contributing to this article by sharing unpublished data: C. Korner, K. von Figura, D. Vestweber, T. Marquardt and their research teams are kindly acknowledged for providing data on the newly defined LAD I1 syndrome patient, M. Kawakita and N. Ishida for informations on Had-l mutants, C. Hirschberg and P. Berninsone for information on the C. eleyans nucleotide sugar transporter, and H. Bakker for information on the plant UDP-

32

2 Nucleotide Sugur Trunsporters

galactose transporter. We thank C. Hirschberg, A. Miinster, A. Cook, and K. Baker for helpful discussions and critical remarks on the manuscript. Work in the authors’ laboratory was supported by grants from the Deutsche Forschungsgemeinschaft and the European Community.

References 1. Abeijon C, Mandon EC, Hirschberg CB. Transporters of nucleotide sugars, nucleotide sulfate and ATP in the Golgi apparatus. Trends Biochern. Sci. 1997 22:203-207. 2. Hirschberg CB. Transporters of nucleotides and nucleotide derivatives in the endoplasmic reticulum and Golgi apparatus. Soc. Gen. Physiol. Ser. 1996 51:105-120. 3. Hirschberg CB. Transporters of nucleotide sugars, nucleotide sulfate and ATP in the Golgi apparatus membrane: where next? G/ycobio/oqy 1997 7: 169- 171. 4. Hirschberg CB, Robbins PW, Abeijon C. Transporters of nucleotide sugars, ATP, and nucleotide sulfate in the endoplasmic reticulum and Golgi apparatus. Annu. Rev. Biochem. 1998 67:49-69. 5. Kawakita M, Ishida N , Miura N, Sun-Wada GH, Yoshioka S. Nucleotide sugar transporters: elucidation of their molecular identity and its implication for future studies. J. Biochern. ( T o k y o ) 1998 123:717-785. 6. Varki A. Biological roles of oligosaccharides: all of the theories are correct. Glycohioloy/y. 1993 397-130. 7. Zhang JX, Braakman I, Matlack KE, Helenius A. Quality control in the secretory pathway: the role of calreticulin, calnexin and BiP in the retention of glycoproteins with C-terminal truncations. Mol. Biol. Cell 1997 8:1943- 19.54. 8. Jenkins N, Curling EM. Glycosylation of recombinant proteins: problems and prospects. Enzyme Microb. Technol. 1994 16:354-364. 9. Palade G. Intracellular aspects of the process of protein synthesis. Science 1975 189:347-358. 10. Neutra M, Leblond CP. Synthesis of the carbohydrate of mucus in the golgi complex as shown by electron microscope radioautography of goblet cells from rats injected with glucose-H3. J. Cell Bid. 1966 30:119-136. 1 1 . Cacan R, Duvet S, Kmiecik D, Labiau 0, Mir AM, et al. ‘Glyco-deglyco’processes during the synthesis of N-glycoproteins. Biochirnie 1998 80:59-68. 12. Van den Steen P, Rudd PM, Dwek RA, Opdenakker G. Concepts and principles of 0-linked glycosylation. Crit.Rev. Biochern. Mol. Biol. 1998 33:151-208. 13. Munro S. Localization of proteins to the Golgi apparatus. Trends Cell Bid. 1998 8:11-15. 14. Kean EL. Sialic acid activation. Glycohiologv 1991 1:441-447. 1.5. Miinster AK, Eckhardt M, Potvin B, Miihlenhoff M, Stanley P, et al. Mammalian cytidine 5’monophosphate N-acetylneuraminic acid synthetase: a nuclear protein with evolutionarily conserved structural motifs. Proc. Nut1 Acad. Sci. U.S.A. 1998 95:9140-9145. 16. Hirschberg CB, Snider MD. Topography of glycosylation in the rough endoplasmic reticulum and Golgi apparatus. Annu. Rev. Biockeni. 1987 56:63-87. 17. Abeijon C, Orlean P, Robbins PW, Hirschberg CB. Topography of glycosylation in yeast: characterization of GDPmannose transport and lumenal guanosine diphosphatase activities in Golgi-like vesicles. Proc. Nut1 Acad. Sci. U.S.A. 1989 86:6935-6939. 18. Capasso JM, Hirschberg CB. Effect of nucleotides on translocation of sugar nucleotides and adenosine 3’-phosphate 5’-phosphosulfate into Golgi apparatus vesicles. Biochirn. Biophys. Actu 1984 777:133-139. 19. Capasso JM, Hirschberg CB. Mechanisms of glycosylation and sulfation in the Golgi apparatus: evidence for nucleotide sugar/nucleoside monophosphate and nucleotide sulfate/nucleoside monophosphate antiports in the Golgi apparatus membrane. Proc. Nut1 Acad. Sci. U.S.A. 1984 81:7051 -5. 20. Waldman BC, Rudnick G. UDP-GlcNAc transport across the Golgi membrane: electroneutral exchange for dianionic UMP. Biochemistry 1990 29:44-52.

21. Yanagisawa K, Resnick D, Abeijon C, Robbins PW, Hirschberg CB. A guanosine diphosphatase enriched in Golgi vesicles of Saccharomyces cerevisiae. Purification and characterization, J. Riol. Chern. 1990 265:19351-19355. 22. Abeijon C, Yanagisawa K, Mandon EC, Hausler A, Moremen K, et al. Guanosine diphosphatase is required for protein and sphingolipid glycosylation in the Golgi lumen of Saccharomyces cerevisiae. J. Cell Biol. 1993 122:307-323. 23. Berninsone P, Miret JJ, Hirschberg CB. The Golgi guanosine diphosphatase is required for transport of GDP-mannose into the lumen of Saccharomyces cerevisiae Golgi vesicles. J. Biol. Chem. 1994 269:207-21 I . 24. Gao XD, Jigami Y. Yeast YNDZ gene encodes a new nucleoside diphosphatase required for maturation of N - and 0-glycosylation. G1j~cohiolo;qy1997 7: 1044. 25. Gao XD, Kaigorodov V, Jigami Y. Y N D 1 , a homologue of CDAZ, encodes membrane bound apyrase required for Golgi N- and 0-glycosylation in Succhnron?jws wrerisiue. J. Biol. C!7ern. 1999 274:21450-21456. 26. Traynor AJ, Hall ET, Walker G, Miller WH, Melancon P, et al. Inhibition of UDP-Nacetylglucosamine import into Golgi membranes by nucleoside monophosphates. J. Med. Client 1996 39:2894-2899. 27. Pels Rijcken WR, Overdijk B, Van den Eijndcn DH, Ferwerda W. The effect of increasing nucleotide-sugar concentrations on the incorporation of sugars into glycoconjugates in rat hepatocytes. Biochem. J. 1995 305:865-870. 28. Stanley P. Membrane mutants of animal cells: rapid identification of those with a primary defect in glycosylation. Mol. Cell Bid. 1985 5923 929. 29. Hara T, Endo T, Furukawa K, Kawakita M. Kobata A. Elucidation of the phenotypic change on the surface of Had-1 cell, a mutant cell line of mouse FM3A carcinoma cells selected by resistance to Newcastle disease virus infection. J . Biochem. (Tokyo) 1989 106:236-247. 30. Hara T, Hattori S, Kawakita M. Isolation and characterization of mouse FM3A cell mutants which are devoid of Newcastle disease virus receptors. J. Virol. 1989 63:182-188. 3 I . Stanley P, Ioffe E. Glycosyltransferase mutants: key to new insights in glycobiology. FASEB J. 1995 91436- 1444. 32. Deutscher SL. Nuwayhid N, Stanley P, Briles EI, Hirschberg CB. Translocation across Golgi vesicle membranes: a CHO glycosylation mutant deficient in CMP-sialic acid transport. Cell 1984 39:295- 299. 33. Deutscher SL, Hirschberg CB. Mechanism of galactosylation in the Golgi apparatus. A Chinese hamster ovary cell mutant deficient in translocation of UDP-galactose across Golgi vesicle membranes. .I. Bid. Chcm 1986 261:96 100. 34. Taki T: Ogura K, Rokukawa C, Hara T, Kawakita M, et al. Had-I, a uridine 5’diphosphogalactose transport-defective mutant of mouse mammary tumor cell FM3A: composition of glycolipids, cell growth inhibition by lactosylceramide, and loss of tumorigenicity. Cancer Rex 1991 51:1701-1707. 35. Abeijon C, Mandon EC, Robbins PW. Hirschberg CB. A mutant yeast deficient in Golgi transport of uridine diphosphate N-acetylglucosamine. J. Biol. Clzern. 1996 271:8851 8854. 36. Dean N, Zhang YB, Poster JB. The VRG4 gene is required for GDP-mannose transport into the lumen of the Golgi in the yeast, Saccharomyces cerevisiae. J. Bid. Cheni. 1997 27,7:3190831914. 37. Tabuchi M, Tanaka N, Iwahara S, Takegawa K. The Schizosaccharomyces pombe gmsl+ gene encodes an UDP-galactose transporter homologue required for protein galactosylation. Biochem. Biophys. Res. Cominun. 1997 232: 121 125. 38. Descoteaux A, Luo Y, Turco SJ, Beverley SM. A specialized pathway affecting virulence glycoconjugates of Leishmania. Science 1995 269: 1869-1 872. 39. Herman T, Hartwieg E, Horvitz HR. sqv mutants of caenorhabditis elegans are defective in vulval epithelial invagination. Proc. Not1 Acud Sci. U.S.A. 1999 Y6:968- 913. 40. Herman T, Horvitz HR. Three proteins involved in Caenorhabditis elegans vulval invagination are similar to components of a glycosylation pathway. Puoc. i V d Acud. Sci. U.S.A. 1999 96:974-979. 41. Berninsone P, Hirschberg CB. Pcrsonal communication. -

34

2 Nucleotide Sugar Transporters

42. Abeijon C, Robbins PW, Hirschberg CB. Molecular cloning of the Golgi apparatus uridine diphosphate-N- acetylglucosamine transporter from Kluyveromyces lactis. Proc. Nutl Acud. Sci. U.S.A . 1996 Y35963-5968. 43. Eckhardt M, Muhlenhoff M, Bethe A, Gerardy-Schahn R. Expression cloning of the Golgi CMP-sialic acid transporter. Proc. Natl Acad. Sci. U.S.A . 1996 93:7572-7576. 44. Eckhardt M, Gerardy-Schahn R. Molecular cloning of the hamster CMP-sialic acid transporter. Eur. J. Biochem. 1997 248:187-192. 45. Ishida N, Miura N, Yoshioka S, Kawakita M. Molecular cloning and characterization of a novel isoform of the human UDP-galactose transporter, and of related complementary DNAs belonging to the nucleotide-sugar transporter gene family. J. Biochem. (Tokyoj 1996 120:1074-1078. 46. Miura N, Ishida N, Hoshino M, Yamauchi M, Hara T, et al. Human UDP-galactose translocator: molecular cloning of a complementary DNA that complements the genetic defect of a mutant cell line deficient in UDP-galactose translocator. J. Biochem. (Tokyo) 1996 120:236241. 47. Ma D, Russell DG, Beverley SM, Turco SJ. Golgi GDP-mannose uptake requires Leishmania LPG2. A member of a eukaryotic family of putative nucleotide-sugar transporters. J. Biol. Chem. 1997 272:3799-3805. 48. Poster JB, Dean N. The yeast VRG4 gene is required for normal Golgi functions and defines a new family of related genes. J. Bid. Chem. 1996 271:3837-3845. 49. Eckhardt M, Muhlenhoff M, Bethe A, Koopman J, Frosch M, et al. Molecular characterization of eukaryotic polysialyltransferase-1. Nature 1995 373:715-718. 50. Frosch M, Gorgen I, Boulnois GJ, Timmis KN, Bitter-Suermann D. NZB mouse system for production of monoclonal antibodies to weak bacterial antigens: isolation of an IgG antibody to the polysaccharide capsules of Escherichia coli K1 and group B meningococci. Proc. Nutl Acud. Sci. U.S.A. 1985 82:1194-1198. 5 1. Baumann H, Brisson JR, Michon F, Pon R, Jennings HJ. Comparison of the conformation of the epitope of alpha(27-8) polysialic acid with its reduced and N-acyl derivatives. Biochemistry 1993 32:4007-4013. 52. Jennings, H. J. N-Propionylated Group B Meningococcal Polysaccharide Glycoconjugate Vaccine against Group B Meningococcal Meningitis. Znt. J. Infect. Dis.1997 1:158-164. 53. Puglielli L; Mandon EC, Rancour DM, Menon AK, Hirschberg CB. Identification and Purification of the Rat Liver Golgi Membrane UDP-N-acetylgalactosamine Transporter. J. Biol. Chem. 1999 274:4474-4479. 54. Berninsone P, Eckhardt M, Gerardy-Schahn R, Hirschberg CB. Functional expression of the murine Golgi CMP-sialic acid transporter in saccharomyces cerevisiae. J. Biol. Chem. 1997 272:12616-12619. 55. Sun-Wada GH, Yoshioka S, Ishida N, Kawakita M. Functional expression of the human UDP-galactose transporters in the yeast Saccharomyces cerevisiae. J. Biochem. (Tokyoj 1998 123:912-917. 56. Guillen E, Abeijon C, Hirschberg CB. Mammalian Golgi apparatus UDP-N-acetylglucosamine transporter: molecular cloning by phenotypic correction of a yeast mutant. Proc. Nutl Acud. Sci. U.S.A. 1998 95:7888-7892. 57. Bakker H. Personal communication. 58. Eckhardt M, Gotza B. Gerardy-Schahn R. Membrane topology of the mammalian CMP-sialic acid transporter. J. Biol. Chem. 1999 27453779-8787. 59. Eckhardt M, Gotza B, Gerardy-Schahn R. Mutants of the CMP-sialic acid transporter causing the Lec2 phenotype. J. Biol. Chern. 1998 273:20189-20195. 60. Mandon EC, Milla ME, Kempner E, Hirschberg CB. Purification of the Golgi adenosine 3’phosphate 5’-phosphosulfate transporter, a homodimer within the membrane. Proc. Nut1 Acad. Sci. U.S.A. 1994 YI:10707-10711. 61. Schroers A, Burkovski A, Wohlrab H, Kramer R. The phosphate carrier from yeast mitochondria. Dimerization is a prerequisite for function. J. Bid. Chem. 1998 273: 14269- 14276. 62. Sprong H, Kruithof B, Leijendekker R, Slot JW, van Meer G. et al. UDP-ga1actose:ceramide galactosyltransferase is a class I integral membrane protein of the endoplasmic reticulum. J. Biol. Chem. 1998 273:25880-25888.

ReJiwnces

35

63. Yoshioka S, Sun-Wada GH, Ishida N, Kawakita M. Expression of the human UDP-galactose transporter in the Golgi membranes of murine Had-I cells that lack the endogenous transporter. J. Biochem. (Tokyo) 1997 12269 1-695. 64. Ishida, N., Ito, M., Yoshioka, S., Sun-Wada, G.H., and Kawakita, M. Functional expression of human golgi CMP-sialic acid transporter in the Golgi complex of a transporter-deficient Chinese hamster ovary cell mutant. J. Biochmz. (Tokyo) 1998 124:171-178. 65. Ishida N, Yoshioka S, Iida M, Sudo K, Miura N , Aoki K, Kawakita M. Indispensability of transmembrane domains of Golgi UDP-Galactose transporter as revealed by analysis of genetic defects in UDP-Galactose transporter-deficient murine Hud-I mutant cell lines and construction of deletion mutants. J. Biochem. (Tokyo) 1999 126:1107-1117. 66. Kurt von Figura, Institut fur Biochmie Universitlt Gottingen, Germany. Dietmar Vestweber, Institut fur Zellbiologie, Universitat Munster, Germany. Thorsten Marquardt, Klinik und Polyklinik fur Kinderheilkunde, Universitiit Munster, Germany. 67. Etzioni, A., Frydman, M., Pollack, S., Avidor, l., Phillips, M.L., Paulson, J.C., and GershoniBaruch, R. Brief report: recurrent severe infections caused by a novel leukocyte adhesion deficiency. N. Engl. J. M a ! 1992 327: 1789- 1792. 68. Marquardt T, Brune T, Luhn K, Zimmer P, Korner C, Fabritz L, vander Werft N, Vormoor J, Freeze HH, Louwen F, Biermann B, Harms E,von Figura K, Vestweber D, Koch HG. A new patient with LeukocyteAdhesion Deficiency (LAD) IT Syndrome. J. Pediutr. 1999 f34:681-688. 69. Marquardt T, Luhn K, Srikrishna G, Freeze HH, Harms E, Vestweber D. Correction of leukocyte adhesion deficiency type I1 with oral fucose Blood 1999 943976-3985, 70. Korner C, Linnebank M, Koch HG, Harms E, von Figura K, Marquardt T. Decreased availability of GDP-L-fucose in a patient with LAD I1 with normal GDP-D-mannose dehydratase and FX protein activities J. Leukoc. Biol. 1999 66:95-98. 7 1. Liibke T, Marquardt T, von Figura K, Kdrner C. A new type of carbohydrate-deficient glycoprotein syndrome due to a decreased import of GDP-fucose into the Golgi. J. Bid. Chem. 1999 27425986-25989. 72. Marquardt T, Luhn K, Srikrishna G, Freeze HH, Harms E, Vestweber D. Manuscript submitted. 73. Huang MC, Zollner 0, Moll T, Maly P, Thall AD, Lowe JB, Vestweber D. Manuscript in preparation. 74. Fukuda M. Possible roles of tumor-associated carbohydrate antigens. Cancer Res. 1996 56:2237-2244. 75. Hakomori, S. Tumor malignancy defined by aberrant glycosylation and sphingo(g1yco)lipid metabolism. Cancer Res. 1996 56:5309-5318. 76. Noda K, Miyoshi E, Uozumi N, Yanagidani S, Ikeda Y, et al. Gene expression of alphal-6 fucosyltransferase in human hepatoma tissues: A possible implication for increased fucosylation of alpha-fetoprotein. Heputoloyy 1998 28:944-952. 77. Borges E, Pendl G, Eytner R, Steegmaier M, Zollner 0, et al. The binding of T cell-expressed P-selectin glycoprotein ligand-1 to E- and P-selectin is differentially regulated. J. Bid. Chem. 1997 272:28786-28792. 78. Minch SL, Kallio PT, Bailey JE. Tissue plasminogen activator coexpressed in Chinese hamster ovary cells with alpha(2,6)-sialyltransferase contains NeuAc alpha(2,6)Gal beta(l,4)Glc-NAcR linkages. Biotechnol. Proy. 1995 I I :348-35 1. 79. Grabenhorst E, Hoffmann A, Nimtz M. Zettlmeissl G, Conradt HS. Construction of stable BHK-21 cells coexpressing human secretory glycoproteins and human Gal(beta 1-4)GlcNAc-R alpha 2,6-sialyltransferase alpha 2,6-linked NeuAc is preferentially attached to the Gal(beta l-4)GlcNAc(beta I-2)Man(alpha 1-3)-branch of diantennary ohgosaccharides from secreted recombinant beta-trace protein. Eur. J. Biochem. 1995 232:7 18-725. 80. Toma L, Pinhal MA, Dietrich CP, Nader HB, Hirschberg CB. Transport of UDP-galactose into the Golgi lumen regulates the biosynthesis of proteoglycans. J. Bid. Chem. 1996 271 :38973901, 81. Greenberger, L.M.. Collins, K.I., Annable, T., Boni, J.P., May, M.K., Lai, F.M., Kramer, R., Citeralla, R.V., Hallett, W.A., and Powell, D. alpha-(3,4-dimethyoxyphenyl)-3,4-dihydro6,7-dimethoxy-alpha-[(4-methylphenyl)thio1-2( 1H)-isoquinolineheptanenitrile(CL 329,753): a

36

82. 83.

84.

85.

2 Nucleotide Sugar Transporters novel chemosensitizing agent for P-glycoprotein-mediated resistance with improved biological properties compared with verapamil and cyclosporine A. Oncol. Res. 1996 K:207-218. Kramer R. Analysis and modeling of substrate uptake and product release by prokaryotic and eukaryotic cells. A h . Biochem. Eny. Biotechnol. 1996 54:31-74. Klingenberg M. Dialectics in carrier research: the ADP/ATP carrier and the uncoupling protein. J. Bioenery. Biomembr. 1993 25:441-451. Kainuma M, Ishhida N, Yoko-o T, Yoshioka S, Takeuchi M, Kawakita M, Jigami Y. Coexpession of al,2 galactosyltransferase and UDP-galactose transporter efficiently galactosylates N- and 0-glycans in Saccharomyces cerevisiae. Glycohioloyy 1999 9:133-142 Zollner 0, Vestweber D. The E-selectin ligand-1 is selectively activated in Chinese hamster ovary cells by the alpha(l,3)-fucosyltransferasesIV and VII. J. Biol. Clzeni. 1996 271:3300233008.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

3 Biosynthesis of Oligosaccharyl Dolichol Slzuron S. Kray

3.1 General Overview Oligosaccharyl dolichol is a biosynthetic intermediate in the process of attaching an oligosaccharide to particular asparagine residues in proteins via an N-glycosidic bond. These glycoproteins, often referred to as N-linked glycoproteins or asparagine-linked glycoproteins, generally have 1-20'%) carbohydrate by weight which represents 1- 10 oligosaccharides attached to the protein backbone. N-linked glycoproteins are membrane-associated or soluble products of a biosynthetic system found in the secretory pathway of eukaryotic cells. These glycoproteins are diverse in structure and function and often have other co-/post-translational modifications such as lipids or sugars linked via 0-glycosidic bonds. For example, the insulin receptor, low-density lipoprotein receptor, ovalbumin, thyroglobulin, transferrin, immunoglobulins, lysosomal hydrolases, HMG-CoA reductase, and many Golgi glycosy1 transferases are N-linked glycoproteins. The function of the oligosaccharide moiety on these N-linked glycoproteins varies from one N-linked glycoprotein to another. These functions include having a role in the folding of the protein, being important for the biological half-life, being an important component of ligand-receptor interactions, affecting three-dimensional structure. and affecting solubility (see [ 11 and later Chapters of this volume). The cotranslational addition of oligosaccharide from oligosaccharyl dolichol to protein is necessary for life, as will be discussed below in this Chapter. Oligosaccharyl transferase, a multi-subunit, ER enzyme, catalyzes the transfer of oligosaccharide from oligosaccharyl dolichol to asparagine residues in an Asn-X-Ser(Thr) consensus sequence of protein (see next chapter and [2, 31). Thus, the core of each oligosaccharide of N-linked glycoproteins is not synthesized by a step-wise addition of monosaccharides onto a protein substrate, as is the case for oligosaccharides on many types of glycoproteins. Instead, the oligosaccharide core of N-linked glycoproteins is first assembled step-wise on a lipid carrier, transferred en bloc to protein, and then modified by numerous glycosidases, glycosyltransferases, and sulfo-

38

3 Biosynthesis of' Oligosaccharyl Dolichol

Figure 1. Structure of oligosaccharyl dolichol. The structure of this amphipathic phosphopolyisoprenyl glycolipid is given in this Figure. As mentioned in the text, the major isomer of dolichol in hamsters is N = 17 (4) while the major isomer of dolichol in S. pombe is N = 14,15 ( 5 ) . M = mannose; GlcNAc = N-acetylglucosamine; G = glucose.

transferases (most of which are subjects of other chapters in this volume) in order to reach its final. mature structure.

3.2 Oligosaccharyl Dolichol Oligosaccharyl dolichol is a large, amphipathic phosphopolyisoprenyl glycolipid; its structure is shown in Figure 1. This glycolipid has solubility properties similar to large, complex glycosphingolipids, in that it is not soluble in the standard lipid solvent, chlorofomxmethanol 2 : 1 but is soluble in a more polar solvent, chloroform: methanokwater 10 : 10 : 3 . The lipid portion of oligosaccharyl dolichol in any particular species is a mixture of different lengths of isoprene units. For example, in hamster cells, the dolichol moiety in oligosaccharyl dolichol ranges from 17 to 22 isoprene units with the predominant chain length being 19 isoprenyl units [4],while in Schizosacchavomyces pombe the dolichol moiety ranges from 14 to 18 isoprene units with the predominant chain length being 16 or 17 isoprene units [ 5 ] . Oligosaccharyl dolichol contains two phosphate molecules, in a pyrophosphoryl linkage. Oligosaccharyl dolichol also contains two N-acetylglucosamine (GlcNAc) residues, nine mannose (Man) residues, and three glucose (Glc) residues (Figure 1). The linkages and anomeric configurations of these sugars are identical to those found in N-linked glycoproteins that have undergone little or no processing [6]. Synthesis of the lipid moiety dolichol occurs in the cytoplasm and endoplasmic reticulum, beginning with steps identical to those of cholesterol biosynthesis, up to and including the formation of farnesyl pyrophosphate [7]. At this point, the synthesis pathways of cholesterol and dolichol branch, with dolichol synthesis proceeding with the addition of isoprenyl units to farnesyl pyrophosphate by cisprenyltransferase, yielding polyprenylpyrophosphate as the product. An important step in the synthesis of dolichol is the reduction of polyprenol, with a double bond in every isoprene unit, to dolichol, in which the ct isoprene, the one nearest the hydroxyl unit, is saturated (Figure 1 and [7]). The importance of the reduction of polyprenol to dolichol, catalyzed by polyprenol reductase, is clear from the study of Chinese hamster ovary (CHO) cells that lack polyprenol reductase activity [ 81. These cells only synthesize polyprenol and utilize polyprenol rather than dolichol for oligosaccharyl lipid and for the various

3.3 Kejs Enzytnatir. Steps in the Assembly Process

39

monoglycosylated lipids involved in the assembly of oligosaccharyl lipid (see below). The molecular basis of the lack of polyprenol reductase activity in these CHO mutants is not known yet, but the assumption is that the functional allele for polyprenol reductase is defective in these mutants; CHO cells are functionally haploid at many loci [9]. The result of a lack of polyprenol reductase activity is that glycosylation of N linked proteins is altered. Both non-glycosylation of certain sites and altered mature structures on certain sites of N-linked glycoproteins are observed in these mutants [lo]. The level of oligosaccharyl polyprenol is reduced in these mutants, due to the inability of the first enzyme in the assembly pathway (see below) to use polyprenyl phosphate [ 1 I]. Finally, the structure of the oligosaccharyl moiety of the most abundant oligosaccharyl polyprenol is truncated in these mutant cells due to differences in the abilities of various biosynthetic enzymes in the assembly pathway to use polyprenol derivatives [ 81. Recently, observations in two other systems have been reported that corroborate the importance of the structure of the lipid carrier in N-linked glycoprotein biosynthesis. A glycosylation mutant of Tryprrnosoma brucei has been shown to have reduced polyprenol reductase activity [ 121. Also, cells from patients with Carbohydrate Deficient Glycoprotein Syndrome (CDGS) were shown to have reduced polyprenol reductase activity [13]. Interestingly, in both these systems, the amount of dolichol and polyprenol was about equal, suggesting that only one of the two alleles of polyprenol reductase was mutated, and that polyprenol reductase activity is a rate-limiting step in dolichol synthesis [ 141. As expected from substrate specificity studies with membranes of CHO cells, the lipid utilized in these systems as a carrier of the oligosaccharide moiety was dolichol, since the first enzyme in the assembly pathway strongly prefers dolichyl phosphate to polyprenyl phosphate as a substrate [ 1 11. Nonetheless, these Trypuno.somi and CDGS cells displayed altered glycosylation of N-linked glycoproteins, perhaps because they had lower amounts of oligosaccharyl dolichol. The synthesis of the oligosaccharide of oligosaccharyl dolichol, detailed below, involves membrane-associated enzymes using either sugar nucleotides (UDPGlcNAc or GDP-Man) or monoglycosylated dolichols (dolichyl phosphate mannose, DPM, or dolichyl phosphate glucose, DPG) as sugar-donating substrates. The use of two types of sugar donors for the assembly of oligosaccharyl dolichol is reminicent of the biosynthesis of complex glycans in bacteria, such as peptidylglycan and lipopolysaccharide (see Chapter 26 in this volume). The metabolism of sugars and sugar nucleotides has been covered in Chapter 1 in this volume. Different enzymatic activities involved in the biosynthesis of oligosaccharyl dolichol occur on different faces of the endoplasmic reticulum of cells, and the challenges of this topology of the biosynthetic pathway will be discussed below.

3.3 Key Enzymatic Steps in the Assembly Process There are at least 14 enzymatic steps directly involved in the assembly of the oligosaccharide onto dolichyl phosphate to make oligosaccharyl dolichol. The levels

40

3 Biosynthesis of Oligosuccliuryl Dolichol

of the glycosyl transferase enzymes are thought to be very low, and the transferases are all membrane-associated. Thus, none of the transferases involved in these reactions have been studied as purified proteins from the ER membrane. The information we have about these transferases and the reactions they catalyze comes from genetics, the use of specific inhibitors, and in vitro reactions utilizing membrane preparations. The assembly begins with the addition of GlcNAc-P from UDP-GlcNAc to dolichyl phosphate, generating dolichol-P-P-GlcNAc. This reaction is catalyzed by the enzyme UDP-N-acetylglucosaminy1:dolichyl phosphate N-acetylglucosaminyl phosphoryl transferase (L-G 1PT). Studies on L-G 1PT have been performed primarily using membrane preparations and techniques such as in vitro mutagenesis, epitope tagging, peptide antibodies, and plasmid shuffling since quantities of purified, stable enzyme have thus far been impossible to obtain 11.51. Sequences of the cDNA of L-GlPT are known from hamster and mouse, and the sequences and/or structure of the gene are known from hamster, mouse, Saccharomyces cerevisiae, Schizosaccharomyces pombe, and Leishmania amazonensis ([ 161 and references therein). L-G1PT is a hydrophobic protein with its active site on the cytoplasmic face of the ER membrane 1171. L-G1PT has at least five transmembrane regions 118, 191, three residues known to be essential for enzymatic activity (arginine 303, aspartic acid 252, and asparagine 185) [18, 191, and five regions, conserved in L-GlPT and two bacterial enzymes with analogous functions, which appear important for activity and/or stability of the protein 1201. L-GlPT oligomerizes within the membrane, an unusual property for a protein with multiple transmembrane domains As mentioned above, L-G1PT has a strict substrate preference for dolichyl phosphate [ 1 I]. Its activity is stimulated specifically in vitro by mannosylphosphoryldolichol (MPD) and by phospholipids [22], although the function of this stimulation is not yet clear. Although L-G1PT appears to be constitutively expressed in all cells, its level is higher in actively growing cells. The amount of L-G1PT and its activity and the amount of L-GIPT mRNA increases when B-cells are activated by LPS [23] and in mammary gland tissue during lactation 1241; the amount of mRNA increases when growth-arrested Saccharomyces cerevisiae begin to proliferate [25]. LTGIPT is inhibited by a fungal antibiotic tunicamycin (TM), which has a structure that suggests it is a bisubstrate analog [26]. Therefore, TM has been used to inhibit N-linked glycosylation in a large number of systems since it is specific for this initial enzyme in the assembly pathway of oligosaccharyl dolichol [27]. TM is cytotoxic to cells, suggesting that N-linked glycosylation is crucial for cell viability 1281. Haploid S. cerevisiae cells or Schizosaccharomyces pombe cells with a disruption in the gene for L-G1PT are not viable [16, 291. Transgenic mice with a disrupted L-G1PT have been generated (see Chapter 55 in this volume), and embryos die between days 4-5 post-fertilization, again indicating the essential role of LGlPT in cell viability [30]. Mammalian cells treated with TM for as short a time as 4 h die by apoptosis, suggesting that altered glycosylation may be an endogenous signal for induction of apoptosis [ 3 11. The next six steps in the pathway, resulting in the elongation of dolicho1-P-P-

3.3 Key Enzymatic Steps in the Assembly Process

41

GlcNAc to dolichol-P-P-GlcNAc2M an5, are catalyzed by membrane-associated transferases that utilize sugar nucleotides as the sugar donors. Intermediates in this pathway normally do not accumulate, as one rarely detects any labeled products smaller than dolichol-P-P-GlcNAcZMan5 during metabolic labeling experiments with [2-3H]Man in parental cells. What is known about these transferases comes from studies of the yeast mutants akq1 (accumulates dolichol-P-P-GlcNAc2) and alg2 (accumulates dolichol-P-P-GlcNAc:Man*) [32]. Deletion of these genes is lethal. The addition of the final four Man residues to dolichol-P-P-GlcNAcZMans to form dolichol-P-P-GlcNAc2Mang is catalyzed by membrane-associated mannosyl transferases, products of ALQ, ALG9, and ALGl2 [32]. These genes are not essential for cell viability. The mannosyl transferases encoded by these genes utilize DPM as the sugar donor. DPM is synthesized in cells by DPM synthase, an endoplasmic reticular membrane enzyme, that catalyzes the addition of Man to dolichyl phosphate from GDP-Man. The structure of DPM synthase appears to be either a single polypeptide of about 26 KDa (encoded by D P M l ) in Saccharomyces cevevisiae, T. brucei and Ustilage maydis [33] or two polypeptides, dpmlp and dpm2p in Schizosaccharomyces pombe, humans, hamster, mouse, and nematode [33, 341. Dpmlp presumably contains the catalytic site. Sequence analysis of human D P M l revealed a conserved serine moiety and neighboring conserved residues that met criteria for a consensus site for phosphorylation by CAMP-dependent kinase [33], consistent with earlier biochemical studies [ 351 that suggested that the activity of DPM synthase was stimulated by phosphorylation. These biochemical studies. done with membrane fractions, showed a two-fold increase in DPM synthase activity in membranes either isolated from isoproterenol-treated cells or incubated in vitro with CAMP-dependent protein kinase. Dpm2p is a hydrophobic protein of only 84 amino acids and may function to correctly localize and stabilize dpmlp [36]. It is important to note that cells in culture can grow without DPM synthase activity. For example, Thy1 thyoma cells, with a mutation in DPMl [34], and Lecl5.1 (B4-2-1 [58])or Lec15.2 fibroblasts [36], with mutations in DPM2, survive, although the expression of cell-surface molecules (Thy antigen or mannose 6phosphate receptor) in these cells is altered. B4-2-1 cells do die at elevated temperatures [31]. However, while DPM synthase activity is not necessary for cell viability, lack of DPM synthase activity has serious consequences at the organism level. Cells of CGDS patients lacking DPM synthase activity have been found [37]. The final three steps in the assembly of oligosaccharyl dolichol are catalyzed by three different membrane-associated glucosyltransferases that utilize DPG as the sugar donor. A lack of these glucosyltransferase activities either in yeast (alg6, aly8, alglO, references [32, 381) or in mammalian cells [39, 591 does not result in cell death. Again, however, there appears to be serious consequences to the organism of a lack of glucosyltranferase activity, as cells of CGDS patients lacking glucosyltransferase activity have been found [40, 411. A dogma in the field had been that the presence of glucose in oligosaccharyl dolichol was essential for the transfer of oligosaccharide from oligosaccharyl dolichol to protein [42]. While in vivo studies using mutants synthesizing a truncated oligosaccharyl dolichol showed a preference for the transfer of glucosylated oligosac-

42

3 Biosynthesis of Oligosaccharyl Dolichol

charides [43, 441, it is clear from the work on alg6 and MI8-5 mutants that transfer of non-glucosylated GlcNAcZMang is only slightly less (no more than two-fold) than glucosylated oligosaccharide [38, 391. The function of the presence of glucose in oligosaccharyl dolichol is not clear at the present time, but glucose residues may be involved in determining the level of oligosaccharyl dolichol.

3.4 Topology of the Assembly Process One of the intriguing, yet complicating aspects of the assembly of oligosaccharyl dolichol is that it occurs on both faces of the ER [45]. That is, the first seven steps of the assembly process are thought to occur on the cytoplasmic face [46].For example, as mentioned above, the catalytic site of L-G1PT has been localized to the cytoplasmic side of the ER [ 171. The intermediate dolichol-P-P-GlcNAczMan5 is thought to be translocated through the bilayer to face the lumenal side of the endoplasmic reticulum, and subsequently elongated to dolichol-P-P-GlcNAc2MangGlc3. The oligosaccharyl transferase active site is thought to be on the lumenal side of the ER [2]. In addition, other molecules involved in the assembly process must be translocated. For example, DPM and DPG are synthesized on the cytoplasmic face and then translocated to the lumenal face. Dolichyl pyrophosphate or dolichyl phosphate, generated on the lumenal face by transfer of the oligosaccharide from oligosaccharyl dolichol to protein, must be translocated back to the cytosolic face in order to be used in the L-G1PT reaction to begin the assembly process again. While it is generally thought that these translocation events are protein-mediated, little experimental data exists to document these activities. The best work in this area is being done with water-soluble analogs of dolichol by Waechter and coworkers [47, 481. Their data are consistent with the presence of a membrane-associated “flippase” that mediates transbilayer movement of the saccharyl dolichols.

3.5 Utilization of Oligosaccharyl Dolichol As has been mentioned earlier in this chapter, oligosaccharyl dolichol is a substrate for oligosaccharyl transferase, generating glycoprotein and dolichyl pyrophosphate [2]. The latter is hyrolyzed by a phosphatase, regenerating dolichyl phosphate for use in subsequent assembly cycles [49]. The level of dolichyl phosphate is clearly important in the regulation of the amount of assembly of oligosaccharyl dolichol that occurs [7, 501. Oligosaccharyl dolichol is the major dolichol form in growing cells, with DPM, GPM, dolichol, and dolichyl phosphate being present in lower amounts [51]. Besides being a substrate for oligosaccharyl transferase, oligosaccharyl dolichol is

References

43

also catabolized to free oligosaccharide in the lumen of the ER [52, 531. Shorter, non-glucosylated oligosaccharyl dolichols, such as dolichol-P-P-GlcNAc2Mans, are substrates for a pyrophosphatase present on the cytoplasmic face of the ER, yielding free oligosaccharidyl phosphates [ 531. Thus it is clear that oligosaccharyl dolichol is catabolized by at least two pathways. The function of this catabolism is perhaps to regulate the level of oligosaccharyl dolichol in cells [54]. There is a transport system for non-glucosylated free oligosaccharides out of the ER [55]. Cytoplasmic catabolism of oligosaccharides is rapid, yielding GlcNAcMan5 [ 561, which is specifically transported into the lysosome [57].

Acknowledgment

The author gratefully acknowledges support from NIH grant R01 CA20421. References 1. 2. 3. 4. 5.

A. Varki, Glycobioloyy, 1993, 3, 97-130. S. Silberstein, and R. Gilmore, FASEB J . , 1996, 10. 849-858. R. Knauer and L. Lehle, Biochim. Biophys. Actu, 1999, 1426, 259-273. J . Stoll, A.G. Rosenwald, and S.S. Krag, J. B i d Chem., 1988, 263, 10774-10782. G.J. Quellhorst, Jr., J.S. Piotrowski, S.E. Steffen, and S.S. Krag, Biochenz. Bioplzys. Research Conzm., 1998, 244, 546-550. 6. E. Li, 1. Tabas, and S. Kornfeld, J. Biol. Clzem. 1978, 253, 7762-7770. I. A. Kaiden and S.S. Krag, Trends G1yco.sc.i. Glycoteclinol., 1991, 3, 275-287. 8. S.S. Krag, Biochem. Biophys. Rex Comm., 1998, 243, 1-5. 9. G.M. Adair and M.J. Siciliano, Somutic Cell Mol. Gen., 1986, 12, 11 1-1 19. 10. A.G. Rosenwald, P. Stanley, and S.S. Krag, Mol. C d . Biol., 1989, 9, 914-924. 11. K.R. McLachlan and S.S. Krag, Glycohioloyj~,1992, 2, 313-319. 12. A. Acosta-Serrano, K.-Y. Hwa, G.J. Jr. Quellhorst, H. Lei, S.S. Krag, and P.T. Englund, Mol. Parrcsit. Mtg VIII, 1997, Woods Hole, MA. 13. T. Ohkura, K. Fukushima, A. Kurisaki, H. Sagami, K. Ogura, K. Ohno, S. Hara-Kuge, and K. Yamashita, J. Bid. Chtw., 1997, 272, 6868-6875. 14. A.G. Rosenwald, P. Stanley, K.R. McLachlan, and S.S. Krag, Glycohiology, 1993,3, 481-488. 15. K. Shailubhai. B. Dong-Yu, E.S. Saxena, and I.K. Vijay, .I. Biol. Chem., 1988, 263, 1596415972. 16. J. Zou, J.R. Scocca, and S.S. Krag, Arch. Biochem. Biophys., 1995, 317, 487-496. 17. E.L. Kean, J. Biol. Chem., 1991,266, 942-946. 18. N. Dan, R.B. Middleton, and M.A. Lehrman, J. Biol. Chem.. 1996,271, 30717-30724. 19. J.R. Scocca and S.S. Krag, Glycohiokocjy, 1997, 7, 1181-1191. 20. A.R.D. Nogare, N. Dan, and M.A. Lehrman, Gljmhiology, 1998, 8, 625-632. 21. N . Dan and M.A. Lehrman, J. Bid. Clzem., 1997, 272, 14214-14219. 22. E.L. Kean, J. Bid. Chern., 1985,260, 12561-12571. 23. D.C. Crick, J.R. Scocca, J.S. Rush, D.W. Frank, S.S. Krag, andC.J. Waechter, J. Bid. Chenz., 1994,269, 10559-10565. 24. J. Ma, H. Saito, T. Oka, and I.K. Vijay, J. Biol. Chem., 1996, 271, 11197-11203. 25. M.A. Kukuruzinska and K. Lennon-Hopkins, Bioclzim. Biophys. Acta, 1999, 1426, 359-372. 26. A. Takatsuki, K. Kawamura, M. Okina, Y. Kodama, T. Ito, and G. Tamura, Agric. Biol. Chem., 1977,41,2307-2309. 27. Y.T. Pan and A.D. Elbein, 1995, in Glycoproteins, Montreuil, J., Schachter, H., and Vliegenthard, J.F.G., eds. Elsevier Science, Amsterdam, pp. 415-454.

44 28. 29. 30. 31. 32. 33.

3 Biosynthesis of Oligosaccharyl Dolichol

B.A. Criscuolo and S.S. Krag. J. Cell Biol., 1982, 94, 586-591. M.A. Kukuruzinska and P.W. Robbins, Proc. Nut1 Acad. Sci. USA, 1987,84, 214552149, K.W. Marek, I.K. Vijay, and Marth, J.D., Glycohiology, 1999, 9, 1263-1271. B.K. Walker, H. Lei, and S.S. Krag. Biochem. Biophys. Rex Comrnun., 1998. 250, 264-270. P. Burda and M. Aebi, Biochim. Biophys. Acta, 1999, 1426, 239-257. P.A. Colussi, C.H. Taron, J.C.Mack, and P. Orlean, Proc. Nut1 Acad. Sci. USA, 1997, 94, 7873-7878. 34. S. Tomita, N. Inoue, Y. Maeda, K. Ohishi, J. Takeda, and T. Kinoshita, J. Biol. Chem., 1998, 273, 9249-9254. 35. D.K. Banerjee, E.E. Kousvelari, and B.J. Baum, Proc. Vat1 Acad. Sci. USA, 1987, 84, 63896393. 36. Y. Maeda, S. Tomita, R. Watanabe, K. Ohishi, and T. Kinoshita, EMBO J., 1998, 17, 49204929. 37. S. Kim, D. Mehta, G. Srikrishna, S. Murch, and H. Freeze, Glycobioloyy, 1998,8, abstract 153. 38. G. Reiss, S. te Heesen, J. Zimmerman, P.W. Robbins, and M. Aebi, Glycohioloyy, 1996, 6, 493-498. 39. G.J. Quellhorst, Jr., J.L. O’Rear, R. Cacan, A. Verbert, and S.S. Krag, GIycohiology, 1999, 9: 65-72. 40. P. Burda, L. Borsig, J. de Rijk-van Andel, R. Wevers, J. Jaeken, H. Carchon, E.G. Berger, and M. Aebi, J. Clin. Invest., 1998, 102, 647 652. 41. T. Imbach, P. Burda, P. Kuhnert, R.A. Wevers, M. Aebi, E.G. Berger, and T. Hennet. Proc. Natl Acad. Sci. USA, 1999, 96, 6982-6987. 42. L.A. Murphy and R.G. Spiro, J. Biol. Chem., 1981,256, 7487-7494. 43. S. Kornfeld, W. Gregory, and A. Chapman, .I. Biol. Chew?.,1979,254, 11649-11654. 44. J. Stoll, R. Cacan, A. Verbert, and S.S. Krag, Arch. Biochem. Biophys., 1992, 299, 225-231. 45. C.B. Hirschberg and M.D. Snider, Annu. Rev. Biockenz., 1987, 56, 63-87. 46. C. Abeijon and C.B. Hirschberg, J. Bid. Chern., 1990,265, 14691-14695. 47. J.S. Rush and C.J. Waechter, J. Cell Biol., 1995, 130, 529-536. 48. J.S. Rush, K. van Leyen, 0. Ouerfelli, B. Wolucka, and C.J. Waechter, GljXdJiohgy,1998, 8, 1195-1205. 49. D. W. Frank and C.J. Waechter, J. Bid. Chern., 1998, 273, 11791-11798. 50. A.G. Rosenwald, J. Stoll, and S.S. b a g , J. Biol. Chem., 1990, 265, 14544-14553. 51. A. Kaiden and S.S. Krag, Biochem. Cell Biol., 1992, 70, 385S389. 52. M.J. Spiro and R.G. Spiro, J. Bid. Chem., 1991, 266, 5311-5317. 53. R. Cacan, C. Villers, M. Belard, A. Kaiden, Krag, S.S. and A. Verbert, Glycobioloyy, 1992. 2, 127- 136. 54. R. Cacan and A. Verbert, Trends Gl.ycosci. Glycotechn., 1997, 9, 365-377. 55. S.E.H. Moore, C. Bauvy, and P. Codogno. EMBO J.. 1995,14, 6034-6042. 56. D. Kmiecik, V. Herman, C.J.M. Stroop, J.-C. Michalski, A.-M. Mir, 0. Labiau, A. Verbert, and R. Cacan, Glycobiology, 1995. 5, 483-494. 57. A. Saint-Pol, C. Bauvy, P. Codogno, and S.E.H. Moore, J. Cell Biol., 1997, 136, 45-59. 58. L. Pu, J.R. Scocca, and S.S. Krag, unpublished. 59. G.J. Quellhorst, Jr., J.L. O’Rear, R. Cacan, A. Verbert, and S.S. Krag, unpublished.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

4 Biochemistry and Molecular Biology of the N-Oligosaccharyltransferase Complex Roland Knauer and Ludwig Lehle

4.1 Introduction Protein glycosylation is one of the most common types of eukaryotic protein modification [I-51. There is clear evidence that it also occurs in many archaea [6, 71, whereas in eubacteria it is found only exceptionally [8]. Both N-linked (9-1 11 and O-linked [ 121 glycans are essential for cell viability and can have a profound role in the biological function and in physicochemical properties of many secreted and integral membrane proteins [ 131. The first commited step of the N-glycosylation pathway is the en bloc transfer of the preassembled high mannose core oligosaccharide Glc3MangGlcNAc2 from dolichyl pyrophosphate onto selected Asn-XSer/Thr acceptor sites of nascent polypeptide chains during or shortly after their translocation into the lumen of the endoplasmic reticulum. It is catalyzed by the hetero-oligomeric membrane-bound enzyme complex N-oligosaccharyltransferase (OST) (Figure 1). It is these processes, the assembly of the lipid-linked oligosaccharide and the so called “core glycosylation” reaction of the protein at the ER, that are highly conserved during evolution, whereas subsequent diversification of the primary glycoprotein conjugate occurs along the secretory pathway, mainly in the Golgi, in a species and cell type specific manner. Previous attempts to purify and characterize OST have been hindered by the lability of the enzyme upon solubilization [ 14-16] and probably also because of its hetero-oligomeric nature. Meanwhile, however, OST complexes have been purified from different sources, such as dog [17, 181, yeast [19-231, hen [24], human [25, 261 and pig [27]. The subunit composition of the various isolated complexes and the protein sequences so far obtained reveal indeed a high degree of conservation. In this review we will try to summarize the current knowledge of biochemistry and molecular biology of the OST from lower and higher eukaryotes. Sequence search in databases indicate also homologous proteins in archaea that may be involved in the N-glycosylation of these organisms. The reader is also referred to previous reviews dealing with this topic 128-3 11.

46

4 Biochemistry and Molecular Biology Ribosome

OST

m

Cytosol

ER lumen

Man

Man

Man I G,Ic G,Ic Glc

Figure 1. The glycosylation reaction catalyzed by oligosaccharyltransferase. The unique core oligosaccharide Glc3MangGlcNAcz is transferred from dolichyl pyrophosphate to selected asparagine residues in an Asn-X-Ser/Thr consensus sequon of nascent polypeptides. The dolichol moiety contains an saturated a-isoprenoid unit and may vary in chain length. The OST complex from yeast is shown. Subunits in bold face are essential for yeast growth; SEC61 represents the yeast protein translocation complex.

4.2 Biochemistry of OST OST catalyzes an unusual reaction, rare in mechanistic enzymology, in which the nucleophilicity of the amide nitrogen of the asparagine residue must be enhanced to make a covalent bond with the oligosaccharide (see below). Before purified OST became available, basic biochemical characteristics of the enzyme were determined using microsomal membranes or detergent solubilized extracts. OST exhibits a requirement for divalent cations with a preference for Mn2+ over Mg2+ [15, 16, 27, 32-34]. The pH optimum of the enzyme is between 6.5 and 7.5 [15, 27, 341. OST from different sources has been shown to be strongly stabilized by addition of phos-

4.2 Biochemistry of OST

47

pholipids upon detergent solubilization [17, 21, 23, 27, 35, 361, sucrose or glycerol [15, 271. OST activity is determined in vitro using an appropriate lipid-linked saccharide as glycosyl donor [ 151 and a peptide acceptor containing the Asn-X-Thr/Ser recognition motif, whereby X can be every amino acid except proline [37].

4.2.1 Lipid-Saccharide Donor Full length core oligosaccharide DolPP-GlcNAczMang Glc3 is the optimal glycosyl donor both in vitro [15, 27, 33, 38-40] and in vivo [41-441. The minimal glycosyl donor in vitro is DolPP-GlcNAcz; no transfer is observed from DolPP-GlcNAcl [ 151. The glucose residues enhance glycosyl transfer presumably by adopting a favourable conformation contributing to oligosaccharide recognition by the OST [45, 461 and thus affecting the apparent binding affinity for the acceptor substrate [15, 271. For the yeast enzyme a ten times lower K, value for the acceptor hexapeptide YNLTSV was determined, when DolPP-GlcNAc2MangGlc~is the donor compared to DolPP-GlcNAc2 [15]. Thus OST may display two substrate recognition sites, the chitobiose site and a glucose binding domain [ 15, 461. Transfer of incomplete core oligosaccharides was originally observed in vitro [15, 47-49], but later shown also to occur in vivo [50]. In alg (asparaginelinked glycosylation) yeast mutants [9, 50, and reviewed in 511 which have a defect in lipid-linked oligosaccharide assembly, shorter glycan chains are transferred to proteins: e.g. Man*-zGlcNAc2 in a1q2 [52, 531; MansGlcNAc2 in 4 3 [42, 43, 541; MangGlcNAcz in alg.5 and alg6 [55-571; Man7GlcNAcz in alg12 [58]; Glcl MangGlcNAcz in a478 [59]; Man6GlcNAc2 in aly9 [60] or GlczMangGlcNAc2 in alglO [61]. Depending on the protein, not only truncated saccharides, but also fewer chains may be transferred, and to selected glycosylation sites [42, 43, 50, 62651. The basis for this underglycosylation appears to be the reduced affinity of the OST towards truncated oligosaccharides. Glucosylation is not essential for viability of yeast cells under laboratory growth conditions, but double mutants with a simultaneous defect in glucose addition and OST activity [59, 661 may display a temperature-sensitive growth phenotype. Nonglucosylated oligosaccharides are transferred to protein in trypanosomatid protozoa, which lack DolP-Glc synthase, and depending on the strain, may also have a defect in lipid-oligosaccharide mannosylation [67, and reviewed in 681, or in mammalian mutants that synthesize primarily Man5GlcNAcz (lack of DolP-Man synthase) [44, 691 or MangGlcNAc2 (lack of glucosyltransferase) [70]. The dolichol moiety of lipid-intermediates is not homogeneous but a mixture of 14-20 isoprene units [71-731. For the yeast OST it was shown that the enzyme hardly discriminates specific isoprene chain lengths [ 741. Whether unsaturated polyprenols, preferred in eubacterial cell wall synthesis, will be accepted is not known. However, the formation of the DolPP-GlcNAc, i.e. the first step in lipid-linked precursor formation shows an absolute dependency for saturation of the w-isoprenoid unit, i.e for dolichol-type polyprenols [74],what seems to be generally true for other eukaryotic glycosyltransferases involved in lipid-linked sugar formation [75, 761.

48

4 Biochemistry and Molecular Biology

4.2.2 Acceptor Specificity of OST Insights into the structural requirements for the peptidyl substrate were obtained both by using synthetic peptides or by engineering novel sequons into recombinant proteins. An Asn-X-Thr/Ser tripeptide is a sufficient in vitro acceptor provided that both termini are blocked [77, 47-49]. Only rare exceptions from the consensus “sequon” Asn-X-Thr/Ser occur, e.g. glycosylation of Asn-X-Cys [ 78-80] and Asn-GlyGly-Thr [XI] have been described. In vitro, peptides with Asn-X-Thr are about 40fold better substrates for the OST than peptides containing an Asn-X-Ser sequon [82]. This holds also true in vivo, even though the preference for Thr is less pronounced [81, 831. Peptides with Pro in the position X do not function as acceptors [82, 841 and fail to inhibit co-translational glycosylation in cell free systems [8S-87]; also introduction of Pro into recombinant sequons blocks glycosylation [86, 881. Pro in the flanking region is inhibitory when present C-terminal to the sequon, whereas OST function is not influenced when Pro immediately preceeds the sequon [XI, 82, 85, 891. Besides Pro, also large hydrophobic amino acids (e.g. Trp) or negatively charged ones (e.g. Asp and Glu) in the position Ximpair glycosylation [86, 89, 901. Finally, the amino acid following a sequon was found to be an important determinant of glycosylation efficiency [89). Sequons near the C-terminus of a protein are less likely used by the transferase [XI, 91, 921, perhaps because prior or ongoing folding of the protein imposes physical constraints on glycosyl transfer. The minimum distance of an acceptor site from the lumenal end of a transmembrane segment was determined to be 12-14 residues, suggesting that the active site of the OST is positioned 30-40 A above the ER membrane surface [93]. N-glycosylation of multispan membrane proteins reveals that glycosylated sequons occur in the first extracytosolic loop, provided that it is greater than 30 residues in size and spaced at least 10 amino acids from the transmembrane domain [94]. It has been speculated that in polytopic membrane proteins, only one polypeptide segment can be optimally occupied by the OST during protein synthesis P41. Since about 10-30% of potential sequons are either not glycosylated at all, or are not efficiently glycosylated, additional, as yet not well defined factors. influence N-glycosylation in vivo [XI, 951. Protein folding is assumed to favour or hinder OST action [96-981, or may influence core glycosylation at a further sequon. Thus, deletion of the Am297 in gpSS of Friend spleen focus-forming virus leads to glycosylation of a downstream sequon that is normally not used [79, 96, 99, 1001. Conversely, glycosylation can influence the local structure of the glycosylation site [ 101-1 031 by altering the solvation properties, by participating in hydrogen bonding interactions with the polypeptide, or for sterical reasons. NMR studies on glycopeptides provided evidence that derivatization with N-acetylchitobiose induces conformational switching into a p-turn structure, similar to the structure found in many native proteins [101, 102, 1041. On the other hand it is clear also that other types of secondary structures fulfil the spatial requirement of having the glycan exposed at the surface of proteins [103, 81, 9.51, and the widely believed assumption that the glycosylated tripeptide sequence has to adopt a p-turn conformation no longer holds true (see also below).

4.2 Biochemistry of OST

49

4.2.3 Catalytic Mechanism of OST

Since N- and C-terminal capped tripeptides are the simplest substrates for Nglycosylation, local turn motifs, such as a p-turn or an Asx-turn (Figure 2) seem to be a sufficient secondary structural requirement, rather than an a-helix or P-sheet which would require longer peptides for complete formation. Using a series of conformationally constrained peptide substrates Imperiali et al. 130, 1051 suggest the Asx-turn as the favourable conformation and propose a plausible reaction mechanism involving carboxamide tautomerization in which hydrogen-bond contacts are formed between the p-carbonyl group of asparagine and the a-NH and P-OH group, respectively, of the hydroxyamino acid. This enhances a proton dissociation from the p-nitrogen generating an imidate yielding the reactive nucleophile required for glycosyl transfer (Figure 2, model A). An alternate mechanistic model by Bause et al. [106-108], which can also explain both specificity and enhanced amide nitrogen nucleophilicity, proposes that the pamide nitrogen is the hydrogen-bond donor generating an anionic species (Figure 2, model B). The essential threonine/serine hydroxy group would act in this case as the acceptor upon deprotonation by an enzyme-bound base. However, this mechanism requires a p-turn or loop conformation, what, as discussed above, seems to be rather unlikely. For a comprehensive discussion that addresses the catalytic mechanism see 130, 3 1, 109, 110, 1061. Independent of this, the unique ability of asparagine residues to adopt a specific loop structure may explain, why e.g. a Gln-X-Thr is not glycosylated, although this sequence is found in nature. Using photoprobes containing the Asn-X-Thr sequence, in which X wasp-benzoylphenylalanine, a protein could be labelled in yeast microsomes that was identified as Ostlp indicating that this subunit may recognize the peptide glycosylation site in nascent chains [ 11 11. Similarly, it was shown that an epoxyethylglycyl acceptor peptide irreversibly inactivates pig liver OST and simultaneously leads in the presence of radioactive DolPP-GlcNAcz to the labelling of the 66 kDa ribophorin I subunit (the homolog of Ostlp) and OST48 (the homolog of Wbplp) by covalent attachment via the epoxy-group [ 1081. The reason why two subunits become labelled is not clear at present. Perhaps it may indicate that the active site is placed at the interface between the two components. For yeast OST circumstantial evidence exists in favour of Wbplp being also a subunit involved in catalysis. Thus, the thiol-modifying reagent biotinoyl(aminoethane)-thiosulfonate was found to inactivate enzyme activity and to label Wbpl p. Since furthermore DolPP-GlcNAcz was able to protect the inactivation of the enzyme, it was proposed that WbpIp may contain a binding site for the lipidlinked oligosaccharide [21]. On the other hand, none of the three Cys residues in WbpIp is conserved in the corresponding eukaryotic OST48 sequences; hence a direct role of Cys residues in the catalysis is doubtful. The development of peptidyl OST inhibitors has also been reported [112]; they may contribute to the structural and functional analysis of the enzyme, or being helpful diagnostic tools to evaluate the role of protein glycosylation and glycoproteins in vivo.

50

4 Biochemistry and Moleculur Biology

A) Y

CH3

*d Enzyme

Model A

Model B

P-Turn

ASX-Turn

Figure 2. Proposed reaction mechanisms for oligosaccharyltransferase. (A) Model A by Imperiali et al. [39, 31, 1051 proposes an Asx-turn conformation for the peptide backbone. Hydrogen bonding interactions occur between the asparagine p-carbonyl group and both the backbone amide and hydroxyl group of threonine or serine. Enzyme-mediated deprotonation at the asparagine nitrogen by a basic residue in the active site induces tautomerization of the carboxdmide to an imidol species. The generated nucleophilic species could then react with the electrophilic lipid-linked oligosaccharide. Model B by Bause et a]. [106-108] favours the existence of a 0-turn or other loop structures stabilized by hydrogen bonds between the P-amide nitrogen and the hydroxyl group of the hydroxy amino acid leading to amide deprotonation. (B) Comparison of an Asx-turn and a p-turn conformation for the Ac-Asn-X-Thr-NHz tripeptide.

4.3 Isolation of OST C'orwpkexes)om Different Sources

51

4.2.4 Regulation of OST Activity There is little information available dealing with this aspect. Hormones are known to influence N-glycosylation of proteins. But in most cases the type of regulation is unclear. The observed enhancement could be indirect by stimulating lipidintermediates [ I 13, 1141 or due to accelerated protein synthesis rather than to an increase of OST action. A marked stimulation of OST activity (4.2-fold) was reported in porcine thyroid cells stimulated by thyrotropin 11 131. In human lymphocytes OST activity increased ten-fold by interleukin-2 mitogen activation and remained elevated during long-term culture in the presence of interleukin-2 1261. A different type of regulation of N-glycosylation was suggested to occur by OST transferring the oligosaccharide to water as acceptor, if appropriate protein synthesis is not available [ 115-1 17, 401. Moreover, also transcriptional regulation could take place. S W P l and also WBPl from yeast were reported to be regulated in a cell proliferation-dependent manner. Together with the ALG genes, they are down regulated when cells enter the Go-phase, and they are induced following growth stimulation in absence of de novo protein synthesis [ 1181.

4.3 Isolation of OST Complexes from Different Sources The first isolation of an active OST complex, composed of three subunits (ribophorin I and 11, and OST48) succeeded from dog pancreas cells [17]. The ribophorins had been characterized earlier as possible receptors for ribosomes and thought to be involved in protein translocation [ 119- 1211. Since OST activity could be depleted from detergent solubilized extract by anti-ribophorin I antibodies, it became clear that ribophorins are indeed constituents of OST (171. They are no longer believed to be involved in ribosome binding, but in accord with the cotranslational nature of N-glycosylation, the identification of ribophorins as part of the OST may indicate a close location of the enzyme to the protein translocation channel. A breakthrough in the understanding of composition and molecular biology of OST was achieved upon isolation of two essential genes from yeast, WBPl and S W P l , and the demonstration that they are required for N-glycosylation in vitro and in vivo [122, 1231. Subsequently, independent OST purifications have been reported from Succharornyces cerevisiue yielding enzymatically active complexes consisting of either four polypeptides: Ostlp (64/62 kDa), Wbplp (47 kDa), Ost3p (34 kDa), and Swplp (30 kDa) 119, 211, or five ([22] in addition Ost2p; 16 kDa), or of six subunits (1201 in addition Ost2p and Ost5p; 9.5 kDa). Apparently, Ost2p and Ost5p are dispensable for enzyme activity in vitro. In the meantime, yeast-specific genetic approaches, such as high copy number suppression and synthetic lethality, or searches of the yeast protein sequence database, have identified STT3, OST4 and OST6 as additional OST genes. Studies using epitope-tagged OST proteins or specific antibodies against various subunits [ 124, 1251, or investigating the complex by

52

4 Biochemistr,v and Molecular Biology

Table 1. Comparison of OST complex from different organisms. Yeast

Mol. wt. Essential Dog (k) pancreas

Ostlp

64/62

Stt3p Wbplp Ost3p Ost6p Swplp

-60 47 34 32 30

Ost2p Ost5p Ost4p

16 9.5 3.6

+ + -

Hen Chicken Pig oviduct liver liver

Ribophorin I 65-1 (66)

65-1

66

~

~

~

OST48 (48) ~

~

50 ~

50 ~

+

65-11 ~

65-11

65-1 ._

65-1 ~

48 (40*) 50 ~

~

Ribophorin I1 (63/64) DADl (12)

Human Human liver lymphocytes

6 1/63

...

-

65-11

50 ~

~

65-11

~

-x

The numbers designate the apparent molccular weight of the subunits as determined by gel electrophoresis. : ts, necessary for growth at elevated temperatures. *: An additional 40 kDa band of unknown origin was reported for pig liver.

blue native polyacrylamide electrophoresis [23] indicate that these subunits are indeed constituents of the complex and necessary for optimal glycosyl transfer. They seem to have been lost during purification (for further discussion see below). There is also evidence that some of the subunits are not easily detectable by protein staining methods or reveal an abnormal migration behaviour in SDS-PAGE [23, 1241. In Table 1 a compilation of OST complexes from different organisms and the homology between the various subunits is depicted. In most cases the genes have been cloned or partial sequence information is available to allow their classification. Compared to yeast, the mammalian enzymes seem to be composed of fewer subunits. In the light of the discussion above. one may predict that additional subunits may be detected that have escaped detection so far. In fact, it was subsequently found that the trimeric OST complex from dog pancreas [17], contains in addition DADl (defender of apoptotic death) [18] which is 40% identical to Ost2p from yeast. The isolation of “incomplete”, but active complexes indicates that the respective subunits may be sufficient to constitute a “catalytical core” unit.

4.4 Molecular Biology of OST From the nine yeast genes encoding OST subunits, five are essential for growth of yeast under laboratory conditions (Table 1). In Figure 3 the presumed membrane topology of the derived proteins is depicted. Due to the high conservation of the

4.4Molecuhr Biology Dog pancreas

of OST

53

Yeast

N=--=fl=N ER membrane

Ribophorin I

Ostlp Ost5p

Ribophorin II

SWPl P

OST48

WbPl P

DAD1

Ost2p

Stt3p

C Ost4p N

Ost3p Ost6p

? = N-linked oligosaccharide 0

= signal sequence

I

I

I

I

I

0

100

200

300

400

ER lumen

1 -

500 0

100 cytosol

Figure 3. Predicted membrane topology of the oligosaccharyltransferase subunits from yeast. The topology is based partially on direct or indirect experimental evidence (N-terminal sequencing, protease accessibility, glycosylation sites, HIS4 fusions) as discussed. The presentation is not to scale. For comparison, the corresponding honiologues of the OST complex from dog pancreas are depicted which have the same predicted topology (not shown).

pathway, it is not unanticipated that both primary sequences and also their membrane topology are conserved. In the following, guided by the known OST genes from yeast, corresponding homologues from higher eukaryotes and homologous sequences found in databases will be discussed along with their possible properties and functions.

54

4 Biochemistry and Molecular Biologj,

4.4.1 WBPl /OST48 WBPl (wheat germ agglutinin binding protein) was originally identified as a yeast gene encoding an essential ER type I glycoprotein with the apparent molecular mass of 47 kDa and it received its name on account of its lectin binding property [126]. Later it was shown that W B P l is required for OST activity in vitro and in vivo 11221 and represents a component of the enzyme 119, 20, 231. Depletion of Wbplp in response to a regulatory promotor, or conditional wbpl mutations lead to an underglycosylation of yeast glycoproteins or to a reduced oligosaccharide transfer to an acceptor hexapeptide 11221. The membrane topology of Wbplp was determined using the HIS4C fusion technique [126]. The protein contains a short cytoplasmically exposed carboxy-terminal domain with the ER-retrieval motif KKXX, which, however, is not absolutely required for proper Wbplp function and localization [ 1271. It is assumed that OST complex formation is primarily responsible for ER retention of Wbplp. A genetic interaction between the wbpl and kar2 mutations was found (KAR2 encodes BIP in yeast) 11281. W B P l has 25% identity to OST48 from dog pancreas 11291 and to the 48/50 kDa subunit of the OST complex isolated from pig 11071 and human [25]liver, hen oviduct [24] and human lymphocytes 1261. The homology among the mammalian sequences is higher than 90%. In contrast to Wbplp, the OST48 protein is not glycosylated [17, 1291. Homologous sequences to W B P l are also available from nematode, insect and Schizosacchavomyces pombe. Genomic OST48 clones from human and mouse were isolated and shown that the gene is organized into 11 exons expanding about 9 kb. Fluorescence in situ hybridization localizes the human gene to chromosome lp36.1 [ 1301.

4.4.2 SWPl/Ribophorin I1

Since overexpression of Wbplp did not result in increased OST activity, it was assumed that Wbplp is a non-limiting component of a larger complex [123]. This led to the identification of S W P l (suppressor of W B P ) by using a high copy number suppression approach. S W P l was isolated as an allele-specific suppressor of the temperature sensitive wbpl-2 mutation and encodes a 30 kDa, essential transmembrane protein that can be chemically crosslinked to Wbplp 11231. By immunodetection [19] and N-terminal sequencing 120, 211 Swplp was also identified as a component of the isolated complex. No sequence homology to other proteins was originally detected in searches of the databases [ 123). However, a direct comparison of Swplp with the C-terminal half of rat and human ribophorin 11, i.e. about the size of Swplp, reveals 46% similarity and 24% identity. Whether the function of the N-terminal half of ribophorin I1 is dispensable in Swplp, or whether other subunits of yeast OST serve this function, remains an open question. Characteristic of both Swplp and ribophorin I1 is the arrangement of three hydrophobic stretches at the C-terminus. Depletion of Swplp leads to underglycosylation of proteins in vivo and to reduced OST activity in vitro, similar to the lack of Wbplp. Ribophorin I1 as a component of other OST complexes has been documented (see Table 1).

4 . 4 Moleculur Biolnyj, of OST

55

4.4.3 OSTl/Ribophorin1 The OSTl gene is the yeast homologue of ribophorin I with 58% sequence similarity and 28°/0 identity and has been identified in all OST preparations so far (Table 1). Protein sequence information from the purified 64 kDa subunit [19, 1321 led to the identification of OSTl [131]; in an independent isolation of this gene it was termed N L T l (for H-Linked oligosaccharyl rransferase) [ 1321. Ostlp is an essential type I membrane glycoprotein with a cleavable signal peptide of 22 amino acids, a large lumenal domain and five potential glycosylation sites, maximally four of which are used. Ostlp is incompletely glycosylated in wild-type cells and migrates as 64/62/60 kDa glycoforms containing four, three or two core-oligosaccharides. Incomplete glycosylation is also true for ribophorin I. The fact that Ostlp and Wbplp (and also Stt3p, see below) are glycoproteins is fascinating but also puzzling and means that the mature complex must be able to process some of its own components. The most striking difference between yeast Ostlp and the canine ribophorin I is that the yeast protein has only a short 9 amino acid cytoplasmic domain at the C-terminus compared to the 150 residues in ribophorin I. Interestingly, a truncated version of ribophorin I, missing the cytoplasmic segment, was shown to be rapidly degraded in vivo (1331. This may indicate that this part is not necessary in yeast, or that its function is served by another protein. Defects in OSTI cause a pleiotropic underglycosylation of soluble and membrane-bound glycoproteins in temperature sensitive ostl mutants, and lead to a reduced OST activity in vitro that correlates with the glycosylation index in vivo [ 1311.

4.4.4 OST3/OST6 OST3 [ 131, 231 and OST6 [23] from yeast encode the 34 kDa and 32 kDa, respectively, subunits. They reveal only a weak sequence identity of 21%, but according to a hydropathy analysis they display a striking similar topology with four putative Cterminal membrane spanning segments (Figure 3). These proteins have not been identified yet in other OST preparations, but homologous proteins with the same membrane topology exist in Caenovhahditis elegans and man. The human gene, designated N33, was identified as a candidate tumor suppressor gene [ 1341 and has 21% identity to Ost3p/Ost6p. The relationship of OST3, OST6 or N33 to a neoplastic phenotype is obscure, but altered glycosylation of cell surface proteins is a well known feature of tumor cell lines [ 135, 1361. Neither OST3 nor OST6 gene is essential for yeast growth. A single disruption of OST3 [ 137, 231 or of OST6 [23] leads to only a minor defect in N-glycosylation both in vivo and in vitro. However, an Aost3Aost6 double mutant displays a synthetic phenotype leading to a severe underglycosylation with a similar pattern as that observed for mutants with a defect in the essential WBPI or OSTl genes [23]. Nevertheless an Aost3Aost6 double mutant is viable and hardly displays a temperature sensitive phenotype up to 37°C. Thus the degree of underglycosylation cannot be considered a direct predictor of a temperature sensitive growth, nor can it explain the essential role of N-glycosylatjon. Overexpression of Ost6p rescues most of the severe underglycosylation

56

4 Biochemistry and Molecular Biology

defect of proteins in the double mutant and also the milder defect in host3. On the other hand, each of the two genes seems to have also a specific function, since agents affecting cell wall biogenesis reveal different growth phenotypes in the respective null mutants. It was found that an Aost6, but not an host3 null mutant is sensitive to caffeine, Calcofluor White and SDS. In the case of SDS the host3 null mutant was even more resistant compared to an isogenic wild-type strain [23]. 4.4.5 OST5 This gene encodes the 9.5 kDa subunit of the yeast complex; homologous sequences are found in man, mouse and C. elegans, but the corresponding proteins have not been identified in OST preparations from higher eukaryotes. OST5 was identified in a genetic screen [ 1381 combining the observation that ost mutations, in combination with a deficiency in DolP-Glc biosynthesis (halg5), resulting in incomplete lipidoligosaccharide precursor, reveal a synthetically lethal growth phenotype [ 591. OST5 itself is not essential for growth, but its depletion results in a reduction in OST activity to about half that of wild-type levels. Experimental evidence for its identity with the 9.5 kDa subunit of the complex was obtained by comparison of the deduced and experimentally determined amino acid sequence of the N-terminus of the purified protein. Genetic analysis also shed light on its possible role in the OST complex [138]. It was observed that overexpression of OST5, but not of genes for other OST components, specifically suppresses the ostl-5 mutation. Since the Ostl p level is normal in Ost5p-depleted strains, and no instability of OST activity is observed in vitro, the small hydrophobic protein was suggested to provide a proper membrane requirement for optimal activity and optimal OST complex assembly. The latter point is experimentally underlined by the finding that an stt3-3Aost5 double mutant is inviable. Stt3p (see below) has been shown to influence OST assembly [66]. 4.4.6 OST4 Genetic screens for temperature-sensitive mutants with a defect in glycosyl transfer to an [ 1251]-labelledpeptide acceptor [ 1391 as well as for vanadate resistant mutants [140] led to the identification of a further gene, designated OST4. It encodes an unusually small protein of 36 amino acids [141]. host4 mutants grow poorly at 30 "C and are inviable at 37 "C, demonstrating that Ost4p is necessary for growth at elevated temperatures. OST activity is greatly diminished both in vitro and in vivo [ 1411. Hydropathy analysis leads to the prediction that Ost4p lies entirely in the hydrophobic core of the membrane. Ost4p was originally not detected in any of the purified OST complexes. But recent co-immunoprecipitation data, obtained with in vivo labelled OST subunits, suggest that it is indeed a component of the enzyme [124]. The specific function of Ost4p is not clear. A five-fold increased level of Kar2p (BIP) was noted even at the permissive temperature in Aost4 pointing to a defect in protein folding [141]. Kar2p induction, however, is not a general response

related to defects in OST function. Alternatively, Ost4p may facilitate or maintain OST assembly. OST4 mutants were reported to be subject to multicopy suppression by the MEGI gene, the nature and function of which is not known [ 1411. ESTs with sequence similarities to 0ST4 exist in man and mouse. 4.4.7 OST2/DAD1

The essential OST2 encodes the 16 kDa subunit of the yeast OST [142]. It is 40'Yo identical to DADl (dejender against upoptotic death), a highly conserved protein found in many higher eukaryotes, such as man, mouse, chicken, hamster, pig. Xenopus Zurvis, C. eleyans and several plant organisms. More recently, D A D l was identified now as the 12 kDa subunit of the OST from dog pancreas by demonstrating its copurification with the other subunits [ 181. Cross-linking studies [ 181 and a yeast two-hybrid assay [143] indicate a physical interaction with OST48. DADl seems to be required for the structural integrity of the mammalian enzyme. In a temperature sensitive cell line DADl was no longer detectable upon shift to the restrictive temperature, and the steady state levels of the other three OST subunits (OST48, ribophorin I and 11) were drastically reduced [ 1441 without affecting other ER components, e.g. the translocon. Experimental evidence suggests for Ost2p a membrane topology as shown in Figure 3. Ost2p and DADl lack an N-terminal cleavable signal sequence. A third predicted hydrophobic segment at the C-terminus is probably of insufficient length and hydrophobicity to span the membrane. In the case of the mammalian DAD1, both the N- and C-terminus could be released upon mild digestion with proteinase K [ 1451. Analysis of conditional ost2 strains demonstrates a pleiotropic underglycosylation of soluble and membrane-bound glycoproteins because of a defect in OST activity. Lipid-linked oligosaccharide assembly was normal in these strains [ 1421. D A D l was originally isolated as a cDNA complementing a t s BHK cell line that dies by apoptosis at its non permissive temperature [ 1461. The loss of DADl function probably leads to apoptosis indirectly by disturbing cellular functions, what occurs when the essential process of N-linked glycosylation is impaired. Thus, inhibition of N-glycosylation by tunicamycin was also reported to induce apoptosis of HL-60 cells [ 1471. However, in another report contrary to this result, tunicamycin addition did not lead to apoptosis [ 1451; the reason for this discrepancy remains obscure at the moment. Surprisingly, the region of highest sequence identity between Ost2p and DADl extends into the predicted membrane spanning segments [ 1421, and interestingly, the point mutations interfering with Ost2p function are also located in this region. Since Ost2p lacks a significant lumenal hydrophilic extension (Figure 3), and because OST activity is located in the lumen, it was speculated that the possible function of Ost2p is less likely to catalyze the OST reaction, but rather serves a structural role in the assembly of the complex [ 1421, as has been seen for DA D l (see above). Consistent with this is also the finding that Ost2p, like Swplp, can function as an allele-specific suppressor of wbpl-2 when overexpressed. This finding may indicate a physical and functional interaction among these three subunits. This is underlined by the observation that in vivo labelling of yeast cells and

58

4 Biochemistry und Moleculur Biology

immunoprecipitation of solubilized enzyme extract using an anti-Wbpl p antibody leads to coprecipitation of Wbpl, Swpl and Ost2p [161]. Why the lack of most of the essential Ost2p in the “tetrameric” OST complex leads to an active enzyme is not clear [19]. Perhaps Ost2p is only needed for catalytic activity in vivo, but not in vitro. DAD1 homologs in Arahidopsis thaliana [148] and rice [149] were reported to rescue the ts dad1 mutant of hamster cells from apoptotic death.

4.4.8 STT3 This gene encodes an essential transmembrane protein, which is the most conserved among proteins of the OST. Homologous sequences occur also in archaea. STT3 (gaurosporine and femperature sensitive) was originally identified in a screen for temperature-sensitive yeast mutants with an increased sensitivity towards the protein kinase C ( P K C I )inhibitor staurosporine [150] and later found to be involved in protein glycosylation [151]. PKCl has been shown to have a regulatory function in cell wall biogenesis of Saccharomyces cerevisiue, and pkcl mutations result in rapid cell lysis and an altered cell wall [152-1541. Since N-glycosylation has a key function in cell wall formation, it is not surprising that STT3 was detected in an independent screen for mutants synthetically lethal in combination with whpl [66] or with Aalg.5 (leading to incomplete, non-glucosylated lipid precursor oligosaccharide) [138].Yeast STT3 encodes a 78 kDa transmembrane protein containing a cleavable signal peptide, 12 putative transmembrane segments [ 1S O ] , three potential glycosylation sites (some of which seem to be used [ 1241) and a hydrophilic C-terminus covering about one third of the protein [66, 1501. In spite of its essential function in glycosyl transfer, Stt3p was not identified as a component in previous purified OST complexes. More recently, however, co-immunoprecipitation studies using epitopetagged Stt3p or Ost3p [ 124, 1251, or analysis of the complex by blue native electrophoresis [23] clearly demonstrate that Stt3p is indeed part of the complex. One of the reason it may have escaped detection in purified complexes is that Stt3p has an anomalous gel mobility and migrates as a diffuse 60 kDa band (the calculated molecular mass is 78 kDa) that is not well resolved from Ostlp, and stains poorly with Coomassie or silver [124, 125, 231. Protein sequence comparison reveals homologies of more than 50% identity between Stt3p and eukaryotic proteins from human, mouse, zebrafish, C. eleguns, Schizosaccharomyces pombe and Schistosoma mansoni. ORFs encoding homologous proteins with a lower identity (around 25%), but again with a similar transmembrane topology, exist also in archaea: in Methanococcus junaslzii, Archaeoglobus fulgidus, Pyrococcus horikoshii and Methanobucterium thermoautotrophicum. Most interestingly, the highest similarity between the proteins is the hydrophilic C-terminal domain, the area where all temperature-sensitive stt3 mutations have been located [ 12.51. N-linked glycosylation is not restricted to eukaryotes, but was demonstrated to occur also in archaea, even though the type of linkage (Asn-Glc-Hex,, and AsnGalNAc-Hex,) is different [6, 71. A mutation in the STT3 locus influences the substrate specificity of the OST complex [66]. In the stt3-3 mutant, transfer from truncated lipid-oligosaccharides

4.5 Strirc~turalOryanizution

of

the OST Complex

59

(DolPP-GlcNAc2 or DolPP-GlcNAc2Mans) is almost completely abolished, whereas glycosylation from the full length precursor DolPP-GlcNAc2Man~Glc~ is scarcely affected. Depletion of Stt3p causes also a loss of OST activity as well as a deficiency in the assembly of the complex [66]. The incorrectly-assembled, suboptimal complex may no longer be able to accept incomplete lipid-linked oligosaccharides as substrate. Moreover, genetic interactions between STT3 and other OST components, such as O S T 3 and OST5 were reported [138]. Whereas the double mutant stt3-3Aost3 is viable and results solely in a temperature sensitive phenotype at 37 "C, a combination of .stt3-3 and Aost5 is lethal.

4.5 Structural Organization of the OST Complex Although the sequence homology between the mammalian OST subunits and their yeast counterparts in most cases amounts to only 20-30% identity, biochemical and genetic evidence indicates that the composition and organisation of the complexes, as well as the interaction of the various subunits appear to be conserved between these phylogenetically distant organisms. As pointed out, in contrast to previous findings, all genetically defined OST genes from yeast can be detected as constituents of the enzyme complex, if it is analyzed by a quick and gentle method, such as blue native electrophoresis [23]. One may predict that components lacking so far in other eukaryotic OST complexes, will be identified in the future. Studies in yeast demonstrating genetic interactions among different OST genes, either by using a high copy number suppression approach or by constructing double mutants with a synthetic phenotype have led to the suggestion that OST subunits can be sorted into three groups: 1) Ostlp-Ost5p; 2) Wbplp-Swplp-Ost2p; and 3) Stt3p-Ost4p-Ost3p-Ost6p [ 124, 125, 231. Complementary biochemical studies (chemical cross-linking, co-immunoprecipitation) indicate a direct physical interaction between Wbplp, Swplp [123], and also with Ost2p [161], as well as between Stt3p, Ost3p and Ost4p [124] or between Ostlp and Wbplp [155]. Similarly, analogous to the interaction among Wbplp. Swplp and Ost2p, a physical interaction between the respective mammalian homologs OST48, ribophorin I1 and DAD 1 was demonstrated by using the yeast two-hybrid system, and also by biochemical means [ 1431. The interaction occurred between the short cytoplasmic tail of OST48 and the N-terminal region of DADl, whereas luminal domains were found to be responsible for the interaction between OST48 and ribophorin 11. OST48 interacted also with ribophorin I, but no direct interaction was observed between the two ribophorins suggesting that OST48 may serve as a bridge. Interestingly, no glycosylation of the ribophorins was necessary for interaction. At present it is not clear, however, whether the proposed subcomplexes comprise intermediates in the assembly of a fully functional complex, or whether they are autonomous in vivo pools that may assemble into OST complexes with slightly different composition and function. Recently, three complexes for protein translocation into the ER have been identified [156] that in part share subunit components and which can be recruited to

60

4 Biochemistry and Moleculuv Biology

complexes serving different functions (i.e. SRP-dependent and SRP-independent mode of translocation). This raises the intriguing possibility that slightly different OST complexes may exist, not yet detected by currently applied analytical methods. Now that the composition of one of the most complex enzymes in nature as well as some of the interactions between subunits have been defined, further work needs to concentrate on a number of issues. These include the specific functions of the various subunits of the complex in the glycosylation process, the regulation of the enzyme and the coupling of OST to lipid-linked oligosaccharide precursor formation, to protein translocation and to protein folding. One may also predict defects in this enzyme that may lead to abnormal glycosylation as it occurs in the so-called carbohydrate deficient glycoprotein syndrome, a severe human disorder with a multisystemic clinical picture [ 1 57- 1601.

Acknowledgments This work has been supported by the Deutsche Forschungsgemeinschaft (SFB 52 l/ A l ) and by the Fonds der Chemischen Industrie. References 1. R. Kornfeld, S. Kornfeld, Annu. Rev. Biochem. 1985, 54, 631-664. 2. A. Herscovics, P. Orlean, FASEB J. 1993, 7, 540-550. 3. L. Lehle, W. Tanner, 1995, in G1ycoprotein.s (Montreuil, J., Vliegenthart, J.F.G. and Schachter, H., eds). 29a, pp 475-509 Elsevier, Amsterdam. 4. W. Tanner, L. Lehle, Biochim. Biophys. Actu 1987, 906, 81-99. 5. H. Schachter, Glycobioloqy 1991, 1. 453-461. 6. J. Lechner, F. Wieland, Annu. Rev. Biocheni. 1989, 58, 173--194. 7. M. Sumper, F. T. Wieland, 1995, in Glycoproteins (Montreuil, J., Vliegenthart, J.F.G. and Schachter, H., eds), 29a, pp 455-473 Elsevier, Amsterdam. 8. P. Messner, U.B. Sleytr, Glycobioloqy 1991, 1, 545-551. 9. T. C . Hufdker, P. W. Robbins, J. Biol. C/iem. 1982, 257, 3203-3210. 10. E. Ioffe, P. Stanely, Proc. Nut1 Acud. Sci. U S A 1994, 91, 728-732. 1 1 . M. Metzler, A. Gertz, M. Sarkar, H. Schachter, J. W. Schrader, J. D. Marth, E M B O J. 1994, 13, 2056-2065. 12. S. Strahl-Bolsinger, M. Gentzsch, W. Tanner, Biochim. Biophys. Actu 1999, 1426, 297-307. 13. A. Varki, Glycobioloqy 1993, 3, 97-130. 14. H. A. Kaplan, J . K. Welply, W. J. Lennarz, Biochim. Biophys. Acta 1987, 906, 161-173. 15. C . B. Sharma, L. Lehle, W. Tanner, Eur. J. Biochem. 1981, 116, 101-108. 16. R. C . Das, E. C. Heath, Proc. Natl Acad. Sci. USA 1980, 77, 3811-3815. 17. D. J. Kelleher, G. Kreibich, R. Gilmore, Cell 1992, 69, 55-65. 18. D. J. Kelleher, R. Gilmore, Proc. Nut1 Acad. Sci. U S A 1997, 94, 4994 4999. 19. R. Knauer, L. Lehle, FEBSLett. 1994, 344, 83-86. 20. D. J. Kelleher, R. Gilmore, J. Bid. Chem. 1994, 269, 12908-12917. 21. R. Pathak, T. L. Hendrickson, B. Imperiali, Bioc/?emistry 1995, 34, 4179-4185. 22. R. Pdthdk, B. Imperiah, Arch. Biochem. Biophys. 1997, 338, 1-6. 23. R. Knauer, L. Lehle, J. Bid. Chem. 1999, 274, 17249-17256. 24. V. Kumar, S. Heinemann, J. Ozols, J. Bid. Chem. 1994, 269, 13451-13457. 25. V. Kumar, G . Korza, F. S. Heinemann, J. Ozols, Arch. Biochem. Biopliys. 1995, 320, 217-223.

R t jerences J

61

V. Kumar, F. S. Heinemann, J. Ozols, Biochem. Biophys. Res. Comrn. 1998. 247. 524-529. W. Breuer, E. Bause, Eur. J. Biochern. 1995, 228, 689-696. S. Silberstein, R. Gilmore, FASEB J. 1996, 10, 849-858. R. Knauer, L. Lehle, Biochim. Biophys. Actu 1999, 1426, 259-273. B. Imperiali, T. L. Hendrickson, Bioorg. M e d Cheni. 1995, 3. 1565-1578. B. Imperiali, Ace. Chem. Res. 1997, 30, 452-459. T. L. Hendrickson, B. Imperiali, Biochmi~~try 1995, 34, 9444-9450. R. B. Trimble, J. C. Byrd, F. Maley, J. Biol. Chrm.1980. 255, 11892 11895. M. J. Spiro, R. G. Spiro. V. D. Bhoyroo. J. B i d Chern. 1979, 254, 7668-7674. R . J. Chalifour. R. G. Spiro, J. Bid. Cliem. 1988, 263, 15673-15680. J. L. Franc, S. Bouchilloux, Biochim. Biopliys. Acta 1984, HOI), 166-170. R. D. Marshall, Annu. Reu. Biochem. 1972, 41. 673-702. S. J. Turco, B. Stetson, P. W. Robbins. Proc. Nut1 Acad. Sci. U S A 1977, 74, 441 1-4414. L. A. Murphy, R. G. Spiro, J. Bid. C'hem. 1981, 256, 7487-7494. K. R. Anumala, R. G. Spiro, J. Biol. Chcm 1983. 258, 15274-15282. K. W. Runge, T. C. Huffaker, P. W. Robbins, J. Bid. Chem. 1984, 259, 412-417. M. F. Verostek, P. H. Atkinson, R. B. Trimble, J. Biol. Chem. 1991. 266. 5547-5551. M. F. Verostek, P. H. Atkinson. R. B. Trimble, J. Bid. Chem.1993. 268, 12095-12103. J. Stoll, R. Cacan, A. Verbert. S.S. Krag, Arch. Biochem. Biophys. 1992, 299, 225-231. E. Alvarado. T. Nukada, T. Ogawa, C. E. Ballou, Biochemistry 1991, 30, 881-886. A . J. Petrescu, T. D. Butters, G . Reiukensmeier, S. Petrescu, F. M. Platt, R. A. Dwek, M. R. Wormald, EMBO J. 1997, 16, 4302--4310. 47. E. Bause, H. Hettkamp, FEBS Lett. 1979, 108, 341-34419. 48. E. Bause, L. Lehle, Eur. J. Biocheni. 1979, 101. 531-540. 49. G. W. Hart, K. Brew, G. G. A. Grant. R . A. Bradshaw, W. Lennarz, J. B i d Chcw. 1979, 254, 9747-9753. SO. T. C. Huffaker, P. W. Robbins, Proc. Nut1 Acad. Sci. USA 1983, KO, 7466-7470. 51. P. Burda, M. Aebi, Biochim. Biophys. act^ 1999, 1426, 239-257. 52. B. J. Jackson, C. D. Warren, B. Bugge, P. W. Robbins, Arch. Biochem. Biophys. 1989. 272, 203-209. 53. B. J. Jackson, M. A. Kukuruzinska, P. W. Robbins, Glycohiology 1993, 3, 357-365. 54. M. Aebi, J. Gassenhuber, H. Domdey, S. te-Heesen, Glycohiohgy 1996, 6 , 439-444. 5 5 . S. te Heesen, L. Lehle, A. Weissmann, M. Aebi, Eur. J. Bioclzem. 1994. 224, 71-79. 56. G. Reiss, S. te Heesen, J. Zimmerman, P. W. Robbins, M. Aebi, Glycohiology 1996, 6 , 493498. 57. K. W. Runge. P. W. Robbins, J. Biol. Cliern. 1986, 261, 15582-15590. 58. P. Burda. C. A. Jacob, J. Beinhauer, J. H. Hegemann, M. Aebi, Gl.vcohioloyy 1999, Y, 617625. 59. J. Stagljar, S. te Heesen, M. Aebi, Proc. Natl Acad. Sci. USA 1994, Y l , 5977-5981. 60. P. Burda, S. te Heesen, A. Brachat, A. Wach, A. Diisterhoft, M. Aebi, Proc. Nut1 Acad. Sci. USA 1996, Y3, 7160-7165. 61. C.A. Jakob, P. Burda, S. te Heesen, M. Aebi. J. Roth, Clycohioloy~1998, 8 , 155-164. 62. L. Ballou. P. Gopal. B. Krummel, M . Tammi, C.E. Ballou, Proc. Natl Acad. Sci. USA 1986, 83, 3081-3085. 63. M. D. Munoz, L. M . Hernandez, R. Basco, E. Andaluz, G . Larriba, Biochim. Biophys. Acta 1994, 1201, 361-366. 64. R. Cueva, M. D. Munoz, E. Andaluz, R. D. Basco, G. Larriba, Biochim. Biophys. Acta 1996, 1289, 336-342. 65. R Cueva, C. Cotano, G. Larriba, Yeust 1998, 14, 773-781. 66. R. Zufferey, R. Knauer, P. Burda, 1. Stagljar, S. te Heesen. L. Lehle, M. Aebi, EMBO J. 1995, 14, 4949--4960. 67. M. Bosch, S. Trombetta, U. Engstrom, A.J. Parodi, J. Biol. Chem. 1988, 263, 17360-17365. 68. A. J. Parodi, Glycobiology 1993, 3, 193-199. 69. S. Kornfeld, W. Gregory, A. Chapman, J. Biol. Chem. 1979, 254, 9176-9177. 70. G. J. Quellhorst jr, J. L. O'Rear, R. Cacan, A. Verbert, S. S. Krag, Glycohiology 1999, 9, 6572.

26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46.

62

4 Biochemistry and Moleculur Biology

71. P. Jung, W. Tanner, Eur. J. Biochern. 1973, 37, 1-6. 72. F. W. Hemming, 1974, in Biochen?istry of Lipids (Goodwin, T.W., ed), pp 39-97, Butterworth, London. 73. T. Chojnacki, G. Dallner, Biochemistry 1988, J. 251, 1-9. 74. G. Palamarczyk, L. Lehle, T. Mankowski, T. Chojnacki, W. Tanner, Eur. J. Biochem. 1980, 105, 517-523. 75. L. Jaenicke, K. Van Leyen, H.-U. Siegmund, Biol. Chem. H o p p Seyler 1991, 372, 10211026. 76. J. S. Rush, J. G. Shelling, N. S. Zingg, P. H. Ray, C. J. Waechter, J. Bid. Chem. 1993, 268, 13110-13117. 77. C. Ronin, S. Bouchilloux, C. Granier, J. van Rietschoten, FEBS Letters 1978, 96, 179-182. 78. J. P. Miletich, G. J. J. Broze, J. Bid. Chem. 1990, 265. 11397-11404. 79. B. W. Grinnell, J. D. Walls, B. Gerlitz, J. Bid. Chenz. 1991, 226, 9778-9785. 80. K. Titani, S. Kumar, K. Takio, L.H. Ericsson, R.D. Wade, K. Ashida, K.A. Walsch, M.W. Chopek, J.E. Sadler, K. Fujikawa, Biochemistry 1986, 25, 3171-3184. 81. Y. Gavel, G. von Heijne, Protein Eng. 1990,3, 433-442. 82. E. Bause, Biochenz. J. 1983, 209, 331-336. 83. L. Kasturi, J. R. Eshleman, W. H. Wunner, S. H. Shakin-Eshleman, J. Biol. Chem. 1995,270, 14756-14761. 84. L. Lehle, E. Bause, Biochim. Biophys. Actu 1984, 799. 246-251. 85. T. Roitsch, L. Lehle, Eur. J. Biochem. 1989, 181, 525-529. 86. S. H. Shakin-Eshleman, S. L. Spitalnik, L. Kasturi, J. Bid. Chem. 1996, 271, 6363-6366. 87. J. T. Y. Lau, J. K. Welply, P. Shenbagamurthi, F. Naider, W. Lennarz, J. Biol. Clzem. 1983, 258, 15255-15260. 88. D. T. Berg, B. W. Grinnell, Protein Sci. 1993, 2, 126-127. 89. J. L. Mellquist, S. L. Kasturi, S. H. Spitalnik, S. H. Shakin-Eshleman. Biochem. 1998, 37, 6833-6837. 90. L. Kasturi, H. Chen, S. H. Shakin-Eshleman, Biochenz. J. 1997, 323, 415-419. 91. K. Nakai, M. Kanehisa, J. Biochem. 1988, 104, 693-699. 92. S. H. Shakin-Eshleman, W. H. Wunner, S. L. Spitalnik, Biochemistry 1993, 32, 9465-9472. 93. I. Nilsson, G . von Heijne, J. Bid. Chem. 1993, 268, 5798-5801. 94. C. Landolt-Marticorena. R. A. F. Reithmeier, Biochem. J. 1994, 302, 253-260. 95. I. Mononen, E. Karjalainen, Biochim. Biophys Acta 1984, 788, 364-367. 96. N. J. Bulleid, R. S. Bassel-Duby, R. B. Freedman, J. F. Sambrook, M. J. Gething, Biochein. J. 1992,286, 275-280. 97. R. W. Doms, R. A. Lamb, J. K. Rose, A. Helenius, Viro/oqy 1993, 193, 545-562. 98. A. S. Vidal, G. Mottet, D. Kolakofsky, L. Roux, J. Virol. 1989, 63, 892-900. 99. L. W. McGinnes, T. G. Morrison, Virology 1994, 200, 470-483. 100. S. Rau, R. Geyer, R. W. Friedrich, J. Gen. Viro/. 1993, 74, 699-705. 101. A . H. Andreotti, D. Kahne, J. Am. Clzem. Soc. 1993, 115, 3352-3353. 102. S. E. O’Connor, B. Imperiali, Chemistry and Biology 1998, 5, 427-437. 103. A. Imberty, S. Perez, Protein Eng. 1995, 8, 699-709. 104. B. Imperiali, K. W. Rickert, Proc. Nut1 Acud. Sci. USA 1995, 92, 97-101. 105. B. Imperiali, K. L. Shannon, K. W. Rickert, J. Am. Chem. Soc. 1992, 114, 7942-7944. 106. E. Bause, G. Legler, Biochem. J. 1981, 195, 639-644. 107. E. Bause, W. Breuer, S. Peters, Biochem. J. 1995, 312, 979-985. 108. E. Bause, M. Wesemann, A. Bartoscheck, W. Breuer, Biochem. J. 1997, 322, 95-102. 109. T. Xu, J. K. Coward, Biochemistry 1997. 36, 14683-14689. 110. J. Lee, J. K. Coward, Biochemestry 1993, 32, 6794-6801. 111. Q. Yan, G. D. Prestwich, W. Lennarz, J. Bid. Chem. 1999,274, 5021-5025. 112. C. Kellenberger, T. L. Hendrickson, B. Imperiali, Biochemistry 1997, 41, 12554-12559. 113. S. Desruisseau, A. Valette, J.-L. Franc, 0. Chabaud, 0. Mol. CeL Endocrinoll996, 122, 223228. 114. M. Konrad, W. E. Merz, J. Biol. Chem. 1994,269, 8659-8666. 115. R. Cacan, C. Villers, M. Belard, A. Kaiden, S. S. Krag, A. Verbert, Glycohiology 1992, 2, 127-136.

Rcjerences

63

116. R. Cacan, S. Duvet, D. Kmiecik, 0. Labiau, A. M. Mir, A. Verbert, Biochimie 1998. 80, 5968. 117. M. J. Spiro, R. G. Spiro, J. B i d . CIZLW.1991, 266, 5311-5317. 118. K . Lennon. R. Pretel, J . Kesselheim, S. te Heesen, M. A. Kukuruzinska, Glycohioloyy 1995, 5, 633-642. 119. G. Kreibich. B. L. Ulrich, D. D. Sabatini, J. Cell Biol. 1978, 77, 464-487. 120. E. E. Marcantonio, A. Amar-Costesec, G. Kreibich. J. Cell Biol. 1984, 99, 2254-2259. 121. Y. Yu, D. D. Sabatini, G. Kreibich, J. Cdl Bid. 1990, I l l , 1335-1342. 122. S. te Heesen, B. Janetzky, L. Lehle, M. Aebi, EMBO J. 1992, 11, 2071-2075. 123. S. te Heesen, R. Knauer, L. Lehle. M. Aebi. EMBO J. 1993, 12, 279-284. 124. D. Karaoglu, D. J. Kelleher, R. Gilmore, J. Bid. C'hrm. 1997, 272, 32513-32520. 125. U. Spirig, M. Glavas, D. Bodmer, G. Reiss, P. Burda, V. Lippuner, S. te Heesen. M. Aebi, Mol. Gen. Genet. 1997, 256, 628-637. 126. S. te Heesen, R. Rauhut, R. Aebersold, J . Abelson, M. Aebi, M. W. Clark, Eur. J. Cell Biol. 1991, 56, 8-18. 127. E. C. Gaynor, S. te Heesen, T. R. Graham, M . Aebi. S. D. Emr. J. Cell Biol. 1994, 127, 653665. 128. S. te Heesen, M. Aebi, Eur. J. Biocheni. 1994, 222, 631-637. 129. S. Silberstein, D. J. Kelleher, R. Gilmore, J . Biol. CIiem. 1992, 267, 23658-23663. 130. T. Yamagata, T. Tsuru, M. Y . Monioi, K. Suwa, Y. Nozaki, T. Mukasa, H. Ohashi, Y. Fukushima, T. Momoi, Genomics 1997, 45, 535-540. 131. S. Silberstein, P. G. Collins, D. J. Kelleher, P. J. Rapiejko, R. Gilmore, J. Cell Biol. 1995, 128, 525-536. 132. R. Pathak, C. S. Parker, B. Imperiali. FEBS Lett. 1995, 362, 229-234. 133. Y. S. Tsao, N . E. Ivessa, M. Adesnik, D. D. Sabatini, G. Kreibich, J. Cell Bid. 1992, 116. 5767. 134. D. MacGrogan, A. Levy, G. S. Bova, W. B. Isaacs. R. Bookstein, Genomics 1996, 35, 55-65. 135. J. W. Dennis, 1992. in Cell Surfuce Curbohydruntes rind Development (Fukuda. M., ed.) pp. 161-194, CRC Press, London, UK. 136. A. Kobata, 1996, in Glycoproteins und Diseuse (Montreuil, J., Schachter, H., and Vliegenthart, J.F.G., eds.), New Comprehensive Biochemistry 30, pp. 21 1-228, Elsevier Science. Netherlands. 137. D. Karaoglu, D. J. Kelleher, R. Gilmore, J. Cell B i d . 1995, 130, 567-577. 138. G. Reiss, S. te Heesen, R. Gilmore, R. Zufferey, M Aebi, EMBO J. 1997, 16, 1164-1172. 139. J. Roos, R. Sternglanz, W. J. Lennarz, Proc. Nut1 Acad. Sci. U S A 1994, 91, 1485-1489. 140. N. Dean, 1995, 92, 1287-1291. 141. J. H. Chi, J. Roos, N. Dean, J. Bid. Clietn. 1996, 271, 3132-3140. 142. S. Silberstein, P. G. Collins, D. J. Kelleher, R. Gilmore, J. Cell Biol. 1995, 131, 371-383. 143. J. Fu, M. Ren, G. Kreibich, J. Biol. Chetn. 1997, 272, 29687729692, 144. A. Sanjay, J . Fu, G. Kreibich. J. Bid. Chern. 1998, 273, 26094-26099. 145. T. Makishima, T. Nakashima, K. Nagata-Kuno, K. Fukushima, H. Iida. M. Sakaguchi, Y. Ikehara, S. Komiyama. T. Nishimoto. Genes Cells 1997, 2 , 129-141. 146. T. Nakashima, T. Sekiguchi, A. Kuraoka, K. Fukushima, Y. Shibata, S. Komiyama, T. Nishimoto, Mol. Cell. Biol. 1993, 13, 6367 -6374. 147. D. Perez-Salva, F. Mollinedo, J. Cell. Physiol. 1995, 163, 523-531. 148. P. Gallois, T. Makishima, V. Hecht, B. Despres, M. Laudie, T. Nishimoto, R. Cooke. Plunt J. 1997,II, 1325-1331. 149. Y. Tanaka. T. Makishima, M. Sasabe, Y. Ichinose. T. Shiraishi, T. Nishimoto, T. Yamada, Plant Cell Physiol. 1997, 38, 379-383. 150. S. Yoshida, E. Ikeda, I. Uno. H. Mitsuzawa, Mol. Gen. Genet. 1992, 231, 337-344. 151. S. Yoshida. Y. Ohya, M. Goebl, A. Nakano. Y. Anraku, J. Biol. C/iem. 1994, 269, 11661171. 152. D. E. Levin, E. Bartlett-Heubusch, J. C d Bid. 1992, 116, 1221-1229. 153. G. Paravicini, M. Cooper, L. Friedli, D. J. Smith. J.-L. Carpentier, L. S. Klis, M. A. Payton, Mol. Cell. Biol. 1992, 12, 4896-4905. 154. T. Roemer, G. Paravicini, M. A. Payton, H. Bussey, J. Cell B i d . 1994, 127, 567-579.

64

4 Biochemistry and Molecular Biology

155. I. Stagljar, C. Korostensky, N. Johnson, S. te Heesen, Proc. Nut1 Acud. Sci. USA 1998, 95, 5187-51 92. 156. T. A. Rapoport, B. Jungnickel, U. Kutay, Annu. Rev. Biochem. 1996, 65, 271-303. 157. J. Jaeken, H. Carchon, H. Stibler. Clycobiology 1993, 3, 423-428. 158. J. Jaeken, H. Schachter, H. Carchon, P. De Lock. B. Coddeville, G. Spik, Arch. Dis. Child. 1994, 71, 123-127. 159. C. Korner, R. Knauer, U. Holzbach, F. Hanefeld, L. Lehle, K. von Figurd, Proc. Natl Acad. Sci. USA 1999, 95, 13200-13205. 160. P. Burda, L. Borsig, J. de Rijk-van Andel, R. Wevers, J. Jaeken, H. Carchon. E. F. Berger, M. Aebi, J. Clin. Innest. 1999, 102, 647-652. 161. Knauer, Lehle, unpublished.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

5 Processing Enzymes Involved in the Deglucosylation of N-Linked Oligosaccharides of Glycoproteins: Glucosidases I and I1 and Endomannosidase Robert G. Spiro

5.1 Introduction Subsequent to the discovery that dolichol-linked oligosaccharides are involved in the N-glycosylation of proteins in eukaryotic cells [ 11, it became apparent that the carbohydrate portion of these molecules contains peripheral residues of glucose linked to polymannose-di-N-acetylchitobiose[2]. Further structural investigations demonstrated that the glucose residues occur in a trisaccharide sequence [3] in which a-glycosidic bonds link the constituents to each other, as well as to the polymannose unit [4] as shown in Figure 1. It moreover became evident that these glucose constituents are essential for the cotranslational en bloc transfer of the oligosaccharide to asparagine in the Asn-XSer (Thr) constellations in the polypeptide chain [ 5 , 6, 71. Since glucose is generally not observed in mature N-linked oligosaccharides, their presence in newly glycosylated proteins prompted the suggestion that trimming enzymes which can remove these saccharides must be present as components of the intracellular processing apparatus [S]. Indeed, reports from a number of laboratories soon indicated that a-glucosidases which can cleave the glucose residues from the N-linked parent oligosaccharide occur in microsomal fractions from various sources [4, 9, 101. Subsequently it became apparent that two distinct processing a-glucosidases coexist in the endoplasmic reticulum (ER) which release the terminal a 1 1 2 and the two more internal wl43-linked glucose residues, respectively [ 1I], and were accordingly designated as glucosidase I and glucosidase 11 (Figure 1). Concurrently or subsequently, ER and Golgi situated a-mannosidases trim the N-linked oligosaccharide to ManSGlcNAcI [12] which in turn is acted upon by a number of glycosyltransferases to yield the large variety of complex carbohydrate units which have been observed in eukaryotes. In some instances, however, mannose excision is limited and untrimmed or incompletely trimmed polymannose oligosaccharides can be found in the mature glycoprotein, as in the polymannose oligosaccharides of thyroglobulin or the hybrid units of ovalbumin. In any case, glucose removal is a pre-

66

5 Processing Enzymes Involved in the Deglucosylution

-

a1-2

Man Glc

- --

k"

Glucosidase II

\

a1-2

Ma+l-8

ala+lan Mana-l;? Man

h !

Glucosidase I Glc

Glc

Glc

a13

/

Glc

Man 'GlcNAc

\

a13

a1-2

Man -Man

a1-2

-.+

01-4

--GlcNAc

A -Asn

413

Man

Endomannosidase

\

Glc (af-3) Man

Figure 1. Trimming enzymes involved in the deglucosylation of asparagine-linked oligosaccharides. The sites of action and the products released are shown.

requisite to the formation of the vast array of N-linked oligosaccharides which are present on membrane and secreted glycoproteins. In 1987 an endomannosidase was discovered [13] which can bring about deglucosylation by a quite distinct mechanism based upon the unusual capacity of this enzyme to effect an internal cleavage between the glucose-substituted mannose and the remainder of the polymannose oligosaccharide with the release of G l c a l j 3 M a n (Figure 1). Indeed, by its endoglycosidase action this processing enzyme differs from the other trimming enzymes which operate by a sequential release of monosaccharide residues. The wide distribution of the enzyme in members of the chordate phylum provides an alternate processing route which is particularly important in glucosidase-inhibited or glucosidase-deficient cells [ 141. It is the aim of this review to describe the enzymatic and molecular properties as well as the intracellular location of these three glucose-removal enzymes (see Table 1) as well as to describe their physiological role when acting in concert with the other processing enzymes in normal cells and in mutants. Moreover, this Chapter will evaluate some recent studies which indicate that monoglucosylated polymannose N-linked oligosaccharides may play an important role in the quality control of proteins due to their interaction with chaperones. The biological consequences of glucosidase inhibition in regard to protein degradation and viral replication will also be examined.

5.2 Glucosidase I The a-glucosidase I which is the first processing enzyme to act on the triglucosylated polymannose oligosaccharides by specifically cleaving the a1 t 2 linked glucose residue (Figure 1) has been observed in all eukaryotes examined, ranging from yeast to higher mammals as well as plants. Under physiological circumstances the enzyme acts with great speed and during pulse-chase radiolabeling of intact tissues or cells, triglucosylated N-linked oligosaccharides (Glc3Man9GlcNAcz) can

5.2 Glucosiduse I

67

Table 1. Properties of processing glycosidases involved in the removal of glucose from N-linked oligosaccharides of glycoproteins”. Property

Glucosidase I

Predominant action Subcellnlar location Subunit Mr (kDa) Cation requirement pH optimum Inhibitors‘ Deficient mutant‘ N-Linked carbohydrate N-Glycosylation consensus sequenceg Ph ylogenetic distribution Source of purified enzyme

None 6.5-7.0 DNJ, CST, N-Me & N-Bu-DNJ Lec 23 (CHO) glsl-1 (S. cerevisiae) Yes

Glucosidase I1

Endomannosidase

G2 I M Y + M ~

GI M9-StMX-4 IC, cis/medial Golgi 56

ER, 1C‘ u 100-123 p 58--80* None 6.5-7.0 DNJ, CST, MDL

None

7.0 Glcul+3DMJ

PHAR2.7 (mouse lymphoma) Yes

No

One

Two

None

Eukaryotes

Eukaryotes

Chordates’

Liver, mammary gland, mung bean, yeast

Liver, kidney, mammary gland T-cells, soybean, yeast Yes Pig and rat liver, mouse T-cells’ CHO cells Other glucohydrolases

Rat liver

Cloned Clone library

Yes Human hippocampus,

Expressed Sequence homology

COS- 1 None

(U)h

None

Yes Rat liver

E. coli None

For references to information in this table see text. The abbreviations for the predominant reaction are G, glucose; M, mannose. All oligosaccharides terminate in a di-N-acetylchitobiose sequence by which they are linked to Asn in the peptide chain. ‘ZC, intermediate compartment refers to the membranes interposed between the ER and Golgi compl ex. The values given refer to recent reports which indicate that the enzyme has an a catalytic peptide chain as well as a P-subunit [34, 351. Several other reports have not identified the latter polypeptide. Moreover, early studies underestimated the M, of the catalytic subunit due to proteolytic cleavage during isolation [37]. Abbreviations for inhibitors are DNJ, I-deoxynojirimycin; CST, castanospermine; N-Me and NBu, A‘-methyl and N-butyl-DNJ, respectively; M D L , 2,6-dideoxy-2,6-imino-7-O-(~-~-glucopyranosy1)-D-glycerol-L-glucoheptitol;DMJ, 1deoxymannojirimycin. The entry in parenthesis refers to the cell type in which the mutation has been observed. Refers to the Asn-X-Ser(Thr) sequence. Present on the a-subunit. i In addition to the various classes of the phylum Choru’utu, the enzyme has been observed in Mu/luscu [42]. ’The a-subunit has also been identified in the genome of yeast and man [34]. a

68

5 Processing Enzymes Involced in the Deglucosylation

only fleetingly be detected [15, 161. As measured in vitro, the enzyme acts optimally on oligosaccharides which are not processed by mannosidase action beyond the Glc3MangGlcNAc2 stage [ 111. The enzyme is believed to be an ER-situated integral membrane protein with a transmembrane anchor and a luminally oriented catalytic domain and a short cytosolic tail [ 17, 181. Glucosidase I has been purified from several tissues including calf and pig liver [ 19, 201, bovine mammary gland [21], mung bean seedlings [22] and Saccharomyces cereuisiae [23] utilizing various chromatographic procedures, among which the affinity matrix containing the N-carboxypentyl-1 -deoxynojirimycin proved to be the most useful [19-211. There is general agreement that the enzyme consists of a single polypeptide chain in the 85-97 kDa molecular mass range [ 17-22], which may associate into a tetramer [21]. Several reports indicate that the enzyme contains a single N-linked oligosaccharide, which on the basis of its endo-P-Nacetylglucosaminidase H (endo H) susceptibility, appears to be of the polymannose type [17, 18, 20, 221, as would be anticipated in an ER residential protein. Indeed, a single N-glycosylation site has been identified at Asn 655 in the cloned enzyme from human hippocampus, which became substituted with an endo H-labile oligosaccharide when expressed in COS-I cells [18]. A comparison of the glucosidase I cDNA from human hippocampus with the cDNA of other processing enzymes revealed no homology [18]. The enzyme has no metal requirements and can be effectively inhibited by a number of agents (Table 1) which however also block the action of glucosidase I1 to various extents [24]. The in vivo effect of the glucosidase inhibitors on glycoprotein processing is only partially effective due to the alternate deglucosylation pathway provided by endomannosidase (see below). Since the glucosidase I does not act on aryl a-glucosides, its assay has routinely relied on the use of radiolabeled Glc3Man9GlcNAc as substrate followed by quantitation of the cleavage products subsequent to their separation by various thin layer, paper, or column chromatographic procedures [17-23, 251. In these assays it is important to measure the GlczMan9GlcNAc product, as the excised monosaccharide could be a product of the combined action of glucosidase I and contaminating glucosidase 11. A spectrophotometric assay for glucosidase I has been reported in which Glcal + 2Glcal + ~ G ~ C - O - ( C H ~ ) ~ C O is OC used H ~as a substrate, but since this procedure only measures glucose release it also cannot be used when the additional presence of glucosidase I1 is suspected [26].

5.3 Glucosidase I1 The second a-glucosidase in the usual processing sequence has the capacity to cleave both a1+3 linked glucose residues despite the fact that these two inner residues are attached to glucose and mannose respectively (Figure 1). This dual action of the enzyme has been attributed to two distinct active sites [27]. Glucosidase 11 has strong specificity for the oligosaccharide with untrimmed mannose branches (Glc2-1MansGlcNAcz) [ 111. This interaction of the enzyme with the 6’-

69

5.3 Gliccosiduse II

Table 2. Comparison of the specificities of rat liver ghcosidase 11, endomannosidase and calreticulin”. Oligosaccharide

Glucosidase 11

Glcl MangGlcNAc Glcl ManzGlcNAc Glcl Man7GlcNAc Glcl ManbGlcNAc Glc, MansGlcNAc

100 21

9 5 3

Endomannosidase

Relative activity” I00 I20 150 ND‘ 220

Calreticulin

100 81

63 62 62

Glucosidase I1 values are taken from [ 1I], while the data for endoniannosidase is derived from [41] and that for calreticulin was obtained by [46]. GlqMan4GlcNAc was a good substrate for the endomannosidase but had no affinity for calreticulin. Relative glycosidase (glucosidase I1 and endomannosidase) and binding (calreticulin) activities. Not determined.

pentamannosyl branch of the substrate determines that excision of even one mannose residue results in a steep fall in activity (Table 2). In vivo the Glc2ManqGlcNAcz is a very transient intermediate, while substantial amounts of the monoglucosylated polymannose oligosaccharide accumulate [ 15, 161. Although this finding suggests that the innermost glucose is removed much slower than the more peripheral residue, this interpretation is complicated by the possibility that the UDP-g1ucose:glycoprotein glucosyltransferase can regenerate Glcl Mans GlcNAcz from the deglucosylated polymannose oligosaccharides [28]. Glucosidase I1 has been shown by immunoelectron microscopy of hepatocytes to be located in the ER and transitional elements between the ER and Golgi complex [29], which have more recently been ascribed to the intermediate compartment [30]. Furthermore, the enzyme has been detected in the nuclear envelope, but not in the Golgi apparatus [29, 301. In contrast to glucosidase I, it has become evident that the glucosidase I1 is a soluble luminal or loosely associated membrane enzyme [ 3 1, 321 and indeed its catalytic subunit as determined from its cDNA does not contain a transmembrane domain, nor has an ER retention signal been found [33-351. The enzyme has been detected in all eukaryotic cells which have been examined, including yeast and higher plants and has been purified from rat 134. 36, 371 and pig [33]liver, bovine mammary gland [38], pig kidney [31] and mouse T-cells [35]. Early studies on the purified enzyme from rat liver [36] as well as bovine mammary gland [38] reported that the catalytic subunit had a molecular mass of about 65 kDa and indicated that this occurred in the form of a tetramer. However, a subsequent study on rat liver indicated that the enzyme has an M, of 123 kDa which readily undergoes proteolytic cleavage to a 73 kDa fully active fragment [37]. The importance of including protease inhibitors during isolation of glucosidase 11 was emphasized in this investigation [37]. Indeed more recent reports of the enzyme isolated from pig kidney [31]. mung bean seedlings [39], mouse T lymphocytes [35] and rat [34] as well as pig [33] liver have assigned M, values of 100-1 16 kDa to the a-glucosidase 11 catalytic subunit. Furthermore, studies on rat liver [34] and mouse T lymphocytes

70

5 Processing Enzymes Involved in the Deglucosylution

1351 have indicated that the glucosidase I1 is a heterodimer in which the catalytic subunit (a) is associated with another polypeptide (subunit P) with M, values of 58 kDa in liver and 80 kDa in T-lymphocytes. It has been proposed that this noncatalytic subunit may promote interaction of the enzyme with other proteins, such as calnexin and calreticulin or that it is responsible for the ER localization of the enzyme [34, 351; the latter possibility is supported by the fact that the P subunit has an HDEL retention signal, which is a feature not found on the a-chain 134, 351. Glucosidase I1 like glucosidase I is believed to contain a single endo Hsusceptible and concanavalin A-reactive N-linked oligosaccharide [31-40]. From its pig liver cDNA a single N-glycosylation site has been identified close to the amino terminus of the enzyme [ 331, while two potential glycosylation consensus sequences have been reported to occur in the rat liver a-subunit, as inferred from its cDNA 1341. The enzyme has been cloned from rat liver [33, 341 and mouse T-lymphocytes [35] and these studies have indicated that the catalytic subunit has no transmembrane domain nor ER retention signals, which is consistent with a soluble protein. Although glucosidase I1 shows no homology to glucosidase I, it manifests a high degree of homology with other glucohydrolases, including lysosomal a-glucosidase, sucrase-isomaltase and the yeast family 3 1 glucosidases [33-351. Gene disruption experiments indicated that a S. cevevisiae gene is a functional homologue of mammalian glucosidase I1 [34]. The P-subunit of the liver glucosidase I1 showed no sequence homology to other known proteins [34, 3.51. In view of the broad homology of the catalytic subunit of glucosidase I1 it is not surprising that this enzyme, in contrast to glucosidase I, readily cleaves aryl a-glucosides and this property has been widely employed to assay this enzyme. However, this facile approach of measuring the enzyme has its pitfalls since the presence of non-processing a-glucosidases which are widely distributed could give misleading results unless a highly purified ER membrane is being assayed. The use of readily prepared radiolabeled Glcl MangGlcNAc as substrate [25, 381 can provide more definitive results. Glucosidase I1 has no cation requirements and a number of effective inhibitors of its activity are available (Table 1) but all of these agents also block glucosidase I action to various extents [24]. Of the various inhibitors currently available, castanospermine (CST) has proven to be highly effective in both in vitro and intact cell studies. In the latter case, as already mentioned, the glucosidase blockade is usually incomplete due to the presence of an alternate processing route provided by endomannosidase.

5.4 Endo-a-mannosidase The internal scission which endomannosidase brings about by cleaving the a1 4 2 linkage between the glucose substituted mannose and the remainder of the 3’trimannosyl branch of the N-linked oligosaccharides (Figure 1) gives it a unique place among the processing glycosidases. Although the primary physiological

5.4 End~-~-munnosiduse

71

action of the enzyme is to act on monoglucosylated oligosaccharides with the release of Glcul-t3Man, it also has the capacity to remove Glcal +3Glcal+3Man and Glcal+2Glcul+3Glca1+3Man from di- and triglucosylated polymannose species, respectively [41]. On the basis of differential and sucrose density centrifugation, it was determined that the endomannosidase of liver is present in the Golgi complex [ 131 and this has been recently confirmed by immunoelectron microscopy which demonstrated its localization in the cislmedial cisternae of this apparatus [ 301. The latter studies, moreover indicated that endomannosidase also occurs in the intermediate compartment but in contrast to the glucosidases it is not present in the ER [30]. Furthermore, it became apparent from confocal microscopy and double immunogold labeling of hepatocytes that although endomannosidase and glucosidase I1 are both present in the intermediate compartment, the two enzymes are spatially separated in these vesiculo-tubular elements [30]. The distribution of endomannosidase is limited to the phylum Chorduta, including placental and marsupial mammals, birds, reptiles, amphibians and fish, with the single exception of Molluscu, where it has been detected in three distinct classes [42]. This late evolutionary appearance of the enzyme. in contrast to the processing glucosidases, may reflect the more prominent biological role which complex oligosaccharides take on in higher organisms. The presence of an alternate processing pathway, in addition to the highly conserved glucosidase trimming route, would insure that no incompletely deglucosylated oligosaccharides are present on cell surfaces or secreted glycoproteins. Furthermore, as discussed below, the endomannosidase may be a component of a more sophisticated protein quality control machinery. The enzyme has been found in all mammalian cell lines so far assayed 125. 42-44] with the sole exception of Chinese hamster ovary cells (CHO) [25]. Purification of endomannosidase from rat liver Golgi membranes by affinity chromatography on a Glccll+3Man-O-(CH2)8CONH-Affi-Gel matrix yielded a polypeptide doublet (M, 60 kDa and 56 kDa) upon examination by SDS-PAGE in which the two components were present in approximately equal amounts [45]. Subsequent studies revealed on the basis of sequence analyses and immunoreactivity that the larger of the two components was calreticulin while all the enzyme activity was present in the 56 kDa component [46]. The copurification of the endomannosidase with the chaperone on a matrix containing the G l c a l i 3 M a n ligand suggested a biological relationship of the components which will be discussed (see below). The endomannosidase has recently been cloned from a rat liver cDNA library and the deduced open reading frame was found to encode a protein of 451 amino acids which corresponded to a molecular mass of 52 kDa [47]. Homology with other processing glycohydrolases or indeed any other known protein was not evident [47]. No N-glycosylation consensus sequence was identified and indeed, consistent with the latter, it was found that the electrophoretic mobility of the endomannosidase was not affected by N-glycanase treatment [47]. The enzyme was effectively expressed in Escherichiu coli yielding a protein which could be purified to homogeneity in high yield by affinity chromatography with a specific activity which was actually several fold higher than that obtained from rat liver Golgi membranes [47]. Indeed, since E. coli does not contain processing glycosidases, a lysate of these

72

5 Processing Enzymes Involved in the Deglucosylation

cells after transfection manifests endomannosidase as the sole enzyme with a capacity to trim N-linked oligosaccharides. Northern blotting revealed the presence of endomannosidase mRNA in all rat tissues examined [47]. Endomannosidase like the other deglucosylating enzymes has no cation requirements and moreover its activity is unaffected by any of the a-glucosidase or amannosidase inhibitors. By preparing a number of Glcal+3Man derivatives, several potent inhibitors of endomannosidase have been found [25] among which Glcal+3( 1-deoxy)mannojirimycin was the most effective ( I C ~= O 1.7 pM). While this inhibitor could block endomannosidase action in vitro as well as in some intact cell lines [25], in the latter case the possibility that the disaccharide could be cleaved by endogenous a-glucosidases necessitated the inclusion of CST in the medium. The assay of endomannosidase is routinely carried out in buffers other than Tris, which is highly inhibitory [41], with radiolabeled Glcl MangGlcNAc as substrate in the presence of glucosidase and mannosidase inhibitors [e.g., CST and 1-deoxymannojirimycin (DMJ)] followed by the thin layer chromatographic separation and quantitation of the released G l c a l i 3 M a n component [13, 25, 411; no inhibitors are of course required in the purified or E. coli expressed enzyme. Since concanavalin A-sepharose binds the polymannose product and substrate, but not the released disaccharide, assays can be carried out using the immobilized lectin, but the reliability of this procedure has not been found to be adequate when dealing with unexplored and unfractionated membrane systems. Recently it has been observed that the cloned enzyme can be used as an effective tool in assessing the glucosylation of radiolabeled glycoproteins or oligosaccharide lipids, since endomannosidase cleaves these constituents irrespective of whether their polymannose oligosaccharides are in the tri-, di-, or monoglucosylated state [69].

5.5 Concerted Action of Deglucosylation Enzymes Since the requirement of a triglucosyl sequence (Figure 1) on the donor oligosaccharide-lipid is a clear requirement for N-glycosylation in almost all eukaryotes, subsequent glucose removal becomes a necessity for the production of the vast array of complex oligosaccharide units which occur on these cells. The enzymes iniolved in these early trimming reactions consequently play an important role and indeed function efficiently, since under normal conditions glucose residues are not seen as constituents of N-linked oligosaccharides. It may even be conjectured that since glucose is not a constituent of either complex or polymannose N-linked carbohydrate units, it commends itself as a recognition signal for a variety of biological events. While the predominant deglucosylation route is believed to be mediated by the sequential action of glucosidase I and glucosidase 11, it has become apparent that trimming of mannose residues, instead of being relegated to the Golgi apparatus as it was initially believed, commences in the ER through the action of resident a1,2mannosidases [ 121 which at least in mammalian cells can extensively degrade the 6’-

5.5 Concerted Action of Deglucosylation Enzyrnes

73

pentamannosyl branch of the polymannose oligosaccharide. Since glucosidase I1 acts very poorly on monoglucosylated oligosaccharides which have been trimmed by a-mannosidase action (Table 2), these species would become suitable substrates for endomannosidase which in fact favors carbohydrate units with truncated mannose chains (Table 2). Indeed, in any case, endomannosidase which is situated distal to the location of the processing glucosidases would remove any glucose which had escaped scission by the latter enzymes. It has been calculated that in mouse lymphoma cells under normal conditions, about 16% of the total deglucosylated oligosaccharides are generated by endomannosidase action [48]. Numerous studies with cultured cells from various sources have indicated that in the presence of glucosidase inhibitors, such as CST, formation of complex N-linked oligosaccharides ranging from 10Y0to 700/0of normal (see [ 141) continued to occur. Since deglucosylation is an essential prelude to the formation of these mature carbohydrate units, the existence of an alternate glucosidase-independent processing pathway, for which endomannosidase is ideally suited, became evident. Indeed studies on HepG2 cells [ 141 indicated that in the presence of CST there was a release of glucosylated mannose oligosaccharides (primarily Glc3Man) which closely correlated on a molar basis with the number of N-linked complex oligosaccharides produced (Figure 2). Since the latter accounted for about 50% of the complex carbohydrate units produced under normal conditions, it is clear that the endomannosidase route can accommodate a substantial traffic [ 141. As anticipated

J COMPLEX

Figure 2. Pathways for glucose removal from N-linked triglucosylated polymannose oligosaccharides in normal, glucosidase-inhibited and glucosidase-deficient cells. Glucosidase I and I1 as well as endomannosidases can participate in the deglucosylation of normal cells. When the action of the two glucosidases is inhibited by an agent such as castanospermine ( C S T ) endomannosidase provides the only route for glucose removal. In the mouse lymphoma glucosidase 11-deficient mutant cell line (PHAR 2.7) endomannosidase circumvents the block subsequent to the release of the terminal glucose residue. Deglucosylation permits the formation of complex carbohydrate units. The abbreviations used are G, Glc; M , Man; and GN, GlcNAc.

74

5 Processing Enzymes Involved in the Deglucosylation

from the internal cleavage action of endomannosidase, Man9 GlcNAc was not present among the polymannose processing intermediates in the glucosidaseinhibited cells (Figure 2) and indeed in the presence of a mannosidase inhibitor, Mans GlcNAc and to a lesser extent Man~GlcNAcand MansGlcNAc, originating from Glq-1MangGlcNAc, Glq-1MangGlcNAc, and GIq-1Man~GlcNAc,were observed [14]. The endomannosidase pathway has now been identified in a number of cells and shown to lead during CST-induced glucosidase blockade to various endo H-resistant serum glycoproteins secreted by HepG2 cells [49] as well as the endo H-resistant G protein in several vesicular stomatitis virus (VSV)-infected cell lines [44]. It has been shown moreover that the addition of Glca1-3DMJ to CSTtreated cells inhibited the alternate pathway so that glycoproteins with complex endo H-resistant N-linked oligosaccharides ceased to be formed 1251. While it has been reported that the processing glucosidases can act on lipid-linked oligosaccharides in vitro 14, 501, evidence has emerged through studies in thyroid slices that these enzymes may mediate a glucosyltransferase-glucosidase shuttle in vivo which helps to control the level of dolichyl triglucosylated polymannose available for the N-glycosylation of proteins 1511.

5.6 Mutants A glucosidase 11-deficientmouse lymphoma cell line (PHAR2.7)has been described [ 521. Although the predominant N-linked oligosaccharides from these cells were in the diglucosylated form (GlczMan9GlcNAcz), a substantial portion (35-40%) were processed to complex oligosaccharides. It was demonstrated [ 5 3 , 541 that the endomannosidase pathway was responsible for circumventing the genetically determined glucosidase blockade with the release of GlczMan, as shown in Figure 2. As in the CST-inhibited cells, Mans GlcNAc was not present among the processing intermediates (Figure 2). A CHO cell line deficient in glucosidase I activity has been reported 1551. The observation that this mutant (Lec23) was highly resistant to wheat germ agglutinin and yet hypersensitive to concanavalin A suggested a block in the formation of N-linked complex oligosaccharides and indeed in radiolabeled VSV-infected cells, triglucosylated polymannose oligosaccharides were found to be the predominant carbohydrate attached to the G protein 1551. The addition of DMJ furthermore resulted in the presence exclusively of Glc3Man9GlcNAcz. Quite small amounts of complex oligosaccharides was reported to be present in these cells, but their occurrence is unlikely to be attributable to the alternate endomannosidase pathway as CHO cells are quite unique among cultured mammalian cells in not having detectable activity of this enzyme, as evaluated either in vitro or whole cell assays 125, 441. More specifically, endomannosidase assays of Lec23 postnuclear membranes were found to be negative and furthermore, subsequent to [ 3sS]methionine labeling, the G protein of VSV migrated to the same position on polyacrylamide gel electrophoresis as the one from CST-treated wild type cells; this G protein was sensitive to

5.7 Role of Monoylucosyluted N-Linked Oligosucchurides

75

in vitro treatments with either endo H or endomannosidase as would be anticipated if it occurred in an unprocessed state 1691. A glucosidase I deficient mutant has also been found in yeast [56]. Although this S. cevevisiue glsl-1 mutation results in the production of glycoproteins containing N-linked oligosaccharides with the intact triglucosyl sequence, the presence of these glucose residues does not appear to interfere with the formation of mature carbohydrate units. The defect in glucose excision in yeast does not result in the profound processing block which is noted in mammalian cells as in the former only the terminal mannose of the middle chain has to be removed to permit the addition of outer chains [57]. No significant effect on growth properties have been noted in either the glsl-1 or the Lec23 mutants [ 5 5 , 561.

5.7 Role of Monoglucosylated N-Linked Oligosaccharides and Glucose Trimming Enzymes in Regulating Quality Control of Glycoproteins In recent years important new insights into the biological role of glucose residues on N-linked oligosaccharides have emerged. The first evidence for a regulatory role of the glucose components on the fate of newly formed glycoproteins came from a number of in vivo studies with cultured liver cells indicating that inhibitors of glucosidase I and I1 action, which keep the triglucosyl sequence intact, delayed secretion of various glycoproteins and/or promoted their degradation in an ERassociated site [49, 58, 591. Indeed it was noted that the reduction in secretion of various serum glycoproteins by HepG2 cells in the presence of CST was directly related to their number of N-linked oligosaccharides [49]. A more definitive indication of the effect of glucose trimming inhibition on the degradation of glycoproteins was observed in the CMT-cKd1 cell line which produces unassembled major histocompatibility complex (MHC) class I molecules which never reach the Golgi compartment. Upon exposure of these cells to CST the MHC heavy chains underwent a rapid degradation during a 60-min chase in contrast to control cells in which they remained stable during this period [60]. These findings of the effect of CST on MHC class I molecules were also noted in BW thymoma cells [61] and moreover it was noted that the a-subunit of T-cell antigen receptor in these cells underwent accelerated degradation in the presence of the glucosidase inhibitor and indeed also in the mutant PHAR2.7 [62, 701. The molecular basis of the instability of proteins with an untrimmed triglucosyl sequence began to become apparent from studies which indicated that the ER-situated transmembrane chaperone, calnexin, selectively associates itself in a transient fashion with incompletely folded glycoproteins produced by HepG2 cells [63]. In a series of reports, Helenius and coworkers (for review see [64]) provided evidence that the association between an ER-situated chaperone, whether calnexin or calreticulin, and a glycoprotein represents at least initially a lectin-like interaction with monoglucosylated N-linked polymannose oligosaccharide (i.e., GlqMansGlcNAc) on the latter molecule. The transient nature of this interaction and its role in protein quality control was clarified by a model in which

76

5 Processing Enzymes Involved in the Deylucosylution

Calreticulin .......... ............... ................ ...................

.............. ..........

Fhdomannosidase

-------

A

E m Calreticulin

Folded

w

B

Figure 3. Schematic representation of the role by which trimming of monoglucosylated N-linked oligosaccharides can regulate the interaction of glycoproteins with calreticulin during quality control. In the model shown in scheme B, which is based on the studies of Helenius and coworkers [64] the ER-situated monoglucosylated N-linked oligosaccharide with as yet untrimmed mannose residues interacts with the chaperone and subsequently undergoes a cyclic deglucosylation by glucosidase 11 and reglucosylation by UDP-G1c:glycoprotein glucosyltransferase until folding of the polypeptide chain has been achieved, at which point the latter enzyme ceases to function. In scheme A , which is based on the work of Spiro et al. [46]. variable degrees of trimming of the 6’-pentamannosyl branch has occurred by ER and Golgi mannosidases so that the specificity of glucosidase IT precludes its action. This more processed glycoprotein-calreticulin complex then enters or is formed in the intermediate and Golgi compartments where endomannosidase is situated. This enzyme readily cleaves even extensively trimmed monoglucosylated oligosaccharides with the release of G l c a l i 3 M a n ( G - M ) and thereby dissociates the glycoprotein-chaperone complex. If the protein has properly folded it can continue to traverse the secretory channels, while if it remains incompletely folded it will undergo degradation. The closed squares).( represent mannose residues while the open squares (0) indicate mannose which in various degrees has been excised by ER and Golgi mannosidases.

glucosidase I1 works in tandem with the UDP-G1c:glycoprotein glucosyltransferase to bring about deglucosylation followed by reglucosylation until proper folding and/or oligomerization has taken place (Figure 3, scheme B). The termination of this cycle in this proposal is brought about by the observation that the ER-situated glucosyltransferase acts solely on denatured substrates [28]. Binding studies with radiolabeled oligosaccharides indicated that calnexin [65] and calreticulin [46] had a high specificity for monoglucosylated oligosaccharides; indeed, the finding that the tri- and di-glucosylated polymannose units completely failed to bind to these chaperones provided a basis for understanding the results of

5.8 Effect qf Glucosiclase Inhibitors on Viral Proliferation

77

the in vivo studies in which accelerated degradation of glycoproteins was noted in the presence of glucosidase inhibitors. Examination of a series of monoglucosylated oligosaccharides in which the mannose residues of the 6’-pentamannosyl branch had been trimmed indicated that the lectin-like binding activity of calreticulin was retained to a substantial extent (Table 2). Excision of the a1+6 linked mannose from GlcIMansGlcNAc however resulted in a complete loss of interaction with the chaperone [46]. Since it is known that glucosidase I1 acts very poorly on trimmed monoglucosylatcd oligosaccharides (Table 2) it would appear that the chaperone-glycoprotein disassociation mechanism depicted in Scheme B of Figure 3 would not be effective if excision of mannose residues from the 6’-pentamannosyl branch had taken place through the action of the ER mannosidases. Since endomannosidase, however. has the distinct capacity to cleave even extensively trimmed monoglucosylated carbohydrate units (Table 2) it has been proposed [46] that this enzyme could readily achieve this disassociation when the glycoprotein-chaperone complex reaches the intermediate compartment or &/medial Golgi location where this glycosidase is known to be located (Scheme A, Figure 3 ) . This model was prompted by the finding that endomannosidase and calreticulin can be copurified in a approximately 1 : 1 ratio from rat liver Golgi membranes by affinity chromatography on a G l c a l b 3Man-containing matrix [45]. Furthermore it has recently been shown by immunomicroscopy that calreticulin, while primarily present in the ER, is also situated in the intermediate compartment and the cislmedial Golgi stacks [30], which is consistent with its properties as a soluble chaperone and its known role in intracellular trafficking. Indeed it was demonstrated that endomannosidase and calreticulin colocalize in these more distal compartments [ 301. The proposed dissociative capacity of endomannosidase in the more distal location would make it possible for properly folded or oligomerized glycoproteins to continue their journey through the secretory channels while glycoproteins which had not reached their mature configuration would be susceptible to degradation with the release of their oligosaccharides into the cytosol [66].

5.8 Effect of Glucosidase Inhibitors on Viral Proliferation A number of reports have appeared which indicate that several glucosidase I and I1 blocking agents can function as inhibitors of human immunodeficiency virus (HIV) proliferation and syncytium formation (for review see [67]). These antiviral effects which were noted when the glucosidase inhibitors were added to infected cultured cells have however not been observed with some other viruses and this might be attributable to the unusually high degree of N-glycosylation which is a characteristic of the HIV gp 120 envelope glycoprotein [67]. Retention of the outer two glucose residues of N-linked oligosaccharides of viral envelope glycoproteins would interfere with their association with the lectin-like chaperones, calnexin and calreticulin, and this could lead to their enhanced degradation due to improper folding or oli-

78

5 Processing Enzymes Involved in the Deglucosylution

gomerization. Indeed it has been shown that a variety of envelope glycoproteins including HIV gp 160 bind to calreticulin and/or calnexin [64, 681. Since it has been reported that viral glycoproteins (e.g. the G protein of VSV) can be processed by the endomannosidase pathway in many cells [44], it can be conjectured that the maximum effect of a glucosidase blockade on viral replication should be evident in cells in which the glycoprotein under study is not a substrate for this alternate deglucosylation route or if an endomannosidase inhibitor (e.g. Glcal+3DMJ) is added to the infected cells in addition to the glucosidase blocking agent. Acknowledgments

The author’s studies were supported by Grants DK17325 and DK17477 from the National Institutes of Health. References 1. Behrens, N.H., Parodi, A.J., and Leloir, L.F. (1971) Proc. Nut/ Acud. Sci., U.S.A., 68, 28572860. 2. Spiro, R.G., Spiro, M.J., and Bhoyroo, V.D. (1976) J. Biol. Chern. 251, 6409-6419. 3. Li., E., Tabas, I., and Kornfeld, S. (1978) J. Biol. Chem. 253, 7762-7770. 4. Spiro, R. G., Spiro, M. J., and Bhoyroo. V. D. (1979) J. Bid. Chem. 254, 7659-7667. 5. Turco, S.J., Stetson, B., and Robbins, P.W. (1977) Proc. Nut1 Acud. Sci., U.S.A., 74, 44114414. 6. Spiro, M. J., Spiro, R. G., and Bhoyroo, V. D. (1979) J. Biol. Chem. 254, 7668-7674. 7. Murphy, L. A,, and Spiro, R. G. (1981) J. Biol. Chem. 256, 7487-7494. 8. Spiro, M. J., Spiro, R. G., and Bhoyroo, V. D. (1976) J. Biol. Chem. 251, 6400-6408. 9. Ugalde, R.A., Staneloni, R.J., and Leloir, L.F. (1979) Biochem. Biophys. Res. Commun. 91, 1 174-1 181. 10. Chen. W.W., and Lennarz, W.J. (1978) J. Bid. Chem. 253, 5780-5785. 11. Grinna, L. S., and Robbins, P.W. (1980) J. B i d Cllenz. 255, 2255-2258. 12. Daniel, P.F., Winchester, B., and Warren, C.D. (1994) Glycobiology 4, 551-566. 13. Lubas, W.A., and Spiro, R.G. (1987) J. Bid. Chem. 262, 3775-3781. 14. Moore, S.E.H., and Spiro, R.G. (1990) J. Biol. Chern. 265, 13104-13112. 15. Godelaine, D., Spiro, M. J., and Spiro, R. G . (1981) J. B i d Chem. 256, 10161-10168. 16. Hubbard, S.C., and Robbins, P.W. (1979) J. Biol. Chem. 254, 4568-4576. 17. Shailubhai, K., Pukazhenthi, B.S., Saxena, E.S., Varma, G.M., and Vijay, I.K. (1991) J. Biol. Chem. 266, 16587-16593. 18. Kalz-Fuller, B., Bieberich, E., and Bause, E. (1995) Eur. J. Bioclzern. 231, 344-351. 19. Hettkamp, H., Legler, G., and Bause, E. (1984) Eur. J. Biochem. 142, 85-90. 20. Bause, E., Schweden, J., Grass, A,, and Orthen, B. (1989) Eur. J. Biochem. 183, 661-669. 21. Shailubhai, K., Pratta, M.A., and Vijay, I.K. (1987) Biochem. J. 247, 555-562. 22. Zeng, Y.C., and Elbein, A.D. (1998) Arch. Biochem. Biophys. 355, 26-34. 23. Kilker, R.D., Jr., Saunier, B., Tkacz, J.S., and Herscovics: A. (1981) J. Biol. Chem. 256, 52995303. 24. Elbein, A.D. (1991) FASEB J. 5, 3055-3063. 25. Hiraizumi, S., Spohr, U., and Spiro, R.G. (1993) J. Biol. Chem. 268, 9927-9935. 26. Neverova, I., Scaman, C.H., Srivastava, O.P., Szweda, R., Vijay, I.K., and Palcic, M.M. (1994) Anal. Biochern. 222, 190-195. 27. Alonso, J.M., Santa-Cecilia, A,, and Calvo, P. (1991) Biochem. J. 278, 721-727. 28. Sousa, M.C., Ferrero-Garcia, M.A., and Parodi, A.J. (1992) Biochemistry 31, 97-105.

RLJ'erences

79

29. Lucocq, J.M., Brada, D.. and Roth, J. (1986) J. Cell Bid. 102, 2137-2146. 30. Zuber, C., Spiro, M.J., Guhl, B., Spiro, R.G., and Roth, J. Manuscript submitted for publication. 31. Brada, D., and Dubach, U.C. (1984) Eur. J. Biochem. 141, 149-156. 32. Strous, G.J, Van Kerkjof, P., Brok, R., Roth, J., and Brada, D. (1987) J. Bid. Chern. 262, 3620 ---362S. 33. Flura, T., Brada, D., Ziak, M. and Roth, J. (1997) Glycobiology 7, 617-624. 34. Trombetta, E.S.. Simons, J.F., and Helenius, A. (1996) J. Biol. Cliem. 271, 27509-27516, 35. Arendt, C.W., and Ostergaard, H.L. (1997) J. Biol. Chem. 272, 13117-13125. 36. Burns, D.M., and Touster, 0 . (1982) J. B i d . Clienz. 257, 9990- 10000. 37. Hino, Y., and Rothman, J.E. (1985) L3ioclieinistr.y 24, 800-805. 38. Saxena, S., Shailubhai, K., Dong-Yu, B., and Vijay, I.K. (1987) Biochem. J. 247, 563-570. 39. Kaushal. G.P., Pastuszak, I., Hatanaka, K.; and Elbein, A.D. (1990) J. Biol. Chem. 265, 16271-16279. 40. Saunier, B., Kilker, R.D., Jr., Tkacz, J.S., Quaroni, A,, and Herscovics, A. (1982) J. B i d . Cheni. 257, 14155-14161. 41. Lubas, W.A., and Spiro, R.G. (1988) J . Biol. Clirm. 263, 3990-3998. 42. Dairaku, K., and Spiro, R.G. (1997) G/~vcobiology7, 579-586. 43. Weng. S., and Spiro, R.G. (1993) J. Rid. Chem. 268, 25656-25663. 44. Karaivanova, V.K., Luan, P., and Spiro, R.G. (1998) Glycohiology 8, 725-730. 45. Hirdizumi, S., Spohr, U., and Spiro, R.G. (1994) J. Biol. Chem. 269, 469774700, 46. Spiro, R.G., Zhu, Q., Bhoyroo, V, SBling, H.-D. (1996) J. Biol. Cheni. 271, 11588-11594. 47. Spiro, M.J., Bhoyroo, V.D., and Spiro, R.G. (1997) J. Biol. Chem. 272, 29356-29363. 48. Weng, S., and Spiro, R.G. (1996) Glyc,ohzo/oyy 6, 861-868. 49. Rabouille, C., and Spiro, R.G. (1992) J. Biol. Chem. 267, 11573-11578. 50. Eking, J.J.. Chen, W.W., and Lennarz, W.J. (1 980) J. Biol. Chen7. 255, 2325--233I . 51. Spiro, M.J., and Spiro, R.G. (1991) J. Bid. C/iem. 266, 5311-5317. 52. Reitman, M.L., Trowbridge, I.S., and Kornfeld, S. (1982) J. Bid. Cliem. 257, 10357-10363, 53. Fujimoto, K., and Kornfeld, R. (1991) J. B i d . Clzem. 266, 3571-3578. 54. Moore, S.E.H., and Spiro, R.G. (1992) J. Biol. Chern. 267, 8443-8451. 55. Ray, M.K., Yang, J., Sundaram, S., and Stanley, P. (1991) J. Biol. Chen7. 266, 22818-22825. 56. Esmon. B., Esmon, P.C., and Schekman, R. (1984) J. Biol. Chem. 259, 10322-10327. 57. Herscovics, A,; and Orlean, P. (1993) FASEB J . 7, 540-550. 58. Lodish, H.F., and Kong, N. (1984) J. Cdl Bid. 98, 1720-1729. 59. Parent, J.B., Yeo, T.-K., Yeo, K.-T., and Olden, K. (1986) Mol. Ce//. L3iocht.m. 72, 21-33. 60. Moore, S.E.H., and Spiro, R.G. (1993) J. Riol. Chem. 268, 3809-3812. 61. Balow, J.P., Weissman, J.D., and Kearse, K.P. ( 1 995) J. Biol. Chem. 270, 29025-29029. 62. Kearse, K.P.. Williams, D.B., and Singer, A. (1994) E M B O J. 13, 3678-3686. 63. Ou, W.-J., Cameron, P.H.. Thomas, D.Y., and Bergeron, J.J. (1993) Nature 364, 771-776. 64. Helenius, A,, Trombetta, E.S., Hebert, D.N., and Simons, J.F. (1992) Tretid~sCell Biol. 7. 193200. 65. Ware, F.E., Vassilakos, A,, Peterson, P.A., Jackson, M.R., Lehrman, M.A., and Williams, D.B. (1995) J. Biol. Clzem. 270, 4697-4704. 66. Moore, S.E.H., and Spiro, R.G. (1994) J. Bid. C/iern. 269, 12715-12721. 67. Feizi, T., and Larkin, M. (1990) G/ywhio/ogy 1, 17-23. 68. Otteken, A,, and Moss, B. (1996) J. Bid. C h m . 271, 97-103. 10 (in press). 69. Spiro, M.J., and Spiro, R.G. (2000) Gl,ycohio/o(gj~ 75. Karaivanova, V.K., and Spiro, R.G. (2000) G/iwhio/ogy 10 (in press).

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

6 a-Mannosidases in Asparagine-linked Oligosaccharide Processing and Catabolism Kelley W. Moremen

6.1 Overview The maturation of Asn-linked oligosaccharides is initiated in the ER and continues in the Golgi complex by the trimming of the Glc3MangGlcNAcz oligosaccharide structure down to a Man3 core that is necessary for maturation to complex type oligosaccharides. Several distinct enzymes are required for the trimming steps including several glucosidases and several mannosidases. Mannosidase activities are also found in other compartments of the cell, including the cytosol and lysosomes, where they are involved in glycoprotein and oligosaccharide catabolism. A comparison of the biochemical characteristics and protein sequences of the mannosidases involved in glycoprotein maturation and catabolism has revealed that they can be divided into two families with several subgroups within each family. Two subgroups of the Class 1 mannosidases are found in the ER and Golgi where they are involved in early trimming events in oligosaccharide processing. One of the other Class 1 mannosidase subgroups are comprised of secreted fungal mannosidases of unknown function. The last of the Class 1 mannosidase subgroups is conserved from yeast to mammals, but members of thus subgroup also have no known function. The Class 2 mannosidases have members that are found in the ER, Golgi, cytosol, and lysosomes. Four subgroups have been identified with distinctive functions in glycoprotein biosynthesis or catabolism. One of the subgroups of Class 2 mannosidases has members conserved between eubacteria, archaea, and eukaryotes. This review summarizes the biochemical and sequence relationships between the members of the two mannosidase classes and their roles in glycoprotein biosynthesis and catabolism.

82

6 a-Mannosiduses in Asparugine-linked Oliyosucchuvide Processing

6.2 Introduction 6.2.1 Roles of N- and 0-Linked Glycans and Compartmentalization of Biosynthetic and Catabolic Reactions Protein-linked carbohydrate structures are among the most complex and diverse set of post-translational modifications on intracellular and secreted proteins. Large numbers of intracellular and extracellular proteins contain N-linked oligosaccharides including enzymes, cell surface receptors, secreted proteins, hormones, immunoglobulins, and viral antigens [I]. Although there are many examples of the contribution of N-glycans to the bioactivity, folding, localization, and immunogenicity of the attached polypeptide [2], the functional roles of individual oligosaccharide structures on a given glycoprotein are difficult to predict. At the cellular level, N - and O-glycan structures have been shown to contribute to several types of biological recognition events including cell adhesion during development, immune surveillance, inflammatory reactions, hormone action, viral infection, arthritis, and metastasis of oncogenically transformed cells [ 1, 3-51, The pathways leading to the synthesis and catabolism of glycoproteins are highly compartmentalized in eukaryotic cells [6-81. Biosynthesis of cellular and secreted glycoproteins occur in the membranes of the secretory pathway including the ER, the Golgi complex, secretory vesicles, and the transport vesicles between these compartments. Glycoprotein catabolism was originally considered to be the exclusive domain of the endosomal/lysosomal system [9], but recent data have indicated that a considerable amount of glycoprotein degradation also occurs in the cytoplasm as a result of a failure in glycoprotein folding in the ER [lo]. Glycoproteins that fail to fold properly are translocated from the lumen of the ER into the cytosol where the protein portion is degraded by the proteosome. The released oligosaccharides are partially cleaved in the cytosol and further degraded in the lysosome. Thus, the compartmentalization of the synthesis and catabolism of glycoproteins into discrete and non-overlapping regions of the cell allows an independent regulation of the two pathways. 6.2.2 Processing of Asn-Linked Oligosaccharides

The synthesis and maturation of N-linked oligosaccharides occurs in four stages in the membranes of the ER and Golgi complex and several of these stages are completely, or at least partially, conserved in all eukaryotic organisms. The first stage of the pathway involves the synthesis of the precursor oligosaccharide by the stepwise addition of monosaccharides to the polyisoprenoid lipid, dolichol (summarized in Chapter 3 in this volume), with the early reactions occurring on the cytoplasmic face of the ER and the later reactions occurring on the luminal face of the membrane [ 11, 121. The final Glc3MangGlcNAc;!-P-P-dolichol structure is conserved in all eukaryotic organisms, with the only known exceptions being the Trypanosoma-

6.2 Introduction

83

tids which have been shown to synthesize and transfer non-glucosylated lipid-linked oligosaccharide structures [ 13, 141. The second stage of oligosaccharide biosynthesis is the transfer of the lipid-linked precursor to the polypeptide through the action of a highly conserved multi-subunit oligosaccharide transferase complex (Figure 1, OST [ 151) that recognizes Asn-X-Ser/Thr sequons on newly synthesized polypeptides as they emerge on the luminal side of the ER membrane (summarized in Chapter 4 in this volume). The third stage of N-linked oligosaccharide maturation involves the trimming of the transferred oligosaccharide and invariably involves the removal of all three glucose residues (Glc I and Glc I1 in Figure 1) and a variable degree of mannose trimming [7, 16, 171. In all eukaryotic organisms examined thus far, with the possible exception of the fission yeast Schi~osaccharomycespombe [ 181, mannose trimming is initiated in the ER by the removal of a single mannose residue to produce a specific MangGlcNAcz isomer (ManxGlcNAcz isomer B in Figure 1) structure. In Saccharomyces cerevisiae, the single mannose removal is the sole mannose cleavage step [ 171, whereas oligosaccharide processing in vertebrate organisms results in the removal of six of the original nine mannose units [7]. The glucose trimming steps are summarized in Chapter 5 in this volume, while the trimming of mannose residues will be the subject of this review. The final stage of oligosaccharide maturation involves the branching and extension of the oligosaccharides by Golgi glycosyltransferases [ 19,201 (summarized in numerous Chapters in this volume). The final elaboration of N-glycans is the most varied stage in eukaryotic organisms and is the major cause of the extreme diversity in mature oligosaccharides between organisms, between different cells in the same organism, and between different glycoproteins in the same cell [20]. The conserved nature of the early steps in the pathway for the synthesis of Nglycans suggests an important role for the mature lipid-linked oligosaccharide and the trimmed intermediates in the efficient N-glycosylation and maturation in the ER. Sequence similarities have been noted between the enzymes involved in the synthesis of the lipid-linked precursor oligosaccharide in eukaryotes, the enzymes for glycoconjugate synthesis in archea, and the enzymes catalyzing lipopolysaccharide biosynthesis in eubacteria [ 111. Indeed, archea have also been found to synthesize glycoproteins containing Asn-linked oligosaccharides attached to Asn-XSer/Thr sequons that are generated as lipid-linked precursors [211. In eukaryotes, the full Glc3MangGlcNAc2 oligosaccharide structure is strongly preferred for efficient transfer to protein acceptor sites [ 111 suggesting that full glucosylation may act as a signal for recognition by the oligosaccharide transferase. Smaller oligosaccharide structures can be transferred to proteins in yeast [22] and mammalian cells [23], either as a result of mutations in the enzymatic machinery for the synthesis of the glucosylated oligosaccharide or as a potential alternative pathway for oligosaccharide transfer, as in mouse teratocarcinoma cells [24], but the efficiency of transfer is significantly compromised. The pathological consequences of genetic defects in the synthesis of the full-length lipid-linked oligosaccharide have been observed in human patients with type I CDGS [25] (see below), where incomplete lipid-linked oligosaccharide synthesis leads to an underglycosylation of cellular and secreted glycoproteins.

9

B1,4-Man

Legend

fi1,4-Gal o2.6-NeuNAc

I

I

Glycoprotein Catabolism

Cytosol

Man

Figure 1. The biosynthesis and catabolism of glycoproteins in mammalian cells. The glycosidases and glycosyltransferases involved in glycoprotein maturation and degradation are indicated by a bold solid arrow with the following enzyme abbreviations: OST, oligosaccharide transferase; Glc I, glucosidase I; Glc TI, glucosidase 11, Glc T, UDP-G1c:glycoprotein glucosyltransferase; ER Man I, ER mannosidase I; ER Man 11, ER mannosidase TI (the ER form of the cytosolic/ER mannosidase 11); Endo a-Man, endo a-mannosidase; Golgi Man IA/IB, Golgi niannosidases IA and IB; GnTI, GlcNAc transferase I; Golgi Man 11, Golgi mannosidase IT; GnTII, GlcNAc transferase 11; Golgi Man III/IIx (??), either a-mannosidase 111or Golgi mannosdiase IIx; Cytosolic Man, Cytosolic mannosidase (the cytosolic form of Cytosolic/ER mannosidase 11); Lysosomal Man, broad specificity lysosomal a-mannosidase; Lys ul.6-Man, lysosomal ul,6-mannosidase; Lys P-Man, lysosomal P-mannosidase. Positions where processing inhibitors can act to block enzyme reactions are indicated by a thin arrow with the following abbreviations: dMNJ, 1-deoxymannojlrimycin; Kif, kifunensine; Sw, swainsonine. The legend for the oligosaccharide structures displayed in the figure is indicated in the lower left. Solid boxes indicate the membrane boundaries of the ER, Golgi, and lysosomes as labeled in the figure. The plasma membrane in indicated by the bold dotted line.

Plasma Membrane

. I '

: 99

-: p @l,4-GlcNAc 9 ul,2-Glc : 4 o1.3-Glc

=-

..I

Glvcom-otein Biosvnthesis

6.2 Introduction

85

6.2.3 Early Trimming Events: importance for quality control glycoprotein degradation and anteriograde transport The transfer of the Glc3MangGlcNAcl structure to protein is followed by the rapid and immediate removal of the glucose residues from the oligosaccharide by glucosidases I and I1 (Figure 1) 17, 16, 171. A role for glucosylated oligosaccharides has been proposed as a part of the quality control process of glycoprotein folding in the ER 126-281. Proteins that are glycosylated as they are extruded into the lumen of the ER undergo a concerted folding process that is facilitated by the interaction with a variety of molecular chaperones including BiP, GRP94, calnexin, and calreticulin 1291. The latter two chaperonins have been shown to contain a lectin activity and initiate their interaction with unfolded glycoproteins by specifically binding to Glcl MangGlcNAcz oligosaccharides 1261. Subsequent interactions between the chaperonin and the polypeptide have been proposed to promote polypeptide folding. Correctly folded glycoproteins are released from calnexin or calreticulin through the action of glucosidase I, which removes the last glucose residue, and the glycoconjugate will then proceed through the secretory pathway to the Golgi complex and other parts of the cell. Proteins that have not folded correctly will remain transiently associated with the chaperonins until glucosidase I removes the last glucose residue. Unfolded glycoproteins in the ER containing MangGlcNAcz structures can then be acted upon by UDP-G1c:glycoprotein glucosyltransferase to replace the Glc residue to the same position and linkage where it was removed by glucosidase I1 130, 311, thus allowing the protein to re-bind to the chaperonins. Since the glucosyltransferase recognizes only unfolded glycoprotein intermediates, it has been proposed that the enzyme acts as the “sensor” of the folding status of the newly synthesized glycoproteins [ 3 I ] . The cycle of binding to the chaperonins, sugar cleavage by glucosidase 11, release from the chaperonins, sugar addition by the glucosyltransferase, and re-binding to the chaperonins can proceed through multiple rounds in an effort to effectively fold the newly synthesized glycoproteins [26]. Under normal conditions glycoproteins that fail to fold after a defined period of time are targeted for translocation into the cytosol via the Sec6lp translocon complex followed by degradation by the cytosolic proteosome [32, 331. While many of the components involved in glycoprotein folding have been identified, the components that recognize glycoproteins which have failed the folding process and target them for degradation are not well understood. It was originally proposed that mannose trimming could act as a timing step for the quality control cycle [ 341 by interrupting one or more steps in the glucosylation/de-glucosylation cycle and targeting glycoproteins with trimmed oligosaccharides for degradation. Several lines of evidence now indicate that mannose trimming, especially to a specific Man8GlcNAc2 structure (isomer B), plays an important role in targeting unfolded proteins for degradation. The role of a mannose cleavage step in targeting the degradation of a mutant form of carboxypeptidase Y in S. cerruisiur has been shown by blocking the formation of the MangGlcNAcz isomer B structure. When glucose or mannose trimming was abrogated through gene disruptions in the processing w-glucosidases or ER mannosidase I, or when the synthesis of the lipid-linked

86

6 a-Munnosidases in Asparagine-linked Oligosaccharide Processing

precursor was blocked by gene disruption of the oligosaccharide biosynthetic enzymes, misfolded carboxypeptidase Y accumulated in the ER [22, 351. In mammalian cells, the degradation of a misfolded variant form of a,-antitrypsin [36, 371 or the T cell receptor subunit CD3-6 [38] was blocked by treatment of the cells with DMJ indicating the involvement of ER mannosidases. Treatment of cells with kifunensine, a selective inhibitor of ER mannosidase I but not ER mannosidase 11, also blocked the degradation of a1 -antitrypsin implicating a requirement for the formation of the MangGlcNAc2 isomer B structure by ER mannosidase I in targeting glycoproteins for degradation [37]. It is noteworthy that ER mannosidase 1 is the last step in oligosaccharide processing that is fully conserved from yeast to mammals. In S. cerevisiue oligosaccharide structures are extended into mannan structures after the single mannose cleavage step by ER mannosidase I [17, 391. In mammals, trimming to the MansGlcNAq isomer B structure is followed by further mannose trimming and extension into complex type structures [7]. ER mannosidase I does not appear to be essential for growth in S. cerevisiue [40] or extension of mannan structures (411, but evidence described above indicates that it does appear to be involved in the timing of glycoprotein folding [22]. Inhibition of ER mannosidase 1 and subsequent Golgi trimming by treatment of mammalian cells in culture with kifunensine does not appear to cause alterations in cell morphology or growth, but does have an effect on the rate of degradation of unfolded glycoproteins in the ER [36, 371. Two models have been proposed for the recognition and degradation of unfolded glycoproteins containing the MansGlcNAcz isomer B structure. In one model, the MansGlcNAcl isomer B structure can be acted upon by the glucosyltransferase to make Glcl MansGlcNAcz, but the product may be cleaved less efficiently by glucosidase I1 [37]. This would lead to a prolonged interaction with calnexin/calreticulin and potential targeting to the Sec6lp translocon. An alternative model proposes that an, as of yet unidentified, lectin may recognize the MansGlcNAc2 isomer B structure in the context of an unfolded protein to target the glycoconjugate for translocation [22]. In each model discrimination between folded and unfolded substrates must be made and a low efficiency of recognition of the oligosaccharide on unfolded proteins would be predicted since transiently unfolded nascent polypeptides and correctly folded glycoproteins should not be targets for translocation and degradation. Thus, the conservation of early processing events in oligosaccharide maturation appears to contribute to the quality control recognition processes in the ER, and these processes are independent of the roles that the oligosaccharides may play in later compartments of the secretory pathway. A role for high mannose structures has also been proposed for facilitating glycoprotein transport through the secretory pathway. ERGIC-53, a protein that had previously been identified as a marker for the intermediate compartment between the ER and Golgi, has recently been found to act as a binding protein specific for high mannose oligosaccharides [42, 431. This protein has been hypothesized to mediate glycoprotein transport between the ER and Golgi complex through a direct interaction with high mannose oligosaccharides on newly synthesized glycoproteins [43]. A dominant mutant form of the protein that is retained in the ER has been expressed in HeLa cells and results in the mislocalization of cathepsin C, while other

6.2 Intvoduction

87

lysosomal and cell surface glycoprotein markers were unaffected [44]. A human genetic disease characterized by a combined deficiency in coagulation factors V and VIII has recently been shown to be linked to mutations that result in truncated or “null” expression of ERGIC-53 [45, 461. The levels of these coagulation factors are 5% and 30% of normal, respectively, in the serum of affected patients indicating that ERGIC-53 may be involved in the transport of a subset, but presumably not the majority, of glycoproteins through the secretory pathway.

6.2.4 Glycoprotein Catabolism: multiple routes for glycoprotein breakdown Lysosomes contain a wide range of proteolytic and glycosidic enzyme activities for the full degradation of glycoprotein substrates to monosaccharides and amino acids. At least three potential transport routes converge in the lysosome for the final degradation of glycoprotein oligosaccharides. Two routes, including classical uptake of glycoproteins by receptor mediated endocytosis and autophagy result in the delivery of substrates to lysosomes by vesicle-mediated fusion or engulfment 191. In contrast, soluble oligosaccharides can gain direct access to the lumen of the lysosome by transport from the cytoplasm. Free or protein-associated oligosaccharides can be released into the cytosol from the ER, cleaved to smaller structures, and transported directly into lysosomes via an ATP-dependent transporter in the lysosoma1 membrane [47, 481. Soluble oligosaccharides can be produced from a variety of sources in the ER. These sources include: 1) the release of Man5GlcNAc2-P from Man5GlcNAcz-P-P-Dol on the cytosolic face of the ER membrane by phosphodiesterases [lo]; 2) the release of Glc3-oMang-8GlcNAc2 by the hydrolysis of Glc3MansGlcNAc2-P-P-Do1, presumably through the transfer to water in an oligosaccharide transferase reaction 149, 501, and subsequent export of the oligosaccharide into the cytoplasmic compartment via an ATP- and calcium-requiring process [51]; and 3) the translocation of unfolded glycoproteins through the Sec61 translocon complex into the cytoplasm [33, 521 where they are acted upon by a cytosolic peptide-N-glycosidase 153, 541 to yield Mans-8GlcNAcl oligosaccharide structures [ 10, 551 and degradation of the protein component by the cytoplasmic proteosome. Evidence has also been presented for the release of oligosaccharides in the lumen of the ER by an endo-p-N-acetylglucosaminidasein thyroid microsomal membranes, but the in vivo requirements for substrate recognition are unknown 1491. Cleavage of the core of these soluble oligosaccharides to a single GlcNAc residue is accomplished by a cytosolic neutral chitobiase 156, 571 followed by the selective cleavage of the mannose residues to a distinctive Man5GlcNAcl isomer by a cytosolic neutral mannosidase [ 5 8 ] (see below). The sequential nature of the cleavage pathway is indicated by the requirement for cleavage to a GlcNAcl core prior to the mannose cleavage steps in the cytoplasm [58]. Thus, the routes for oligosaccharide catabolism can come from cellular uptake, autophagy, or through release of oligosaccharides from lipid or glycoprotein intermediates in the ER. The pathways for biosynthesis and degradation of glycoproteins diverge in the ER based on the folding-based decision of quality control. Folded glycoproteins which are no longer

88

6 a-Munnosiduses in Aspurugine-linked Oligosucchuride Processing

recognized by the glucosyltransferase, disengage from calnexin/calreticulin in the ER, and are packaged for anteriograde transport at exit sites in the ER for transport down the secretory pathway to the Golgi complex. Unfolded glycoproteins are cleaved to a Glcl-oMan&lcNAcz isomer B structure and, in a process that is still not understood, engage the Sec6lp complex for export from the ER into the cytoplasm for degradation.

6.2.5 Consequences of Genetic Defects in Oligosaccharide Biosynthesis and Catabolism The physiological importance of maintaining the balance of the biosynthetic and catabolic pathways can most effectively be seen in the variety of human genetic diseases where defects in the enzymes involved in either glycoprotein synthesis or degradation cause disruption of the respective pathway. In the heterogeneous collection of human genetic diseases characterized by the defective expression or targeting of lysosomal enzymes, termed lysosomal storage diseases, the enzyme defects cause an extreme accumulation of the normally degraded substrates [9]. The pathologic consequences of the oligosaccharide storage material are commonly a variety of neural, skeletal, and immune system defects that can result in death before the age of ten. In contrast to the severe clinical consequences of lysosomal storage diseases, genetic heterogeneity of the enzymes involved in glycoprotein biosynthesis can have varied consequences [20, 591. The enzymes that are involved in the maturation of N-linked and 0-linked glycans can generally be divided into two classes, those that are involved in the maturation of the oligosaccharide core and those that generate the terminal oligosaccharide capping structures. Genetic heterogeneity in the enzymes that generate the terminal capping reactions on N- and 0-glycans is surprisingly common in the human population, as evidenced by the variability in the carbohydrate-based blood group substances on cellular and secreted glycoconjugates [ 191. Animal models of genetic defects in some of the terminal transferases have been shown to cause a variety of immune system defects [60-621, yet these enzymes are not usually required for early development. Gene disruption in the enzymes catalyzing terminal capping reactions commonly results in the formation of alternate capping structures or compensation by other enzymes with a similar specificity. A hypothesis has been presented that the genetic and structural variation in terminal glycosylation reactions may reflect the genetic remnants of a selection process for resistance to natural pathogens by interrupting pathogen interaction with cell surface glycoconjugates [ 11. More severe pathological effects are commonly correlated with genetic defects in enzymes that are involved in the synthesis of the glycoprotein core. GlcNAc transferase I appears to be essential for mammalian development since mice lacking the enzyme die by day 10.5 in embryonic development [63, 641. A heterogeneous collection of genetic diseases termed carbohydrate-deficient glycoprotein syndrome (CDGS) [65] are caused by defects in either the synthesis of the lipid-linked precursor of N-linked oligosaccharides (type I CDGS[66]) or the early maturation of

6.3 Munnosiduses in GIj.coproteirz Processing und Catubolism

89

the glycoprotein core (GlcNAc transferase I1 deficiency in type I1 CDGS[67]). The outcome is a collection of multisystemic diseases with major nervous system involvement. In contrast, HEMPAS disease is a heterogeneous genetic disease characterized in some individuals by a genetic defect in Golgi mannosidase TI, a n enzyme that acts between GlcNAc transferases I and 11. Patients with this enzyme defect have relatively mild symptoms of anemia and mild secondary tissue effects [68, 691. Analysis of glycoprotein structures derived from patient samples [69] and tissues isolated from a mouse model for the Golgi mannosidase I1 deficiency [70] indicate that there is a previously undetected backup pathway for the trimming to the Man3 core. Multiple routes for oligosaccharide processing may also be available in the early glucose and mannose trimming steps. No naturally occurring genetic diseases have been identified involving the early oligosaccharide trimming steps indicating that each of these steps are either essential for development or that the overlapping specificities of alternate enzymes mask the pathological effects of an enzyme deficiency. There can also be an overlap in the processing of glucose and mannose residues between the exo-glucosidases (glucosidase I and 11), an endomannosidase activity that excises the terminal glucose residues along with an underlying mannose residue. and the collection of exo-mannosidases that lead to the trimming down to the MansGlcNAcr core [ 161. Thus, early steps in the synthesis, transfer, and trimming of N-linked oligosaccharides appear to be critical for normal development and physiology of vertebrate organisms, at least in the few cases where non-overlapping specificities can be demonstrated.

6.3 Mannosidases in Glycoprotein Processing and Catabolism The major goal of this review is to describe the most recent literature on the role of inannose cleavage reactions in oligosaccharide maturation and catabolism in mammalian systems. Reference will also be made to glycoprotein processing in other organisms when relevant. Several previous reviews have described the biochemistry and molecular biology of the processing and catabolic mannosidases in fungal and vertebrate systems [6, 7, 16, 17, 391 and additional relevant details and references may be found in these sources.

6.3.1 Classification of Mannosidases Mannosidases involved in glycoprotein maturation and catabolism have been divided into three broad classes based on their distinctive substrate specificities, responses to inhibitors, cation requirements, protein molecular weights, subcellular localizations, and enzyme mechanisms (Table 1) [7, 161. This classification, based on biochemical characteristics, has been reinforced by the sequence comparison of a growing number of family members (Figure 2). Two of the mannosidase classes are exo-glycosidases, cleaving mannose residues from the non-reducing terminus of the oligosaccharide. The third mannosidase class cleaves an internal glycosidic

al,2-Man

ER Man I1 al,2-Man

Golgi Man IA a1,2-Man

Golgi Man IB a1,3/1,6-Man

Golgi Ma nII ?

Golgi ManII"

DMJ, KIF None 6.5 Mouse, Human. Pig

Class 1

DMJ, DIM

pNP-Man

6.5 Rat

Class 2

DMJ, KIF

None

6.5-7.0 Human

Class 1

Caf2

Golgi

ER None

ER Ca+'

Class 1

6.5 Mouse, Human

None

DMJ, KTF

Golgi Ca+'

Class 2

5.6 Mouse, Human

pNP-Man

Sw, Manstat

Golgi None

Class 2

Human, Mouse

?

Lysosomal Man

Cytosol Co+' activates

116 kDa

Class 2

6.0 Rat

Class 2

4.5 Mouse, Human, Cat, Cow

pNP-Man

sw

Lysosomes None

111 kDa

retaining

Man~GlcNAc?" Manpl,4 GlcNAc

a1,2/1,3/1.6-Man al,2/1.3/1,6-Man

Cytosolic Man

DMJ, DIM, Sw(poor) pNP-Man pNP-Man

sw

Golgi None

MansGlcNAcl MansGlcNAcz ManSGlcNAcz* Man5GlcNAc2 * GlcNAcMan3- ? GlcNAcz inverting inverting (yeast) 82 kDad 76 kDa 13 kDa 73 kDa 131 kDa I29 kDa

al,Z-Man

ER Man I

"The nomenclature and abbreviations for the enzyme names as well as the references for the indicated characteristics are found in the text. Hypothesized in vivo limit digestion products. KIF, kifunensine; DMJ, deoxymannojirimycin; DIM, I ,4-imino-1,4-imino-D-mannitol; Sw, swainsonine, Manstat, mannostatin. Subunit Mr for ER mannosidase I1 is based on the SDS-PAGE mobility of the protein detected in an ER membrane fraction by Western blotting [ 1061. *Golgi Man IA cleaves the ManRB isomer to MansGlcNAcz while it cleaves MangGlcNAq to MansGlcNAc2 and with low efficiency to MansGlcNAcz. # Cytosolic Man cleaves ManTGlcNAc to a ManSGlcNAc isomer that is distinctive from the digestion product of Golgi Man IA or IB.

Synthetic substrates pH optimum Cloned mammalian cDNAs or genes Mdnnosidase Class

specificity Limit digestion product Glycosidase mechanism Subunit M, (not including glycosylation) Localization Metal requirement Inhibitors'

Linkage

Characteristics

Table I . Mammalian Processing and Catabolic a-Mannosidases"

I?

ic

s g.

C

s.

h.

R

0

3 2

S'

ic

2

b +3

3

D

&

0;

-.

S S 0

5

8

o\

W

0

(BAA758171

Figure 2. Dendrograms for members of the two mannosidase classes. Full length members of the Class 1 and Class 2 mannosidase families that are present in the GenBank sequence database were identified by BLAST searches (194) and the sequences were edited to reflect the conserved core sequence in each family (-440-510 amino acid region for Class 1 mannosidases and -180 amino acid region for Class 2 mannosidases). Multiple sequence alignments were performed using the ClustalW program (195) and a neighbor-joining tree was generated within this program using 1000 bootstrap trials. The tree was displayed as a radial unrooted tree using the program Treeview (196). Sequence and species designations and their corresponding GenBank accession numbers (in parentheses) are indicated at the end of each branch. Those sequences that have quotation marks in the title indicate sequences that are proposed to have the corresponding activity based on sequence similarity to other members of the subgroup, but the catalytic activity has not been demonstrated by direct enzyme assays. Those sequences that have only gene designations, but no putative mannosidase designation in the title (as indicated by a "Man" in the title) are genes with unknown function. The outer oval with the indicated labels show the designations of the subgroups within the Class 1 and Class 2 mannosidase families based on the sequence similarity analysis.

ER

Gold Mannosidase I1

Unknown function

I

(Glycosylhydrolase Family 38)

(Glycosylhydrolase Family 47)

Epididy mal/sperm Mannosidase

Class 2 Mannosidases

Class 1 Mannosidases

92

6 x-Mannosidases in Asparagine-linked Oligosaccharide Processing

linkage between two internal al,2-linked mannosyl residues, and is designated an endo-a-mannosidase [ 7 1, 721. Although the latter mannosidase class has only a single member, it is unique in that it is the sole enzyme involved in glycoprotein maturation that catalyzes an endoglycosidase reaction. The endo-a-mannosidase cleaves Glc+]Man&ilcNAcz structures by the removal of Glc3-lMan to yield a MangGlcNAcz isomer A structure (Figure 1) [71]. The enzyme is found predominately in chordate organisms, with the exception of several classes of molluscs [73], and is believed to play a role in ensuring the complete trimming of glucose residues from glycoproteins that have completed folding and have exited the ER prior to full excision of the terminal glucose residues [73]. A potential role of the endo-amannosidase in the dissociation of folded proteins from calnexin and calreticulin in the early Golgi complex has also been proposed [74]. Since the role of the endo-amannosidase is predominately for the removal of residual glucose residues, the discussion of the biochemistry and molecular biology of the enzyme is described in Chapter 5 , glucosidases in oligosaccharide processing, in this volume. The other two classes of mannosidases that will be the subject of the remainder of this review are exo-mannosidases that are involved in glycoprotein biosynthesis and catabolism. The original proposal for two exo-mannosidase classes was based on both the biochemical characteristics of the enzymes purified from natural sources and on sequence data that had been obtained on a handful of representative members of each of the two classes 171. Since that time, numerous additional related sequences have been obtained and additional biochemical characteristics have been determined for enzymes purified from either natural sources or as recombinant enzymes purified from heterologous expression systems. The Class 1 mannosidases are now distinguished by several characteristics (7, 161 including: 1) similarities in sequence including a conserved 440-5 10 amino acid catalytic domain; 2) specificity for cleaving a1,2-mannose linkages; 3 ) in most cases a requirement for Ca2+ for catalytic activity; 4) sensitivity to inhibition by the pyranose substrate mimics deoxymannojirimycin (DMJ) and kifunensine (KIF) [6]; and 5 ) cleavage of the glycosidic linkage by inversion of configuration of the released mannose residue [75, 761. This classification contrasts them with the more heterogeneous collection of processing and catabolic mannosidases, termed Class 2 mannosidases, that are present in the ER, Golgi, lysosomes, and cytosol of mammalian cells, but have additional homologs in a widely diverse array of organisms including eubacteria and archea (see below). A more restricted sub-region of sequence is conserved among the Class 2 mannosidases but this region contains an acidic amino acid that is known to be part of the active site [77]. Class 2 mannosidases are larger (1 10-135 kDa), do not require cations for catalytic activity, are sensitive to inhibition by the furanose transition state analogs swainsonine (Sw) and 1,4-dideoxy-1,4-imino-r>-mannitol (DIM) [6], and cleave glycosidic linkages by retention of anomeric configuration of the released monosaccharide [78]. The nomenclature for the multiple members of the mannosidase families can appear to be quite confusing, especially with multiple Class 1 and Class 2 enzymes in the ER and Golgi. Fortuitously, the nomenclature for the enzymes is in agreement with the nomenclature of the classes. Thus, ER mannosidase 1 and Golgi mannosidases IA and IB are Class 1 mannosidases. ER mannosidase 11, Golgi mannosidase 11, and Golgi mannosidase Ilx are Class 2

6.3 Mannosiduses in Glycoprotein Processing and Catabolism

93

mannosidases. Other Class 2 enzymes, including the lysosomal mannosidase and the cytosolic mannosidase, do not follow the same nomenclature, but so far they are the only mannosidases in each respective compartment. A web-based database (URL: afmb.cnrs-mrs.fr/-pedro/CAZY/ghf.html)has collected data on the characteristics of classes of glycosylhydrolases based primarily on sequence similarity and has also separated the two enzyme classes into different families. (Class 1 = family 47; Class 2 = family 38) [79-811. The remainder of this review will summarize the recent data on the members of the two mannosidase classes in an effort to emphasize their roles in glycoprotein biosynthesis and catabolism.

6.3.2 Class 1 Mannosidases: enzymes of the ER and Golgi The Class 1 mannosidases were originally defined based on the biochemical characteristics of a collection of mammalian Golgi processing mannosidases and an oligosaccharide processing mannosidase isolated from S. cerevisiae [7].We now know considerably more about the enzymes involved in early glycoprotein processing including their specificities and roles in individual processing steps. Comparison of the sequences of Class 1 mannosidases makes it possible to define subgroups within the members of the family (Figure 2). When the presently available full-length Class 1 protein sequences are compared, at least four subgroups can be identified and three of the four groups contain mammalian sequence homologs (Figure 2). Additional partial sequences can also be found in DNA sequence databases, including numerous additional mammalian EST sequences, but these sequences will not be considered as a part of this discussion.

ER mannosidase I subfamily As mentioned above, Class 1 mannosidases were first identified in S. cerevisiae and mammalian systems, but significant differences in substrate specificities were identified between the enzymes from the two sources [7]. The yeast enzyme was shown to cleave only a single mannose residue from MangGlcNAcz to form the Manx GlcNAcz isomer B structure, without further cleavage [82-85]. This organism does not process their oligosaccharides beyond MansGlcNAcz before extension into mannan structures [17, 391. The enzyme from S. cerrvisiae has been the most completely studied of the Class 1 mannosidases, both as an enzyme purified from natural sources and by characterization of the recombinant enzyme 140, 41, 76, 82-90]. Yeast ER mannosidase I is an ER resident transmembrane protein containing a lumenally oriented catalytic domain [88]. Like the other Class 1 mannosidases the yeast enzyme can not cleave aryl-a-mannosides, is inhibited by dMNJ and KIF, but not Sw, and requires Ca+2 for catalytic activity [ 17, 83, 841. Contrary to prior hypotheses, the yeast enzyme has recently been shown not to bind the cation via the putative EF-hand-like consensus sequence that was found in several of the Class 1 enzymes [ 871. Mutagenesis studies have also identified several acidic residues that are required for Ca+2 binding and catalytic activity [87]. The S. cerevisiae enzyme has been overexpressed in Pichia pastoris and the recombinant enzyme was the first

94

6 r-Munnosiduses in Aspuragine-linked Oligosncchuride Processing

of the Class 1 enzymes in which the mannosidase reaction was shown to proceed with an inversion of the anomeric configuration of the released monosaccharide [76]. The recombinant yeast ER mannosidase I has been purified in milligram quantities, crystallized, and an X-ray diffraction data set has been obtained [90] indicating that this enzyme will most likely be the first glycoprotein processing hydrolase to have its structure determined. The initial characterization of the early processing mannosidases in mammalian systems was far more coniplicated than in yeast as a result of the multiple enzyme activities present in different subcellular compartments in the secretory pathway. Early studies on the processing mannosidases took advantage of the ability to identify glycoprotein processing intermediates by pulse labeling cultured cells with radioactive monosaccharide precursors, either in the presence of processing mannosidase inhibitors or following treatment with ionophores, such as CCCP, that blocked the transport of glycoproteins between the ER and Golgi [91-991. The early literature on the ER processing mannosidases is full of confusing and contradictory data indicating the presence of DMJ-sensitive and DMJ-resistant mannosidase activities producing MansGlcNAcl and smaller structures [91-93, 981. In addition, mannosidases were detected, purified, or cloned that had substrate specificities that did not match the substrate specificities predicted by the metabolic labeling studies in cultured cells. It is now clear that at least some of the prior confusion was the result of the presence of multiple unanticipated mannosidase activities in the ER. Recent data now indicate that there are at least three principal mannosidase activities in the ER of mammalian cells. The first mannosidase activity is the aforementioned endo-a-mannosidase that cleaves Glc3-l Man9GlcNAcz to MangGlcNAc2 isomer A [71, 72, 100-1041. The second activity, termed ER mannosidase I, cleaves MangGlcNAcl to Man8GlcNAcl isomer B, analogous to the processing mannosidase in S. cereuisiae [ 105-1071. A third activity, termed ER mannosidase 11, cleaves MansGlcNAcz to MangGlcNAc2 isomer C [ 106-1081. The latter enzyme is a Class 2 mannosidase, whereas ER mannosidase I is a Class 1 enzyme. In vitro enzyme assays using ER membrane preparations have been used to examine ER mannosidase I activity in rat liver. This activity was shown to be sensitive to inhibition by DMJ, KIF, and EDTA, but not Sw or DIM [106, 1071. Human ER mannosidase I has recently been cloned, expressed, and characterized and found to have a high degree of sequence similarity to yeast ER mannosidase I [ 1051. In addition, two putative open reading frames were identified in the Cuenorhubditis elegans genome that have a high degree of sequence similarity to the yeast and human enzymes indicating that they may be nematode orthologs [ 1051. Sequence analysis indicates that, like yeast ER mannosidase I, the human and C. elegans proteins have a single transmembrane domain near their NH2-terminus. The proposed cytoplasmic tails and the “stem regions” of the enzymes are quite variable in length and in sequence, but the COOH-terminal 440-504 amino acid region that contains the catalytic domain of the yeast enzyme is highly conserved. These four sequences constitute the presently known members of the ER mannosidase I subgroup of Class 1 mannosidases (Figure 2), although all organisms between

6.3 Munnosiduses in Gljicoprotein Processing unu' Cutuholisni

95

yeast and mammals would be predicted to contain an equivalent of this Class 1 a-mannosidase sub-group. Recombinant human ER mannosidase I has been expressed and purified and the enzyme has many catalytic characteristics in common with the yeast enzyme, including substrate specificity, the requirement for Ca+2 for catalytic activity, and inhibition by DMJ and KIF, but not Sw [ 1051. A recombinant epitope-tagged form of the human enzyme is also localized to the ER of mammalian cells indicating a conservation of localization as well as function of this ER enzyme [105]. The position of ER mannosidase I as the last fully conserved step in glycoprotein processing between yeast and mammals argues strongly in favor of a critical role for this enzyme in order to maintain selective pressure during eukaryotic evolution. The proposal that the enzyme acts as the timing step in determining the failure of glycoprotein folding in ER and targeting the substrates for degradation [22, 371 is a reasonable hypothesis for this essential role. The mechanism and components involved in this timing step remain to be elucidated. Golgi mannosidase I sub-family The second subfamily of the Class I mannosidases includes enzymes that act predominately in the Golgi to cleave Mang-gGlcNAcz structures to MansGlcNAcz [7]. There are multiple members of this class that have been identified from mammalian, insect, and nematode origin (Figure 2), although the latter sequences are putative open reading frames based on sequence similarities alone. The nomenclature for the mammalian forms of the enzymes in this sub-class has been quite confusing. Sequence alignments indicate that there are at least two separate lineages of mammalian Golgi mannosidase I-like activities. One subgroup, designated Golgi mannosidase IA, has been cloned from human [109], mouse [110], rabbit [ 1 101, and pig [ 1111 sources. Members of the second sub-group are termed Golgi mannosidase IB and are described further below. The mouse cDNA encoding Golgi mannosidase IA has been expressed in mammalian cells [ 1101 and in P. pastoris [75] where it has been shown to encode an enzyme that can efficiently cleave Mans GlcNAc2 to MansGlcNAc2, but slowly cleaves the final ul,2-mannose residue to form Man5GlcNAcz. The released monosaccharide in the cleavage reaction is inverted in anomeric configuration indicating that the mammalian enzyme shares a common enzymatic mechanism with yeast ER mannosidase I [75]. Cleavage of MansGlcNAcz by the enzyme produces a distinct series of isomers and the w1,2mannose residue that was produced by the final cleavage step was found to be the sole mannose residue that is recognized and cleaved by the ER mannosidase I. These data indicate that there is a complementarity and nearly non-overlapping specificity in substrate recognition between ER mannosidase I and the Golgi processing mannosidases. Since ER mannosidase I is localized earlier in the secretory pathway it would be expected that the Golgi mannosidases would more commonly encounter substrate glycoproteins containing MangGlcNAcz isomer B structures. The difference in substrate recognition between ER mannosidase I and members of the Golgi mannosidase I subfamily is striking considering their sequence similarity

96

6 a-Munnosiduses in Aspurugine-linked Oliyosuccharide Processing

as Class 1 mannosidases and their presumed similarity in protein structure. Recombinant murine Golgi mannosidase IA expressed in Pichiu has been purified in milligram quantities, crystalized, and X-ray diffraction data has been obtained [112]. Comparison of the structure of this enzyme with the structure of yeast ER mannosidase I will elucidate the structural features of the enzymes that specify their substrate recognition. The cDNAs encoding the human [ 1091 and pig [ 11 11 equivalents of Golgi mannosidase IA have also been isolated. These enzymes have previously been called Many-mannosidases, but this nomenclature remains confusing since several enzymes can cleave mannose residues from Man9GlcNAcz. Surprisingly, immunoEM studies using antibodies raised to the purified pig enzyme indicated that this enzyme is present in ER, not the Golgi, of pig hepatocytes [113]. Substrate specificity studies demonstrated that the pig and calf forms of this enzyme were able to cleave MangGlcNAc? to ManhGlcNAc2 [114]. The combination of an ER localization and the substrate specificity data led to a proposal that the enzyme was distinct from the Golgi-localized enzymes that were supposed to cleave down to MansGlcNAc2. With the cloning of the mouse, rabbit, pig, and human cDNAs it is now clear that they are all mammalian orthologs of the same enzyme activity, but with potentially varied localizations. The basis for the localization difference of the pig enzyme is presumably inherent to the primary sequence of the enzyme since parallel studies on the transfection and localization of the pig and human mannosidase cDNAs indicated an ER localization of the pig enzyme in contrast to a Golgi localization of the human enzyme [ 11I]. Mouse L-cells stably transfected with mouse Golgi mannosidase IA cDNA also resulted in a Golgi pattern of immunofluorescence [ 1 101. Cell and tissue specific variation in localizations of Golgi mannosidase IA within the Golgi and post-Golgi compartments of the secretory pathway have been previously seen in rat tissues [ 1151, and comparison of the localization of the porcine and human forms of Golgi mannosidase IA now indicates the possibility of species-specific variation in localization of these homologous enzymes in a given cell or tissue type. These data also indicate that the nomenclature distinguishing “ER” and “Golgi” mannosidases is misleading since there can be an overlap in localization. Despite the potential for confusion in nomenclature, the enzymes in the ER mannosidase I sub-family and the Golgi mannosidases I subfamily have historically employed this nomenclature and it will continue to be used here. Multiple forms of Golgi mannosidase I-like activity have been previously separated from rat liver Golgi membrane extracts by ion exchange chromatography and gel filtration [116, 1171. These forms were designated Golgi mannosidases IA and IB, and the enzymes were found to share many biochemical and enzymatic characteristics, including inhibition by DMJ and EDTA, and cross reactivity with antibodies raised to rat liver Golgi mannosidase IA. The two forms were distinguished by their chromatographic behavior and a low but detectable level of activity by Golgi mannosidase IB toward pNP-Man. In agreement with the presence of two enzymatic forms of Golgi mannosidase I-like enzymes in rat tissues, a second mammalian cDNA was isolated from a murine cDNA library that was distinct from Golgi mannosidase IA [ 1181. The protein encoded by this cDNA was des-

6.3 Mannosiduses in Glj~coj~roteiiz Processing und Cutaholisin

97

ignated Golgi mannosidase IB by analogy to the two forms of mannosidase activity that were separated and isolated from rat liver Golgi membranes. It is not clear whether there is a direct correspondence between the enzymes separated from rat liver extracts and designated Golgi mannosidase IA and IB and the cDNAs with the same name. At the protein level murine Golgi mannosidases IA and IB are approximately 65% identical within the catalytic domain [ 1 181. The human cDNA and gene encoding Golgi mannosidase IB have also been isolated recently and the human protein is 94% identical to the murine Golgi mannosidase IB [119]. Expression of Golgi mannosidase IA and IB in Pichia has allowed a detailed comparison of their substrate specificities [75]. Both enzymes efficiently cleave MangGlcNAcl to ManhGlcNAc2 and slowly cleave the last al,2-mannose residue to yield ManjGlcNAcz. Differences were detected in the order of residues cleaved when MangGlcNAc2 was the substrate, but a similar order and rate of cleavage of the a1,2-mannose residues were detected when MansGlcNAc? was the substrate. An epitope-tagged form of Golgi mannosidase 1B was localized to the Golgi by immunofluorescence in transfected COS cells similar to murine Golgi mannosidase IA [ 1181. The transcripts encoding both Golgi mannosidase IA and IB were detected in Northern blots from most mouse [ 1181. rat [ 1 101, and human [ 1191 tissues, but there were significant differences in the ratios of the two transcripts. Differences in expression at the cellular level were also detected by immunofluorescence in the rat reproductive tract [ 1201. These data indicate that the similar Golgi mannosidase Ilike forms are most likely not redundant enzymes but that ancestral gene duplication and divergence has resulted in tissue-specific expression patterns. Fungal secreted mannosidases

Two fungal mannosidases have been purified and cloned that have sequence similarity to Class 1 mannosidases, but differ in that they both contain cleavable signal sequences, are found as secreted proteins, appear to not require Ca+' for catalytic activity, and have relatively lower pH optimums. The enzymes cloned from Aspergillus saitoi [ 121-1251 and Pennicilium citrinunz [ 126, 1271 appear to form a separate sub-group of the Class 1 mannosidases based on sequence comparisons (Figure 2). The enzymes have been shown to cleave al,2-mannose residues from high mannose substrates, and the detailed substrate specificity of the Aspergillus al,2-mannosidase has been recently determined [ 1241. Surprisingly, this fungal a-mannosidase has a substrate specificity that is quite similar to the mammalian Golgi mannosidases IA and IB and contains a bound Ca2+ despite prior data indicating that Ca2+ was not required for activity. Site-directed mutagenesis of the Aspergillus al,2-mannosidase identified several residues that are required for catalytic activity [ 1251. The Pennicilium enzyme has also been extensively characterized and cloned [126, 1271. Treatment of the Pennicilium enzyme with a carbodiimide reagent specific for acidic amino residues resulted in an inactivation of the enzyme and a single acidic amino acid residue was found to be protected by preincubation with DMJ [ 1271. The single residue is conserved among the 23 Class 1 mannosidases (Asp375 in the Penniciliunf enzyme) and indicates that this residue may be involved in the active site.

98

6 a-Munnosiduses in Asparugine-linked Oligosacchuride Processing

New genes with unknown functions Sequence similarity searches have also identified a broad collection of mannosidase I-like genes whose functions are not clear. Members from this sub-group include putative open reading frames from S. cerevisiae (two distinct gene products), Schizosaccharomyces pombe, C. eleyans (three distinct gene products), Drosophilu melunogaster; and Homo supiens (at least one gene product). Members of this subgroup are unique among the Class 1 mannosidases in that they contain a sequence extension on the COOH-terminal end of the protein that is not conserved among the other Class 1 mannosidases and has no similarity to other known sequences. In most cases the sequence of the extension is not very well conserved even among members of this subgroup. Members of the subgroup generally contain an NH2terminal hydrophobic sequence, but it is not clear whether these sequence are membrane spanning domains or cleavable signal sequences. One member of this collection (Saccharomyces cerevisiae Y LR057w) is predicted to be an integral membrane protein and contains an extended insertion within the center of the conserved Class 1 mannosidase catalytic domain sequence with a corresponding deletion of part of the conserved sequence. Since no predicted processing mannosidase activities are unaccounted for among the other members of the Class 1 enzymes it is not clear what function these proteins play in vivo. Their identification in the genomes of organisms ranging from yeast to mammals indicate that they most likely play an important role in some biochemical process in the cell, but as of yet none of the members of this subgroup have been shown to contain a catalytic activity.

6.3.3 Class 2 Mannosidases: enzymes of the cytosol, ER, Golgi, and lysosomes The abilities of mammalian Class 2 mannosidases to cleave the synthetic aryl-amannoside substrates, including pNP-a-D-mannoside or 4-methylumbelliferyla-D-mannoside, were instrumental in the early identification, characterization, and distinction of these enzymes. These early studies examined the pH optimum, subcellular fractionation, and effects of cations and EDTA on the rate of substrate cleavage [ 128-1301. It was determined that mammalian tissues contained at least three easily distinguished activities that could act on the synthetic substrate [ 1301. The lysosomal mannosidase was distinguished by a low pH optimum (pH -4.5), stimulation by Zn+2, and co-fractionation with lysosomal enzyme markers. The soluble or cytosolic mannosidase was distinguished by a more neutral pH optimum , fractionation with the cytosolic fraction. The (pH -6.5), stimulation by C O + ~and Golgi mannosidase was distinguished by an intermediate pH optimum (pH - 5 4 , lack of an effect by cations, and fractionation with Golgi membranes. Each of these enzymes has subsequently been purified and cloned from multiple mammalian sources, and sequence comparisons have confirmed the distinctions made on biochemical grounds (Figure 2). Enzymes that fall within the category of Class 2 mannosidases are distinguished from Class 1 mannosidases by similarities in size of their catalytic region (110116 kDa). Although the sizes of the proposed “catalytic domains” of the Class 2

6.3 Mannosiduses in Clycoprotein Processing and Catabolism

99

mannosidases are large, the portion of this sequence that is most highly conserved among the family members is more restricted. Sequence comparisons used to assemble the dendrogram in Fig. 2 employed an -182 amino acid region that is nested to the NHz-terminal end of the catalytic region. Among the 24 independent full-length sequences that were identified in the sequence databases, phylogenetic analysis indicates that the enzymes are clustered into four subgroups, with three of the sub-groups corresponding to the mammalian enzymes that were originally defined based on biochemical and cell fractionation characteristics. Golgi mannosidase I1 The identification of a pNP-mannoside-cleaving mannosidase in the Golgi complex and its distinction from enzymes in the soluble fraction or lysosomes was one of the first indications that cleavage of oligosaccharide intermediates during glycoprotein maturation might be catalyzed by enzymes in the Golgi complex [128]. Metabolic radiolabeling studies indicated that glucose and mannose cleavage were a necessary prerequisite for extension into complex type structures [131, 1321 and the pNP-mannose cleaving activity in the Golgi complex was the most likely candidate for the processing reactions. It was only years later that it was discovered that several mannosidase activities were required for the oligosaccharide trimming steps and the enzyme that catalyzed the pNP-mannoside cleavage in vitro was capable of selectively cleaving only specific al,3- and al,6-mannose residues from the MangGlcNAcz processing intermediate [ 1 161. The natural substrate specificity was surprisingly restricted considering the ability of the enzyme to cleave the synthetic substrate. Only the two non-reducing terminal a1,3- and al,6-mannose residues on GlcNAcManSGlcNAcz could be cleaved by the enzyme and no direct cleavage of Man5GlcNAcz was detected. The substrate specificity indicated that there was an overlap in the trimming and elongation phases of the oligosaccharide biosynthetic pathway with Golgi mannosidase I activities cleaving to a MansGlcNAc2 structure followed by the action of GlcNAc transferase I and subsequent action by the enzyme designated Golgi mannosidase I1 to form GlcNAcMan3GlcNAcz [ 1331351. Enzymes with similar substrate specificities to mammalian Golgi mannosidase I1 have been identified in mung beans [136] and insect cell lines [137]. In contrast, oligosaccharide trimming in Dictyostelium discoideum appears to not require the prior action of GlcNAc transferase I [138, 1391. A Sw-sensitive direct cleavage of MansGlcNAcz to smaller structures was found in this organism. The human [140] and mouse [141] forms of Golgi mannosidase I1 have been cloned and were shown to be type I1 transmembrane proteins with a small (5 amino acid) NHz-terminal cytoplasmic tail, a single transmembrane domain, a region of sequence on the luminal side of the membrane that is dispensable for catalytic activity (-80 amino acids, generally termed the “stem domain”), and a large luminal catalytic domain (-1017 amino acids) [ 1411. This structural topology is common with other Golgi and ER processing mannosidases and glycosyltransferases involved in glycoprotein extension [7, 1421. It is noteworthy that only the luminal catalytic domain, excluding the cytoplasmic tail, transmembrane domain, and stem domain, has sequence similarity to the other Class 2 mannosidases. Clearly the process that

100

6 a-Munnosiduses in Asparugine-linked Oligosuccharide Processing

has led to the sequence duplication and divergence between Golgi mannosidase 11, the lysosomal mannosidase, and the cytosolic/ER mannosidase I1 also led to the acquisition of additional NHz-terminal sequences that were unique to the Golgi enzyme. Evidence has been presented that the transmembrane domains and flanking sequences of these Golgi type I1 transmembrane processing activities may be involved in the retention of the enzymes in the Golgi complex [ 1431. cDNAs or genes encoding sequence homologs of the mammalian Golgi mannosidase I1 sequences have been identified in insect systems (Spodoptera [ 1441 and two independent Drosophilu sequences, one of which is published [ 1451) and C. elegans (2 independent sequences) (Figure 2). To this date only the cDNA from Spodoptera has been shown to possess activity toward pNP-mannoside [ 1441, but the natural substrate specificities of all of these additional gene products remain to be determined. A second cDNA and gene that are similar, but not identical, to Golgi mannosidase I1 have also been isolated from human [ 1401 and mouse sources. The corresponding enzyme has been designated Golgi mannosidase IIx [ 1401. Overexpression of the cDNA in COS cells resulted in a 2.5-fold increase in activity toward pNP-mannose, but further characterization of the oligosaccharide substrate specificity of the enzyme has not yet been published. The pattern of transcript expression for Golgi mannosidase I1 and Ilx in human tissues are overlapping, but there are significant differences in transcript abundance in several tissues. These differences in tissue-specific expression patterns are reminiscent of the differences in expression of the two Golgi processing mannosidases that act earlier in the pathway, Golgi mannosidases IA and IB. A deficiency in Golgi mannosidase I1 has been implicated in a human genetic disease characterized by ineffective erythropoiesis, bone marrow erythroid multinuclearity, and secondary tissue siderosis [ 691. The autosomal recessive disorder, termed congenetal dyserythropoietic anemia type I1 or HEMPAS, results from an incomplete processing of Asn-linked glycoproteins, including two erythrocyte membrane glycoproteins, called bands 3 and 4.5. These two glycoproteins are normally processed to complex type oligosaccharides with extended polylactosamine structures. In HEMPAS, the oligosaccharides on these glycoproteins are processed to hybrid type structures and the polylactosamine structures that are normally found on bands 3 and 4.5 are found on glycolipids instead. Patients with this disease suffer from life-long anemia, although recent literature suggests that the disease is often asymptomatic and may be more common than previously expected [ 1461. The genetic basis of the disease is complex. Biochemical analysis of the oligosaccharides accumulated on erythrocytes from patients indicates an enzyme defect in either the GlcNAc transferase I1 or the Golgi mannosidase I1 step [69]. Consistent with this hypothesis, two patients were found to have defects at the Golgi mannosidase I1 locus [68]. In contrast, other patients from Italy appear not to show linkage to GlcNAc transferase 11, Golgi mannosidase 11, or Golgi mannosidase IIx, but, instead, show a genetic linkage to a locus on chromosome 20q11.2 [ 147, 1481. A mouse model for HEMPAS disease was created by the inactivation of the Golgi mannosidase I1 gene and many of the characteristics of the human disease were reproduced [70]. The mice show anemia, splenomegaly, and immature erythrocytes or reticulocytes in the peripheral blood. Lectin blots and cell sorting

6.3 Munrzosidases in Glycoprotein Processing and Cutaholisnz

101

indicated an absence of complex type structures on erythrocytes in the Golgi mannosidase I1 null mice. In contrast, oligosaccharide structural analysis in splenocytes or fibroblasts and enzyme assays from a variety of mouse tissues indicated the presence of an alternate processing activity that could bypass the Golgi mannosidase 11 step, but with lowered efficiency in most tissues. This activity was termed a-mannosidase I11 [70], but it remains to be resolved whether this bypass activity represents Golgi mannosidase IIx or an alternate activity that was detected in the tissue extracts that directly cleaves MansGlcNAc2 to Man3GlcNAc2 without the prior activity of GlcNAc transferase I. This latter activity was found to be stimulated by C O + and ~ weakly inhibited by Sw and had many of the catalytic features of an enzyme previously characterized from BHK cells and rat liver [ 14915 I]. The latter enzyme has been purified from rat liver microsomes and can cleave Man9-4GlcNAc2 to Man3GlcNAc2, is weakly inhibited by Sw, and is stimulated by Cof2 [150]. The cDNA or gene encoding this enzyme has not yet been isolated so comparison of its sequence with the other Class 2 mannosidases has not been performed. Future studies on the gene disruption of Golgi mannosidase 1Ix and the proposed a-mannosidase 111 activity that acts directly on MansGlcNAc2 should help to resolve the contributions of each of these enzymes in glycoprotein processing in mammalian tissues.

Lysosomal mannosidase Catabolism of mannose terminal oligosaccharide structures occurs predominately in lysosomes, with the additional contribution of cytosolic mannose trimming of oligosaccharides released from glycoproteins and glycolipids in the ER [SO]. Oligosaccharides from glycoproteins that enter the lysosomal compartment through endocytosis are cleaved from the non-reducing terminus by exo-mannosidases and from the reducing terminus by a glycosylasparaginase and, in some species, by a chitobiase, to result in the complete hydrolysis of oligosaccharide structures to monosaccharides [ 1521. Cytosolic oligosaccharides are cleaved by a distinct cytosolically-localized chitobiase [ 1531 and a cytosolic mannosidase [ 1541 (see below) to a distinct isomer of Man5GlcNAc before transport across the lysosomal membrane [48] for further degradation into monosaccharides. The cleavage of the a-linked mannose residues in lysosomes is largely accomplished by a broad specificity lysosoma1 mannosidase that is capable of cleaving crl,2/a I ,3/al ,6-mannose linkages [1551. The substrate specificity and order of mannose removal has been well described for the broad specificity enzyme and in most instances unique isomers are produced in the cleavage series [6]. The enzyme is capable of readily cleaving 7 of the 8 a-linked mannose residues on MangGlcNAc2, but cleaves the last a1,6mannose with relatively low efficiency. The lysosomal mannosidase has been purified from a number of species [ 130, 156-1601 and the subunit molecular weights vary significantly between species. The human enzyme can be separated into two forms by ion exchange chromatography, termed A and B, but significant evidence indicates that the two enzymes are differently modified versions of the same gene product [156]. Pulse-chase studies have indicated that the initial translation product of the human gene is a 110 kDa pre-

102

6 ti-Mannosidases in Asparagine-linked Oligosaccharide Processing

cursor that is converted to subunits of 62, 58, and 26 kDa by proteolysis, presumably in the lysosome [ 1611. Fragments of different sizes have also been isolated from human tissues or cells indicating significant heterogeneity in processing of the initial precursor polypeptide [ 1601. The native apparent molecular mass of the lysosomal mannosidase A is 280 kDa and mannosidase B is 260 kDa [ 1561. The cDNAs and genes encoding the broad specificity mannosidase have been cloned from a variety of mammalian species including human [160, 162, 1631, mouse [164, 1651, cat [158], and cow [159], and the genes encoding slime mold (Dictyosteliurn discoideurn) [ 1661, protozoan parasite (Trypanosoma cvuzi) [ 1671, plant (Arabidopsis thalaina), and nematode (C. elegans) enzymes have also been identified. The latter two genes were identified in recent genome sequencing projects and functional demonstration of the encoded protein has not yet been performed. The D. discoideum enzyme was the first of the lysosomal mannosidases to be cloned and the developmental expression of the protein has been extensively studied [ 168, 1691. All of the genes encode a protein of similar size (-900-1000 amino acids) and those enzymes that have been studied have a similar broad specificity toward oligosaccharide substrates. The genomic organization of the nematode, plant, and mammalian genes are quite complex containing 19-28 introns [159, 163, 1651. Many of the intron positions are conserved among the diverse set of organisms with an extensive similarity in intron positions among the mammalian species. The complex gene organization in the higher organisms is in contrast with the single intron found in the Dictyostelium a-mannosidase gene [166] and the absence of introns in the T. cruzi gene [ 1671. In some species a second lysosomal a-mannosidase activity is expressed that is capable of cleaving the final al,6-mannose residue linked to the p-mannose core [6, 170, 1711, but this activity will not cleave other a-mannosyl residues on MangGlcNAcz-1. Cleavage of the core a1,6-mannose by this selective a1,6mannosidase appears to require the prior activity of the glycosylasparaginase and chitobiase since the enzyme strongly prefers oligosaccharides with a single GlcNAc at the reducing terminus [172]. Similar to other Class 2 mannosidases, including the broad specificity lysosomal mannosidase and Golgi mannosidase 11, the ul,6mannosidase is inhibited by Sw and DIM, but in contrast the enzyme acts poorly toward aryl-a-mannoside substrates [6]. The detection of the a1,6-mannosidase activity in a restricted number of mammalian species shows a surprising concordance with the species that express the lysosomal chitobiase [6, 1731. The chitobiase gene has been shown to be present both in species that express chitobiase activity, including humans, rodents, rabbits, chickens and slime mold, as well as in species that do not express detectable activity, including cats and cattle. In the latter species it appears that translation is suppressed [ 1731. Similarly, ul,6-mannosidase activity was detected in tissues or cell lines from rats and humans, but not cats or cattle [6]. It has been hypothesized that the cooperative specificity of the lysosomal a1.6mannosidase activity after chitobiase cleavage of the substrate has provided an additional route for oligosaccharide catabolism in some species, but that these enzymes are not required for glycoprotein catabolism. In this manner the enzymes are proposed to act as “accessory glycosidases” to aid in the efficient cleavage of the oligosaccharide core. For a detailed discussion of the relationship between the

6.3 Munnosickises in GIjwprotein Processing and Catabolism

103

al,6-mannosidase and the chitobiase see the review in [6]. Confirmation that the al,6-mannosidase is a Class 2 enzyme and comparison of the mannosidase and chitobiase coordinate expression will await the cloning of the cDNA or gene encoding the al,6-mannosidase. A human lysosomal storage disease characterized by a defect in the broad specificity lysosomal mannosidase. a-mannosidosis, results in varied clinical features including hepatomegaly, recurrent infections, skeletal dysplasia, deafness, severe mental deterioration, facial coarsening, eye opacities, and commonly death before ten years of age (174). A milder form of the disease has been described in some patients where hearing loss and mental retardation is evident from early childhood, but the onset of the other clinical features may be varied or absent. Analysis of residual enzyme activities in cells from patients with severe and milder forms of the disease indicated that the clinical severity was directly related to the degree of enzyme impairment [ 1751. Patients accumulate linear and branched oligosaccharides consistent with the catabolic pathways described in Figure 2 [176]. The oligosaccharides are predominately linear Man4-l GlcNAcl structures consistent with catabolic intermediates derived from cytosolic oligosaccharide catabolism and cleavage by the lysosomal al,6-mannosidase. Larger branched oligosaccharides are also found, but these oligosaccharides represent a lesser percentage of the total accumulated structures. The genetic defect in patients with a-mannosidosis generally do not appear to involve the alterations in the transcription or translation of the mannosidase polypeptide [162]. but recent data indicate that a human mutations can include splicing, missense, or nonsense mutations, as well as small insertions or deletions [177-1791. Animal models for the enzyme deficiency have also been identified in Persian cats [ 1581 and in Angus and Galloway cattle [ 1591. The defect in the feline disease results from a 4 bp deletion resulting in a frame shift at codon 583 and a premature termination at codon 645 [158]. Each of the bovine mutations involve different missense mutations causing amino acid substitutions that influence the either the level of protein expression or enzyme stability [ 1591. Epididymal/sperm mannosidase a-Mannosidase activities have been identified in rat, mouse, and pig sperm or epididymal tissues that have characteristics of Class 2 mannosidases [ 180-1841. The rat sperm a-mannosidase has activity toward al,2-, a1,3- and a1,6-mannose linkages and can cleave Man9-4GlcNAc2 to Manc3GlcNAc2 [ 1801. The enzyme could also act on GlcNAcMan5GlcNAc2, releasing at least one mannose residue. The sperm enzyme does not cross-react with the antibody to the ER/cytosolic a-mannosidase or lysosomal mannosidases and evidence has been published that the enzyme is first expressed as a plasma membrane protein in testicular germ cells and subsequently on the external face of the plasma membrane in mature sperm where it has been hypothesized to take part in sperm-egg interactions [180, 1841. An enzyme with similar biochemical characteristics has been purified from porcine epididymal fluid and the cDNA encoding the latter enzyme has been cloned [185]. The pig enzyme was found to be synthesized and secreted by the principal cells between the distal

104

6 a-Mannosidases in Asparagine-linked Oligosaccharide Processing

caput and proximal corpus of the epididymus with low expression in other parts of the epididymus and testicular tissue [186]. In contrast, a sequence homolog from mouse tissues was found to be expressed primarily in type A spermatogonia (stages IX-XI) and transcripts continue to be detectable through the formation of type B spermatogonia (stage VI) [ 1861. The relationship between the secreted porcine epididymal mannosidase and the plasma membrane associated rat sperm mannosidase remains to be established. Comparison of the polypeptide sequence encoded by the pig and mouse cDNAs indicate that these enzymes are Class 2 mannosidases, but significant distinctions from the other Class 2 enzymes result in their placement as a separate group distinct from the lysosomal, Golgi, or cytosolic mannosidases. The purified pig enzyme could cleave pNP-mannose in contrast to the inability of the rat sperm plasma membrane enzyme to cleave this substrate [185]. In contrast both enzymes could cleave oligosaccharide substrates [ 180, 183, 185, 1861. The effects of alkaloid inhibitors such as Sw, DIM, DMJ, and KIF have not been reported for either of the enzymes.

Heterogeneous cluster of mannosidase homologs among eukarya, eubacteria, and archaea The final subgroup of the Class 2 mannosidases are the most complex and unique in that they have sequence homologs among eubacteria and archaebacteria in addition to the eukaryotic family members. The first representative member of this subgroup that was identified was the vacuolar mannosidase from Saccharomyces cerevisiue [187,1881. This enzyme is localized to the internal face of the vacuolar membrane where it was originally proposed to play a role in glycoconjugate catabolism in the fungal vacuole. Suggestive evidence in support of this role was indicated by the reduction in the rate of turnover of metabolically labeled glycoconjugates in a yeast strain disrupted in the vacuolar mannosidase gene [lSS]. No other phenotype was detected in the disrupted strain. Subsequent sequencing of the yeast gene indicated that the encoded polypeptide contains no transmembrane domain or NH2-terminal signal sequence for entrance into the secretory pathway at the ER membrane [187]. Pulse-chase studies using the wild type mannosidase gene or fusions with genes encoding other reporter proteins resulted in the targeting of the protein into the vacuole in a manner that apparently involves entry into the secretory pathway within the Golgi complex or in some compartment downstream from the ER [189]. Yeast transpprt mutants (see mutants) that block transport within the ER or between the ER and Golgi had no effect on vacuolar targeting while some mutants that interrupt transport within the Golgi blocked delivery of the enzyme to vacuoles. None of the N-glycosylation sites on the enzyme or sites on a mannosidase-invertase fusion were glycosylated despite appropriate targeting of the proteins to the vacuole. These data indicate that the yeast a-mannosidase was delivered to the vacuole in a novel pathway separate from the secretory pathway. The gene encoding a putative vacuolar a-mannosidase activity was also isolated from the filamentous fungi, Aspergillus nidulans (now Emericella nidulans) [190]. The translation of this coding region has a high degree of sequence similarity to the S. cerevisiae vacuolar mannosidase (Figure 2) and gene disruption also had no discernable phenotype. A mammalian mannosidase has also been purified and cloned that has a high

6.3 Munnosiduses in Glycoprotein Processing and Cutuholism

105

degree of sequence similarity to the yeast vacuolar enzyme [ 191, 1921. In contrast to the yeast enzyme, the mammalian enzyme is found predominately in the cytosol [ 1921. The mannosidase is a tetramer of 1 16 kDa subunits that can cleave pNPa-mannose and oligosaccharide substrates and is inhibited by Sw and DIM, but not DMJ [106]. The oligosaccharide substrate specificity is unusual in that it requires a single GlcNAc at the reducing terminus of the oligosaccharide substrate and selectively cleaves specific a1,2-, al,3-, and al,6-linkages to produce a unique MansGlcNAc2 isomer (see Figure 2) [55, 581. This specificity indicates that the enzyme plays a role in generating an intermediate digestion product in the catabolism of oligosaccharides in the cytosol before their delivery to lysosomes for completion of the degradation process. The cDNA encoding the enzyme has been isolated from a rat cDNA library and was found to encode a polypeptide of 1040 amino acids containing no transmembrane domain or NH2-terminal signal sequence [ 1911. Transfection of COS cells with an expression construct containing the rat cDNA resulted in a 400-fold overexpression of enzyme activity. Northern blots indicated that transcripts encoding the cytosolic mannosidase were ubiquitously expressed in adult rat tissues. Initial studies on the processing mannosidase activities in the ER in the early 1980s indicated the presence of an oligosaccharide processing enzyme that cleaved MangGlcNAc2 to MansGlcNAc2 [91-941. These studies eventually led to the identification of ER mannosidase I [ 105, 1061. In an early attempt to purify ER mannosidase I, an ER-associated a-mannosidase activity was identified that cleaved pNP-mannose and was relatively less sensitive to inhibition by DMJ than Golgi mannosidase IA [91-93]. Surprisingly, the characteristics of the ER-associated activity were strikingly similar to the cytosolic a-mannosidase. When the cDNA encoding the cytosolic mannosidase was cloned, over-expression of the enzyme in COS cells also resulted in an over-expression of the ER-associated a-mannosidase activity [ 19I]. Although the substrate specificity of the cytosolic a-mannosidase did not match the activity indicated for ER mannosidase I, the hypothesis was presented that the cDNA encoded an a-mannosidase that was translocated into the ER by an unknown process and that the enzyme associated with the ER membrane had an altered substrate specificity. Recent work now indicates that at least some of the original hypothesis was correct. Studies by the Spiro lab have demonstrated that the cytosolic a-mannosidase is immunologically-related to an a-mannosidase activity in the ER, termed ER mannosidase I1 [ 1061. Western Blots have demonstrated that the ER-associated activity is smaller (-82 kDa) than the cytosolic form (-105 kDa) and antibodies directed to the COOH-terminus of the cytosolic form reacted with the ER form while antibodies to the NH2-terminus did not. These data suggest that the ER form is derived from the cytosolic form by a translocation process that involves proteolytic cleavage. The cytosolic and ER forms of the enzyme share common characteristics of inhibition by DIM, DMJ, weak inhibition by Sw, and lack of inhibition by KIF [ 106, 1081. Both enzymes produce the MansGlcNAc isomer C from MangGlcNAc, but the cytosolic a-mannosidase appears to efficiently cleave the ManxGlcNAc structure further to MansGlcNAc, especially in the presence of the cation activator, Co2+.The ER enzyme appears to have lost the capacity for activation by Co'+ and is able to cleave past MansGlcNAcz to Man;iGlcNAcz only after prolonged incu-

106

6 a-Mannosiduses in Aspurugine-linked Oligosaccharide Processing

bation. Comparison of the ER mannosidase I1 enzyme activity with other oligosaccharide processing activities, including ER mannosidase I, endomannosidase, and glucosidase 11, in various cell lines indicated that the ER mannosidase I1 is present at significant levels in all of the cell types and that the ratio of ER mannosidase I1 to ER mannosidase I could vary as much as ten-fold [ 1071. These data suggest that the product of the cytosolic/ER mannosidase I1 gene may play a separate role in each of the two cellular compartments. The ER form may catalyze an alternate pathway for early mannose trimming while the cytosolic form plays a role in oligosaccharide catabolism. In addition to the two fungal mannosidases and the mammalian cytosolic/ER mannosidase I1 sequences, three other sequences with similarity to Class 2 mannosidases were identified in database searches. Two of these sequences were most similar to each other, but were more distantly related to the fungal vacuolar and mammalian cytosolic/ER mannosidase I1 sequences. One of the genes was from a unicellular cyanobacterium, Synechocystis strain PCC6803, a eubacterium, and the second was from a hyper-thermophilic archaebacterium, Pyrococcus horikoslzii OT3. The third sequence, from the eubacteria Mycobacterium tuberculosis, appeared to be more similar to the fungal vacuolar a-mannosidases, the mammalian cytosolic/ER mannosidase 11, and the cyanobacterium and archaebacterium sequence homologs than the other Class 2 mannosidases. The degree of sequence similarity of the Mycobacterium putative a-mannosidase sequence to all at the other Class 2 mannosidases was more remote and the coding region was missing -40 amino acids of the most highly homologous sequence of the Class 2 enzymes. These data call into question whether the gene encodes a protein with a-mannosidase activity. Other mannosidase enzyme activities have been identified in mammalian systems linkages. Noteworthy among that are capable of cleaving al,2-/a1,3-/a1,6-mannose these enzymes is a catalytic activity partially purified from rat brain microsomes that can cleave Mang-4GlcNAc2 to Man3GlcNAc2 [193]. The enzyme is not inhibited by Sw or dMNJ, and has marginal activity toward pNP-mannose. Surprisingly, this enzyme cross-reacts with an antibody to the rat cytosolic/ER mannosidase I1 indicating that the enzyme is likely to be a Class 2 mannosidase. The enzyme activity termed mannosidase I11 (see discussion above) also has hallmarks of a Class 2 mannosidase. This enzyme has been found in rat liver microsomes [149%151]and in tissues from mice that have a disruption in the Golgi mannosidase I1 gene locus [70]. cDNAs or genes encoding the rat brain enzyme or a-mannosidase I11 have not been isolated, so the relationship to other Class 2 enzymes remains to be resolved.

6.4 Conclusions and Future Prospects Several mannosidases have been identified that are involved in glycoprotein biosynthetic reactions and in glycoprotein catabolism. Comparisons of biochemical characteristics and sequence similarities have allowed the exo-mannosidases to be divided into two classes. Within each of the classes are subgroups that have been

References

107

defined based on greater similarities in sequences, unique substrate specificities, and subcellular localizations. Among the Class 1 mannosidases two subgroups have been identified that are involved in early glycoprotein trimming steps, yet their substrate specificities indicate that they have a distinct recognition of non-reducing terminal al,2-mannose residues on Man9 GlcNAq oligosaccharides. The structure determination of members of these two subgroups in the near future should provide a considerable insight into the molecular basis of their substrate recognition. A separate subgroup of Class 1 mannosidases is defined by the secreted enzymes from Aspergillus saitoi and Pennicilium citrinum. The roles of these enzymes in glycoprotein maturation or catabolism are unclear. The fourth subgroup of the Class l mannosidases is a novel subgroup that contains a region of sequence that is conserved among the Class 1 mannosidases and a second region extending from the COOH-terminus that has no obvious similarity to any known protein. Although members of this subclass are conserved from yeast to mammals, the function of the proteins remain obscure. Defining the role of this novel class of mannosidase-like proteins will be a major goal for the future. The Class 2 mannosidases have a diverse set of members, including mammalian enzymes that are involved on glycoprotein maturation in the Golgi complex as well as enzymes involved in glycoprotein catabolism in the cytosol and lysosomes. The roles of the Golgi and lysosomal subgroups of enzymes appear to be well defined, but the presence of additional activities in both compartments are indicated by the detection of an ctl,6-mannosidase in lysosomes and a-mannosidase I11 in the secretory pathway catalyzing a proposed alternate pathway for glycoprotein processing. The unusual features of dual localization of the mammalian cytosolic/ER mannosidase I1 and its fungal vacuolar homologs indicate that further study will be required to determine the mechanism of transport of these enzymes into their respective membrane bound compartments. In addition, the structural features of this broad class of enzymes that allow their discrimination in substrate specificity and localization should provide additional questions for investigation. The detection of genes encoding Class 2 mannosidase homologs in eubacteria and archeabacteria indicates that this subclass of mannosidases has a highly ancestral origin and could possibly represent the ancestral progenetor gene from which the other Class 2 mannosidases were derived by gene duplication and divergence. Confirmation of this hypothesis will require the identification and sequencing additional family members from other organisms and determination of their substrate specificities.

Acknowledgments

The work described from the author’s laboratory is supported by U.S. Public Health Service Grants GM-47533 and RR-05351 References I . Varki, A,, Biological Roles Of Oligosaccharides-All Of The Theories Are Correct. Glycohio/ogJ). 1993, 3, 97 - 130.

108

6 a-Munnosiduses in Asparugine-linked Oligosucchuvide Processing

2. Cumming, D. A., Glycosylation of recombinant protein therapeutics: control and functional implication, Glycohiology, 1991, 1, 115-130. 3. Brandley, B. K., S. J. Swiedler, and P. W. Robbins, Carbohydrate ligands of the LEC cell adhesion molecules, Cell, 1990, 63, 861-863. 4. Feizi, T., Demonstration by monoclonal antibodies that carbohydrate structures of glycoproteins and glycolipids are onco-developmental antigens, Nature, 1985, 314, 53-57. 5. Feizi, G., and R. A. Childs, Carbohydrate structures of glycoproteins and glycolipids as differentiation antigens, tumour-associated antigens and components of receptor systems, TIBS, 1985, 24-29. 6. Daniel, P. F., B. Winchester, and C. D. Warren, Mammalian a-mannosidases--multiple forms but a common purpose?, Glycohiology, 1994, 4, 551-566. 7. Moremen, K. W., R. B. Trimble, and A. Herscovics, Glycosidases of the asparagine-linked oligosaccharide processing pathway, Glycobiology, 1994, 4, 1 13-125. 8. Kornfeld, R., and S. Kornfeld, Assembly of asparagine-linked oligosaccharides, Ann. Rel;. Biochem., 1985, 54, 631 -664. 9. Neufeld, E. F., Lysosomal storage diseases, Ann. Rev. Biochem., 1991, 60, 257-280. 10. Cacan, R., and A. Verhert, Free and N-Linked Ohgomannosides as Markers of the Quality Control of Newly Synthesized Glycoproteins, Biochem. Biophys. Res. Commun., 1999,258, I 5. 11, Burda, P., and M. Aebi, The dolichol pathway of N-linked glycosylation, Biochim. Biophys. Acta, 1999, 1426, 239-257. 12. Hirschberg, C. B., and M. D. Snider, Topography of glycosylation in the rough endoplasmic reticulum and Golgi apparatus, Annu. Rev. Biochem., 1987, 56, 63-87. 13. Parodi, A. J., Reglucosylation of glycoproteins and quality control of glycoprotein folding in the endoplasmic reticulum of yeast cells, Biochim. Biophys. Acta, 1999, 1426, 287-295. 14. Parodi, A. J., N-glycosylation in trypanosomatid protozoa, Glycobiology, 1993, 3, 193-199. 15. Knauer, R., and L. Lehle, The oligosaccharyltransferase complex from yeast, Biochim. Biophys. Acta, 1999, 1426, 259-273. 16. Herscovics, A,, Glycosidases of the asparagine-linked oligosaccharide processing pathway, 1999, Pages 13-35, in Comprehensive Natural Products Chemistry, B. M. Pinto, ed, Elsevier, New York. 17. Herscovics, A., Processing glycosidases of Saccharomyces cerezjisiae, Biochim.Biophys. Actu, 1999, 1426, 275-285. 18. Ziegler, F. D., T. R. Gemmill, and R. B. Trimble, Glycoprotein synthesis in yeast. Early events in N-linked oligosaccharide processing in Seliiro.saccharomyn);cesponibe, J. Biol. Chem., 1994,269, 12527-12535. 19. Brockhausen, I.; J. Schutzbach, and W. Kuhns, Glycoproteins and their relationship to human disease, Actu Anat., 1998, 161, 36-78. 20. Varki, A., Factors controlling the glycosylation potential of the Golgi apparatus, Trends Cell Biol., 1998, 8, 34-40. 21. Lechner, J., and F. Wieland, Structure and biosynthesis of prokaryotic glycoproteins, Annu. Rev. Biochem., 1989, 58, 173-194. 22. Jakob, C. A,, P. Burda, J. Roth, and M. Aebi, Degradation of misfolded endoplasmic reticulum glycoproteins in Saccharomyce.~ceretlisiue is determined by a specific oligosaccharide structure, J. Cell Biol., 1998, 142, 1223-1233. 23. Quellhorst. G. J., Jr., J. L. O’Rear, R. Cacan, A. Verbert, and S. S. Krag, Nonglucosylated oligosaccharides are transferred to protein in MIS-5 Chinese hamster ovary cells, Glycobiology, 1999, 9, 65-72. 24. Romero, P. A,, and A. Herscovics, Transfer of nonglucosylated oligosaccharide from lipid to protein in a mammalian cell, J. Biol. Chem., 1986, 261, 15936- 15940. 25. Jaeken, J., G. Matthijs, R. Barone, and H. Carchon, Carbohydrate deficient glycoprotein (CDG) syndrome type I. J. Med. Genet., 1997, 34, 73-76. 26. Trombetta, E. S., and A. Helenius, Lectins as chaperones in glycoprotein folding, Curr. Opin. Struct. Biol., 1998, 8, 587-592. 27. Helenius, A., How N-linked oligosaccharides affect glycoprotein folding in the endoplasmic reticulum, Mol. Biol. Cell., 1994, 5, 253-265.

References

109

28. Williams, D. B., The Merck Frosst Award Lecture 1994/LA conference Merck Frosst 1994, Calnexin: a molecular chaperone with a taste for carbohydrate, Biochem. Cell B i d , 1995, 73, 123-132. 29. Hammond, C., and A. Helenius, Quality control in the secretory pathway, Curr. Upin. Cell B i d . 1995, 7, 523-529. 30. Sousa, M. C., M. A. Ferrero-Garcia. and A. J. Parodi, Recognition of the oligosaccharide and protein moieties of glycoproteins by the UDP-G1c:glycoprotein glucosyltransferase, Biochemistry, 1992, 31, 97-105. 31 Sousa, M. C., and A. J. Parodi, The interaction of the UDP-GLC:glycoprotein glucosyltransferase with the acceptor glycoprotein, Cell Mol. Biol., 1996, 42, 609- 6 16. 32 Wiertz, E. J., T. R. Jones, L. Sun, M. Bogyo, H. J. Geuze, and H. L. Ploegh, The human cytomegalovirus US1 1 gene product dislocates MHC class I heavy chains from the endoplasmic reticulum to the cytosol, Cell, 1996, 84, 769- 779. 33 Wiertz, E. J., D. Tortorella, M. Bogyo, J. Yu, W. Mothes, T. R. Jones, T. A. Rapoport, and H. L. Ploegh, Sec61-mediated transfer of a membrane protein from the endoplasmic reticulum to the proteasome for destruction, Nuture, 1996, 384, 432-438. 34 Hammond, C., I. Braakman, and A. Helenius, Role of N-linked oligosaccharide recognition, glucose trimming, and calnexin in glycoprotein folding and quality control, Proc. Nutl Accid. Sci. U. S. A , , 1994, 91, 913-917. 35 Knop, M., N. Hauser, and D. H. Wolf, N-Glycosylation affects endoplasmic reticulum degradation of a mutated derivative of carboxypeptidase yscY in yeast, Yeast, 1996, 12, 12291238. 36 Liu, Y., P. Choudhury, C. M. Cabral, and R. N. Sifers, lntracellular disposal of incompletely folded human alphal-antitrypsin involves release from calnexin and post-translational trimming of asparagine-linked oligosaccharides, J. Bid. Chem., 1997,272, 7946-795 1. 31 Liu, Y., P. Choudhury, C. M. Cabral, and R. N. Sifers, Oligosaccharide modification in the early secretory pathway directs the selection of a misfolded glycoprotein for degradation by the proteasome, J. Biol. Chem., 1999,274, 5861-5867. 38. Yang, M., S. Omura, J. S. Bonifacino, and A. M. Weissman, Novel aspects ofdegradation of T cell receptor subunits from the endoplasmic reticulum (ER) in T cells: importance of oligosaccharide processing, ubiquitination, and proteasome-dependent removal from ER membranes, J. E.xp. Med., 1998, 187, 835-846. 39. Herscovics, A,, and P. Orlean, Glycoprotein biosynthesis in yeast, FASEB J . , 1993, 7, 540550. 40. Camirand, A., A. Heysen, B. Grondin, and A. Herscovics, Glycoprotein biosynthesis in Sacckuromyces cereuisiae. Isolation and characterization of the gene encoding a specific processing a-mannosidase, J. Biol. Chem., 1991. 266, 15120-15127. 41. Puccia, R., B. Grondin, and A. Herscovics, Disruption of the processing a-mannosidase gene does not prevent outer chain synthesis in Succhuromyces cerevisiae, Biochem. J., 1993, 290, 21-26. 42. Arar, C., V. Carpentier, J. P. Le Caer, M. Monsigny, A. Legrand, and A. C. Roche, ERGIC53, a membrane protein of the endoplasmic reticulum-Golgi intermediate compartment, is identical to MR60, an intracellular mannose-specific lectin of myelomonocytic cells, J. Bid. Chem., 1995,270, 3551-3553. 43. Itin, C., A. C. Roche, M. Monsigny, and H. P. Hauri, ERGIC-53 is a functional mannoseselective and calcium-dependent human homologue of leguminous lectins: Mol. Biol. Cell, 1996, 7,483-493. 44. Vollenweider, F., F. Kappeler, C. Itin, and H. P. Hauri, Mistargeting of the lectin ERGIC-53 to the cndoplasmic reticulum of HeLa cells impairs the secretion of a lysosomal enzyme, J. Cell B i d , 1998, 142, 377-389. 45 Neerman-Arbez, M., K. M. Johnson, M. A . Morris, J. H. McVey. F. Peyvandi, W. C. Nichols, D. Ginsburg, C. Rossier, S. E. Antonarakis, and E. G. Tuddenham, Molecular analysis of the ERGIC-53 gene in 35 families with combined factor V-factor VIII deficiency, Blood, 1999, 93, 2253-2260. 46 Nichols. W. C.. V. H. Terry, M. A. Wheatlev, A. Y a w . A. Zivelin. N . Ciavarella. C. Stefanile, T. Matsushita, H. Saito, N. B. de Bosch,A . Ruiz-Saez, A. Torres, A. R. Thompson, D. I.

1 10

6 a-Munnosiduses in Aspuragine-linked Oligosucchuride Processing

Feinstein, G. C. White, C. Negrier, C. Vinciguerra, M. Aktan, R. J. Kaufman, D. Ginsburg, and U. Seligsohn, ERGIC-53 gene structure and mutation analysis in 19 combined factors V and VIII deficiency families, Blood. 1999, 93, 2261-2266. 47. Saint-Pol, A,, C. Banvy, P. Codogno, and S. E. Moore, Transfer of free polymannose-type oligosaccharides from the cytosol to lysosomes in cultured human hepatocellular carcinoma HepG2 cells, J. Cell Biol., 1997, 136, 45-59. 48. Saint-Pol, A,, P. Codogno, and S. E. Moore, Cytosol-to-lysosome transport of free polymannose-type oligosaccharides. Kinetic and specificity studies using rat liver lysosomes, J. Bid. Chem., 1999,274, 13547-13555. 49. Anumula, K., and R. Spiro, Release of glucose-containing polymannose oligosaccharides during glycoprotein biosynthesis. Studies with thyroid microsomal enzymes and slices., J. Biol. Chem., 1983,258, 15274-15282. SO. Cacan, R., S. Duvet, D. Kmiecik, 0. Labiau, A. M. Mir, and A. Verbert, ‘Glyco-deglyco’ processes during the synthesis of N-glycoproteins, Biochimie, 1998, 80, 59-68. 51. Moore, S. E., Transport of free polymannose-type oligosaccharides from the endoplasmic reticulum into the cytosol is inhibited by mannosides and requires a thapsigargin-sensitive calcium store, Glycohioloyy, 1998, 8 , 373-381. 52. Hughes, E. A,, C. Hammond, and P. Cresswell, Misfolded major histocompatibility complex class I heavy chains are translocated into the cytoplasm and degraded by the proteasome, Proc. Nut1 Acud. Sci. U. S. A . , 1997, Y4, 1896-1901. 53. Suzuki, T., K. Kitajima, S. Inoue, and Y. Inoue, N-glycosylation/deglycosylation as a mechanism for the post-translational modification/remodification of proteins, Glycoconj. J., 1995, 12, 183-193. 54. Suzuki, T., H. Park, K. Kitajima, and W. J. Lennarz, Peptides glycosylated in the endoplasmic reticulum of yeast are subsequently deglycosylated by a soluble peptide: N-glycanase activity, J. Bid. Chem., 1998, 273, 21526-21530. 5 5 . Duvet, S., 0. Labiau, A. M. Mir, D. Kmiecik, S. S. Krag, A. Verbert, and R. Cacan, Cytosolic deglycosylation process of newly synthesized glycoproteins generates oligomannosides possessing one GlcNAc residue at the reducing end, Biochem. J., 1998, 335, 389--396. 56. Pierce, R. J., G. Spik, and J. Montreuil, Cytosolic location of an endo-N-acetyl-b-D-glucosaminidase activity in rat liver and kidney, Biochem. J., 1979, 180, 673-676. 57. Pierce, R. J., G. Spik, and J. Montreuil, Demonstration and cytosolic location of an endo-Nacetyl-0-D-glucosaminidase activity towards an asialo-N-acetyl-lactosaminic-type substrate in rat liver, Biochem. J . , 1980, 185, 261-264. 58. Grard, T., V. Herman, A. Saint-Pol, D. Kmiecik, 0. Labiau, A. M. Mir, C. Alonso, A. Verbert, R. Cacan, and J. C. Michalski, Oligomannosides or oligosaccharide-lipids as potential substrates for rat liver cytosolic a-D-mannosidase, Biochem. J., 1996, 316, 787-792. 59. Marth, J., Will the transgenic mouse serve as a Rosetta Stone to glycoconjugate function?, Glycoconj. J . , 1994, 11, 3-8. 60. Hennet, T., D. Chui, J. C. Paulson, and J. D. Marth, Immune regulation by the ST6Gal sialyltransferase, Proc. Nut1 Acud. Sci. U. S. A , , 1998, YS, 4504-4509. 61. Maly) P., A. Thall, B. Petryniak, C. E. Rogers, P. L. Smith, R. M. Marks, R. J. Kelly, K. M. Gersten, G. Cheng, T. L. Saunders, S. A. Camper, R. T. Camphausen, F. X. Sullivan, Y. Isogai, 0. Hindsgaul, U. H. von Andrian, and J. B. Lowe, The a(l,3)fucosyltransferase FucTVII controls leukocyte traflicking through an essential role in L-, E-, and P-selectin ligand biosynthesis, Cell, 1996, 86, 643-653. 62. Ellies, L. G., S. Tsuboi, B. Petryniak, J. B. Lowe, M. Fukuda, and J. D. Marth, Core 2 oligosaccharide biosynthesis distinguishes between selectin ligands essential for leukocyte homing and inflammation, Immunity, 1998, 9, 881-890. 63. Metzler, M., A. Gertz, M. Sarkar, H. Schachter, J. Schrader, and J. Marth, Complex asparagine-linked oligosaccharides are required for morphogenic events during post-implantation development, EMBO J., 1994, 13, 2056-2065. 64. Campbell, R., M. Metzler, M. Granovsky, J. Dennis, and J. Marth, Complex asparaginelinked oligosaccharides in Mgatl-null embryos., Glycobiology, 1995, 5, 535-543. 65. Jaeken, J., H. Carchon, and H. Stibler, The carbohydrate-deficient glycoprotein syndromes: pre-Golgi and Golgi disorders?, Glycobiology, 1993, 3, 423-438.

References

11 1

66. Burda, P., L. Borsig, J. de Rijk-van Andel, R. Wevers, J. Jaeken, H. Carchon, E. G. Berger, and M. Aebi, A novel carbohydrate-deficient glycoprotein syndrome characterized by a deficiency in glucosylation of the dolichol-linked oligosaccharide, J. Clin. Invest., 1998, 102, 647652. 67. Jaeken, J., H. Schachter, H. Carchon, P. De Cock, B. Coddeville, and G. Spik, Carbohydrate deficient glycoprotein syndrome type 11: a deficiency in Golgi localised N-acetyl-glucosaminyltransferase 11, Arch. Dis. Child., 1994, 71, 123~-127. 68. Fukuda, M. N., K. A. Masri, A. Dell, L. Luzzatto, and K. W. Moremen, Incomplete synthesis of N-glycans in congenital dyserythropoietic anemia type I1 caused by a defect in the gene encoding a-mannosidase 11, Proc. Nut1 Acud. Sci. U. S. A , , 1990, 87, 7443-7447. 69. Fukuda, M. N., HEMPAS disease: genetic defect of glycosylation, Glycobioloqy, 1990,1,9-15. 70. Chui, D., M. Oh-Eda, Y. F. Liao, K. Panneerselvam, A. Lal, K. W. Marek, H. H. Freeze, K. W. Moremen, M. N. Fukuda, and J. D. Marth, a-Mannosidase-I1 deficiency results in dyserythropoiesis and unveils an alternate pathway in oligosaccharide biosynthesis, Cell, 1997, 90, 157- 167. 71. Lubas, W., and R. Spiro, Golgi endo-a-D-mannosidase from rat liver, a novel N-linked carbohydrate unit processing enzyme, J. Biol. Cliem., 1987,262, 3775- 378 1. 72. Lubas, W., and R. Spiro, Evaluation of the role of rat liver Golgi endo-a-D-mannosidase in processing N-linked oligosaccharides, J. Bid. Chem., 1988,263, 3990-3998. survey of endomannosidase indicates late evolution73. Dairaku. K.. and R. SDiro. Phvlogenetic . ary appearance of this N-linked oligosaccharide processing enzyme, Glycohioloqy, 1997, 7, 579-586. 74. Spiro, R., Q. Zhu, V. Bhoyroo, and H. Soling, Definition of the lectin-like properties of the molecular chaperone, calreticulin, and demonstration of its copurification with endomannosidase from rat liver Golgi, J. Bid. Chem., 1996, 271, 11588-1 1594. 75. Lal, A,, P. Pang, S. Kalelkar, P. A. Romero, A. Herscovics, and K. W. Moremen, Substrate specificities of recombinant murine Golgi al,2-mannosidases IA and IB and comparison with endoplasmic reticulum and Golgi processing al,2-mannosidases, Glycohioloqy, 1998, 8, 98 1995. 76. Lipari, F., B. J. Gour-Salin, and A. Herscovics, The Succhuromyces cereuisiue processing u1,2mannosidase is an inverting glycosidase, Biochern. Biophys. Res. Commun., 1995, 209, 322326. 77. Howard, S., S. He, and S. G. Withers, Identification of the active site nucleophile in jack bean a-mannosidase using 5-fluoro-P-L-gulosyl fluoride, J. Bid. Cliem., 1998, 273, 2067-72. 78. Howard, S., C. Braun, J. McCarter, K. W. Moremen, Y. F. Liao, and S. G. Withers, Human lysosomal and jack bean a-mannosidases are retaining glycosidases, Biochem. Biophys. Rex Commun., 1997,238, 896-898. 79. Henrissat, B., and A. Bairoch, Updating the sequence-based classification of glycosyl hydrolases, Biochem. J., 1996, 316, 695-696. 80. Henrissat, B., and G. Davies, Structural and sequence-based classification of glycoside hydrolases, Cum Opin. Struct. Biol., 1997, 7, 637-644. 8 1. Henrissat, B., Glycosidase families, Biochem. Soc. Trans., 1998, 26, 153-1 56. 82. Byrd, J. C., A. L. Tarentino, F. Maley, P. H. Atkinson, and R. B. Trimble, Glycoprotein synthesis in yeast. Identification of Man8GlcNAc2 as an essential intermediate in oligosaccharide processing, J. Biol. Chem.. 1982.257, 14657714666, 83. Jelinek-Kelly, S., T. Akiyama, B. Saunier. J. S. Tkacz, and A. Herscovics, Characterization of a specific a-mannosidase involved in oligosaccharide processing in Succhuromyces cereuisiue, J. Bid. Cliem., 1985, 260, 2253-2257. 84. Jelinek-Kelly, S., and A. Herscovics, Glycoprotein biosynthesis in Succhuromyces crrevisiue. Purification of the a-mannosidase which removes one specific mannose residue from Man9GlcNAc, J. Biol. Chem., 1988,263, 14757-14763. 85. Ziegler, F. D., and R. B. Trimble, Glycoprotein biosynthesis in yeast: purification and characterization of the endoplasmic reticulum Mans processing a-mannosidase, Glycobioloqy, 1991, 1, 605-614. 86. Lipari, F., and A. Herscovics, Production, purification and characterization of recombinant yeast processing al,2-mannosidase, Glycohioloqy, 1994, 4, 697-702.

1 12

6 a-Mannosidases in Asparagine-linked Oligosaccharide Processing

87. Lipari, F., and A. Herscovics, Calcium binding to the class I a-1,2-mannosidase from Succhuromyces cerevisiue occurs outside the EF hand motif, Biochemistry, 1999, 38, 1 1 11-1118. 88. Grondin, B., and A. Herscovics, Topology of ER processing a-mannosidase of Sacclzuromyces cerevisiae, Glycohioloyy, 1992, 2, 369-312. 89. Burke, J., F. Lipari, S. Igdoura, and A. Herscovics, The Succhuromyces cerehiae processing al,2-mannosidase is localized in the endoplasmic reticulum, independently of known retrieval motifs, Eur. J. Cell Bid.. 1996, 70, 298-305. 90. Dole, K., F. Lipari, A. Herscovics, and P. L. Howell, Crystallization and preliminary X-ray analysis of the class 1 a I ,2-mannosidase from Succhuromyces cereoisiae, J. Struct. Biol., 1997, 120, 69-72. 91. Bischoff, J., and R. Kornfeld, Evidence for an a-mannosidase in endoplasmic reticulum of rat liver, J. Bid. Chem., 1983, 258, 7907-7910. 92. Bischoff, J., and R. Kornfeld, The effect of I-deoxymannojirimycin on rat liver amannosidases, Biocliem. Biophys. Rex Commun., 1984, 125, 324-33 1. 93. Bischoff, J., L. Liscum, and R. Kornfeld, The use of 1-deoxymannojirimycin to evaluate the role of various u-mannosidases in oligosaccharide processing in intact cells, J. Biol. Chem., 1986,261,4766-4774. 94. Godelaine. D., M. J. Spiro, and R. G. Spiro. Processing of the carbohydrate units of thyroglobulin, J. Bid. Chem., 1981,256, 10161-10168. 95. Hickman, S.. J. L. Theodorakis, J. M. Greco, and P. H. Brown, Processing of MOPC 315 immunoglobulin A oligosaccharides: evidence for endoplasmic reticulum and trans Golgi al,2-mannosidase activity, J. Cell Biol. 1984, 98, 407-416. 96. Atkinson, P. H., and J. T. Lee, Co-translational excision of a-glucose and a-mannose in nascent vesicular stomatitis virus G protein, J. Cell Bid., 1984, 98, 2245-2249. 97. Rosenfeld, M. G., E. E. Marcantonio, J. Hakimi, V. M. Ort, P. H. Atkinson, D. Sabatini, and G. Kreibich, Biosynthesis and processing of ribophorins in the endoplasmic reticulum, J. Cell Biol., 1984, 99, 1076-1082. 98. Rizzolo, L. J., and R. Kornfeld, Post-translational protein modification in the endoplasmic reticulum. Demonstration of fatty acylase and deoxymannojirimycin-sensitive a-mannosidase activities, J. Biol. Chem., 1988, 263, 9520-9525. 99. Brown, P. H., and S. Hickman, Oligosaccharide processing at individual glycosylation sites on MOPC 104E immunoglobulin M. Differences in al,2-linked mannose processing, J. Bid. Chern., 1986,261, 2575--2582. 100. Moore, S., and R. Spiro, Demonstration that Golgi endo-a-D-mannosidase provides a glucosidase-independent pathway for the formation of complex N-linked oligosaccharides of glycoproteins, J. Biol. Chem., 1990, 265, 13104-13112. 101. Moore, S., and R. Spiro, Characterization of the endomannosidase pathway for the processing of N-linked oligosaccharides in glucosidase 11-deficient and parent mouse lymphoma cells, J. Biol. Chern., 1992, 267, 8443-8451. 102. Hiraizumi, S., U. Spohr, and R. Spiro, Characterization of endomannosidase inhibitors and evaluation of their effect on N-linked oligosaccharide processing during glycoprotein biosynthesis, J. Biol. Chem., 1993, 268, 9921-9935. 103. Spiro, M.; V. Bhoyroo, and R. Spiro, Molecular cloning and expression of rat liver endo-amannosidase, an N-linked oligosaccharide processing enzyme, J. Biol. Chem., 1997, 272, 29356-29363. 104. Karaivanova, V., P. Luan, and R. Spiro, Processing of viral envelope glycoprotein by the endomannosidase pathway: evaluation of host cell specificity, Glycobiology, 1998, 8, 725-730. 105. Gonzalez, D. S., K. Karaveg, A. S. Vandersall-Nairn, A. Lal, and K. W. Moremen, Identification, expression, and characterization of a cDNA encoding human ER mannosidase I, the enzyme that catalyzes the first mannose trimming step in mammalian Asn-linked oligosaccharide biosynthesis, J. Biol. Chem., 1999, 274, 21375-21386. 106. Weng, S., and R. Spiro, Endoplasmic reticulum kifunensine-resistant a-mannosidase is enzymatically and immunologically related to the cytosolic u-mannosidase, Arch. Biochem. Biophys., 1996,325, 113-123. ~

References

1 13

107. Weng, S . , and R. Spiro, Demonstration that a kifunensine-resistant a-mannosidase with a unique processing action on N-linked oligosaccharides occurs in rat liver endoplasmic reticulum and various cultured cells, J. B i d . Chenz., 1993, 268, 25656-25663. 108. Weng, S., and R. Spiro, Evaluation of the early processing routes of N-linked oligosaccharides of glycoproteins through the characterization of MangGlcNAcz isomers: evidence that endomannosidase functions in vivo in the absence of a glucosidase blockade, Glycohioloyy, 1996, 6, 861-868. 109. Bause. E., E. Bieberich, A. Rolfs, C. Volker, and B. Schmidt, Molecular cloning and primary structure of Man9-mannosidase from human kidney, Eur. J. Biochem., 1993, 21 7, 535-540. 110. Lal, A., J. S. Schutzbach, W. T. Forsee, P. J . Neame, and K. W. Moremen, Isolation and expression of murine and rabbit cDNAs encoding an al,2-mannosidase involved in the processing of asparagine-linked oligosaccharides, J. Bid. Chem., 1994,269, 9872-9881. 1 1 1 . Bieberich, E., K. Treml, C. Volker, A. Rolfs, B. Kalz-Fuller, and E. Bause, Man9mannosidase from pig liver i s a type-I1 mcmbrane protein that resides in the endoplasmic reticulum. cDNA cloning and expression of the enzyme in COS I cells, Eur. J. Biochem., 1997, 246, 68 1-689. 112. Valee, F., A. Lal, K. W. Moremen, and P. L. Howell, Purification, crystallization and preliminary X-ray crystallographic analysis of recombinant murine Golgi mannosidase IA, a class 1 a-mannosidase involved in Asn-linked oligosaccharide maturation, Actu Cryst., 1999, 0 5 5 , 571-573. 1 1 3. Roth, J., D. Brada, P. M. Lackie, J. Schweden, and E. Bause, Oligosaccharide trimming Man9-mannosidase i s a resident ER protein and exhibits a more restricted and local distribution than glucosidase 11, Eur. J . Cell Biol.. 1990, 53, 131 14 1. 114. Bausc, E., W. Breuer, J. Schweden, R. Roeser, and R. Geyer, Erect of substrate structure on the activity of Man9-mannosidase from pig liver involved in N-linked oligosaccharide processing, Eur. J. Biochem., 1992,208, 451-457. 115. Velasco, A,, L. Hendricks, K. W. Moremen. D. R. P. Tulsiani, 0. Touster, and M. G. Farquhar; Cell type-dependent variations in the subcellular distribution of a-mannosidase I and 11, J. Cell Biol., 1993, 122, 39-51. 116. Tulsiani, D. R., S. C. Hubbard, P. W. Robbins, and 0. Touster, a-D-Mannosidases of rat liver Golgi membranes. Mannosidase I1 is the GlcNAcMANs-cleaving enzyme in glycoprotein biosynthesis and mannosidases Ia and IB are the enzymes converting Man9 precursors to Mans intermediates, J. B i d . C/zem., 1982, 257, 3660-3668. 117. Tulsiani, D. R., and 0. Touster, The purification and characterization of mannosidase IA from rat liver Golgi membranes, J. Bid. Chem., 1988, 263, 5408-5417. 1 18. Herscovics, A,, J. Schneikert, A. Athanassiadis, and K. W. Moremen, Isolation of a mouse Golgi mannosidase cDNA, a member of a gene family conserved from yeast to mammals, J. B i d . Chew?.,1994, 269, 9864-9871. 119. Tremblay, L. O., N. Campbell Dyke. and A. Herscovics, Molecular cloning, chromosomal mapping and tissue-specific expression of a novel human ul,2-mannosidase gene involved in N-glycan maturation, Glycohioloyy, 1998. A', 585 -595. 120. Igdoural, S. A,, A. Herscovics. A. Lal. K. W. Moremen, C. R. Morales, and L. Hermo, aMannosidases involved in N-glycan processing show cell specificity and distinct subcompartmentalization within the Golgi apparatus of cells in the testis and epididymus. Euv. J. Cell B i d , 1999, in press. 121. Ichishima, E., M. Arai, Y. Shigematsu, H. Kuinagai, and R. Sumida-Tanaka, Purification of an acidic a-D-mannosidase from Asprryillus suitoi and specific cleavage of 1;2-a-Dmannosidic linkage in yeast mannan, Biocliirn. Biophys. Actu, 1981, 658, 45-53. 122. Yamashita, K., E. Ichishima, M. Arai, and A. Kobata, An alpha-mannosidase purified from Aspergillus saitoi i s specific for al,2 linkages, Biocheni. Bioph-vs. Res. Comrnun., 1980, 96: 1335 1342. 123. Inoue, T., T. Yoshida, and E. Ichishima, Molecular cloning and nucleotide sequence of the l u ~ and expression of the gene in yeast 1.2-a-D-mannosidase gene, msdS, from A ~ p r r ~ j i l .suitoi cells, Biochini. Biophys. Actci, 1995, 1253, 141- 5. 124. Ichishima, E., N. Taya, M. Ikeguchi, Y. Chiba, M. Nakamura, C. Kawabata, T. Inoue, K. Takahashi, T. Minetoki, K. Ozeki, C. Kumagai. K. Gomi, T. Yoshida, and T. Nakajima, -

114

6 a-Mannosiduses in Asparagine-linked Oligosaccharide Processing

Molecular and enzymic properties of recombinant 1,2-a-mannosidase from Asperqillus saitoi overexpressed in Aspergillus oryzue cells, Biochem. J.. 1999, 339, 589-597. 125. Fujita, A,, T. Yoshida, and E. Ichishima, Five crucial carboxyl residues of 1,2-a-mannosidase from Aspergillus saitoi ( A . phoenicis), a food microorganism, are identified by site-directed mutagenesis, Biochem. Biophys. Rrs. Commun., 1997,238, 719-783. 126. Yoshida, T., and E. Ichishima, Molecular cloning and nucleotide sequence of the genomic DNA for 1.2-a-D-mannosidase gene, msdC from Penicillium citrinum, Biochim. Biophys. Acta, 1995, 1263, 159-162. 127. Yoshida, T., K. Maeda, M. Kobayashi, and E. Ichishima, Chemical modification of Penicillium 1,2-a-D-rnannosidase by water- soluble carbodi-imide: identification of a catalytically important aspartic acid residue, Biochem. J., 1994, 303, 97-103. 128. Dewald, B., and 0. Touster, A new a-D-mannosidase occurring in Golgi membranes, J. B i d . Chem., 1973,248, 7223-7233. 129. Tulsiani, D. R., D. J. Opheim, and 0. Touster, Purification and characterization of u-Dmannosidase from rat liver golgi membranes, J. Biol. Chem., 1977, 252, 3227-3233. 130. Opheim, D. J., and 0. Touster, Lysosomal a-D-mannosidase of rat liver. Purification and comparison with the golgi and cytosolic a-D-mannosidases, J. Biol. Chenz., 1978, 253, 10171023. 131. Spiro, R. G., M. J. Spiro, and A. M. Adamany, Enzymic assembly of the carbohydrate units of glycoproteins and the role of lipid intermediates, Biochem. Soc. Symp., 1974, 37755. 132. Tabas, I., S. Schlesinger, and S. Kornfeld, Processing of high mannose oligosaccharides to form complex type oligosaccharides on the newly synthesized polypeptides of the vesicular stomatitis virus G protein and the IgG heavy chain, J. Biol. Chem., 1978,253, 716-722. 133. Harpaz, N., and H. Schachter, Control of glycoprotein synthesis. Bovine colostrum UDP-Nacety1glucosamine:alpha-D-mannosidep 2-N-acetylglucosaminyltransferase I. Separation from UDP-N-acety1glucosamine:a-D-mannoside p 2-N-acetylglucosaminyltransferase 11, partial purification, and substrate specificity, J. Biol. Chem., 1980, 255, 4885-4893. 134. Harpaz, N., and H. Schachter, Control of glycoprotein synthesis. Processing of asparaginelinked oligosaccharides by one or more rat liver Golgi a-D-mannosidases dependent on the prior action of UDP-N-acetylglucosamine: a-D-mannoside p 2-N-acetylglucosaminyltransferase I, J. Biol. Chrm., 1980, 255, 489444902, 135. Tabas, I., and S. Kornfeld, The synthesis of complex-type oligosaccharides. 111. Identification of an a-D-mannosidase activity involved in a late stage of processing of complex-type oligosaccharides, J. Biol. Chem., 1978, 253, 7779-1186. 136. Kaushal, G. P., T. Szumilo, I. Pastuszak, and A. D. Elbein, Purification to homogeneity and properties of mannosidase I1 from mung bean seedlings, Biochemistry, 1990, 29, 2168-21 76. 137. Altmann, F., and L. Marz, Processing of asparagine-linked oligosaccharides in insect cells: evidence for a-mannosidase 11, Glycocoi~.J . , 1995, 12, 150-155. 138. Sharkey, D. J., and R. Kornfeld, Developmental regulation of processing a-mannosidases and “intersecting” N-acetylglucosaminyltransferase in Dictyostrlium discoideum, J. Biol. Chem., 1991,266, 18477-18484. 139. Sharkey, D. J., and R. Kornfeld, Developmental regulation of asparagine-linked oligosaccharide synthesis in Dictyostelium discoideum, J. Biol. Chem., 1991, 266, 18485- 18497. 140. Misago, M., Y. Liao, S. Kudo, S. Eto, M. Mattei, K. Moremen, and M. Fukuda, Molecular cloning and expression of cDNAs encoding human a-mannosidase I1 and a previously unrecognized a-mannosidase IIx isozyme, Proc. Nut1 Acud. Sci. U.S. A , , 1995, 92, 11766-1 1770. 141. Moremen, K. W., and P. W. Robbins, Isolation, characterization, and expression of cDNAs encoding murine a-mannosidase 11, a Golgi enzyme that controls conversion of high mannose to complex N-glycans, J. Cell B i d , 1991, 115, 1521 1534. 142. Paulson, J. C., and K. J. Colley, Glycosyltransferases: Structure, localization, and control of cell type-specific glycosylation, J. B i d . Chem., 1989, 264, 17615-17618. 143. Colley, K. J., Golgi localization of glycosyltransfera~es:more questions than answers, Glycobiology, 1997, 7, 1-13. 144. Jarvis, D. L., D. A. Bohlmeyer, Y. F. Liao, K. K. Lomax, R. K. Merkle, C. Weinkauf, and K. W. Moremen, Isolation and characterization of a class I1 a-mannosidase cDNA from lepidopteran insect cells, Glycohioloqy, 1997, 7, 1 13-127. ~~

References

1 15

145. Foster, J. M., B. Yudkin, A. E. Lockyer, and D. B. Roberts, Cloning and sequence analysis of GmII, a Drosophila melanogaster homologue of the cDNA encoding murine Golgi amannosidase 11, Gene, 1995, 154, 183- 186. 146. Iolascon, A., G. D’Agostaro, S. Perrotta, P. Izzo. R. Tavano, and B. Mirdglia del Giudice, Congenital dyserythropoietic anemia type 11: molecular basis and clinical aspects, Haematologira, 1996, 81, 543-559. 147. Iolascon, A., E. Miraglia del Giudice, S. Perrotta, M. Granatiero, L. Zelante, and P. Gasparini, Exclusion of three candidate genes as determinants of congenital dyserythropoietic anemia type I1 (CDA-II), Blood, 1997, YO, 4197-4200. 148. Gasparini, P., E. Miraglia del Giudice, J. Delaunay, A. Totaro, M. Granatiero, S. Melchionda, L. Zelante, and A. Iolascon, Localization of the congenital dyserythropoietic anemia I1 locus to chromosome 20q I 1.2 by genomewide search, Am. J. Hum. Genet., 1997, 61, 1 112 11 16. 149. Monis, E., P. Bonay, and R. C. Hughes, Characterization of a mannosidase acting on a1-3and al-6-linked mannose residues of oligoniannosidic intermediates of glycoprotein processing, Eur. J. Biochem., 1987, 168, 287 294. 150. Bonay, P., and R. C. Hughes, Purification and characterization of a novel broad-specificity (al-2, al-3 and a-1-6) mannosidase from rat liver, Eur. J. Biochem., 1991, 197, 229-238. 151. Bonay, P., J. Roth, and R. C. Hughes, Subcellular distribution in rat liver of a novel broadspecificity (al-2: al-3 and al-6) mannosidase active on oligomannose glycans, Eur. J. Biochem., 1992,205. 399-407. 152. Aronson, N. N . , Jr., and M. J. Kuranda, Lysosomal degradation of Asn-linked glycoproteins, FASEB J., 1989, 3. 2615-2622. 153. Cacan, R., C. Dengremont, 0. Labiau, D. Kmiecik, A. M. Mir, and A. Verbert, Occurrence of a cytosolic neutral chitobiase activity involved in oligomannoside degradation: a study with Madin-Darby bovine kidney (MDBK) cells, Bioclrem. J . , 1996, 313, 597-602. 154. Haeuw, J. F., G. Strecker, J. M. Wieruszeski, J. Montreuil, and J. C. Michalski, Substrate specificity of rat liver cytosolic a-D-mannosidase. Novel degradative pathway for oligomannosidic type glycans, Eur. J. Biochen?., 1991, 202, 1257-1 268. 155. a1 Daher, S., R. de Gasperi, P. Daniel, N. Hall, C. D. Warren, and B. Winchester, The substrate-specificity of human lysosomal alpha-D-mannosidase in relation to genetic amannosidosis, Biochem. J., 1991, 277, 743-75 I. 156. Cheng, S. H., S. Malcolm, S. Pemble, and B. Winchester, Purification and comparison of the structures of human liver acidic a-D-mannosidases A and B, Biochem. J . , 1986, 233, 6572. 157. Okumura, T., and I. Yamashina, Further purification and characterization of a-mannosidase from hog kidney, J. Biochem., 1973, 73, 131-138. 158. Berg, T., 0. K. Tollersrud, S. U. Walkley, D. Siegel, and 0. Nilssen, Purification of feline lysosomal a-mannosidase, determination of its cDNA sequence and identification of a mutation causing a-mannosidosis in Persian cats. Biochem. J., 1997, 328, 863-870. 159. Tollersrud, 0. K., T. Berg, P. Healy, G. Evjen, U. Ramachandran, and 0. Nilssen, Purification of bovine lysosomal a-mannosidase, characterization of its gene and determination of two mutations that cause a- mannosidosis, Euu. J. Biochem., 1997, 246, 410-419. 160. Nilssen, O., T. Berg, H. M. Riise, U. Ramachandran, G. Evjen, G . M. Hansen, D. Malm, L. Tranebjaerg, and 0. K. Tollersrud, a-Mannosidosis: functional cloning of the lysosomal amannosidase cDNA and identification of a mutation in two affected siblings, Hum. Mol. Genet., 1997. 6, 717--726. 161. Pohlmann, R., A. Hasilik, S. Cheng, S . Pemble, B. Winchester, and K. von Figura, Synthesis of lysosomal a-mannosidase in normal and mannosidosis fibroblasts, Biochem. Biophys. Rex Commun., 1983, 115, 1083-1089. 162. Liao, Y.-F., A. Lal, and K. W. Moremen, Cloning, Expression, Purification, and Characterization of the Human Broad Specificity Lysosomal Acid a-Mannosidase, J. Biol. Chem.; 1996, 271, 28348--28358. 163. Riise, H. M., T. Berg, 0. Nilssen, G. Romeo, 0. K. Tollersrud, and I. Ceccherini, Genomic structure of the human lysosomal a-mannosidase gene (MANB), Genomics, 1997, 42, 200207. ~

1 16

6 a-Mannosiduses in Asparugine-linked Oligosuccharide Processing

164. Merkle, R. K., Y. Zhang, P. J. Ruest, A. Lal, Y. F. Liao, and K. W. Moremen, Cloning, expression, purification, and characterization of the murine lysosomal acid a-mannosidase. Biochim. Biophys. Actu, 1997, 1336, 132- 146. 165. Gonzalez, D. S., Y. Kagawa, and K. W. Moremen, Isolation and characterization of the gene encoding the mouse broad specificity lysosomal a-mannosidase, Biochim. Biophys. Acta, 1999, 1445, 177-183. 166. Schatzle, J.; J. Bush, and J. Cardelli, Molecular cloning and characterization of the structural gene coding for the developmentally regulated lysosomal enzyme, a-mannosidase, in Dictyostelium discoideum, J. Bid. Chem., 1992, 267, 4000-4007. 167. Vandersall-Nairn, A. S., R. K. Merkle, K. O’Brien, T. N. Oeltmann, and K. W. Moremen, Cloning, expression, purification, and characterization of the acid a-mannosidase from Trypunosoma cvuri, Glycohiuloyy, 1998, 8, 1183-1 194. 168. Schatzle, J., J. Bush, S. Dharmawardhane, R. A. Firtel, R. H. Gomer, and J. Cardelli, Characterization of the signal transduction pathways and cis-acting DNA sequence responsible for the transcriptional induction during growth and development of the lysosomal a-mannosidase gene in Dictyostelium discoideum, J. Biol. Chem., 1993, 268, 19632- 19639. 169. Schatzle, J., A. Rathi, M. Clarke, and J. A. Cardelli, Developmental regulation of the amannosidase gene in Dictyostelium cliscoideum: control is at the level of transcription and is affected by cell density, Mol. Cell. Biol., 1991. 11, 3339-3347. 170. Daniel, P. F.. J. E. Evans, R. De Gasperi, B. Winchester, and C. D. Warren, A human lysosomal a(1-6)-mannosidase active on the branched trimannosyl core of complex glycans, Glycohiology, 1992, 2, 321-336. 171. De Gasperi, R., P. F. Daniel, and C. D. Warren, A human lysosomal a-mannosidase specific for the core of complex glycans, J. Bid. Chenz.. 1992, 267. 9706 9712. 172. Haeuw, J. F., T. Grard, C. Alonso, G. Strecker, and J. C. Michalski, The core-specific lysosomal a( I -6)-mannosidase activity depends on aspartamidohydrolase activity. Biochem. J. 1994,297,463-466. 173. Fisher. K. J . , and N. N. Aronson, Jr., Cloning and expression of the cDNA sequence encoding the lysosomal glycosidase di-N-acetylchitobiase, J Biol Chem, 1992, 267, 19607-19616. 174. Autio, S., T. Louhimo, and M. Helenius, The clinical course of mannosidosis, Ann. Clin. Rex; 1982, 14, 93-97. 175. Bennet, J. K., P. P. Dembure, and L. J. Elsas, Clinical and biochemical analysis of two families with type I and type I1 mannosidosis, Am. J. Med Genet., 1995, 55, 21 -26. 176. Yamashita, K.. Y. Tachibana, K. Mihara, S. Okada. H. Yabuuchi, and A. Kobata. Urinary oligosaccharides of mannosidosis, J. Biol. Chem., 1980. 255. 5126-5133. 177. Nilssen, O., T. Berg, H. M. Riise, U. Ramachandran, G. Evjen, G. M. Hansen, D. Malm, L. Tranebjaerg, and 0. K. Tollersrud, a-Mannosidosis: functional cloning of the lysosomal amannosidase cDNA and identification of a mutation in two affected siblings, Hum. Mol. Genet., 1997, 6 , 717-726. 178. Gotoda, Y., N. Wakamatsu, H. Kawai, Y. Nishida, and T. Matsumoto, Missense and nonsense mutations in the lysosomal a-mannosidase gene (MANB) in severe and mild forms of amannosidosis, Am. J. Hum. Genet.. 1998, 63, 1015-1024. 179. Berg, T., 11. M. Riise, G. M. Hansen, D. Malm, L. Tranebjaerg, 0. K. Tollersrud, and 0. Nilssen, Spectrum of mutations in a-mannosidosis, Am. J. Hum. Genet., 1999, 64, 77-88. 180. Tulsiani, D. R., M. D. Skudlarek, and M. C. Orgebin-Crist, Novel a-D-mannosidase of rat sperm plasma membranes: characterization and potential role in sperm-egg interactions, J. Cell Biol., 1989, 109. 1257-1267. 181. Pereira, B. M., A. Abou-Haila, and D. R. Tulsiani, Rat sperm surface mannosidase is first expressed on the plasma membrane of testicular germ cells, Biol. Reprod., 1998, 59, 128% 1295. 182. Tulsiani, D. R., S. K. NagDas, M. D. Skudlarek, and M. C. Orgebin-Crist, Rat sperm plasma membrane mannosidase: localization and evidence for proteolytic processing during epididymal maturation, Dev. Biol., 1995; 167, 584-595. 183. Tulsiani. D. R.; M. D. Skudlarek, S. K. Nagdas, and M. C. Orgebin-Crist, Purification and characterization of rat epididymai-fluid a-D-mannosidase: similarities to sperm plasmamembrane a-D-mannosidase, Biochern. J . , 1993, 290, 427-436.

184. Cornwall, G. A,, D. R. Tulsiani, and M. C. Orgebin-Crist, Inhibition of the mouse sperm surface u-D-mannosidase inhibits sperm-egg binding in vitro, Bid. Reprod,, 1991, 44, 91 3921. 185. Okamura, N., M. Tamba. H. J. Liao. S. Onoe, Y. Sugita, F. Dacheux, and J. L. Dacheux, Cloning of complementary DNA encoding a 135-kilodalton protein secreted from porcine corpus epididymis and its identification as an epididymis- specific u-mannosidase, Mol. Reprod. Dev., 1995, 42, 141-148. 186. Hiramoto, S., M. Tambd, S. Kiuchi, Y. Z. Jin, S. Bannai, Y. Sugita, F. Dacheux, J . L. Dacheux, M. Yoshidd, and N. Okamura, Stage-specific expression of a mouse homologue of the porcine 135kDa a-D-mannosidase (MAN2B2) in type A sperniatogonia, Biochcw. Bioplzys. Rex Cornniun., 1997, 241, 439-445. 187. Yoshihisa, T., and Y. Anrdku, Nucleotide sequence of AMSI, the structure gene of vacuolar a-mannosidase of Sncchuromyces cer iue, Biochcm. Biiylzys. Rex Cornmun., 1989, 163, 908-915. 188. Kuranda, M. J., and P. W. Robbins, Cloning and heterologous expression of glycosidase genes from Saccliarornycc~scerevisine, Proc.. Nut1 Acud. Sci. U. S. A , , 1987, 84. 2585-2589. 189. Yoshihisa, T., and Y. Anraku, A novel pathway of import of u-mannosidase, a marker enzyme of vacuolar membrane, in Succlzurornji~esciwuisiue, J. Biol. Chem. 1990, 265, 2241 822425. 190. Eades, C. J., A. M. Gilbert, C. D. Goodman, and W. E. Hintz, Identification and analysis of a class 2 a-mannosidase from Aspergillus t~id~ilurz.~., Glycobioloyy, 1998, 8, 17-33. 191. Bischoff, J., K. Moremen, and H. F. Lodish, Isolation, characterization, and expression of cDNA encoding a rat liver endoplasmic reticulum a-mannosidase, J. Bid. Chem., 1990, 265, 17110-17117. 192. Shoup, V., and 0. Touster, Purification and characterization of the a-D-mannosidase of rat liver cytosol, J Biol Clzeni, 1976. 251, 384553852. 193. Tulsiani, D. R., and 0. Touster, Characterization of a novel a-D-mannosidase from rat brain microsomes, J. Biol. Chem., 1985,260. I308 I- 13087. 194. Altschul, S. F., T. L. Madden, A. A. Schaffer. J. Zhang, Z. Zhang, W. Miller, and D. J. Lipman, Gapped BLAST and PSI-BLAST: a new generation of protein database search programs, Nucleic Acids R e x , 1997, 25, 3389-3402. 195. Thompson, J. D., D. G. Higgins, and T. J. Gibson, CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice, Nucleic Acids R e x , 1994, 22, 4673 -4680. 196. Page, R. D. M., TREEVIEW: An application to display phylogenetic trees on personal computers, Computer Applic. Biosci., 1996, 12, 351-358.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

7 The Role of UDP-Glcyglycoprotein Glucosyltransferase as a Sensor of Glycoprotein Conformations Armundo J. Purodi

7.1 Introduction N-glycoproteins are first glycosylated in the lumen of the endoplasmic reticulum (ER) upon transfer of an oligosaccharide (Glc3MangGlcNAc2 in most cells) from a dolichol-P-P derivative to Asn residues in nascent polypeptide chains [ 11. Dolichol is a polyprenol lipid having variable number of isoprene residues (18-2 1 in mammalian, 17-20 in plant, 15-18 in fungal and 10- 13 in protozoan cells). The first isoprene is always saturated. It is worth mentioning that Glc3MangGlcNAcl is transferred about 20-fold more efficiently by the oligosaccharyltransaferase than MangGlcNAc2. Mutations that affect formation of fully glucosylated oligosaccharides result in glycoprotein underglycosylation. Processing of protein-linked oligosaccharides in the ER is initiated immediately after transfer. Glucosidase I removes the external glucose unit and glucosidase I1 (GII) excises the two remaining glucoses. Specific mannosidases may remove up to two mannose units in the mammalian cell ER [l]. An additional ER processing reaction is that catalyzed by the UDP-Glcyglycoprotein glucosyltransferase (GT). This enzyme adds a single glucose unit to glucose-free, high mannose-type protein-linked oligosaccharides [2]. GI1 deglucosylates not only the monoglucosylated oligosaccharides generated by partial deglucosylation of the transferred oligosaccharides but also those formed by GT as they have identical structures [2]. Further oligosaccharide processing may occur in the Golgi apparatus. Protein N-glycosylation and oligosaccharide processing reactions occurring in the ER of mammalian cells as well as the reaction catalyzed by G T and the structure of Glc~MangGlcNAc2are depicted in Figure 1. This chapter deals with some peculiar features of GT specificity as well as with the role of this enzyme in sensing glycoprotein conformations.

120

7 The Role o j UDP-Glcyglycoprotein Glucosyltransferase

B

GsMgGNA2-P-P-0

1 G3MgGNA2-Pr

1 G2MgGNA2-Pr

1 G1MgGNA2-Pr

4 4

G,MgGNA2-Pr _L MgGNA2-Pr

C G1M8GNA2-Pr

Man,.,GlcNAc,-Pr

+ UDP-Glc G1M7GNA2-Pr

__

M8GNA2-Pr

1

L M7GNA2-Pr

Figure 1. Glycosylation of glycoproteins in the mammalian cell ER. A: the structure of Glc3Man&ilcNAcz; B: Processing of oligosaccharides in the mammalian ER; C: the reaction catalyzed by GT. D stands for dolichol, Pr for protein, G for glucose, M for mannose and G N for N-acetylglucosamine.

7.2 General Properties GT has been detected in microsomes of mammalian, insect, protozoan, plant, and fungal cells [2, 31. The GT was purified to homogeneity from rat liver, Drosophila melanoyaster and Schizosaccharomyces pombe [ 3-51. The enzyme uses UDP-Glc as glycosyl donor and not ADP-Glc, TDP-Glc or UDP-Gal [4, 51. GT is a rather large (about 160 kDa) soluble protein of the lumen of the ER, and requires Ca2+ ions in millimolar concentrations for activity [4, 51. This last result is consistent with the subcellular location of the enzyme because Ca2+ concentrations in the ER may reach the relatively high values necessary for GT activity. Concentrations of the divalent cation in the cytosol are about three orders of magnitude lower than those in the ER. Further, Ca2+ concentrations in this organelle are regulated by the action of an ion-motif ATPase that pumps calcium into the lumen, and by inositol 1,4,5-triphosphate, that induces release of Ca2+ into the cytosol. It may be speculated, therefore, that variations of Ca2+concentrations in the lumen may affect GT activity.

7.3 GT Recognizes Glycoprotein Conjormations

121

7.3 GT Recognizes Glycoprotein Conformations A striking property of G T purified from either mammalian, insect or S. pomhe cells is that in cell-free assays it glucosylates denatured (misfolded) glycoproteins, but not properly folded ones [2, 61. It was this property that allowed the assay and purification of the enzyme. As is the case with native glycoproteins, glycopeptides containing only one (Asn) or a few amino acids linked to Man7-9GlcNAc2 were not glucosylated by purified G T [6]. This result indicated that the effect of denaturation was not simply to make the oligosaccharides accessible to the transferring enzyme, but rather to allow interaction of protein domains, hidden in native conformations, with GT. The interiors of the water-soluble proteins in their native states are predominantly composed of hydrophobic amino acids whereas the hydrophilic side chains are at the exterior, where they interact with water. Denatured states of a protein have, in general, more hydrophobic side chains at their exterior than native ones do. It could be, therefore, that the effect of denaturation is to create a hydrophobic environment in the vicinity of the oligosaccharide that is recognized by the GT. It was found that under physiological conditions of pH and salt concentration, the G T binds very efficiently to hydrophobic but not hydrophilic peptides, and that the binding of hydrophobic peptides can be inhibited by denatured but not native glycoproteins [7]. Additional experiments showed that GT recognizes not only protein domains exposed in denatured conformations, but also the innermost GlcNAc unit of the oligosaccharide, that is, the residue that is left linked to the protein moiety following treatment with endo-P-N-acetylglucosaminidaseH (endo H) [7].In many native glycoproteins the innermost GlcNAc interacts with neighboring amino acid residues and is not accessible to macromolecular probes such as endo H [8]. In such cases, cleavage of oligosaccharides by the endoglycosidase requires denaturation of the glycoprotein [9]. It may be speculated that proper folding of most glycoproteins would hinder recognition of the innermost GlcNAc unit by the G T and thus prevent glucosylation. Several experiments showed that both recognition elements-the protein domains (probably hydrophobic amino acid patches) and the oligosaccharides-had to be covalently linked [7].This is an important restriction. As there are numerous unfolded, partially folded, and misfolded proteins and glycoproteins in the ER lumen, then if the protein domains recognized by G T and the oligosaccharides were not required to be covalently linked it might be expected that domains exposed in improperly folded species would induce glucosylation of glycoproteins already in their native conformations, as long as the innermost GlcNAc units in the latter species were accessible to GT. G T appeared to be an extremely sensitive sensor of the tertiary structure of glycoproteins, as it was able to recognize two forms of a neoglycoprotein (that is, a glycoprotein formed by a chemical coupling of a protein and an oligosaccharide) having the same secondary but slightly different tertiary structures: a glycopeptide (MangGlcNAcz-Asn) was linked to a Cys residue in position 70 of a full length

122

7 The Role of UDP-Glcyglycoprotein Glucosyltransferase

staphylococcal nuclease or of a truncated version of the same protein lacking the last 14 amino acid residues at the carboxyl terminus. The full length neoglycoprotein was not glucosylated by GT unless previously denatured upon a treatment with 8 M urea. On the other hand, the truncated neoglycoprotein was glucosylated by GT even without any denaturation treatment. Both forms of the neoglycoprotein (full length, not able to be glucosylated and truncated, able to be glucosylated) had the same secondary structure as judged by far UV spectra but they slightly differed in their tertiary structures as the full length species had the same specific nuclease activity as the native enzyme, whereas the truncated neoglycoprotein only showed one third of that activity. In addition, the truncated but not the full length neoglycoprotein was degraded by trypsin under controlled conditions [7]. The excellence of G T as a sensor of tertiary structures of glycoproteins was also revealed by the fact that the folding status of glycoproteins was paralleled by its ability to be glucosylated by the enzyme: soybean agglutinin was denatured with 6 M guanidine hydrochloride, diluted and allowed to renature under controlled conditions. The decrease in glucose acceptor capacity of this glycoprotein when incubated with GT plus UDP-Glc closely paralleled its renaturation, thus confirming the above mentioned conclusion [7].

7.4 The Primary Structure of the UDP-Glcyglycoprotein Glucosyltransferase The primary sequences of G T from two different species (D.melanogaster and S. pombe, GenBank accession numbers U20554 and U384 17, respectively) are known [3, 101. Based initially on partial amino acid sequences obtained from the pure rat liver G T and using current nucleic acid sequencing techniques we have recently derived the entire sequence of the enzyme from a rat liver cDNA open reading frame (a collaboration with D. Y. Thomas from the Department of Anatomy and Cell Biology, McGill University, Montreal). The sequence of an open reading frame (gene F48E3.3’; GenBank U28735) of the Caenorhabditis elegans genome probably corresponds to the gene encoding the G T due to its high level of identity with the other three sequences. The four sequences encode predicted proteins that are similar in size, and which have retrieval sequences typical of soluble ER proteins at their carboxyl terminal ends. All four proteins also have several consensus N glycosylation sites, consistent with the fact that the mammalian and fungal enzymes interact with concanavalin A [4, 51. All four enzymes show extremely high sequence homology in their carboxyterminal domains. A similar homology has been found between those domains and conceptual translations of expressed sequence tags from rice and Arabidopsis thaliana (GenBank D24933 and T23006, respectively). Since a limited but significant sequence homology was observed between the carboxyl terminal domains of the fly, fungal and mammalian enzymes and several bacterial glycosyltransferases that use UDP-Glc or UDP-Gal as donor substrates, it may be speculated that the highly conserved G T carboxyl terminal portions are responsible for UDP-Glc recognition.

7.5 The Role o j Monoglucosyluted Oliqosucchurides in Glycoprotein Folding

123

The amino terminal portions of the G T sequences show a lower degree homology than the carboxyl terminal ones. This is especially noticeable upon comparison of the rat liver and S. pombe sequences. Such variations in the extent of primary sequence similarity in different regions in a family of proteins are reminiscent of those seen in the Hsp70 family of proteins which includes the ER chaperone BiP. Here, their amino termini, containing the ATPase activity have a high degree of homology, whereas the carboxyl terminal portions, containing the hydrophobic peptide binding sites, are much less conserved [ l l ] . As mentioned above, under physiological salt and pH conditions, the GT binds hydrophobic peptides exposed in misfolded glycoproteins. The G T shares this property, therefore, with known chaperones and it may well perform this function in vivo. It may be speculated that the amino-terminal domains of G T are responsible for binding the hydrophobic patches, and since they have to recognize a wide variety of different structures, they may be constituted from quite diverse amino acid sequences in different organisms.

7.5 The Role of Monoglucosylated Oligosaccharides in Glycoprotein Folding Glycoproteins acquire their proper tertiary structure in the ER. This process requires participation of chaperones and other folding-assisting proteins. Glycoproteins that fail to properly fold are retained in the ER and further transported to the cytosol where they are degraded in the proteasomes. On the other hand, properly folded species may continue their transit through the secretory pathway to their final destinations. Two unconventional chaperones (calnexin and calreticulin), that recognize monoglucosylated high mannose-type oligosaccharides, have been described in the ER lumen of mammalian cells [12]. Membrane-bound calnexin and soluble calreticulin display a 35% similarity in their amino acid sequence. It has been shown that interaction of calnexin/calreticulin with monoglucosylated glycoproteins generated either by partial deglucosylation of the transferred oligosaccharide or by reglucosylation of glucose-free compounds by G T facilitates glycoprotein folding in mammalian cells by preventing aggregation and suppressing formation of nonnative disulfide bonds [ 131. Moreover, the above mentioned interaction provides one of the alternative mechanisms by which cells retain misfolded glycoproteins in the ER [14]. In vivo disruption of the calnexin/calreticulin-glycoprotein interaction would occur upon GII-mediated deglucosylation of glycoproteins. Not properly folded species would be reglucosylated by G T and recognized again by calnexin/ calreticulin. On the contrary, no reglucosylation would occur upon proper folding and glycoproteins, finally liberated from their calnexin/calreticulin anchors, would be free to continue their transit through the secretory pathway. The calnexin/ calreticulin interaction with monoglucosylated glycoproteins has been considered to be, therefore, a quality control of glycoprotein folding. Although the bulk of evidence supporting the model of quality control as proposed comes from experi-

124

7 The Role of' UDP-Glcyglycoprotein Glucosyltrunsferuse

/

Glc MxI&IcNAC~-P-P-DOI

, G~~,~~,G~CNAC,@

\ GOLGI

i

DEGRADATION

Figure 2. Schematic representation of the role proposed for monoglucosylated oligosaccharides in glycoprotein folding. Sad and happy faces represent not properly folded and properly folded protein moieties, respectively. CNX: calnexin; CRT: calreticulin; OT: oligosaccharyltransferase; GI: glucosidase 1; GII: glucosidase 11 and GT: UDP-Glcyglycoprotein glucosyltransferase. CNX and CRT interact with monoglucosylated oligosaccharides created either by partial deglucosylation of the transferred oligosaccharide or by GT-mediated reglucosylation of glucose-free oligosaccharides exclusively linked to not properly folded protein moieties. The interaction facilitates proper folding of glycoproteins and prevents their exit to the Golgi when not properly folded. Glycoproteins are liberated from the CNXjCRT anchor upon acquisition of the correct tertiary structure. Permanently misfolded species are bound to degradation in the proteasome.

ments performed with mammalian cell systems, available evidence indicates that the same features of the model occur in the yeast S. pombe, in plants and in protozoa [2, 5 , 10, 15-18]. A schematic representation of the role of monoglucosylated oligosaccharides in glycoprotein folding is depicted in Figure 2. It has been more recently reported that ERp57, an ER lumenal redox enzyme (also called GRP58, ERp61, ER 60, HIP70, Q2 or P58) of the protein disulfide isomerase family specifically interacts with partially deglucosylated glycoproteins, as it catalyzes oxidative refolding of glycoproteins recruited by calnexin or calreticulin [ 19-21]. In addition, it has been reported that these two chaperones mediate the interaction of ERp57 with

7.5 The Role of Monoylucosylated Oligosaccliorides in Glycoprotein Folding

125

soluble and membrane-bound chaperones devoid of cysteine residues. This result suggests that ERp57 might have a chaperone-like role not related to its oxidative catalytic properties. We have obtained two single S. pomhe (a yeast having calnexin but not calreticulin) mutants affected in either one of the two pathways leading to the production of monoglucosylated oligosaccharides [22].Alg6 mutants transferred Man9GlcNAcz to nascent polypeptide chains and y p t l mutants lacked GT. Both single mutants were found to grow normally at 28 " C . On the other hand, gptl/ulg6 double mutant cells grew very slowly and with a rounded morphology at 28 "C and did not grow at 37 " C . The wild type phenotype could be restored upon transfection of double mutants with a GT-encoding expression vector. It was concluded that glycoprotein folding facilitation mediated by the interaction of monoglucosylated oligosaccharides with calnexin is essential for cell viability under conditions of severe ER stress such as underglycosylation of proteins caused by the alg6 mutation and high temperature. A glsZ/alg6 double mutant that transferred Man9 GlcNAc2 and was unable to remove the glucose units added by G T as it lacked GI1 had a wild type phenotype of growth and morphology at 28 "C and was able to grow at 37 "C thus indicating that glycoprotein folding facilitation mediated by the interaction of calnexin and monoglucosylated oligosaccharides does not necessarily require cycles of reglucosylation-deglucosylation catalyzed by G T and GI1 [22]. There is a controversy on whether calnexin/calreticulin behave as lectins that exclusively recognize the above mentioned oligosaccharides or if. alternatively, such recognition is the first and necessary step for an interaction between misfolded glycoprotein protein moieties and calnexin/calreticulin. According to this last interpretation, once the protein-protein interaction is established, recognition of monoglucosylated oligosaccharides would be irrelevant for the stability of the complex. Liberation of glycoproteins from the complex would result from a conformational change in the substrate polypeptides. On the other hand, according to the first interpretation release of glycoproteins from the complex would exclusively occur through the action of GII. In both cases, properly folded species would not be able to be reglucosylated by G T and thus no further interaction with calnexin/ calreticulin would occur [ 121. Evidences supporting the second model (interaction of calnexin/calreticulin first with monoglucosylated oligosaccharides and then with the protein moieties of folding glycoproteins) were provided by experiments in which several transmembrane glycoproteins (murine MHC class I heavy chain Kb and Db, human MHC class I HLA heavy chain, MHC class I1 DRa and DRP and invariant chains, T cell receptor a chain) were found to immunoprecipitate with anticalnexin antibodies even after enzymatic removal of all oligosaccharide chains [23-26]. The possibility exists, however, that as both calnexin and the substrates are transmembrane species, they may have remained trapped in the same detergent micelles after removal of oligosaccharides. On the other hand, a soluble glycoprotein as a1 -antitrypsin was also found to be apparently immunoprecipitated by anticalnexin antiserum after removal of oligosaccharides. In this case the above mentioned possible colocalization of calnexin and the substrate glycoprotein in the same micelles may be ruled out [25]. However, a poor solubility of the yet not properly folded a,-antitrypsin

126

7 The Role Gf UDP-Glcyglycoprotein Glucosyltransjierase

(and also of the above mentioned deglycosylated transmembrane glycoproteins) may explain its appearance in the immunoprecipitate. Evidence for the role of calnexin/calreticulin as lectins that exclusively recognize monoglucosylated oligosaccharides comes from experiments in which the interaction of glycosylated bovine pancreas RNase with calnexin/calreticulin was studied either in a dog pancreas microsome-rabbit reticulocyte translation system or with isolated RNase and calnexin [27, 281. It was concluded in both studies that removal of oligosaccharides with endoglycosidases or of the glucose units present in the oligosaccharides with GII, completely abolished calnexin/calreticulin interaction with monoglucosylated RNase. The main drawback of the above mentioned conclusion is that the model glycoprotein chosen (RNase) is practically devoid of hydrophobic domains as revealed by a Kyte and Doolittle hydrophilicity plot. An extremely weak interaction of the protein moiety of RNase with calnexin/calreticulin would be expected if such interaction involved hydrophobic domains in the substrate glycoprotein as has been described for folding proteins and classical chaperones. Trypanosoma cruzi is a protozoan parasite that belongs to an early branch in evolution and lacks several features of the pathway of protein N-glycosylation and oligosaccharide processing present in the ER of higher eukaryotes but that nevertheless displays GT and GI1 activities. We have found recently that this microorganism also expresses a calreticulin-like molecule. No calnexin-encoding gene was detected. Addition of anticalreticulin serum to extracts obtained from cells pulse-chased with [ 35S]Met plus [ 35S]Cys immunoprecipitated two proteins that were identified as calreticulin and the lysosomal proteinase cruzipain (a major soluble glycoprotein). A treatment of the calreticulin-cruzipain complexes with endo H either before or after addition of anticalreticulin serum completely disrupted calreticulin-cruzipain interaction. In addition, mature monoglucosylated but not unglucosylated cruzipain isolated from lysosomes was found to interact with recombinant calreticulin. It was concluded that T. cvuzi calreticulin is a lectin that exclusively recognizes monoglucosylated oligosaccharides in glycoproteins [29]. Results obtained with this protozoan obviate drawbacks mentioned above for conclusions supporting either one of both models of the calnexin/calreticulin-glycoprotein interaction. Both calreticulin and cruzipain are soluble proteins (i.e., they can not be trapped in micelles) and the substrate glycoprotein has several hydrophobic domains interspersed along the entire molecule. Moreover, the conclusion that calreticulin behaves exclusively as a lectin that recognizes monoglucosylated oligosaccharides was derived from experiments performed in intact cells, using a naturally expressed glycoprotein. Glycoprotein folding facilitation mediated by the interaction of protein-linked monoglucosylated oligosaccharides and ER lectins is, therefore, a process that appeared early in evolution. Acknowledgments Work performed by the author has been mainly supported by grants from the United Nations Development Program/World Bank/World Health Organization Special Program for Research and Training in Tropical Diseases, the National Institutes of Health (USA) and the Howard Hughes Medical Institute.

References

127

References 1. R. Kornfeld, S. Kornfeld. Annu. Rea. Bioc.hcn7. 1985, 55, 631-664. S. E. Trombetta, M. Bosch, A. J. Parodi. Bioc.lzm/~7isfr.y.1989, 28, 8 108-8 1 16. C. G. Parker, L. I. Fessler, R. E. Nelson, J. H. Fessler E M 5 0 .I. 1995, 14 1294--1303. S. E. Trombetta, A. J. Parodi. J . Bid. Clie/n. 1992, 267, 9236-9240. F. Fernindez, S. E. Trombetta, U. Hellman, A. J. Parodi. J. Biol. Chem. 1994, 269. 30701-

2. 3. 4. 5.

30706. 6. M. Sousa, M. A. Ferrero-Garcia. A. J. Parodi. Biochemistry. 1992, 31, 97- 105. 7. M. Sousa, A. J. Parodi. E M 5 0 J. 1995, 14. 4196-4203. 8. A. Dessen, D. Gupta, S. Sabesan, C . F. Brewer. J. C. Sacchettini. Biochemistry. 1995, 34, 4933-4942. 9. R. B. Trimble, F. Maley. A n d . Biocltcwi. 1984, 141, 515-522. 10. F. Fernindez, M. Jannatipour, U . Hellman, L. Rokeach, A. J. Parodi. EMBO J. 1996, 15, 705 713. 11. .IP. . Hendrick, F. U. Hartl. Annu. Reil. Bioclietn. 1993. 62, 349- 384. 12. A. Helenius, E. S. Trombetta, D. N . Hebert, J. F. Simons. Trends Cell Biol. 1997. 7, 193~-200. 13. D. N. Hebert, B. Foellmer, A. Helenius. E M B O J. 1996, 15, 2961-2968. 14. J.-X. Zhang, 1. Braakman, K. E. S. Matlack, A. Helenius. Mol. Bid. Cell. 1997, 8. 1943-1954. 15. M. Jannatipour, L. Rokeach. J. Biol. Clrrn7. 1995, 270, 4845-4853. 16. F. Lupattelli, E. Pedrazzini, R. Bollini, A. Vitale. A. Ceriotti. P l m t Ct4. 1997. 9, 597-609. 17 F. Parlati, D. Dignard, J. J. M. Bergeron, D. Y. Thomas. EMBO J. 1995, 14, 3064-3072. 18. C. Labriola, J. J. Cazzulo: A. J. Parodi. J. Cc4 Biol. 1995, 130, 771-779. 19 J. D. Oliver, F. J. van der Wal. N. J. Bulleid, S. High. Science. 1997, 275. 86-88. 20 J. G. Elliott, J. D. Oliver, S. High. J. Biol. Client. 1997, 272, 13849- 13855. Darby. D. C. Tessier, M . Michalak, J. J. Bergeron, D. Y. Thomas. J. Biol. 21 Chem. 1998, 273, 6009-6012. 22 S. Fanchiotti, F. Fernindez. C. D'Alessio, A. J. Parodi. J . Cdl Bid. 1998. 143. 625-635. 23 B. Arunachalam, P. Cresswell. J. Biol. Chetn. 1995, 270, 2784-2790. 24 M. J. Bennett, J. E. M. Van Leeuwen, K. P. Kearse. J. Biol. Chetn. 1998. 273, 23674423680, 25. F. E. Ware, A. Vassilakos, P. A. Peterson, M. R. Jackson, M. S. Lehrman, D. B. Williams. J. Biol. Chem. 1995. 270, 469774704. 26. Q. Zhang, M. Tector, R. D. Salter. J. Bid. C%e/n.1995, 270, 3944-3948. 27. A. R. Rodan, J. F. Simons. E. S. Trombetta. A. Helenius. E M 5 0 J . 1996, 15. 6921-6930. 28. A. Zapun, S. Petrescu, P. M. Rudd, R. A. Dwek, D. Y. Thomas, J. J. M. Bergeron. Cell. 1997, 88. 29-38. 29. C. Labriola, J. J. Cazzulo, A. J. Parodi. Mol. Biol. CeN 1999, 10, 1381 1394

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

8 Mannosyltransferases Peter Orlean

8.1 Introduction Covalently-linked mannose is present in the glycans, glycoproteins, glycolipids, and sugar nucleotides made by eubacteria, archaea, and eukaryotes. Attachment of mannose can be through 0-glycosidic linkages to other sugars, through 0-linkages to serine or threonine in peptides, through a C-C linkage to tryptophan, and through monophosphates to other sugar residues or polyisoprenoids. Mannosyltransferases (Man-T) are so far known to use one of only two classes of mannosyl donor, GDPMan or polyisoprenoid monophosphate mannose. There is much more diversity in the type of acceptor molecule used and in the type and configuration of linkage formed. Biochemical and genetic studies of glycan synthesis have led to the cloning of many glycosyltransferase genes, and genome sequencing projects have been turning up many more genes encoding homologous proteins. Some of the latter have subsequently been shown to participate in glycan synthesis, whereas the roles of others are as yet unclear. The availability of amino acid sequences of increasing numbers of known and putative glycosyltransferases has in turn started to reveal that these proteins can be grouped into families. Some families have representatives in eubacteria, archaea, and eukaryotes, indicating an early evolutionary origin, but other families so far seem restricted to eukaryotes. Amino acid sequence comparisons have permitted some generalizations to be made about the residues likely to be important for catalysis of glycosyl transfer by certain classes of protein. However, it is not yet possible to correlate conserved primary amino acid sequence features of a glycosyltransferase with the nature of the sugar it transfers or the type of linkage it forms. Much progress has been made in identifying Man-T genes and defining the roles of their products various biosynthetic pathways. Man-T will be discussed in the context of the secretory pathway of eukaryotic cells and the biosynthetic pathways that are localized in the ER and Golgi. Proteins related by amino acid sequence

130

8 Munnosyltransjerases

similarity will be considered as a group. Within this organizational context, reference will be made to related proteins in eubacteria and archaea.

8.2 Occurrence of Covalently-linked Mannose 8.2.1 Eukaryotic Secretory Glycoproteins

Mannose is present in the N - and O-linked glycans and glycosylphosphatidylinositol (GPI) anchors of secretory glycoproteins [ 1-51, In yeast, N-linked saccharides can bear long, branched mannan chains. Other eukaryotic cells also make mannosecontaining polysaccharides, for example, the gluco- and galactomannans of legume cell walls [6,7] and the glucuronoxylomannan capsular polysaccharide of the human pathogenic yeast Cryptococcus fieqforMzuns [S]. At least some of the latter structures may be elaborated on protein-bound glycans, and, consistent with this, the Man-T involved in their synthesis in plants occur in the endomembrane system of the cell. 8.2.2 Glycophospholipids

Man-containing glycophospholipids serve as donors or intermediates in the biosynthesis of various glycans. Particularly widespread are polyisoprenoid-phosphate Man and polyisoprenoid pyrophosphate-linked, Man-containing glycans, which are made by eubacteria, archaea, and eukaryotes. The former are sugar donors in glycan biosynthetic pathways, whereas the latter serve as carriers of larger glycan structures. Free, phosphoinositol-linked, Man-containing glycans such as glycoinositol phospholipids and lipophosphoglycans are anchored in the plasma membrane of protozoa [9], and mannosylated inositol phosphoceramides are made in fungi and plants [lo]. 8.2.3 Eubacterial and Archaeal Mannose-containing Molecules

Some of the earliest studies of Man-Ts were on polyprenol-P-Man made by Micrococcus Zysodeikticus [ 1I]. Other well-studied molecules are the O-antigen repeats of the lipopolysaccharide (LPS) of the outer membrane of Gram-negative bacteria [ 121, exopolysaccharides such as xanthan [ 131, and the lipoarabinomannan of the mycobacterial cell envelope [ 141. At least one archaeon, Thermoplasma acidophilum, makes a protein-linked glycan containing al,2-, a1,3-, and al,6-linked Man [ 151. 8.2.4 C-linked Mannose

Mannose can become directly linked to protein in a C-C linkage between C-1 of Man and the indole moiety of the first tryptophan in the sequence -WXXW- [16,

8.3 Biochemistry of Munnosyl Trunsjijr

131

171. This type of protein mannosylation has been found so far in mammalian RNase 2, but may be more widespread.

8.3 Biochemistry of Mannosyl Transfer 8.3. Many Linkages, Two Donors Mannose is commonly linked to other mannoses, to other sugars, or to inositol in ~ 1 . 2 -al,3-, a1,4-, al,6-, P1,2-, and Pl,4-glycosidic linkages, to polyisoprenoid phosphates in P-linkages, and to serine and threonine hydroxyls, as well as to tryptophan, in a-linkage. Most Man-T transfer single Man residues, but in some cases, polymannose chains can be made. An example of polymerizing activity is the formation of the a1,6-linked backbone of yeast mannan. In this case, a complex of potential Man-Ts is involved (Section 8.4.2), but it is not known whether any one protein transfers all the mannoses, or whether multiple catalytic subunits cooperate. The latter can be achieved in a single protein, MtfAp (or WbdAp), which is involved in 0-9 LPS synthesis in Eschericlzia coli. This protein has two catalytic domains that presumably arose in an earlier gene duplication event, and which allow Man to be transferred to two different acceptor structures in succession. One domain first adds cxl,2-Man to an al,3-linked Man, and the second domain then adds two further al,2-Man to the al,2-Man acceptor the first domain just made [ 18, 191. Use of GDPMan or of polyisoprenoid phosphate-linked Man as donor is not correlated with the type of glycosidic linkage a given Man-T forms: for example, some of the a1,2-, a1,3-, and al,6-Man-Ts involved in assembly of the Dol-PPlinked precursor in N-glycosylation use GDPMan as donor, others, Dol-P-Man. No Man-T is known to use both nucleotide and lipid-linked donors, and the glycosyltransferase families identified so far contain proteins that use only one of the two types of substrate. 8.3.2 Donor Specificity

GDPMan is strongly preferred as donor by sugar nucleotide-requiring Man-T. The specificity seems to be for the GDP moiety, for in vitro, some Man-T can use GDPMan analogs in which the mannose has been modified or replaced with another sugar [20]. The ability of Man-T to recognize analogs of GDPMan has been exploited for photoaffinity labeling of Man-T with GDP-he~anolamine-’~~ Iazidosalicylic acid [21]. So far, however, the structural basis for the specificity of glycosyltransferases for the nucleotide moiety of their donor is not known. Indeed, most families of glycosyltransferases related by amino acid sequence similarity include Man-T as well as enzymes that use other sugar nucleotide donors. Synthesis of the inner core of the LPS of E. coli and Rhizohium leguminosarum provides an example of glycosyltransferases whose natural sugar nucleotide donors

132

8 Munnosyltransjerases

are different, but which can both use the same non-physiological “hybrid” donor. In E. coli, the inner core contains glyceromanno-heptose (Heptose) linked a1,S to ketodeoxy-manno-octulosonic acid (Kdo), whereas in the Rhizohium core, it is Man that is al,5-linked to Kdo. The E. cnli a1,S heptosyltransferase, RfaCp, uses ADPHeptose as donor, and, in vitro, can also use synthetic ADPMan, but not GDPMan. R. leguminosarurn has a GDPMan-utilizing al,5-Man-T activity that recognizes the same acceptor structure as RfaCp does, but which can also use nonphysiological ADPMan as donor [22]. Comparison of the amino acid sequence of the predicted GDPMan-utilizing Rhizohium al,5-Man-T with that of RfaCp may reveal which amino acid residues are important for discrimination between ADP and GDP. In addition to their specificity for the sugar moiety of their donor, the Dol-PMan-utilizing Man-T of the dolichol pathway, as well as the Pmtlp Dol-P-Man: protein O-Man-T, strongly prefer their donor to have a long-chain, lipophilic polyisoprenoid with a p-Man-1 -P attached to a saturated a-isoprene group [23-25]. 8.3.3 Acceptor Specificity Some Man-T, for example, those involved in the ordered assembly pathway for the Dol-PP-linked oligosaccharide [26], can show high specificity for an acceptor glycan of a particular structure, although the protein-glycan interactions involved are as yet unknown. Additional features of some mannosyl acceptors, for example, the nature of their lipid moieties, can have a profound effect on Man-T activity. The influence of isoprenoid chain length and of saturation or unsaturation of the a-isoprenoid has been explored for Man-T that use polyisoprenoid-Ps as substrate. In vitro, Dol-PMan synthase can use shorter chain dolichols, polyprenols, even phenyl phosphate, as mannosyl acceptors [27, 281, but there is a clear preference for a-saturation and for a chain length greater than seven isoprenoid units, the effect of decreasing chain length being to lower the V,, of the reaction. The Man-T of the early part of the dolichol pathway, however, do not require the polyisoprenoid carrier to be long chain dolichol in vitro, for Man5 can be assembled on GlcNAc2 linked to the tetraisoprenoid phytanol [29]. It is not known how these proteins recognize the polyisoprenoid. A photoactivatable dolichol analog has been described that can be phosphorylated, and could thus be used to identify portions of Man-T that are in close proximity to bound Dol-P derivatives [ 301. The Dol-P-Man-utilizing Man-T that adds the first, al,4-Man to GlcN-PI during GPI biosynthesis varies between species in its acceptor specificity. The Trypanosoma enzyme mannosylates GlcN-PI without prior acylation of inositol, whereas the mammalian Man-T requires the 2’-OH of inositol to be acylated [3I , 321. 8.3.4 Structural Features of Man-T

The majority of Man-T are membrane proteins: Man-T activities sediment with membranes, and the available amino acid sequences of Man-T contain one or more predicted transmembrane helices. In a few cases, the transmembrane topography of

8.4 Man-T Families and the P~rthwuysThey Participate in

133

a Man-T has been investigated, and the results are consistent with the accepted notion that the catalytic domains of Dol-P-Man-utilizing Man-T of the endoplasmic reticulum (ER) face the lumen [ S , 331. The GDPMan-utilizing Man-T of the dolichol pathway are believed to catalyze mannosyl transfer at the cytoplasmic face of the ER membrane, whereas mannose transfer from GDPMan in the yeast Golgi takes place in the lumen of that organelle, as expected for these Type I1 membrane proteins. The GDPMan-utilizing Man-T in general have fewer predicted transmembrane segments than the Dol-P-Man-requiring ones, but the greater apparent hydrophobicity of the latter proteins may not necessarily be indicative of multiple transmembrane segments. Thus, experimental determination of the transmembrane organization of Dol-P-Man-utilizing Pig-Bp reveals the protein spans the membrane only once (section 8.4.1). No detailed structure has yet been determined for any Man-T, however, amino acid sequence comparisons have identified motifs conserved in two glycosyltransferase super families that include Man-T. These motifs contain invariant amino acids with acidic side chains that could be important catalytically or for maintenance of structure. A comparison of P-glycosyltransferase sequences by hydrophobic cluster analysis led to the identification of two regions in these proteins in which aspartates are highly conserved [34]. One of these contains what is now recognized as the “DXD” motif, which is found in at least 13 a- and P-glycosyltransferase families that use a sugar nucleotide donor [ 3 S ] , and its presence is an important criterion in the identification of further such enzymes. Site-specific mutagenesis experiments on the two aspartates in Mnnlp, a yeast a1,3 Man-T, showed that both residues are indeed critical for enzyme activity [3S].The second aspartate-containing region [34] includes the sequence -(aliphatic amino acid)4DD-, which is also conserved in members of the phosphoribosyltransferase family, in which the pair of aspartates interacts (via Mg”) with the OH-groups on the ribose moiety of phosphoribosyl pyrophosphate [36]. In glycosyltransferases, these aspartates may likewise interact with the ribose portion of the sugar nucleotide substrate. A motif containing two invariant glutamates, “EXFGXXXXE” (or, more loosely, “E(X7)E”), occurs in a range of a-glycosyltransferases from eubacteria, archaea, and eukaryotes [IS]. Much information is yet needed before we know how a Man-T distinguishes GDPMan from other sugar nucleotides, and what structural features are correlated with the ability of closely-related Man-T to form different a-mannosidic linkages or to transfer Man-P (as can different members of the Ktrlp family (section 8.4.2)).

8.4 Man-T Families and the Pathways They Participate in The Man-Ts that participate in the pathways leading to N-glycosylation, GPI anchoring, and O-mannosylation of protein in the ER will be discussed first, then the Golgi Man-T families involved in the elaboration of yeast mannan. Protein families whose membership includes eubacterial and archaeal proteins will be identified and examples of eubacterial mannosylation pathways will be given.

134

8 Munnosyltrunsjhses

Landmark studies that led to the identification of yeast Man-T genes include the analyses of yeast mannan formation in mnn mutants [37], and the use of [3H]mannose suicide enrichment to isolate yeast a1g mutants in the dolichol pathway [38]. The subsequent finding that late-stage, non-conditional alg mutants have a synthetic phenotype in combination with a defect in oligosaccharyltransferase has permitted the cloning of further ALG genes [3]. The isolation of murine Thy1 lymphoma and Chinese hamster ovary mutants were seminal contributions to our understanding of Man-T and the isolation of mammalian Man-T genes [39, 401. Many Man-T genes that are conserved among eukaryotes were originally identified in Saccharomyces cerevisiae, and yeast also has additional Man-T families whose members may be confined to fungi. Accordingly, the yeast nomenclature for Man-Ts will mainly be used, although the discussion of the ER Man-T is intended to apply to all eukaryotes. Man-Ts will be grouped into “families” for discussion, a “family” being referred to if a given protein has obvious sequence homologs in yeast. 8.4.1 Man-Ts of the ER [l-51

It is generally believed that the Man-T involved in the construction of the Dol-PPlinked oligosaccharide and the GPI precursor are localized in the ER membrane. However, dolichol pathway activities have also been detected in mitochondria [41], but it’s not clear whether known genes or new ones encode the enzymes involved. The Man-T discussed here are regarded as ER proteins. AMP The gene for this p-1,4 Man-T, which transfers mannose from GDPMan to Dol-PPGlcNAc;! at the cytoplasmic face of the ER membrane during assembly of the DolPP-linked precursor, was cloned by complementation of the yeast d g l mutant [38]. Alglp is expressed in E. coli [42]. The protein is conserved among eukaryotes, and although there are no close relatives of this protein in any eukaryote’s genome, Alglp shows limited resemblence to the Alg2p family [3], and has homologs in eubacteria and archaea; however, Algl p lacks the “E(X7)E” signature. AlgZp/Algl l p Alg2p adds either the u1,2- or the u1,6-linked mannose to the p-1,4 Man during assembly of the Dol-PP-linked precursor, presumably at the cytoplasmic face of the ER membrane. Alg2p has a homolog in the yeast protein database, Algl lp, which may therefore also be a GDPMan-dependent transferase that acts early in dolichol pathway. Alg2p and Algl l p are members of a large family of eukaryotic, eubacterial, and archaeal proteins with the “E(X7)E” motif [18]. Most of these proteins seem to be retaining a-glycosyltransferases that use a sugar nucleotide as donor and a sugar or inositol as acceptor. Among the bacterial Man-T in this group are cellobiosyldiphosphoprenyl u-Man-Ts involved in acetan and xanthan biosynthesis [43] and Man-T involved in O-antigen synthesis in Gram negative bacteria (eg. Rhizobium LpcCp and Sulmonella RfbUp) [12, 441.

8.4 Mali-T Funiilirs and the Pathways They Purticipute in

135

DPmlP Dol-P-Man synthase catalyzes transfer of Man from GDPMan to Dol-P with inversion of configuration to form Dol-P-PMan. This enzyme activity is ubiquitous in eukaryotes, and Dol-P-Man serves as a donor in the dolichol pathway 12, 31, in GPI biosynthesis [ 11, in 0-mannosylation [5], and in C-mannosylation [ 161. Related enzyme activities have been detected in eubacteria, for example Mycobacterium [ 141, and in archaea 1451, polyprenol-P and Dol-P being used, respectively, as acceptors. Eukaryotic Dol-P-Man synthases fall into two classes 146, 471. One, the “S. cereuisiae” class, consists of a single catalytic Dpml protein that is expressed in E. coli and which has a COOH-terminal transmembrane domain [48]. The other, “human”, class is a complex of two proteins, a Dpmlp subunit that lacks a COOH terminal domain and which is not active in E. coli, and the small, hydrophobic Dpm2p protein. Dpmlp is catalytic, and Dpm2p is required both for localization of Dpmlp to the ER membrane, and for optimal binding of Dol-P to the complex [49J. S. cerevisiue D P M l complements mammalian mutant cell lines defective in either Dpmlp or Dpm2p, but “human” class D P M l genes do not rescue a lethal disruption of S. cerevisiae D P M l . The amino terminal half of Dpmlp resembles that of Dol-P-Glc synthases 1501, and also shows a high degree of homology to eubacterial and archaeal members of a large family of P-glycosyltransferases I341 (section 8.3.4). Dpmlp is sensitive to the physical state of its membrane environment, and lipid matrices that form destabilized bilayers promote the association of the enzyme with Dol-P and are necessary for optimal activity [51]. More specific modulation of Dol-P-Man synthase activity may be achieved through phosphorylation by CAMPdependent kinase 1521, and the six Dpmlp sequences indeed all have a potential site for phosphorylation by that enzyme [46]. Ak3P This Dol-P-Man-utilizing u1,3 Man-T adds the 6th Man during assembly of the Dol-PP-linked precursor in N-glycosylation [ 31, presumably on the lumenal side of the ER membrane. Alg3p has no obvious homologs in prokaryotes.

Alg9p/PIG-Bp family Members of this group are Dol-P-Man-utilizing transferases that function in either the dolichol pathway or in the GPI anchor assembly pathway. ALG9 and PIG-B were cloned by complementation, respectively, of a yeast N-glycosylation mutant [3] and the murine Thyl-B mutant, which is defective in mannose addition to the GPI glycan core 1331 and genes encoding related proteins were subsequently identified in the S. cereaisiue genome 12, 53, 541. Yeast Alg9p and Algl2p act in the dolichol pathway to add, respectively, an al,2-linked Man to the wl,3-Man transferred by Alg3p, and an a1,6-Man to the GDPMan-derived al,6-Man [3, 261. Neither is essential for viability. Pig-Bp, whose counterpart in S. cerevisiue is the essential Gpi10 protein [32, 53, 541, adds the third, wl,2-linked mannose during assembly of GPI precursor. Intriguingly. the S. cerevisiae genome encodes a fourth Pig-Bp homolog, Smp3p, which like GpilOp is

136

8 Munnosyltrunsj2rases

essential for growth. smp3 mutants are blocked in GPI anchoring and accumulate a Man3-containing GPI [75],suggesting that Smp3p adds the fourth, al,2-Man that has been detected on the complete yeast GPI precursor [55]. Smp3p has a homolog in fission yeast but, unlike the other Alg9p/Pig-Bp family members, Smp3p has no obvious counterpart in Cuenorhabditis elegans or, so far, mammals. If an essential fourth mannosylation step in GPI biosynthesis is confined to eukaryotic microbes and absent from mammalian cells, it could be exploited as a target for antifungal or antiparasitic agents. Hydropathy analyses of Alg9p/Pig-B family members indicate the presence of multiple potential transmembrane segments. However, experimental determination of the transmembrane topology of Pig-Bp reveals the protein has a short cytoplasmic domain, a single membrane-spanning segment, and a large lumenal, catalytic domain [33].Whether the other family members adopt a similar membrane topology is not yet known. Alg9p/Pig-B have no obvious homologs in eubacteria, but weak similarity can be detected to predicted Archaeaglohus and Py~ococcusproteins. Pmtlp family Dol-P-Man-dependent, 0-mannosylation of protein has been intensively studied in yeast [ 5 ] .In this pathway, a first Man is transferred in the ER from Dol-P-Man to serine or threonine, with inversion of configuration to form an a-mannosyl linkage. Up to four further a-Man residues are added in GDPMan-dependent reactions in the Golgi. Transfer of the first Man is catalyzed in the ER lumen by Pmt proteins, seven of which are encoded in the S. cerevisiae genome. The Pmtp’s have multiple transmembrane domains (seven in the case of Pmtlp [56])and a lumenal catalytic domain, and these proteins can form heterodimers. Studies using synthetic peptides indicate that threonine is a better mannosyl acceptor in vitro, but the in vivo acceptor preferences of the individual Pmtps are not yet known. The existence of multiple Pmtps with differing acceptor specificities, coupled with the potential for these proteins to form heterodimers, may permit a wider range of peptide sequences to be recognized as substrates. However, deletion experiments with yeast P M T genes show that in some cases, a given Pmtp may be predominantly responsible for 0-mannosylation of an individual protein, but that in others, several Pmtps have overlapping functions in 0-mannosylation of a protein. 0-mannosylation of protein is essential for viability in S. cereuisicre, and a deficiency in only one Pmt homolog in the fungal pathogen Candidcr ulbicans lowers virulence [57].Were 0-mannosylation confined to fungi, then the Pmtps could be exploited as targets for antifungal agents. However, Pmt-related proteins are encoded in nematode, insect, and mammalian genomes, and 0-linked mannose is found on mammalian proteins, raising the possibility that these organisms also carry out Pmtp-dependent 0-mannosylation. 8.4.2 Golgi Man-Ts and Fungal Mannan Synthesis

In S. cereuisiue, N-linked saccharides can be extended into a “core” structure containing no more than 15 sugars, or into an “outer chain” containing up to 200 mannoses and consisting of an ctl,6-linked polymannose backbone with short al,2-

8.4 Mun-T Futnilirs and the Patlztcuys They Participate in

137

linked Man side-branches that end in a1,3-linked Man [58]. These side chains can also bear Man linked through phosphate [59].Assembly of the outer chain is carried out in the lumen of Golgi compartments by GDPMan-utilizing Man-T, and members of five families of Type TI membrane proteins are involved in these mannosylation reactions. With the exception of the Mnnl 0p family, these proteins have no obvious homologs outside the fungi. An increasingly complete understanding of yeast mannan assembly is emerging [4, 37, 60-651. In brief, mannan formation is initiated on the asparagine-linked GlcNAc2Mans core by addition of an u1,6-Man by Ochlp [60]. A short polymer of al,6-Man is then elongated on this mannose by a complex of Mnn9p and Vanlp (“M-Pol I”) 162, 631. Some a1,2 side branches may be added at this stage, but the protein responsible has not been identified. The backbone is then extended to a final length of about 50 u1,6-Man by M-Pol 11, a complex containing Mnn9p, Anplp, Hoclp, MnnlOp, and M n n l l p that has w1,6 Man-T activity 162, 631. Of the five members of M-Pol 11, related MnnlOp and M n n l l p are responsible for most of the a1,6 mannose polymerizing activity [63]. Further w1,2 side branching residues are added, then extended, by Mnn2p and Mnn5p respectively [64], and the chains are terminated with ul,3-Man by Mnnlp. Man-P can be added to the first u1,2 branching Man by Ktr6p [59]. Five members of the Ktrlp family (Ktrlp, K t d p , Ktr3p, Kre2p, and Yurlp) also seem to participate in formation of some of the u1,2 linkages of the outer chain [65]. Members of two of the Man-T families also participate in the extension of 0linked mannooligosaccharides [65]. Ktr Ip, Ktr3p and Kre2p are involved in adding the second, ul,2-Man to the serine or threonine-linked Man transferred by Pmtps, and these three also participate in addition of the third al,2-Man, though Kre2p’s role is the dominant one at this step. Mnnlp then adds one, and possibly the second of the two ul,3-Man that terminate the oligosaccharide. Ktr6p/Mnn6p can attach a Man in phosphodiester linkage to the second Man of the 0-linked chain [59]. The functioning of all these Man-T in the lumen of the Golgi is dependent on the supply of GDPMan from the cytoplasm, and the essential Vrg4 protein is a strong candidate for the GDPMan transporter [4].Golgi Man-T (including Ochlp, Mnnl p, and Kre2p) have been localized to specific compartments of this organelle, but it seems the signals for their targeting are complex: cytoplasmic, transmembrane, and lumenal domains of these proteins may all have a role in the process [65-671.

Ochlp family Members of this “DXD”-containing group show considerable diversity of function. Ochlp, the outer chain-initiating ul,6-Man-T [60], can use GlcNAcZMang as acceptor, but also transfers Man to GlcNAcZMang. Ochlp has three sequence homologs: Hoclp [4] is a member of the u1,6 mannan-elongating complex M-Pol 11 (section 8.4.2), and Surl p is likely to transfer Man to inositol phosphoceramide 1681. The function of the fourth family member, YBR16lwp, is unknown. Mnn9p family This contains the Mnn9, Anpl (MnnS), and Van1 proteins, which have the “DXD” motif and are members of the M-Pol I and M-Pol I1 complexes that elongate the

1 38

8 Munnosyltrunsferases

a1,6 polymannose backbone and add some al,2-branching mannoses (see above). It is not yet known whether each one of these proteins catalyzes mannosyl transfer. MnnlOp/Mnnl l p These two proteins, which both have a “DXD” motif, are responsible for the majority of the al,6-mannose polymerizing activity of M-Pol 11, because deletion of the gene for either protein abolishes most of the activity of the isolated complex [63]. Intriguingly, MnnlOp and M n n l l p are sequence homologs of a Schizosaccharomyces pombe al,2-galactosyltransferase [4, 631, but the basis for the apparent specificity of the bakers yeast proteins for GDPMan, and the fisson yeast homologs for UDPGal is unknown. MnnlOp/Mnnl l p homologs are also encoded in the C. eleyans and Arabidopsis genomes [35]. Mnnlp family The six members of this “DXD”-containing family of S. cerevisiae Man-T include an al,3-Man-T (Mnnlp) and two al,2-Man-T (Mnn2p and Mnn5p) [ 3 5 , 64, 651. The former adds Man to both N- and O-linked mannose chains, whereas the latter seem specific for the N-linked outer chain. The linkage-specificity and function of the remaining three homologs [65]are as yet unknown. Mnnlp may form multimers in vivo [35]. Ktrlp family

So far, the members of this family of nine proteins exhibit one of two biochemical activities. Ktrlp, K t d p , Ktr3p, Kre2p, and Yurlp have been shown or are likely to be al,2-Man-T [65]. Ktr6p/Mnn6p, however, transfers Man-P to oligosaccharides containing at least one al,2-linked mannobiosyl unit [59]. Of these six proteins, Ktrlp, Ktr3p, Kre2p, and Ktr6p/Mnn6p transfer Man to both N- and O-linked chains, whereas Ktr2p and Yurlp seem to act only on N-linked outer chains. The functions of the remaining three proteins, Ktr4p, KtrSp, and Ktr7p, are as yet unknown. 8.4.3 “Missing” Eukaryotic Man-T

The conserved glycan biosynthetic pathways in eukaryotes include a number of mannosylation steps for which the responsible genes have not been identified. The “missing” Man-Ts of the dolichol pathway include three GDPMan-dependent Man-T and the fourth Dol-P-Man-dependent al,2-Man-T. Genes for the Dol-PMan-utilizing Man-T that add the first, al,4-Man, and the second, a1,6-Man in GPI assembly are yet to be cloned, although the former enzyme’s activity has been extensively characterized [32]. It will be interesting to learn whether these proteins are distant relatives of existing Man-T, or whether they will represent new classes of glycosyltransferases. Subsets of eukaryotes may have Man-T that have specialized functions. Of interest because of the novelty of the linkage they catalyze are, for example, the presumed Dol-P-Man-requiring C-mannosyltransferase, and the five

8.5 coordinating Man Transjer with the Cell Cycle and Morphogenesis

139

P-Man-T activities involved in modifying cell wall mannan of the fungal genus Candida [69], which includes the human pathogen C. albicans. In Saccharomyces cereuisiae, the genes for the Man-T that add additional a-linked Man to proteinbound GPI anchors in the Golgi may yet be found in the MnnIp or KtrIp families. 8.4.4 Eubacterial and Archaeal Man-T So far, potential eubacterial and archaeal Man-T seem to fall into either the “PMan-T” [34] or the “a-Man-T” (“E(X7)E”) [18] classes. Members of both families participate in LPS synthesis in eubacteria as well as in cytoplasmic, sugar nucleotide-dependent steps in the dolichol-linked pathway for N-glycosylation in the ER. Indeed, the similarity between the biochemical steps and enzymes in these two “polyisoprenoid-linked” pathways strongly suggests that they have a common evolutionary origin [3]. GDPMan-dependent a-Man-T serve as donors in the assembly of the polyprenolPP-linked oligosaccharide units that are in turn polymerized-by incorporation of successive, newly generated oligosaccharide repeats at the reducing end of growing polyprenol-PP-linked polysaccharide chains-to form O-antigen repeats [70]and Xanthan gum [13]. The genomes of eubacteria and archaea encode homologs of eukaryotic Dol-P-Man synthase, but little is known about the function of these proteins. An exception is the involvement of polyprenol-P-Man in lipoarabinomannan synthesis in mycobacteria. This process is initiated by two GDPManutilizing Man-T that use PI as acceptor and create PI(Man)z, upon which linear a I ,6 mannan chains are then elongated in polyprenol-P-Man-utilizing reactions to form lipomannan. a1,2-Man branches are transferred to this backbone from GDPMan, and this structure in turn serves as acceptor for transfer of Araffrom polyprenol-P-Araf to complete the polymer [ 141. A candidate for a mycobacterial GDPMan: PI a1,6 Man-T gene, mtjB, encodes a protein of the “E(X7)E” class.

8.5 Coordinating Man Transfer with the Cell Cycle and Morphogenesis Studies in yeast reveal that the levels and activities of Man-Ts and other glycosyltransferases are highly regulated so as to be coordinated with events in the cell cycle. Thus, transcription of ALG genes, as well as of the genes for GDPMan pyrophosphorylase ( P S A 1 ) and the presumed GDPMan transporter (VRG4) is cell cycle-regulated [4, 711, and there are genetic interactions between some ER and Golgi Man-T genes and genes that encode components of the protein kinase C pathway or which regulate morphogenesis [2, 4, 72, 731. The biochemistry that underlies these genetically identified interactions and explains the connection between glycosylation state and morphogenesis is largely unknown. Maintenance of an adequate GDPMan supply is a priori important, and indeed, a critical determinant of Man-T activity in the Golgi. Thus, mutants in GMP kinase (Guklp), in GDPMan pyrophosphorylase (Psa 1p), in the presumed GDPMan transporter

140

8 Munnosyltvunsferuses

(Vrg4p), and in Gdalp (the GDPase that generates GMP, the antiporter for GDPMan transport into the Golgi), all show decreases in Golgi mannosylation [4, 72-74]. Consistent with the notion that GDPMan levels dictate the extent of Golgi Man-T activity, the G M P kinase deficiency can be partially suppressed by overexpressing GDPMan pyrophosphorylase [ 741. Phosphomannose addition to yeast mannan, which occurs at a late stage in cell growth and in response to high osmolarity, is positively regulated by a putative Ser/Thr protein kinase encoded by the stress-responsive MNN4 gene [59].

8.6 Concluding Remarks Man-T have key roles in the biosynthesis of cell surface glycans in eubacteria, archaea, and eukaryotes. Some participate in highly conserved biosynthetic pathways, but others may have assumed specialized functions in the synthesis of glycan structures peculiar to smaller groups of organisms. The availability of recombinant Man-T’s, coupled with advances in chemical synthesis of increasingly elaborate acceptor structures will allow the substrate-specificity of individual Man-T to be analyzed in detail. The challenge will then be establish how different Man-T recognize GDPMan or Dol-P-Man, and how they achieve high specificity for their acceptor glycan structures. The availability of the sequences of many Man-Ts and related proteins with different substrate specificities should now permit the design of site specific mutagenesis and domain-swap experiments to home in on the amino acids important for catalysis and substrate-specificity. Our ability to express large amounts of recombinant Man-T in bacteria should make attempts at determining these proteins’ three-dimensional structures possible. Some of the specialized Man-T of eubacteria and eukaryotic microbes may be essential for cell viability or for construction of cell surface components that are virulence determinants, and as such, these Man-T present good targets for new antimicrobial agents. A detailed knowledge of the structure-function relationships of these Man-T, coupled with the development of assays for use in high through-put screens, should facilitate the design and identification of Man-T inhibitors. Acknowledgments

Work in my laboratory has been supported by grants from the National Institutes of Health and the Burroughs Wellcome Fund, and by a Helen Corley Petit Professorship from the University of Illinois. References I . J. Takeda and T. Kinoshita, GPI anchor biosynthesis, Trends Biochenz. Sci. 1995,ZO; 3677371. 2. P. Orlean, Biogenesis of Yeast Wall and Surface Components, in The Molecular and Cellular Biology of the Yeast Saccharomyces, Cell C j d e and Cell Bidog.v, Volume 3, edited by J.R.

Refirences

I41

Pringle, J.R. Broach, and E.W. Jones, 1997, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York, pp. 229-362. 3. P. Burda and M. Aebi, The dolichol pathway of N-linked glycosylation, Biochim. Biophys. Actu. 1999. 1426. 239-257. 4. N. Dean, Asparagine-linked glycosylation in the yeast Golgi, Biochim. Biophys. Acta 1999, 1426. 309-322. 5 . S. Strahl-Bolsinger, M. Gentzsch, and W. Tanner, Protein 0-mannosylation, Biochim. Biophys. Actn, 1999, 1426, 297-307. 6. G. Piro, A. Zuppa, G. Dallesandro, and D.H. Northcote, Glucomannan synthesis in pea epicotyls: The mannose and glucose transferases, Planta 1993, 190, 206-220. 7. J.S.G. Reid, M. Edwards, M.J. Gidley, and A.H. Clark. Enzyme specificity in galactomannan biosynthesis, Planta 1995, 195, 489-495. 8 . A.K. Bhattacharjee, K.J. Kwon-Chung, and C.P. Glaudemans, Capsular polysaccharides of Cryptococcus neoformuns, Rev. In$ Dis. 1984, 6 , 619-624. 9. M.J. McConville and M.A.J. Ferguson, The structure, biosynthesis and function of glycosylated phosphatidylinositols in the parasitic protozoa and higher eukaryotes, Biochem. J. 1993, 294, 305-324. 10. R.L. Lester and R.C. Dickson, Sphingolipids with inositolphosphate-containing head groups, A&. Lipid. Res. 1993, 26, 253-74. I 1. M. Scher, W.J. Lennarz, and C. Sweeley, The biosynthesis of mannosyl-l-phosphorylpolyisoprenol in Micrococcus lysodeikticus and its role in mannan synthesis, Proc. Natl Acad. Sci. USA 1968,59, 1313-1320. 12. C.R.H. Raetz, Bacterial lipopolysaccharides: a remarkable family of bioactive macroamphiphiles, in Escherichiu coli and Sulmonellu, Cellular and Molecular Biology, 2nd Edition, Volume I, Edited by F.C. Neidhardt, 1996, American Society for Microbiology, Washington, DC, pp. 1035-1063. 13. L. Ielpi, R.O. Couso, and M.A. Dankert, Sequential assembly and polymerization of the polyprenol-linked pentasaccharide repeating unit of the Xanthan polysaccharide in Xunthomonas campestris, J. Bucteriol. 1993, 17.5, 2490 -2500. 14. G.S. Besra, C.B. Morehouse, C.M. Rittner, C.J. Waechter, and P.J. Brennan, Biosynthesis of mycobacterial lipoarabinomannan, J. Biol. Chem. 1997, 272, 18460-1 8466. 15. L.L. Yang and A. Haug, Purification and partial characterization of a procaryotic glycoprotein from the plasma membrane of Thermoplusma acidophilun?, Biochim. Biophys. Acta 1979, 556, 265-277. 16. M.A. Doucey, D. Hess, R. Cacan R., and J. Hofsteenge, Protein C-mannosylation is enzyme catalyzed and uses dolichyl-phosophate-mannose as a precursor, Mol. Biol. Cell 1998, 9, 291300. 17. J. Krieg, S. Hartmann, A. Vicentini, W. Glasner, D. Hess, and J. Hofsteenge, Recognition signal for C-mannosylation of Trp-7 in RNase 2 consists of sequence Trp-x-x-Trp, Mol. Biol. Cell 1998, 9, 301-309. 18. Gereniia, R.A., Petroni, E.A., Ielpi, L.. and Henrissat, R., Towards a classification of glycosyltransferases based on amino acid sequence similarities: prokaryotic a-mannosyltransferases. Biochem. J. 1996, 318, 133-138. 19. N. Kido, T. Sugiyama, T. Yokochi, H. Kobayashi, and Y. Okawa, Synthesis of Escherichia coli 0 9 a polysaccharide requires the participation of two domains of WbdA, a mannosyltransferase encoded within the WB* gene cluster, Mol. Microhiol. 1998, 27, 1213-1221. 20. W. McDowell and R.T. Schwarz, Specificity of GDP-Man:dolichyl-phosphate mannosyltransferase for the guanosine diphosphate esters of mannose analogues containing deoxy and deoxyfluoro substituents, FEBS Lett. 1989, 243,413-416. 21. A. Mudgapalli, S.K. Roy, E.H. Holmes, and Vijay, I.K. Photoidentification of mannosyltransferases of the dolichol cycle in the mammary gland-purification and characterization of GDPMan Man-fi-l,4GlcNAc-PP-dolichol mannosyltransferase, J. Bid. Chem. 1994, 269, 11327-11336. 22. J.L. Kadrmas, K.A. Brozek, and C.R.H. Raetz, Lipopolysaccharide core glycosylation in Rhizohium legurninusarum. An unusual Man-T resembling the heptosyltransferase I of Escherichia coli, J. Biol. Chem. 1996, 271, 32119-32125.

142

8 Munnosyltrunsferases

23. J.S. Rush, J.G. Schelling, N.S. Zingg, P.H. Ray, and C.J. Waechter, Mannosylphosphoryldolichol-mediated reactions in oligosaccharide-P-P-dolicholbiosynthesis. Recognition of the saturated alpha-isoprene unit of the mannosyl donor by pig brain mannosyltransferases, J. Biol. Chem. 1993, 268, 131 10- 131 17. 24. C . D’Souza-Schorey, K.R. McLachlan, S.S. Krag, and A.D. Elbein, Mammalian glycosyltransferases prefer glycosyl phosphoryl dolichols rather than glycosyl phosphoryl polyprenols as substrates for oligosaccharyl synthesis, Arch. Biochem. Biophys. 1994, 308, 497-503. 25. S.B. Dotson, J.S. Rush, A.D. Ricketts, and C.J. Waechter, Mannosylphosphoryl-dolicholmediated 0-mannosylation of yeast glycoproteins: stereospecificity and recognition of the alpha-isoprene unit by a purified mannosyltransferase, Arch. Biochem. Biophys. 1995, 316, 773779. 26. P. Burda, C.A. Jakob, J. Beinhauer, J.H. Hegemann. and M. Aebi, Ordered assembly of the asymmetrically branched lipid-linked oligosaccharide in the endoplasmic reticulum is ensured by the substrate specificity of the individual glycosyltransferases, Glycohioloyy 1999,9, 617-625. 27. G. Palamarczyk, L. Lehle, T. Mankowski, T. Chojnacki, and W. Tanner, Specificity of solubilized yeast glycosyl transferases for polyprenyl derivatives, Eur. J. Biochem. 1980, 105, 5 17523. 28. S. Kato, M. Tsuji, Y. Nakanishi, and S. Suzuki, Phenyl phosphate as a water-soluble analog of dolichyl phosphate in microsomal mannosyltransferase systems, J. Biochem. (Tokyo) 1980, 87. 929--939. 29. I.B.H. Wilson, M.C. Webberley, L. Revers. and S.L. Flitsch. Dolichol is not a necessary moiety for lipid-linked oligosaccharide substrates of the mannosyltransferases involved in in vitro N linked oligosaccharide assembly, Biochem. J. 1995, 310, 909-916. 30. D. Grassi, V. Lippuner, M. Aebi, J. Brunner, and A. Vasella, Synthesis and enzymatic phosphorylation of a photoactivatable dolichol analog, J. Amer. Chem. Soc. 1997, 119, 1099210999. 31. W.T. Doerrler, J.H Ye, J.R. Falck, and M.A. Lehrman, Acylation of glucosaminyl phosphatidylinositol revisited-palmitoyl CoA-dependent palmitoylation of the inositol residue of a synthetic dioctanoyl glucosaminyl phosphatidylinositol by hamster membranes permits efficient mannosylation of the glucosamine residue, J. Biol. Chem. 1996,271, 27031-27038. 32. T.K. Smith, D.K. Sharma, A. Crossman, A. Dix, J.S. Brimacombe, and M.A.J. Ferguson, Parasite and mammalian GPI biosynthetic pathways can be distinguished using synthetic substrate analogs, EMBO J. 1997, 16, 666776675, 33. M. Takahashi, N. Inoue, K. Ohishi, Y. Maeda, N. Nakamura, Y. Endo, T. Fujita, J. Takeda, and T. Kinoshita, PIG-B, a membrane protein of the endoplasmic reticulum, with a large lumenal domain, is involved in transferring the third mannose of the GPI anchor, EMBO J. 1996, 15, 4254-4261. 34. I.M. Saxena, R.M.J. Brown, M. Fevre, R.A. Geremia, and B. Henrissat, Multidomain architecture of P-glycosyltransferases: implications for mechanism of action. J. Bucteriol. 1995, 177, 1419- 1424. 35. C.A.R. Wiggins and S. Munro, Activity of the yeast M N N I a- 1,3-mannosyl-transferase requires a motif conserved in many other families of glycosyl-transferases, Proc. Nut1 Acud. Sci. USA, 1998, 95, 7945-7950. 36. J. H. Kim, D. W o k , K. Haridas, R. J. Parry, J. L. Smith, and H. Zalkin, A stable carbocyclic analog of 5-phosphoribosyl-1-pyrophosphate to probe the mechanism of catalysis and regulation of glutamine phosphoribosyl-pyrophosphate amidotransferase, J. Biol. Chem. 1995, 270. 17394-1 7399. 37. C.E. Ballou, Isolation, characterization, and properties of Succhuromyces cerevisiue mnn mutants with non-conditional protein glycosylation defects, Met. Enzymol. 1990, 185, 440-470. 38. T. C. Huffaker and P.W. Robbins, Temperature-sensitive yeast mutants deficient in asparaginelinked glycosylation, J. Biol. Chem. 1982, 257, 3203--3210. 39. R. Hyman, Somatic genetic analysis of the expression of cell surface molecules, Trends. Genet 1988, 4 , 5-8. 40. J. Stoll, A.R. Robbins, and S.S. Krag, Mutant of Chinese hamster ovary cells with altered mannose-6-phosphate receptor activity is unable to synthesize mannosylphosphoryldolichol, Proc. Natl Acud. Sci. USA 1982, 79, 2296-2300.

References

143

41. C. Levrat, D. Ardail, R. Morelis, and P. Louisot, Study of the N-glycoprotein biosynthesis through dolichol intermediates in the mitochondria1 membranes, Int. J. Biochem. 1989, 21, 265-278. 42. J.R. Couto, T.C. Huffaker, and P.W. Robbins, Cloning and expression in Esclzerichiu coli of a yeast mannosyltransferase from the asparagine-linked glycosylation pathway, J. Biol. Chem. 1984, 10,259, 378-382. 43. E.A. Petroni and L. Ielpi L. Isolation and nucleotide sequence of the GDP-mannose:cellobiosyl-diphosphopolyprenolalpha-mannosyltransferase gene from Acetohacter xylinum, J. Bucteriol. 1996, 178, 4814-4821. 44. J.L. Kadrmas, D. Allaway, R.E. Studholme, J.T. Sullivan, C.W. Ronson, P.S. Poole, and C.R.H. Raetz, Cloning and overexpression of glycosyltransferases that generate the lipopolysaccharide core of Rhizohium leguminosarum, J. Bid. Chem. 1998,273, 26432-26440. 45. B.C.R. Zhu and R.A. Laine, Dolichyl phosphomannose synthase from the Archae Thermoplasma acidophilum, Glycobiology 1996, 6, 8 I 1-8 16. 46. P.A. Colussi, C.H. Taron, J.C. Mack, and P. Orlean, Human and Succhuromyces cereuisiue dolichol phosphate mannose synthases represent two classes of the enzyme, but both function in Sc~iizosuccharomyee~s pun&, Proc. Nut1 Acad. Sci. USA 1991, 94, 7873-7878. 47. S. Tomita, N. Inoue, Y. Maeda, K. Ohishi, J. Takeda, and T. Kinoshita, A homologue of Succharomyces cerevisiue Dpml p is not sufficient for synthesis of dolichol-phosphate-mannose in mammalian cells, J. Biol. Chem. 1998, 272, 9249-9254. 48. P. Orlean. C. Albright, and P.W. Robbins, Cloning and sequencing of the yeast gene for dolichol phosphate mannose synthase, an essential protein, J. Biol. Chem. 1988, 263, 17499-17507. 49. Y. Maeda, S. Tomita, R. Watanabe. K . Ohishi, and T. Kinoshita, DPM2 regulates biosynthesis of dolichol phosphate-mannose in mammalian cells: correct subcellular localization and stabilization of DPM1, and binding of dolichol phosphate, E M B O J . 1998, 17, 49204929. 50. S . Te Heesen, L. Lehle, A. Weissman, and M. Aebi, Isolation of the ALGS locus encoding the UDP-g1ucose:dolichyl-phosphate glucosyltransferase from Succharomyces cereuisiae, Eur. J. Biochenz. 1994,224, 71-79. 5 I , J.W. Jensen and J.S. Schutzbach, Modulation of dolichyl-phosphomannose synthase activity by changes in the lipid environment of the enzyme, Biochemistry 1988,27, 63 15-6320. 52. D.K. Banerjee, E.E. Kousvelari, and B.J. Baum, CAMP-mediated protein phosphorylation of microsomal membranes increases mannosylphosphodolichol synthase activity, Proc. Nut1 Acad. Sci. USA 1987, 84, 6389-6393. 53. E. Canivenc-Gansel, 1. Imhof, F. Reggiori F. P. Burda, A. Conzelmann: and A. Benachour, GPI anchor biosynthesis in yeast: phosphoethanolamine is attached to the al,4-linked mannose of the complete precursor glycophospholipid, Glycohioloyy 1998, 8, 761-770. 54. C. Sutterlin, M.V. Escribano, P. Gerold P, Y. Maeda, M.J. Mazon, T. Kinoshita, R.T. Schwarz, and H. Riezman, Succhmromyce.s cerevisiue GPIIO, the functional hoinologue of human PIG-B, is required for glycosylphosphatidylinositol-anchor synthesis, Biochem. J. 1998, 332, 153-159. 55. G. Sipos, A. Puoti, and A. Conzelmann, Glycosylphosphatidylinositol membrane anchors in Succhnrornyces cerevisiue: absence of ceramides from complete precursor glycolipids, EMBO J. 1994, 13, 2789-2796. 56. S. Strahl-Bolsinger and A. Scheinost, Transmembrane topology of pmtlp, a member of an evolutionarily conserved family of protein 0-mannosyltransferases, J. Biol. Chem. 1999, 274: 9068-9075. 57. C. Timpel, S. Strahl-Bolsinger, K. Ziegelbauer, and J.F. Ernst, Multiple functions of Pmtlmediated protein 0-mannosylation in the fungal pathogen Ccindida alhicans, J. B i d . Chem. 1998,273, 20837-20846. 58. T. R. Gemmill and R. B. Trimble, Overview of N - and 0-linked oligosaccharide structures found in various yeast species, Biochiin. Biophj,s. Actu 1999, 1426, 227-237. 59. Y. Jigami and T. Odani, Mannosylphosphate transfer to yeast mannan, Biochim. Biophys Acta 1999, 1426, 335-345. 60. Y. Nakanishi-Shindo, K. Nakayama, A. Tanaka, T. Toda, and Y. Jigami, Structure of the Nlinked oligosaccharides that show complete loss of a-l,6-polymannose outer chain from ochl,

144

8 Munnosyltran.~eruse,s

ochl mnnl, and ochl mnnl ulg3 mutants of Succharomyces cerevisiue, J. Biol. Chem. 1993,268, 26338-26345. 61. H. Hashimoto and K. Yoda, Novel membrane protein complexes for protein glycosylation in the yeast Golgi apparatus, Biochem. Biophys. Rex Comniun. 1997, 241, 682~686. 62. J. Jungmann and S. Munro, Multi-protein complexes in the cis Golgi of Saccharomyces cereuisiae with u-1,6-mannosyltransferaseactivity, EMBO J. 1998, 17, 423-34. 63. J. Jungmann, J.C. Rayner, and S. Munro, The Succhuromyces cerevisiae protein MnnlOp/ Bedlp is a subunit of a Golgi mannosyltransferase complex, J. Biol. Chem. 1999, 274. 65796585. 64. J.C. Rayner and S. Munro, Identification of the MNN2 and MNN5 mannosyltransferases required for forming and extending the mannose branches of the outer chain mannans of Sacchuromyces cerevisiae, J. Biol. Clzem. 1998, 273, 26836--26843. 65. M. Lussier, A.-M. Sdicu, and H. Bussey, The KTR and M N N l mannosyltransferase families of Saccharomyces cereuisiue, Biochim. Biophys Acta 1999, 1426, 323-334. 66. T.R. Graham and V.A. Krasnov, Sorting of yeast u-1,3 mannosyltransferase is mediated by a luminal domain interaction, and a transmembrane domain signal that can confer clathrindependent Golgi localization to a secreted protein, Mol. Biol. Cell 1995, 6, 809-824. 67. S.L Harris and M.G. Waters, Localization of a yeast early Golgi mannosyltrans-ferase, Ochlp, involves retrograde transport, J. Cell Bid. 1996, 132, 985- 998. 68. T.J. Beeler, D. Fu, J. Rivera, E. Monaghan, K. Gable, and T.M. Dunn, SURl iCSGl/ BCL21), a gene necessary for growth of Succhuromyces ceveuisiue in the presence of high Ca2+ concentrations at 37 "C, is required for mannosylation of inositolphosphoceramide, Mol. Gen. Genet. 1997,255, 570-579. 69. A. Suzuki, N . Shibata, M. Suzuki, F. Saitoh, H. Oyamada, H. Kobayashi, S. Suzuki, and Y. Okawa, Characterization of p-1,2-mannosyltransferasein Cundidu guilliermondii and its utilization in the synthesis of novel ohgosaccharides. J. Biol. Chem. 1997, 272, 16822-16828. 70. P.W. Robbins, D. Bray, M. Dankert, and A. Wright, Direction of chain growth in polysaccharide synthesis, Science 1967, 158, 1536--1542. 71. M.A. Kukuruzinska and K. Lennon-Hopkins, ALG gene expression and cell cycle progression, Biochim. Biophys. Acta 1999, 1426, 359-372. 72. B.K. Benton, S.D. Plump, J. Roos, W.J. Lennarz, and F.R. Cross, Over-expression of S. cereuisiac G1 cyclins restores the viability of ulgl N-glycosylation mutants. Curr. Genet. 1996, 29, 106- 1 1 3. 73. G. Montdesert, D.J. Clarke, and S.I. Reed, Identification of genes controlling growth polarity in the budding yeast Succhuromyces cerevisiue: a possible role of N-glycosylation and involvement of the exocyst complex, Genetics 1997, 147, 421-434. 74. Y. Shimma, A. Nishikawa, B. bin Kassim, A. Eto, and Y. Jigami, A defect in GTP synthesis affects mannose outer chain elongation in Saccharomyces cereuisiue, Mol. Gen. Genet. 1997, 256, 469-480. 75. S. Grimme and P. Orlean, unpublished.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

9 Branching of N-Glycans: N-Acetylglucosaminyltransferases Hurry Schachtev

9.1 Introduction Whereas proteins and nucleic acids, the major biological macromolecules, are linear polymers in which the building blocks, amino acids and nucleotides, are joined together respectively by identical amide and 3’,5’-phosphodiester bonds, glycan polymers are often branched. To add to the structural complexity of glycans, the monosaccharide building blocks may be joined to one another in either alpha or beta glycosidic linkages connecting the anomeric carbon of one sugar to one of several different carbon positions on the adjoining sugar. Protein and nucleic acid polymers are assembled by a template approach in which every new molecule is copied with great accuracy from a pre-existing molecule. However, glycan polymer biosynthesis cannot use such a method and requires an assembly line. Proteinbound glycans are assembled along the endoplasmic reticulum and Golgi apparatus [ 1 ] by a series of glycosyltransferases, processing glycosidases and modifying enzymes (sulfotransferases, acyltransferases, etc). These enzymes are attached to the membranous assembly line with their catalytic domains within the lumen and act sequentially on the growing protein-bound glycan as the glycoprotein moves within the lumen of the endomembrane system from the rough endoplasniic reticulum to one of several destinations (cell surface, secretory product or lysosome). This Chaptcr will focus on one aspect of this assembly line process: the N acetylglucosaminyltransferases which initiate the branches of complex N-glycans. The discussion will deal mainly with relatively recent work on these enzymes and the reader is referred to other sections of this book and to previous comprehensive reviews for other aspects of N-glycan assembly [l-171. All N-glycans share the same pentasaccharide core structure Manu1-6(Manal-3). ManP1-4GlcNAcP 1-4GlcNAc-Asn-X. The oldest structures from an evolutionary point of view are called oligomannose N-glycans in which 1-6 additional Man residues are attached to the core. Complex N-glycans carry 2 5 branches or “antennae” attached to the core structure. Every antenna is initiated by a specific

146

9 Brunching of N-Glycuns: N-Ac.etylglucosuminyltrunsferuses

GnT V

GlcNAcpl-6

GnT VI

GlCNACpl - 4

GnT I1

GlcNAcpl-2-Mjranctl-6

GnT I11

GlcNAcpl-4

GnT I

GlcNAcpl-2-Manctl-3

GnT IV

GlcNAcpl - 4

4

>-p.-4

-R

/

Figure 1. GnT I to VI incorporate GlcNAc residues into the Manctl-6[Manal-3]ManP-R N-glycan core.

N-acetylglucosaminyltransferase (GnT I, 11, IV, V and VI, Figure 1) which adds a GlcNAc in pl-2, pl-4 or pl-6 linkage to one or other of the terminal Man residues of the core. Once the GlcNAc has been added, various types of elongation may occur: i) Sialyla2-6(or a2-3)GalPl-4GlcNAc-; ii) Fucal-2(or Galal-3)Galpl-4GlcNAc-; iii) Galpl-4 may be replaced by Galpl-3 or GalNAcPl-4; iv) Fuc may be added in al-3 or 1x1-4 linkage to GlcNAc; v) antennae may be sulfated or terminated with various antigenic epitopes such as the human blood group ABO, H and Lewis structures [ 181 and the poly-N-acetyllactosamine-containing i and I epitopes [ 191. Structures which carry one or two antennae on the Manal-3 arm of the core and only Man residues on the Manal-6 arm are termed hybrid N-glycans. Both hybrid and complex N-glycans may be “bisected” by a GlcNAc residue linked 01-4 to the p-linked Man residue of the core; this residue is incorporated by GnT 111 (Figure 1) and is not further substituted.

9.2 Processing of N-Glycans within the Endomembrane Assembly Line There are many factors which control and integrate the biosynthetic assembly line [ l , 3 , 20-221: i) synthesis of precursors such as nucleotide-sugars and dolichollinked sugars in the cytoplasm and their transport into the lumen of the endomembrane assembly line; ii) synthesis of the glycosyltransferases, glycosidases and other enzymes attached to the assembly line and their targeting to the endoplasmic reticulum and Golgi apparatus; iii) “substrate-level” control factors such as competition between assembly line enzymes for common substrates, the substrate specificities of these enzymes and factors such as cations and pH which control enzyme activity; iv) “gene-level’’control factors which determine the tissue- and time-specific expression of enzymes on the assembly line; v) the roles of residence time in the endomembrane lumen and of polypeptide and oligosaccharide conformation on the accessibility of glycoprotein to modifying enzymes; and vi) “quality control” mechanisms such as binding of glycoproteins to calnexin and calreticulin.

9.2 Processing oj N-G1ycun.s bcitlzin the Endomembrane Assembly Line

147

The synthesis of N-glycans begins within the ER lumen with the transfer of Glc3MangGlcNAcz from Glc3MangGlcNAcz-pyrophosphate-dolichol to an asparagine residue of the nascent polypeptide chain [6, 71, a reaction catalyzed by a complex enzyme called oligosaccharyltransferase [ 7, 231. The acceptor Asn residue must occur in an Asn-X-Ser/Thr “sequon” where X can be any amino acid except Pro but only about l6Y0 of the potential “sequons” are glycosylated on average. Oligosaccharide processing by glucosidases I and 11, endoplasmic reticulum mannosidase and Golgi mannosidase I forms Man5GlcNAcz-Asn-X (Figure 2), the entry point for the formation of hybrid and complex N-glycans due to the action of UDP-

-

Manal

Oligomannose KGlycans Processing Glycosidases

6‘

IMana\6 Manal

-

1

IM GnT I

Manal

Bisected hybrid KGlycans

6‘

GnT 111

Manal

/3 M a n a K 6

GlcNAcpl-ZManal

/3Ma[-4R

a-mannosidase II (bisecting GlcNAc is a STOP signal)

a-mannosidase II (GnT I is a GO signal) \6 GlcNAcpl-ZManal

/3

Manpl-4R I

1

GnT II

Bi-antennary complex KGlycans

-

GlcNAcpl -2Manal \6

GlcNAcpl-ZManal

/

3Manpl -4R

1

GnT 111 GnT IV GnTV

Multi-antennary complex Kglycans

Figure 2. Conversion of oligomannose N-glycans to hybrid and complex N-glycans. GnT I, UDPGlcNAc:Mana1-3R [GlcNAc to Manal-31 p-I .2-N-acetylglucosaminyltransferase 1. GnT 11, UDPGlcNAc:Manul-6R [GlcNAc to Manal-61 b-I ,2-N-acetylglucosaminyltransferase 11.

148

9 Branching of N-Glycans: N-Acet~l~luco.sumii~yltransfera~ses

GlcNAc:Manal-3R [GlcNAc to Manal-31 ~-1,2-N-acetylglucosaminyltransferase I (GnT I, Figure 2, Table 1). This enzyme step is essential for the subsequent action of several enzymes in the processing pathway (Figures 1 and 2), i.e., a-3/6-mannosidase 11, GnT 11, GnT 111, GnT IV, and the a-1,6-fucosyltransferasewhich adds fucose in a1-6 linkage to the Asn-linked GlcNAc [3, 211. GnT V and VI require the prior action of GnT I1 and therefore, indirectly, GnT I is also required for the subsequent actions of these enzymes. GnT I action is also a prerequisite for the activities of GnT I1 and p1-2-xylosyltransferase in plants [24] and in snail [25]and of the a-1,3fucosyltransferase which adds fucose in al-3 linkage to the Asn-linked GlcNAc in plants and insects [26-301. After the action of a-mannosidase 11, GnT 11, IV, V and VI (Figures 1 and 2) initiate the various complex N-glycan “antennae”. If GnT I11 acts on the product of GnT I before a-mannosidase I1 to form the bisected hybrid structure (Figure 2), the pathway is committed to hybrid structures because a-mannosidase I1 cannot act on bisected oligosaccharides [31]. The reverse order of action leads to complex N glycans. The relative abundance of GnT I11 and a-mannosidase 11 in a particular tissue therefore controls the pathway towards hybrid or complex N-glycans. The route taken at such a divergent branch point is dictated primarily by the relative activities of glycosyltransferases which compete for a common substrate. The insertion of a bisecting GlcNAc by GnT I11 prevents the actions of GnT 11, IV and V, amannosidase I1 and core a-l,6-fucosyltransferaseand is an example of a glycosyl residue acting as a STOP signal whereas the action of CnT I is a G O signal (Figure 2). Competition and STOP and G O signals serve as “substrate-level” controls of the biosynthetic pathways as opposed to “gene-level’’ control factors at the transcriptional or translational levels.

9.3 General Properties of the N-Acetylglucosaminyltransferases The reaction catalyzed by a typical GnT is: UDP-GlcNAc

+ R-OH

4

GlcNAc-0-R

+ UDP

(1)

where the physiological substrate R-OH is a growing protein-bound N-glycan. When these glycosyltransferases are assayed in vitro, R-OH can be a glycopeptide, a free saccharide or a saccharide linked to an aglycone such as a methyl, benzyl or octyl group. Detergent treatment is required for solubilization and full expression of enzymatic activity in vitro. All these enzymes except GnT V require the addition of divalent cation which probably serves to bind the negatively charged nucleotidesugar to the protein. 9.3.1 Domain Structure The GnTs which initiate N-glycan branching (Figure l), like all Golgi-bound glycosyltransferases cloned to date, have an amino acid sequence suggestive of a type

GnT VIII

Not cloned

Not cloned

Initiation of antenna on Man(a1-6) arm

Initiation of antenna on Man(a1-6) arm

2.4.1.155

Novel modification of N-glycan core Novel modification of N-glycan core

GnT VII

GnT VI

GnT V

Initiation of antenna on Man(a1-3) arm

2.4.1.145

Hen oviduct Bovine Human Human GIcNAcPI-6Rat [ GlcNAcP 1-21Mantx 1-6Man P Chinese hamster Chicken GIcNAcP1-4[ G I C N A ~ P ~ - ~ ] [ G ~ C-61N A C PFish ~ Manal-6ManP LEC 14 CHO cells GlcNAcD1-2ManP LECl8 CHO cells GlcNAcP1-6 [ R-ManP 1-4]GlcNAcPl4GlcNAc-Asn-X

GlcNAcP1-4[GlcNAcPl-2]Manal-3ManP

GnT IV

Synthesis of bisected N-glycans

2.4.1.144

GlcNAcfl1-4Mano

Synthesis of biantennary iV-glycans

GnT 111

Human Rat Frog Pig Human Rat Mouse

2.4.1.143

GICNACPI-2Manal-6ManP

GnT I1

Rat Chicken Frog Chinese hamster ovary Baby hamster kidney Cuennrhahditis eleyans

Mouse

M57301 M61829 M55621 X77487-8 M73491 LO7037 D 16302 No Acc.No. No Acc.No. U6579 1-2 AF087456-7 AF082011, AF082010, AF082012 U15128. L36537 U2 1662 No Acc No. YO9537 D13789 D10852 L39373 U66844 Not cloned AB000628 AB0006 16 D17716 L 14284 U62587-8 Not cloned Not cloned

First step towards hybrid and complex N-gl ycans

2.4.1.101

Rabbit Human

GlcNAcPl-2Mana 1-3ManP

GnT I

Acc.No.

Comments

EC No.

Tissue

Enzyme Product

Abbrev.

121 11

121 11

[ 1421 ~071

PI81 [ 1291 ~401 [143, 1441 11451 [21 91 I1701 ~711

P I [ 1081 P 171

[SS] ~711 PI61

[611 ~ 5 1 [2151 PI61 (66I ~ 7 1 [I021

[601

12131 I591 ~141 (631

References

Table 1. N-acetylglucosaminyltransferasesinvolved in N-glycan synthesis. Abbreviations: GnT, N-acetylglucosaminyltransferasc; Acc.No. is the EMBLIGenBank data base accession number.

P W

1

-4

r,

5e

7-,

'",

-d

3

s,

5

a*

c

Q

r,

".

1

z

Q

3-

--% C

-

h

-. 3

Y

%

9

e

2

9 3

LJ

.a

1SO

9 Branching of N-Glycans: N-Acetylylucosaminyltran~~e~uses

I1 integral membrane protein. The enzymes are firmly bound to the Golgi membrane with their C-terminal catalytic domains within the lumen (Nin/Coutorientation). There are four domains [ S , 14, 321, a short amino terminal cytoplasmic domain, a non-cleavable signal/anchor transmembrane domain, a stem region and a long intra-lumenal carboxy-terminal catalytic domain. The amino terminal, transmembrane and stem domains are not required for catalytic activity but are essential for accurate targeting and anchoring of the enzyme to a specific region of the Golgi membrane. 9.3.2 Targeting to the Golgi Apparatus The domains responsible for targeting glycosyltransferases to the Golgi apparatus have been investigated with chimeric cDNAs encoding hybrid proteins in which various domains of a particular glycosyltransferase are connected to reporter proteins not normally retained in the Golgi apparatus [33-371. Following either stable or transient expression of these hybrid constructs in mammalian cells, the intracellular destination of the reporter protein is determined by immunofluoresccnce and immunoelectron microscopy. Although there is a great deal of variation between different glycosyltransferases, the general conclusion from such experiments is that the transmembrane domain is essential for accurate Golgi retention while sequences outside the transmembrane domain appear to play accessory roles [ 33, 38-43]. A possible mechanism for the specific retention of proteins within the Golgi apparatus is the presence of a unique retention signal on every protein and a unique Golgi membrane receptor for every such signal. However, since Golgi retention is not saturable even at the very high levels of expression of transient transfection experiments, a more likely mechanism is retention due to homo-oligomerization of the glycosyltransferase or hetero-oligomer formation with other Golgi proteins. The large aggregate may be unable to enter budding vesicles either because of its size or due to interaction with the Golgi membrane lipid bilayer [44]. Evidence for such hetero-oligomers or “kin oligomers” has been obtained [37, 44-46]. Kin recognition seems to be mediated primarily by the lumenal domain rather than the transmembrane domain [35, 37, 381. The transmembrane domains of plasma membranetargeted proteins are broader and more hydrophobic than those of Golgi-targeted proteins and the length of the transmembrane domain has been shown to be critical for accurate Golgi localization [3S, 47, 481 suggesting that sorting may be mediated by interaction with lipid microdomains of different thicknesses.

9.4 UDP-GlcNAc:Manal-3R [GlcNAc to Manal-31 ~-1,2-N-Acetylglucosaminyltransferase I (GnT I, EC 2.4.1.101) GnT I controls the synthesis of hybrid and complex N-glycans (Figure 2) by initiating the synthesis of the first antenna and this reaction is an essential prerequisite

9.4 UDP-ClcNAc:A4~inal-3R[GlcNAc to Munal-3/

151

for further branching due to the other five N-acetylglucosaminyltransferases (GnT 11 to VI) (Figure 1, Table 1) [ 3 , 5 , 21, 491. Detailed kinetic analysis of highly purified rabbit GnT I [50] has shown that catalysis is by an ordered sequential Bi-Bi mechanism in which UDP-GlcNAc binds first and UDP leaves last. Mn2+ is essential for activity. Although the physiological substrate is Man5GlcNAclAsn-X (Figure 2), the minimum substrate requirement is Manal-3Manpl-R where R can be a 4GlcNAc residue or a hydrophobic octyl group 150, 511. Essential groups on the Manp residue are an unsubstituted equatorial hydroxyl at C-4 and an unsubstituted axial hydroxyl at C-2; modifications at C-6 of the Manp residue caused variations in KM but no major alterations in enzyme activity. As expected, Manal6(2-deoxyManal-3)ManplRis not a substrate; the compound is not a competitive inhibitor indicating that a hydroxyl group at the C-2 position of the Manal-3 residue is essential for binding of the substrate to the enzyme. Removal of the hydroxyl groups at C-3, 4 or 6 of the Manal-3 residue leads either to a poor or inactive is a competitive inhibisubstrate 1521. Manal-6(6-O-methylManal-3)Man~l-octyl tor ( K , = 0.76 mM). The genes for rabbit, human, mouse, rat, Chinese hamster ovary (CHO) cells, baby hamster kidney (BHK) cells, chicken, frog and Cuenorlzubditis eleyuns GnT I have been cloned (Table 1). The amino acid sequences of the mammalian enzymes are over 90% identical. There is limited sequence homology to GnT I1 [220] but not to any other known glycosyltransferase. GnT I has the type I1 integral membrane protein domain structure typical of all cloned glycosyltransferases. The transmembrane segment of GnT I is essential for retention in medial-Golgi cisternae but the other domains also play a role [41, 421. Removal of 106 amino acids from the N-terminus of rabbit GnT I does not inactivate the enzyme whereas removal of a further 14 amino acids results in complete loss of activity 1531. The protein is not N glycosylated because there are no Asn-X-Ser(Thr) sequons but it is O-glycosylated [45, 541. Recombinant GnT I produced in the baculovirus/Sf9 insect cell system 15.51 has been used to convert various derivatives of Manal-b(Mana1-3)Man~l-octylto Manal-6(GlcNAcp1-2Manal-3)Man~l -octyl and these compounds have been used to study the substrate specificity of GnT 11 [56, 571. The human GnT I gene MGATI has been localized to chromosome 5q35 1581. Part of the 5’-untranslated region and all of the coding and 3/-untranslated regions of the human and mouse GnT I genes are on a single 2.5 kb exon (exon 2) [59-61]. The remaining 5’-untranslated sequence of the human GnT I gene is on an exon (exon 1) which is between 5.6 and 15 kb upstream [62].The mouse 1631 and rat [64] GnT I genes have a similar organization except that there are at least two upstream non-coding exons in these species. There is only a single copy of the gene in the haploid human and mouse genomic DNA. There are multiple transcription start sites for human exon 1 compatible with the expression by several human cell lines and tissues of two transcripts, a broad band ranging in size from 2.7 kb to 2.95 kb and a sharper band at 3.1 kb 1621. The 5’flanking region of exon 1 has a G C content of 81?4 and has no canonical TATA or CCAAT boxes but contains potential binding sites for transcription factor Spl . It is concluded that MGATI is a typical housekeeping gene. CAT expression was observed on transient transfection into HeLa cells of a fusion construct containing the CAT gene and a genomic DNA fragment from the 5’-flanking region of human

152

9 Brunching of N-Glycans: N-Acetylglucosuminyltrunsferclses

exon 1. Mouse [60, 631 and rat [64, 651 GnT I mRNA have also shown at least two transcripts with the longer transcript dominant in brain. A recent report has shown that the three rat exons produce five subclasses of tissue-specific GnT I mRNA varying only in their 5'-untranslated regions [64]. The Lecl CHO cell mutant is an L-PHA resistant cell line lacking GnT I activity. The cells synthesize an inactive GnT I protein identical in size to the wild-type CHO enzyme [66]. The Lec 1 GnT I gene has three mutations within the lumenal domain, each resulting in an amino acid substitution. One of the three mutations (C123R) resulted in complete loss of activity whereas the other two mutations had no apparent effect on enzyme activity. The RicR14 BHK cell mutant is a ricin-resistant cell line which also lacks GnT I activity. Analysis of RicR14 cDNA showed two different point mutations, believed to be on separate clones, generating respectively a G320D mutation and a stop codon at position 81 in the open reading frame [67]. The RicR14 G320D GnT I protein both in RicR14 cells and after transfection of the mutant gene into COS cells can be localized by immunocytochemistry to the Golgi apparatus. Introduction of the G320D mutation into the rabbit GnT I sequence and transfection of the mutated gene into COS cells produces in high yield a full-length protein which is enzymatically inactive and localized to the Golgi apparatus. The truncated GnT I protein produced by introduction of the stop codon at position 81 is also localized to the Golgi after transfection of the mutated gene into mouse 3T3 cells. The roles of residues C Y S ' and ~ ~ GIY'~'in GnT I catalysis must await elucidation of the three-dimensional structure.

9.5 UDP-GlcNAc:Manal-6R [GlcNAc to Manal-61 ~-1,2-N-Acetylglucosaminyltransferase I1 (GnT 11, E.C. 2.4.1.143) GnT I1 initiates the first antenna on the Manal-6 arm of the N-glycan core and is therefore essential for normal complex N-glycan formation (Figure 2). GnT I1 has been purified to homogeneity from rat liver [68].Detailed kinetic analysis of the rat enzyme [69] has shown that catalysis is by an ordered sequential Bi-Bi mechanism in which UDP-GlcNAc binds first and UDP leaves last. Mn2+ is essential for activity. The mimimal substrate is Manal-6(GlcNAc~l-2Manal-3)Man~l-R where R can be a hydrophobic aglycone. The 2-deoxyManal-6(GlcNAc~l-2Manwl-3)~ Manpl-R analog is a competitive inhibitor ( K , = 0.13 mM) but the other hydroxyl groups of the Manal-6 residue are not essential for activity [57]. Substitution of the C-4 hydroxyl of the ManP residue but not its removal leads to an inactive substrate. GlcNAc~l-2Manal-3Man~l-octyl is a good inhibitor of the enzyme ( K , = 0.9 mM) indicating that this trisaccharide moiety is required for substrate binding to the enzyme. The human, rat, frog and pig GnT I1 genes have been cloned (Table 1). The enzyme has a typical glycosyltransferase domain structure. There is limited sequence homology to GnT I but not to any other previously cloned glycosyltransferase.

9.6 The Role of GnT I and II in Murnmuliun Developinent

153

Recombinant GnT I1 produced in the baculovirus/Sf9 insect cell system [58] has been used to convert various derivatives of Manal-6(GlcNAc~l-2Manal3)ManPl-octyl to GlcNAcP 1-2Mana 1-6(ClcNAcp1-2Mana 1-3)Manp 1-0cty1; these compounds can be used to study the substrate specificity of GnT 111. IV and V [70]. The human GnT I1 gene (MGAT2) is on chromosome 14q21 [58]. The open reading frame and 3'-untranslated region of the human and rat genes were shown to be on a single exon [58, 711. 5'-RACE (rapid amplification of cDNA ends) and RNase protection analyses of the human gene showed multiple transcription initiation sites at -440 to -489 bp relative to the ATG translation start codon ( + I ) proving that the entire GnT I 1 gene is on a single exon [72]. The gene has three AATAAA polyadenylation sites downstream of the translation stop codon. 3'RACE using RNA from the human cell line LS-180 indicated that all three sites were utilized for transcription termination yielding transcripts at 2.0, 2.7 and 2.9 kb. The gene has a CCAAT box at -587 bp but lacks a TATA box and the 5'untranslated region is GC-rich and contains consensus sequences suggestive of multiple binding sites for Spl; these properties are typical for a housekeeping gene. A series of chimeric constructs containing different lengths of the 5'-untranslated region fused to the chloramphenicol acetyltransferase (CAT) reporter gene were tested in transient transfection experiments using HeLa cells. The CAT activity of the construct containing the longest insert (-1076 bp relative to the ATG start codon) showed a -38-fold increase as compared to that of the control. Removal of the region between -636 and -553 bp caused a dramatic decrease in CAT activity indicating this to be the main promoter region of the gene. Although human GnT I1 is a typical housekeeping gene, it may in some circumstances be under the control of Ets transcription factors [73]. The human GnT I1 promoter contains four putative Ets-binding sites [72]. Co-transfection into HepC2, HeLa or Cos-1 cells of either ets-1 or ets-2 expression plasmids together with chimeric GnT I1 promoter-CAT plasmids [73] results in a 2-3-fold stimulation of promoter activity. Gel mobility shift assays and South-Western blots localized the functional Ets-binding site to one of the four sites in the GnT I1 promoter. Unlike the GnT V promoter (see below), the GnT I1 promoter is not activated by SYC and there are no associated AP1 sites. The pig GnT I1 cDNA is also on a single exon. The gene is located on chromosome lq23-q27 and has a GC-rich TATA-less promoter with multiple transcription start sites, typical of a house-keeping gene 1741.

9.6 The Role of GnT I and I1 in Mammalian Development As discussed above, the synthesis of complex N-glycans can be divided into three distinct stages. The first stage occurs primarily in the cytoplasm and rough endoplasmic reticulum and involves the synthesis of Glc3MangGlcNAcz-pyrophosphatedolichol. The second stage begins with the transfer of Glc3Mans GlcNAc? oligo-

154

9 Branching of N-Glycans: N-Acetyl~lucosaminyltran~erases

saccharide to an Asn residue and involves processing to Man5GlcNAcz-Asn-X, the substrate for GnT I. The third stage completes complex N-glycan synthesis and occurs primarily in the Golgi apparatus. Complex N-glycans are absent from bacteria 1751 and yeast 1761 and are present in very small amounts, if at all, in protozoa [77-791 and Dictyostelium discoideum [80].All of the above organisms except bacteria are capable of making N-glycans of the oligomannose type and therefore express the enzymes of stage one and at least some of the enzymes of stage two of N-glycan synthesis. Complex N-glycans are present in most of the multi-cellular invertebrate and vertebrate animals that have been analyzed, i.e., nematodes [81-831, schistosomes [84-861, molluscs [25, 871, insects 1881, fish [89],birds 1901 and mammals. This indicates that the third stage of N-glycan synthesis appeared in evolution just prior to the development of multicellular organisms probably because complex N-glycans are required for the cellcell interaction process and normal development of multi-cellular animals. The study of mutations in mouse and man provide support for this concept. Somatic CHO cell mutants lacking the GnT I gene show essentially normal growth in tissue culture whereas mouse embryos with a null mutation in this gene do not survive beyond 10.5 days post-fertilization and show severe multi-systemic developmental abnormalities particularly of the brain 191, 921 and no obvious cause of death [93]. Mice with a homozygous null mutation in the gene encoding GnT I1 survive to term but are born stunted with various congenital abnormalities and die within 3-4 weeks after birth 194, 22 11. Carbohydrate-Deficient Glycoprotein Syndrome (CDGS) is a group of autosomal recessive diseases with multisystemic abnormalities including a severe disturbance of nervous system development [95]. Two children with CDGS Type 11 cannot make complex N-glycans due to point mutations in the GnT I1 gene 196-991. These findings indicate that complex N-glycans are essential for normal postimplantation embryogenesis and development, particularly of the nervous system. Since carbohydrates are believed to be important also in pre-implantation development, the survival of GnT I-null mouse embryos for 9.5 days suggested that null mutant blastocysts may have received either GnT I or complex N-glycans from the mother. This idea was supported by the finding that early blastocysts from several Mgatl +/- heterozygous crosses were found to bind the lectins E-PHA 1931 and L-PHA [loo]. E8.5 or E9.5 Mgatl-/- embryos do not bind L-PHA. Analysis of Mgat 1-/- blastocysts, positively identified by polymerase chain reaction of genomic DNA, revealed the presence of wild type Mgatl RNA [loo]. It was concluded that the effects of the Mgatl null mutation are not operative until sometime between implantation and E5.5, due to the continued presence of maternally derived Mgat 1 mRNA in pre-implantation embryos. To circumvent the embryonic lethality of the Mgatl null mouse, WW6 embryonic stem cells with inactivated Mgatl alleles were tracked in chimeric embryos [ 1011. WW6 cells carry an inert beta-globin transgene that allows their identification. It was found that homozygous null Mgatl -/- WW6 cells did not contribute to the epithelial layer in over 99%)of lung bronchi indicating that complex N-glycans are required to form a morphologically recognizable bronchial epithelium. To study the role of complex N-glycans in the development of a simpler organism, we have cloned C. eleyans cDNAs with sequence similarity to mammalian GnT I

9.7 UDP-GlcNAc:RI-Munxl-6/ G I c N A c ~ l - Z M u n r ~ l - 3 ] M u n ~ 1 - 4 R ~155

(three genes designated gly-12, yly-13 and gly-14 [102]) and I1 (one gene, unpublished). All three GnT I cDNAs encode proteins (467, 449 and 437 amino acids respectively) with the domain structure typical of previously cloned Golgi-type glycosyltransferases. Expression in both insect cells and transgenic worms showed that gly-12 and yly-14, but not yly-13, encode active GnT 1. All three genes were expressed throughout worm development (embryo, larval stages L1 to L4 and adult worms). Transgenic worms that overexpress any one of the three genes show no obvious phenotypic defects. Mutants worm strains with deletions in these genes may indicate the roles of complex N-glycans in worm deveopment. Although work will undoubtedly continue on mutant mice, other development models should also be considered. C. eleyans is a particularly attractive model because of the detailed information available on its development, its relatively simple architecture and the availability of the complete genome sequence. Several C. elegans glycosyltransferases in addition to GnT I and I1 are presently under study [103-1061 and others are sure to follow.

9.7 UDP-GlcNAc:RI-Manal-61GlcNAcPl-2Manal-3]Manfll-4Rz [GlcNAc to Manfll-41 ~-1,4-N-Acetylglucosaminyltransferase I11 (GnT 111, E.C. 2.4.1.144)

GnT 111 incorporates a bisecting GlcNAc residue into the N-glycan core. GnT I11 action often serves as a STOP signal, e.g., GnT IT, GnT IV, GnT V and amannosidase I1 cannot act on bisected substrates (Figure 2). Increased levels of GnT 111 are therefore expected to inhibit the synthesis of highly branched sialylated complex N-glycans. Whereas the genes encoding GnT I and I1 are typical housekeeping genes and are expressed in all tissues tested, GnT I11 has a distinct tissue distribution. e.g., it is poorly expressed in liver and strongly expressed in kidney. Rat GnT 111 contains three potential N-glycosylation sites. Removal of any one site reduced enzyme activity, the activity decreased as the number of glycosylation sites decreased and was not detectable when all three sites were absent [ 1071. The genes encoding human, mouse and rat GnT I11 have been cloned (Table 1). The enzyme has a typical glycosyltransferase domain structure. There is no apparent sequence similarity to any known glycosyltransferase. Analysis of the human GnT 111 gene shows that the entire coding region is on a single exon (exon 1) on chromosome 22q.13.1 [108]. The human GnT 111 promoter region has been analyzed [ 109, l 101. There are at least three upstream non-coding exons designated HIS exon-1, H I 5 exon-2 and H20 exon-1 and at least three different mRNA transcripts formed by alternative splicing of these exons [ 1101: H I5 (H1S exons -1 and -2 and exon I), H20 (H20 exon-1 and exon 1) and H204 (exon 1). Assays using luciferase as a reporter protein in a human hepatoblastoma cell line demonstrated promoter activity upstream of transcripts H I 5 and H204 but not H20. None of the promoter regions contained either a TATA or CCAAT box but consensus

156

9 Brunching qf N-Glycuns: N-AcrtylglucosuminyltvLlnsferases

sequences for many putative transcription factor binding sites (Ets, Myb, Myc, etc) were present.

9.7.1 Overexpression of GnT 111 Activity Transfection of the GnT 111 gene into various cell lines has produced interesting biological effects: i) suppression of hepatitis B virus gene expression by a GnT IIItransfected hepatoma cell line [ 1 11, 1 121; ii) suppression of sensitivity of GnT IIItransfected K562 cells to natural killer (NK) cell cytotoxicity with resultant increased spleen colonization by these cells in nude mice [ 1131; iii) suppression of the metastatic potential of GnT III-transfected B16-hm mouse melanoma cells [ 1141 and elevated expression of E-cadherin by the latter cells [ 1151; iv) GnT 111 transfection of a human glioma cell line which expresses the epidermal growth factor (EGF) receptor on its cell surface blocked EGF binding and EGF receptor autophosphorylation [ 1 161; v) overexpression of GnT I11 in rat pheochromocytoma PC12 cells resulted in inhibition of growth response and of tyrosine phosphorylation of the Trk/nerve growth factor (NGF) receptor following addition of N G F [117]; vi) transfection of the GnT I11 gene into a swine endothelial cell line reduced the antigenicity to human natural antibodies presumably by reducing the synthesis of complex N-glycans which carry the Galwl-3Gal epitope that causes rejection of pig xenotransplants [ 1 18-1201; vii) CD44 from GnT I11 transfected mouse melanoma cells showed increased adhesion to hyaluronate with a concomitant increased CD44mediated tumor growth and metastatic development in the spleen after subcutaneous inoculation into mice [ 1211. These effects are presumably mediated by the effect of the bisecting GlcNAc on the conformation of complex N-glycans [ 122-1241 and by the inhibition of complex N-glycan synthesis due to over-expression of GnT 111. Treatment of hepatoma cells with forskolin, an adenylyl cyclase activator, causes a dramatic increase of GnT I11 levels, probably due to enhanced transcription [ 1251, with resultant increase in E-PHA-reactive glycoproteins (indicative of an increase in bisected N-glycans) and decrease in L-PHA-reactive glycoproteins (indicative of a decrease in tri- and tetra-antennary N-glycans probably due to inhibition of GnT IV and V activities). This increase in bisected N-glycans was observed on intracellular glycoproteins whereas cell surface glycoproteins showed a decrease suggesting that the bisecting GlcNAc residue may play a role in intracellular glycoprotein sorting [125]. Overexpression of GnT 111 in transgenic mice reduced the antigenicity of some organs to natural human antibodies presumably by suppression of xenoantigens [ 1261. These mice displayed spleen atrophy. hypocellular bone marrow, pancytopenia and other changes suggesting that N-glycans may have some significant roles in stroma-dependent hemopoiesis [ 1271. GnT I11 activity is elevated during hepatocarcinogenesis in contrast to the undetectable level found in normal hepatocytes and therefore transgenic mice that specifically overexpress GnT 111 in the liver were developed [128]. The hepatocytes of these mice had a swollen oval-like morphology with many lipid droplets due to abnormal glycosylation of apolipoprotein B. Surprisingly, mice in which the GnT I11 gene has been “knocked out” show either no phenotype [129] or a relatively mild phenotype [130].

9.8 UDP-GlcNAc:RlMunxl-3Rz [GlcNAc to Munal-3 1

157

9.7.2 GnT 111 Activity and Cancer GnT 111 has been shown to be elevated in various types of rat hepatoma [ 131 1341, human leukemia [135, 1361 and other cancers [137]. Serum GnT I11 activity in hepatocellular carcinoma patients was significantly higher than that in patients with liver cirrhosis and chronic hepatitis and in normal controls and was reduced in a significant number of hepatocellular carcinoma patients after therapy [ 1381. Since GnT I11 is induced in certain cancers, particularly in hepatic tumorigenesis in the rat (see above), the susceptibility of Mgat3-/- mice (null for the GnT 111 gene) to tumor induction was tested [139]. After a single injection with diethylnitrosamine and subsequent treatment with phenobarbitol for 6 months, Mgat3+/+ and Mgat3+/- mice had grossly enlarged livers with numerous tumors while Mgat3-/- mice had livers of normal size and only 50% of the mice had 1-4 small tumors. Tumors developed in these mice by 10-12 months after diethylnitrosamine injection. Surprisingly, in contrast to the situation in the rat, hepatic tumor formation in the Mgat3+/+ mice was not accompanied by a dramatic increase of GnT 111 activity. The data suggest that a glycoprotein factor with the bisecting GlcNAc facilitates tumor progression in mouse liver and that, in the absence of the bisecting GlcNAc in Mgat3-/- mice. the factor is reduced in activity with consequent severe retardation of tumor progression. -

9.8 UDP-GlcNAc:R1Manal-3R2 [GlcNAc to Manal-31 ~-1,4-N-Acetylglucosaminyltransferase IV (GnT IV, E.C. 2.4.1.145) GnT IV adds a GlcNAc in pl-4 linkage to the Mana1-3Manp- arm of the N-glycan core [ 1401. When the biantennary substrate GlcNAcP I -2Mana1-6[GlcNAcP12Manu1-3]Man~l-4GlcNAc~1-4GlcNAc-Asn-X is presented to a hen oviduct extract, GnT JV acts only on the Manul-3ManP- arm. However, hen oviduct extracts will add a GlcNAc in PI-4 linkage to the Manu residue in the linear oligosaccharides GlcNAcpl-2Manal-6ManP1-R(R = methyl, methoxycarbonyloctyl, onitrophenyl), GlcNAcpl-2Manal-6Glcp1-R (R = o-nitrophenyl, allyl), GlcNAcPI2Mana1-3Man~l-methoxycarbonyloctyl and even GlcNAcPl-2Manal -methyl (but not GlcNAcp1-6Mana-R) [141, 1421. Thus the branch specificity displayed by the biantennary substrate is lost when a linear substrate is used. It is possible that more than one enzyme may be involved in these various activities. Prior action of GnT I is essential and addition o f a bisecting GlcNAc by GnT 111 or addition of a GalP1-4 residue to either the G l c N A c ~ l - 2 M a n a l - 3 M a nor ~ GlcNAcP1-2Mana1-6ManP arms inhibit enzyme action [140, 1431. GnT IV was purified 224,000-fold from bovine small intestine [ 1431. The purified enzyme shows a single band of M, 58,000 and behaves as a monomer. The bovine [ 1441 and human [ 1451 genes have been cloned. The open reading frame of the bovine enzyme has 1605 bp (535 amino acids) and encodes a typical Type I1 membrane protein. There is no homology to known glycosyltransferases. The enzymati-

158

9 Brunching of N-Glycans: N-Acetyl~lucosaminyltransferases

cally active GnT IV isolated from bovine small intestine was truncated and lacked 92 N-terminal amino acids, including the trans-membrane domain, presumably due to proteolysis during purification. Transient transfection of various bovine GnT IV constructs into COS-7 cells verified that deletion of the first 92 N-terminal amino acids did not destroy enzyme activity but showed that deletion of the first 112 and 141 N-terminal residues destroyed activity. Deletion of 152 amino acids from the Cterminus caused no loss of activity but, surprisingly, shorter C-terminal deletions did result in activity loss. None of the three Asn-X-Thr/Ser sequons are essential for activity. A probe based on the bovine GnT IV sequence was used to clone a human GnT IV cDNA [145]. The human cDNA encoded a Type I1 membrane protein of 535 amino acids which is 96% identical to the bovine enzyme. The gene mapped to human chromosome 2q12 in close proximity to the GnT V gene (2q21). Preliminary data indicate that the human gene has multiple exons and that a homologous gene exists in many mammalian species and in chicken [145]. A preliminary report suggested that there may be two human GnT 1V genes [146], GnT IVa with 96% identity to the bovine amino acid sequence (described above) and GnT IVb with 62% homology to the GnT IVa sequence. Both human genes show GnT IV activity on transient expression in COS-7 cells. GnT IVa and IVb show different patterns of tissue expression. Northern blot analysis of 23 human tissues and 8 cancer cell lines with a probe for human GnT IV indicate at least five mRNA sizes [145]. Although the relative distribution of these messages is similar in all tissues, there are significant tissuespecific differences in expression levels. Lymphoid tissues such as spleen, thymus, small intestine, lymph node and peripheral blood leukocytes express large amounts of GnT IV mRNA while liver, bone marrow, lung and many other tissues express very little. The T lineage cell line MOLT-4 has a very high expression level while the B lineage cell line Raji shows very low expression. The myeloid cell line HL-60 shows high expression in contrast to the myeloblastic lymphoma cell line K-562. GnT IV is essential for the production of tri- and tetra-antennary N-glycans and, as discussed in the sections dealing with GnT 111 (above) and GnT V (below), one of the most common alterations in transformed or metastatic malignant cells is the presence of highly branched N-glycans. Indeed, elevated GnT IV activity has been reported in a hepatoma cell line [ 1331, during myelocytic cell differentiation [ 147, 1481 and during chick embryo development [ 1491; also, structural studies on N-glycans have indicated increased GnT IV activity in human hepatocellular carcinoma [ 1501 and choriocarcinoma [ 1511.

9.9 U D P - G ~ C N A C : R I M ~[GlcNAc ~ ~ ~ - ~to R Manald] ~ ~-1,6-N-Acetylglucosaminyltransferase V (GnT V, E.C.2.4.1.155) GnT V adds a GlcNAc in p1-6 linkage to the Manal-6Manp arm of the N-glycan core. Kinetic analysis has shown that, like GnT I and 11, GnT V follows an ordered

sequential Bi-Bi mechanism [ 1521. Several studies have been published which map the substrate requirements of this enzyme [153-1671. The minimal substrate is GlcNAcPl-2Manwl-6Man~l-R where R can be a hydrophobic aglycone although compounds with a biantennary structure are significantly better substrates than those with linear structures. The P-linked Man residue can be replaced by Glc. The enzyme is not inhibited by EDTA [168].GnT V, like GnT 11 and IV, cannot act on substrates with a bisecting GlcNAc residue attached to the C-4 hydroxyl of the Plinked Man residue although methyl substitution of this hydroxyl increases enzyme activity and its removal has minimal effects on enzyme activity. Galactosylation or removal of the C-4 hydroxyl of the GlcNAc residue linked to the Manal-6 residue results in loss of enzyme activity. A substrate analogue, GlcNAcPl-2(6-deoxy)Manal-6GlcPl-octyl, was shown to be an excellent competitive inhibitor ( K , = 0.07 mM) [153, 1561. Rat [ 1681 and human [ 1571 GnT V have been purified and the rat, human and Chinese hamster genes have been cloned (Table 1) and expressed [169-1711. The human gene has been mapped to chromosome 2q21 and contains 17 exons spanning over 155 kb [172]. The gene is expressed in a tissue- and cell type-specific manner and is regulated at the level of transcription by multiple promoters [172]. The Lec4A and Lec4 Chinese hamster ovary (CHO) glycosylation mutants lack N-linked glycans with GlcNAcPl-6Manul-6Man~branches initiated by GnT V [173]. Detergent extracts of Lec4 cells have no detectable GnT V activity but Lec4A extracts have activity equivalent to that of parental cells because Lec4A GnT V activity is localized to the endoplasmic reticulum instead of to Golgi membranes [174]. Lec4 GnT V cDNA contains two insertions and encodes a truncated transferase missing 585 amino acids whereas Lec4A GnT V cDNA possesses a single point mutation from T to G (L188R) [171]. When transfected into Lec4 cells, Lec4 cDNA was unable to restore GnT V activity whereas Lec4A cDNA converted Lec4 cells to the Lec4A phenotype. Thus the L188R point mutation causes mislocalization of a fully active GnT V from the Golgi to the endoplasmic reticulum.

9.9.1 GnT V Activity and Cancer

One of the most common alterations in transformed or metastatic malignant cells is the presence of larger N-glycans due primarily to a combination of increased GlcNAc branching, sialylation and poly-N-acetyllactosamine content [ 175-1 881. GnT V plays a major role in these effects. Cells transformed with polyoma virus, Rous sarcoma virus or T24 H-ras [ 153, 159, 184, 189-1931 or treated with transforming growth factor-P or phorbol ester [ 1941 were shown to have significantly increased GnT V activity. Poly-N-acetyllactosamine chains have been shown to carry cancer-associated antigens and the initiation of these chains is favored on the antenna initiated by GnT V [ 177, 195, 1961. Transfection of the GnT I11 gene into a highly metastatic mouse melanoma cell line resulted in decreased PI-6 branching of N-glycans without altering GnT V enzyme levels [ 1141 due to the fact that GnT V

160

9 Brunching of N-Glycuns: N-Acrtylylutosumin~ltrunsferu.~e,~

cannot act on bisected N-glycans [ 3 ] ;the GnT 111-transfectedcells showed a marked reduction in metastatic potential. Pierce’s group showed that the elevated GnT V activity and mRNA levels following transformation with Rous sarcoma virus could be inhibited by blocking cell proliferation with herbimycin A, demonstrating that Src kinase activity can regulate GnT V expression [ 1971. The GnT V promoter contains AP-1 and PEA3/Ets binding elements and, when co-transfected with a src expression plasmid into HepG2 cells, conferred src-stimulated transcriptional enhancement upon a luciferase reporter gene. This stimulation by src could be antagonized by co-transfection with a dominant-negative mutant of the Raf kinase, suggesting the involvement of Ets transcription factors in the regulation of GnT V gene expression. Ets is a nuclear phosphoprotein transcription factor that binds to purine-rich DNA sequences and is associated with transformation properties [ 1981. The src-responsive element was localized to two overlapping Ets sites. Stimulation of transcription by src was inhibited by co-transfection with a dominant-negative mutant of Ets-2, demonstrating that the effects of the src kinase on GnT V expression are dependent on Ets. Neu-transformed NIH3T3 cells have a three-fold increase in GnT V enzyme activity [ 1991. Promoter/reporter experiments showed that her-2/neu stimulates transcription from the human GnT V promoter and that the her-2/neu response element includes three Ets transcription factor binding sequences. Co-transfections with dominant-negative Raf and Ets expression plasmids demonstrated that the transcriptional activation of the GnT V promoter by neu is mediated by the RasRaf-Ets signal transduction pathway. Two regions of the GnT V promoter have been identified as positive regulatory elements [200, 2011. Both regions contained an Ets consensus sequence and gel mobility shift experiments showed that both regions bound the Ets protein. Cotransfection of luciferase constructs controlled by either of the two putative promoter regions and an Ets expression plasmid showed stimulation of luciferase by the Ets product. It can be concluded from the above experiments that increases in GnT V expression after oncogenic transformation are most likely caused by direct effects on the GnT V promoter by the Ets family of transcriptional activators which are upregulated by a cellular proliferation signaling pathway. This pathway begins with growth factor receptors that activate tyrosine kinases at the cell surface and proceeds through src, ras, and ruf [202]. It is of interest that the promoter regions of both GnT I1 [72] and GnT V [ 172, 1971 have putative binding sites for the products of the Ets and c-Myb families of proto-oncogenes since GnT I1 action is a prerequisite for GnT V action (see above). Transfection of the GnT V gene into premalignant epithelia1 cells resulted in relaxation of growth controls and reduced substratum adhesion [ 1821. Transfection of the GnT V gene into three mouse mammary cancer cell lines increased metastatic potential on injection into the tail veins of mice [203]. Evidence has been obtained for post-translational activation of GnT V by phosphorylation [204]. A peptide encoded by an intron sequence of the GnT V gene has been shown to be a human melanoma-specific antigen recognized by human cytolytic T lymphocytes [205].

9.11 GnT VII and GnT VIII

161

9.10 UDP-GlcNAc:R1(Rz)Manal-6R3 [GlcNAc to Manal-6) p-1,4-N-AcetylglucosaminyltransferaseVI (GnT VI) GnT VI adds a GlcNAc residue in PI-4 linkage to the Manal-6Manp arm of the N-glycan core [ 1421. The minimal substrate for GnT VI is the trisaccharide GlcNAc@l-6[GlcNAc~1-2]Manal-R (R = methyl, 6Manpl-methy1, or 6Manplmethoxycarbonyloctyl). The enzyme therefore requires the prior actions of GnT I, I1 and V. Unlike GnT I to V, GnT VI can act on both bisected and non-bisected substrates. The enzyme has been demonstrated in birds [142, 160. 2061 and fish [207] but not in mammalian tissues. The enzyme has not been purified nor has the gene been cloned.

9.11 GnT VII and GnT VIII Mutagenesis or transfection of CHO cells with large amounts of DNA has allowed the isolation of mutant cell lines which express a glycosyltransferase not expressed by the wild type cell [173]. The first such CHO mutant to be isolated was LEClO which expresses GnT 111 [208]. Several analogous dominant gain-of-function mutants have now been characterized. The dominant CHO mutants, LECl8 and LEC14, are of particular interest because each expresses a GnT activity which creates novel N-glycans that have not previously been described In glycoproteins from any source. LEC18 and LECI4 CHO cells were obtained by screening for resistance to pea lectin. LEC18 cells synthesize complex N-glycans with a GlcNAc linked a or p (not yet determined) to the carbon-6 position of the GlcNAc between the @-linkedMan and Asn-linked GlcNAc residues of the N-glycan core [209] and LEC14 cells synthesize complex N-glycans with a GlcNAc @-linkedto the carbon-2 position of the P-linked Man residue [210]. In both LEC18 and LEC14 cells, GlcNAc transfer is mediated by distinct N-acetylglucosaminyltransferase (GnT) activities termed GnT VIII (LECl8) and GnT VII (LEC14) [21I]. GnT VIII transfers GlcNAc to GlcNAc-j3-I-O-p-nitrophenyl or to GlcNAcpl-

2Man~l-6[GlcNAc~l-2Manctl-3]Man~l-4GlcNAc~l-4[+/-Fuccrl-6]GlcNAc-AsnX . GnT VIII has a sharp pH optimum at 7.0, a broad optimum Mn2’ requirement from 20-60 mM and an absolute requirement for 10 mM ATP similar to a UDPGlcNAc:GlcNAc p-R P 1,4-N-acetylglucosaminyltransferase reported in the snail 12121. The products with both acceptors showed that GnT VIII transfers GlcNAc to the carbon-6 position of the core GlcNAc. GnT VI1 cannot act on low molecular weight acceptors and transfers GlcNAc preferentially to the fucose-containing glycopeptide GlcNAcp 1-2Mana 1-61GlcNAc@1-2Manal-3]Man~l-4GlcNAc@l4[Fucal-6]GlcNAc-Asn-X. The acceptor specificities and other biochemical prop-

162

9 Branching of N-Glycans: N-Acetylglucosuminyltrunsferuses

erties of GnT VII and GnT VIII differ from previously characterized GlcNActransferases indicating that they represent new members of the mammalian GnT family of transferases. The physiological significance of these interesting new enzymes must await the cloning of the GnT VII and VIII genes and subsequent study of the expression of these enzymes in normal tissues. References 1. R. Kornfeld, S. Kornfeld, Assembly of asparagine-linked oligosaccharides, Annu Rev Biochem, 1985, 54, 631-664.

2. T.A. Beyer, J.E. Sadler, J.I. Rearick, J.C. Paulson, R.L. Hill, Glycosyltransferases and their use in assessing oligosaccharide structure and structure-function relationships, Ado Enzym, 1981, 52, 23-175. 3. H. Schachter, Biosynthetic controls that determine the branching and microheterogeneity of protein-bound oligosaccharides, Biochem Cell Biol, 1986, 64. 163-18 I . 4. M.D. Snider, Riosynthesis of Glycoproteins: Formation of N-linked Oligosaccharides, In Biology of’ Carbohydrates, Ginsburg, V.. and Robbins, P.W. (ed), John Wiley and Sons, 1984, New York, N.Y., Vol. 2, pp 163-198. 5 . H. Schachter, Glycosyltransferases involved in the synthesis of N-glycan antennae, In Glycoproteins, J. Montreuil, J.F.G. Vliegenthart, and H. Schdchter (ed), Elsevier, 1995, Amsterdam, The Netherlands, Vol. 29a, pp 153-199. 6. F.W. Hemming, The coenzymic role of phosphodolichols, In Glycoproteins, J. Montreuil, J.F.G. Vliegenthart, and H. Schachter (ed), Elsevier, 1995, Amsterdam, The Netherlands, Vol. 29a, pp 127-143. 7. A. Verbert, From Glc-iMansGlcNAcz-protein to ManjGlcNAc2-protein: transfer “en bloc” and processing, In Glycoproteins, J. Montreuil, J.F.G. Vliegenthart, and H. Schachter (ed), Elsevier, 1995, Amsterdam, The Netherlands, Vol. 29a, pp 145-152. 8. R. Kleene, E.G. Berger, The Molecular and Cell Biology of Glycosyltransferases, Biochim Biophys Acta, 1993, 1154, 283-325. 9. T.W. Rademacher, R.B. Parekh, R.A. Dwek, Glycobiology, Annu Rev Biochem, 1988, 57, 785-838. 10. A. Kobata, Structures and functions of the sugar chains of glycoproteins, Eur J Biochem, 1992,209, 483--501. 11. H. Lis, N. Sharon, Protein Glycosylation---Structural and Functional Aspects, Eur J Biochem, 1993,218, 1-27. 12. G.W. Hart, Clycosylation, Curr Opin Cell Biol, 1992, 4, 1017-1023. 13. D.H. Joziasse, Mammalian glycosyltransferases: genomic organization and protein structure, Glycobiology, 1992,2, 271-277. 14. J.C. Paulson, K.J. Colley, Glycosykrdnsferases. Structure, localization, and control of cell type-specific glycosylation, J Biol Chem, 1989, 264, 17615-17618. 15. H. Schachter, Enzymes associated with glycosylation, Curr Opin Struct Biol, 1991, I , 755-765. 16. J.H. Shaper, N.L. Shaper, Enzymes associated with glycosylation, Curr Opin Struct Biol, 1992, 2, 701-709. 17. D.H. Van den Eijnden, D.H. Joziasse, Enzymes Associated with Glycosylation, Curr Opin Struct Bid, 1993, 3, 71 1-721. 18. W.M. Watkins, Molecular basis of antigenic specificity in the ABO, H and Lewis blood group systems, In Glycoproteins, J. Montreuil, J.F.G. Vliegenthart, and H. Schachter (ed), Elsevier, 1995, Amsterdam, The Netherlands, Vol. 29a, pp 313-390. 19. M. Fukuda, Cell surface carbohydrates: cell-type specific expression, In Molecular Glycohiology, Fukuda, M., and Hindsgaul, 0. (ed), Oxford University Press, 1994, Oxford, UK, pp 152. 20. R.D. Cummings, Synthesis of asparagine-linked oligosaccharides: pathways, genetics and metabolic regulation, In Glycoconjuqates. Composition, Structure and Function, Allen, H.J., and Kisailus, E.C. (ed), Marcel Dekker, Inc., 1992, New York, N.Y., pp 333-360.

21. H. Schachter, The “yellow brick road” to branched complex N-glycans, Glycohiology, 1991, I , 453-461. 22. A. Varki, Factors controlling the glycosylation potential of the Golgi apparatus, Tr Cell Bid, 1998, 8, 34-40. 23. S. Silberstein, R. Gihnore, Biochemistry, molecular biology, and genetics of the oligosaccharyltransferase, FASEB J , 1996, I f ) , 849-858. 24. K.D. Johnson, M.J. Chrispeels, Substrate specificities of N-acetylglucosaminyl-, fucosyl-, and xylosyltransferases that modify glycoproteins in the Golgi apparatus of bean cotyledons, Plant Pliysiol, 1987, 84, 1301Ll308. 25. H. Mulder, F. Dideberg, H. Schachter, B.A. Spronk, M. De Jong-Brink, J.P. Kamerling, J.F.G. Vliegenthart, In the biosynthesis of N-glycans in connective tissue of the snail Lymnaea stagnalis incorporation of GlcNAc by beta 2GlcNAc-transferase I is an essential prerequisite for the action of beta 2GlcNAc-transferase I1 and beta 2Xyl-transferase, Eur J Biochern, 1995, 232,272-283. 26. E. Staudacher, F. Altmann. J. Glossl, L. Marz, H. Schachter, J.P. Kamerling, K. HHrd, J.F.G. Vliegenthart, GDP-fucose: beta-N-acetylglucosamine (Fuc to (Fucalphali6GlcNAc)-Asnpeptide) alpha 1 +3-fucosyltransferase activity in honeybee (Apis mellifica)venom glands-The difucosylation of asparagine-bound N-acetylglucosamine, Eur J Biochem, 1991, I Y 9 , 745-75 I. 27. E. Staudacher, F. Altmann, L. Marz, K. HHrd, J.P. Kamerling, J. Vliegenthart, alphal6(alphal-3)-Difucosylationof the asparagine-bound N-acetylglucosamine in honeybee venom phospholipase A2, Glycoconjugate J , 1992, Y, 82-85. 28. E. Staudacher, T. Dalik, P. Wawra, F. Altmann, L. MHrz, Functional purification and characterization of a GDP-fucose:beta-N-acetylglucosamine (Fuc to Asn linked GlcNAc) alphal,3fucosyltransferase from mung beans, Glycoconj J , 1995, 12, 780-786. 29. E. Staudacher, L. Marz, Strict order of (Fuc to Asn-linked GlcNAc) fucosyltransferases forming core-difucosylated structures; G/ycoconj J , 1998, 15, 355-360. 30. A. Roitinger, H. Leiter, E. Staudacher, F. Altmann, HPLC method for the determination of Fuc to Asn-linked GlcNAc fucosyltransferases, Gl-vcoconjJ , 1998, 15, 89-91. 3 I . N. Harpaz, H. Schachter, Control of glycoprotein synthesis. V. Processing of asparagine-linked oligosaccharides by one or more rat liver Golgi a-D-mannosidases dependent on the prior action of UDP-N-acetylg1ucosamine:a-D-mannosideP-2-N-acety~glucosaminyltransferaseI, J Biol Chem, 1980,255,4894-4902. 32. H. Schachter, Molecular Cloning of Glycosyltransferase Genes, In Molecular Glycohiology, M. Fukuda, and 0. Hindsgaul (ed), Oxford University Press, 1994, Oxford, UK, pp 88162. 33. P A . Gleeson, Targeting of proteins to the Golgi apparatus, Histoehem Cell Bid, 1998, 109, 517-532. 34. P.A. Gleeson, R.D. Teasdale, J. Burke, Targeting of proteins to the Golgi apparatus, Glycoc m j J , 1994; 11; 381-394. 35. S. Munro, An investigation of the role of transmembrane domains in Golgi protein retention, EMBO J , 1995, 14, 4695-4704. 36. S. Munro, Localization of proteins to the Golgi apparatus, Tr Cell Biol, 1998, 8, 11-15. 37. T. Nilsson, C. Rabouille, N. Hui, R. Watson, G. Warren, The role of the membrane-spanning domain and stalk region of N-acetylglucosaminyltransferase I in retention, kin recognition and structural maintenance of the Golgi apparatus in HeLa cells, J Cell Sci,1996, 109, 1975-1989. 38. R.D. Teasdale, F. Matheson, P.A. Gleeson, Post-translational modifications distinguish cell surface from Golgi-retained beta 1,4 galactosyltransferase molecules. Golgi localization involves active retention, Glycohioloyjj, 1994, 4, 9 17-928. 39. S. Munro, Sequences within and adjacent to the transmembrane segment of alpha-2,6sialyltransferase specify Golgi retention, Emho J , 1991, 10, 3577-3588. 40. R.Y. Dahdal, K.J. Colley, Specific Sequences in the Signal Anchor of the beta-galactoside alpha-2,6-sialyltransferase Are Not Essential for Golgi Localization-Membrane Flanking Sequences May Specify Golgi Retention, J Biol Chem, 1993,268, 263 10-263 19. 41. J. Burke, J.M. Pettitt, H. Schachter, M. Sarkar, P.A. Gleeson, The transmembrane and flanking sequences of betal,2-N-acetylglucosaminyltransferaseI specify medial-Golgi localization, J Biol Chem, 1992,267, 24433- 24440.

164

9 Brunching qf N-Glycuns: N-Acetylglucosuininyltrunsft.rases

42. J. Burke, J.M. Pettitt, D. Humphris, P.A. Gleeson, Medial-Golgi Retention of N-Acetylglucosaminyltransferase I-Contribution from All Domains of the Enzyme, J Bid Chem, 1994, 269, 12049-12059. 43. K.J. Colley, Golgi localization of glycosyltransferases: More questions than answers, Glycobiology, 1997, 7 , 1-13. 44. T. Nilsson, P. Slusarcwicz, M.H. Hoe, G. Warren, Kin recognition: a model for the retention of Golgi enzymes, FEBS Lett, 1993, 330, 1-4. 45. T. Nilsson, M.H. Hoe, P. Slusarewicz, C. Rabouille, R. Watson, F. Hunte, G. Watzele, E.G. Berger, G. Warren, Kin Recognition Between medial Golgi Enzymes in HeLa Cells, EMBO J , 1994,13, 562-574. 46. P. Slusarewicz, T. Nilsson, N. Hui, R. Watson, G. Warren, Isolation of a Matrix That Binds Medial Golgi Enzymes, J Cell Bid, 1994, 124, 405-413. 47. A.S. Masibay, P.V. Balaji, E.E. Boeggeman, P.K. Qasba, Mutational analysis of the Golgi retention signal of bovine p-I ,4-galactosyltransferase, J Bid Chem, J993, 268, 9908-991 6. 48. S. Munro, A comparison of the transmembrane domains of Golgi and plasma membrane proteins, Biochem Soc Trans, 1995, 23, 527-530. 49. N. Taniguchi, Y. Ihara, Recent progress in the molecular biology of the cloned Nacetylglucosaminyltransferases, Glycoconj J , 1995, 12, 733-738. 50. Y. Nishikawa, W. Pegg, H. Paulsen, H. Schachter, Control of glycoprotein synthesis. XV. Purification and characterization of rabbit liver UDP-N-acetylglucosamine:a-3-D-mannoside fi-1,2-N-acetylglucosaminyltransferase I, J Bid Chem, 1988, 263, 8270-828 1. 51. G. Moller, F. Reck, H. Paulsen, K.J. Kaur, M. Sarkar, H. Schachter, I. Brockhausen, Control of glycoprotein synthesis: Substrate specificity of rat liver UDP-GlcNAc:Manalpha3R beta2N-acetylglucosaminykrdnsferase I using synthetic substrate analogues, Glycoconj J , 1992, 9, 180- 190. 52. F. Reck, M. Springer, E. Meinjohanns, H. Paulsen, I. Brockhausen, H. Schachter, Synthetic substrate analogues for UDP-GlcNAc: Man alpha 1-3R beta 1-2-N-acetylglucosaminyltransferase I. Substrate specificity and inhibitors for the enzyme, Glycoconj J ; 1995, 12, 747754. 53. M. Sarkar, S. Pagny, U. Unligil, D. Joziasse, J. Mucha, J. Glossl, H. Schachter, Removal of 106 amino acids from the N-terminus of UDP-GlcNAc:alpha-3-D-mannoside beta-I .2-Nacetylglucosaminyltransferase I does not inactivate the enzyme, Gljcoconj J , 1998, 15, 193197. 54. M.H. Hoe, P. Slusarewicz. T. Misteli, R. Watson, C. Warren, Evidence for recycling of the resident medial/trans Golgi enzyme, N-acetylglucosaminyltransferase I, in ldlD cells. J Biol Chem, 1995,270, 25057-25063. 55. M. Sarkar, Expression of recombinant rabbit UDP-GlcNAc:a3-D-mannoside p-I ,2-Nacetylglucosaminyltransferase I catalytic domain in St9 insect cells, Glycoconj J , 1994, 11, 204-209. 56. F. Reck, M. Springer, H. Paulsen, I. Brockhausen, M. Sarkar, H. Schachter, Synthesis of Tetrasaccharide Analogues of the N-Glycan Substrate of p 1-2-N-Acetylglucosaminyltransferase I1 using Trisaccharide Precursors and Recombinant p 1-2-N-Acetylglucosaminy~trdnsferase I, Curhohydr Rex, 1994, 259. 93-101. 57. F. Reck, E. Meinjohanns, M. Springer, R. Wilkens, J.A.L.M. Van Dorst, H. Paulsen, G. Moller, I. Brockhausen, H. Schachter, Synthetic substrate analogues for UDP-GlcNAc: Manal-6R PI-2-N-acetylglucosaminyltransferase11. Substrate specificity and reversible and irreversible inhibitors for the enzyme, Gl-ycoconj J , 1994, 11, 210-216. 58. J. Tan, G.A.F. D’Agostaro, B. Bendiak, F. Reck, M. Sarkar, J.A. Squire, P. Leong, H. Schachter, The human UDP-N-acetylglucosamine: alpha-6-D-mannoside-beta-1,2-Nacetylglucosaminyltransferase 11 gene (MGAT2)-Cloning of genomic DNA, localization to chromosome 14q21, expression in insect cells and purification of the recombinant protein. Eur J Biochem, 1995, 231, 317-328. 59. E. Hull, M. Sarkar, M.P.N. Spruijt, J.W.M. Hoppener, R. Dunn, H . Schachter, Organization and localization to chromosome 5 of the human UDP-N-acetylglucosamine:alpha-3-Dmannoside beta-l,2-N-acetylglucosaminyltransferase I gene, Biochem Biophys Res Commun, 1991, 176, 608-61 5.

References

165

60. S. Pownall, C.A. Kozak. K. Schappert, M. Sarkar, E. Hull, H. Schachter, J.D. Marth, Molecular cloning and characterization of the mouse UDP-N-acetylglucosamine: alpha-3-Dmannoside beta-1 ;2-N-acetylglucosaminyltransferase I gene, Genomics, 1992. 12, 699-704. 61. R. Kumar, J. Yang, R.L. Eddy, M.G. Byers, T.B. Shows. P. Stanley. Cloning and expression of the murine gene and chromosomal location of the human gene encoding N-acetylglucosaminyltransferase I, Glycohioloyy, 1992. 2, 383-393. 62. B. Yip, S.H. Chen. H. Mulder, J.W.M. Hoppener, H. Schachter, Organization of the human beta- 1,2-N-acetyIglucosaminyltransferase1 gene (MGAT I ) , which controls complex and hybrid N-glycan synthesis, Biockmni J , 1997, 321, 465-474. 63. J . Yang, M . Bhaumik, Y. Liu, P. Stanley, Regulation of N-linked glycosylation. Neuronal cell-specific expression of a 5’ extended transcript from the gene encoding N-acetylglucosaminyltransferase I, Glycohiolo~qy,1994, 4, 703-7 12. 64. T. Fukada, N. Kioka, J. Nishiu, S. Sakata, H. Sakai. M. Yamada, T. Komano, Different response to inflammation of the multiple mRNAs of rat N-acetylglucosaminyltransferase I with variable 5’-untranslated sequences, FEBS Lett, 1998, 436. 228-232. 65. T. Fukada, K. Iida, N. Kioka, H. Sakai, T. Komano, Cloning of a cDNA Encoding NAcetylglucosaminyltransferase I from Rat Liver and Analysis of Its Expression in Rat Tissues, Biosri Biotechnol Biocizen?, 1994, 58, 200-201. 66. H. Puthalakath, J. Burke. P.A. Gleeson, Glycosylation defect in Lecl Chinese hamster ovary mutant is due to a point mutation in N-acetylglucosaminyltransferase I gene, J Biol Cizetn, 1996, 271, 27818-27822. 67. A.S. Opat. H. Puthalakath, J. Burke, P.A. Gleeson, Genetic defect in N-acetylglucosaminyltransferase I gene of a ricin-resistant baby hamster kidney mutant. Biochenz J , 1998, 336, 593 598. 68. B. Bendiak, H. Schachter, Control of glycoprotein synthesis. XII. Purification of UDPGlcNAc:a-D-mannoside p1-2-N-acetylglucosaminyl-transferaseI1 from rat liver, J Biol Clzrrn, 1987,262, 5775-5783. 69. B. Bendiak, H. Schachter, Control of glycoprotein synthesis. X111. Kinetic mechanism, substrate specificity, and inhibition characteristics of UDP-GlcNAc:u-D-mannoside p1-2-Nacetylglucosaminyltransferase I1 from rat liver, J Biol Chem, 1987, 262, 5784-5790. 70. F. Reck, E. Meinjohanns, J. Tan, A.A. Grey, H. Paulsen, H . Schachter, Synthesis of pentasaccharide analogues of the N-glycan substrates of N-acetylghcosaminyltransferdses 111. IV and V using tetrasaccharide precursors and recombinant beta-( 1 4 2 ) -N-acetylglucosamiIiyltransferase 11. Curboiiydu Res, 1995. 275, 221 229. 71, G.A.F. D’Agostaro, A. Zingoni. R.L. Moritz, R.J. Simpson, H. Schachter, B. Bendiak, Molecular cloning and expression of cDNA encoding the rat UDP-N-acetylglucosamine:alpha-6-Dmannoside beta-1.2-N-acetylglucosaminyltransferase11,J Biol Chern, 1995,270, 15211- 15221. 72. S.H. Chen. S.H. Zhou, J . Tan. H. Schachter, Transcriptional regulation of the human UDPGlcNAc: alpha-6-D-mannoside beta-l-2-N-acetylg~ucosaminyltransferaseI1 gene (MGAT2) which controls complex N-glycan synthesis, Gljwconj J , 1998, 15. 301-308. 73. W. Zhang, L. Revers. M. Pierce, H. Schachter, Transcriptional regulation of the human beta1,2-N-acetylglucosaminyltransferase I I (GnT 11) gene ( M G AT2) by Ets-l and Ets-2, Glycohioloqy, 1998, 8, Abstract 144. 74. T. Leeb, B. Kriegesmann, B.G. Baumgartner, C. Klett, M . Yerle, H. Hameister, B. Brenig, Molecular cloning of the porcine beta- 1.2-N-acetylglucosaminyltransferase11 gene and assignment to chromosome lq23-q27, Bhu Gm Suhj, 1997, 1336, 361-366. 75. M. Sumper, F.T. Wieland, Bacterial glycoproteins. In Glycoproteins, Montreuil, J.. Vliegenthart. J.F.G., and Schachter, H. (ed), Elsevier, 1995. Amsterdam, Vol. 29a, pp 455-473. 76. L. Lehle, W. Tanner, Protein glycosylation in yeast, In Glycopuoteins, Montreuil, J., Vliegenthart, J.F.G., and Schachter, H. (ed), Elsevier, 1995, Amsterdam, Vol. 29a. pp 475-509. 77. B.K. Hayes, G.W. Hart. Glycoproteins of parasites. Glycoproteins of Trypanosoma cruzi, In Glycopproteins und Diseuse, Montreuil, J., Vliegenthart, J.F.G., and Schachter, H. (ed), Elsevier, 1996, Amsterdam. Vol. 30, pp 99- I I 1. 78. S.J. Turco, Glycoproteins of parasites. Glycoconjugates of Leishmania. In Glycopuoteins and Disease, Montreuil, J., Vliegenthart, J.F.G., and Schachter, H. (ed), Elsevier. 1996, Amsterdam, Vol. 30, pp 113-124.

166

9 Branching of N-Glycans: N-Acetylglucosaminyltransferases

79. A. Dieckmann-Schuppert, P. Gerold, R.T. Schwarz, Glycoproteins of parasites. Glycoproteins of malarial parasites, In Glycoproteins and Disease, Montreuil, J., Vliegenthart, J.F.G., and Schachter, H. (ed), Elsevier, 1996, Amsterdam, Vol. 30, pp 125-158. 80. H.H. Freeze, Dictyostelium discoideum glycoproteins: using a model system for organismic glycobiology, In Glycoproteins ZZ, Montreuil, J., Vliegenthart, J.F.G., and Schachter, H. (ed), Elsevier, 1997, Amsterdam, Vol. 2Yb, pp 89-121. 81. S.M. Haslam, K.-H. Khoo, K.M. Houston, W. Harnett, H.R. Morris, A. Dell, Characterisation of the phosphorylcholine-containing N-linked oligosaccharides in the excretorysecretory 62 kDa glycoprotein of Acanthocheilonema viteae, Mol Biochem Parusit, 1997, 85, 53-66. 82. S.M. Haslam, G.C. Coles; E.A. Munn, T.S. Smith, H.F. Smith, H.R. Morris, A. Dell, Haemonchus contortus glycoproteins contain N-linked oligosaccharides with novel highly fucosylated core structures, J Bid Chem. 1996,271, 30561-30570. 83. A.J. Reason, L.A. Ellis, J.A. Appleton, N. Wisnewski, R.B. Grieve, M. McNeil, D.L. Wassom, H.R. Morris, A. Dell, Novel tyvelose-containing tri- and tetra-antennary N-glycans in the immunodominant antigens of the intracellular parasite Trichinella spiralis, Glycobiology, 1994,4, 593-603. 84. G.J. van Dam, A.M. Deelder, Glycoproteins of parasites. Schistosoma glycoconjugates and their role in host-parasite pathological interactions, In Gl,ycoproteins and Diseuse, Montreuil, J., Vliegenthart, J.F.G., and Schachter, H. (ed), Elsevier, 1996, Amsterdam, Vol. 30, pp 159- 182. 85. K. Nyame, D.F. Smith, R.T. Damian, R.D. Cummings, Complex-type asparagine-linked oligosaccharides in glycoproteins synthesized by ‘Schistosoma mansoni’ adult males contain terminal P-linked ‘N’-acetylgalactosamine, J Biol Chern, 1989, 264(6), 3235-3243. 86. J. Srivatsan, D.F. Smith, R.D. Cummings, Schistosoma mansoni synthesizes novel biantennary Asn-linked oligosaccharides containing terminal beta-linked N-acetylgalactosamine, Glycobiology, 1992, 2, 445-452. 87. J.P. Kamerling, J.F.G. Vliegenthart, Hemocyanins, In Glycoproteins II, Montreuil. J., Vliegenthart, J.F.G., and Schachter, H. (ed), Elsevier, 1997, Amsterdam, Vol. 2Yb, pp 123-142. 88. L. Marz, F. Altmann, E. Staudacher, V. Kubelka, Protein glycosylation in insects, In Glycoproteins, J.Montreui1, J.F.G.Vliegenthart, and H.Schachter (ed), Elsevier, 1995, Amsterdam, Vol. 29a, pp 543-563. 89. S. Inoue, Y. Inoue, Fish glycoproteins, In Glycoproteins ZZ, Montreuil, J., Vliegenthart, J.F.G., and Schachter, H. (ed), Elsevier, 1997, Amsterdam, Vol. 29b, pp 143-161. 90. S. Honda, A. Makino, S. Suzuki, K. Kakehi, Analysis of the oligosaccharides in ovalbumin by high-performance capillary electrophoresis, Anal Biochenz, 1990, 191, 228-234. 91. E. Ioffe, P. Stanley, Mice Lacking N-Acetylglucosaminyltransferase 1 Activity Die at MidGestation, Revealing an Essential Role for Complex or Hybrid N-Linked Carbohydrates, Proc Nut1 Acad Sci USA, 1994, 91, 728-732. 92. M. Metzler, A. Gertz, M. Sarkar, H. Schachter, J.W. Schrader, J.D. Marth, Complex asparagine-linked oligosaccharides are required for morphogenic events during post-implantation development, EMBO J , 1994, 13, 2056-2065. 93. R.M. Campbell, M. Metzler, M. Granovsky, J.W. Dennis, J.D. Marth, Complex asparaginelinked oligosaccharides in Mgat 1-null embryos, Glycohiology, 1995, 5, 535-543. 94. R. Campbell, J. Tan, H. Schachter, B. Bendiak, J. Marth, Targeted inactivation o f the murine UDP-GlcNAc:alpha-6-D-mannoside beta-l,2-N-acetylglucosaminyltransferaseI1 gene, Glycobiology, 1997, 7, 1050. 95. J. Jaeken, H. Carchon, H. Stibler, The Carbohydrate-Deficient Glycoprotein SyndromesPre-Golgi and Golgi Disorders?, Glycobiology. 1993, 3, 423-428. 96. J.H.M. Charuk, J. Tan, M. Bernardini, S. Haddad, R.A.F. Reithmeier, J. Jaeken, H. Schachter, Carbohydrate-deficient glycoprotein syndrome type 11-An autosomal recessive N-acetylglucosaminyltransferase I1 deficiency different from typical hereditary erythroblastic multinuclearity, with a positive acidified-serum lysis test (HEMPAS), Eur J Biochem, 1995, 230, 797-805. 97. J. Tan, J. Dunn, J. Jaeken, H. Schachter, Mutations in the MGAT2 gene controlling complex N-glycan synthesis cause carbohydrate-deficient glycoprotein syndrome type 11, an autosomal recessive disease with defective brain development, An7 J Hum Genet, 1996, 59, 810-817.

Rtlfi.rences

167

98. J. Jaeken, H. Schachter, H. Carchon, P. Decock, B. Coddeville. G. Spik, Carbohydrate deficient Glycoprotein syndrome type 11: A deficiency in Golgi localised N-acetylglucosaminyltransferase 11, Arch Dis C l d l . 1994, 71, 123- 127. 99. J. Jaeken, G. Spik, H. Schachter, Carbohydrate-deficient glycoprotein syndrome Type 11: an autosomal recessive disease due to mutations in the N-acetylglucosaminyltransferase 11 gene, In Glycoproteins and Diseuse, Montreuil, J., Vliegenthart, J.F.G., and Schachter; H. (ed), Elsevier, 1996, Amsterdam, The Netherlands, Vol. 30, pp 457-467. 100. E. Ioffe. Y. Liu, P. Stanley. Complex N-glycans in Mgatl null preimplantation embryos arise from maternal Mgatl RNA, Glycobio/o~/y,1997, 7, 913-919. 101. E. Ioffe, Y. Liu, P. Stanley, Essential role for complex N-glycans in forming an organized layer of bronchial epithelium, Proc Nut/ Acad Sci USA, 1996, 93, 11041-1 1046. 102. S.H. Chen, S.H. Zhou, M. Sarkar, A.M. Spence, H. Schachter, Expression of three Caenorhabditis elegans N-acetylghcosdminyltransferase I genes during development, J Bid Chew, 1999, 274, 288-297. 103. C.E. Warren, P.J. Roy. J.G. Culotti. J,W. Dennis, Beta-6-N-acetylglucosaminyltransferases in Cuenorhuhditis eleyans, Glycobioloyy, 1996, 6, 759. 104. R.A. DeBose-Boyd, A.K. Nyame, R.D. Cummings, Molecular cloning and characterization of an alpha 1,3 fucosyltransferase, CEFT- 1, from Caenorhdbditis elegans, G/ycohio/oyy,1998, 8, 905--917. 105. F.K. Hagen, K. Nehrke, cDNA cloning and expression of a family of UDP-N-acetyl-Dgd1actosamine:polypeptide N-acetylgalactosaminyltransferase sequence homologs from Caenorhabditis elegans, J Bid Chew. 1998,273, 8268-8277. 106. W.A. Lubas, D.W. Frank, M. Krause, J.A. Hanover, 0-linked GlcNAc transferase is a conserved nucleocytoplasmic protein containing tetratricopeptide repeats, J Biol Chem. 1997, 272, 93 16-9324. 107. K. Nagai, Y . Ihara, Y. Wada, N . Taniguchi, N-glycosylation is requisite for the enzyme activity and Golgi retention of N-acetylglucosaminyltransferase 111. Glycohiology, 1997. 7, 769-776. 108. Y. Ihara, A. Nishikawa, T. Tohma, H. Soejima, N. Niikawa, N. Taniguchi, cDNA cloning, expression, and chromosomal localization of human N-acetylglucosaminyltransferase 111 (GnT-III), J Biochem, 1993, 113. 692-698. 109. Y.J. Kim, J.H. Park. K.S. Kim. J.E. Chang, J.H. KO, M.H. Kim. D.H. Chung, T.W. Chung, I S . Choe, Y.C. Lee, C.H. Kim, Sequence analysis of the 5’-flanking region of the gene encoding human N-acetylghcosaminyltransferdse 111, Gene, 1996, 170, 28 1-283. 110. N. Koyama. E. Miyoshi, Y. Ihara, R.J. Kang, A. Nishikawa, N. Taniguchi, Human Nacetylglucosaminyltransferase I11 gene is transcribed from multiple promoters, EW J Biochem, 1996,238, 853-861. 11 I . E. Miyoshi, Y. Ihara, N. Hayashi, H. Fusamoto. T. Kamada. N. Taniguchi, Transfection of N-acetylglucosaminyltransferase 111 gene suppresses expression of hepatitis B virus in a human hepatoma cell line, HB611. J Biol Clleni, 1995, 270, 28311-28315. 112. N . Taniguchi, M. Yoshimura, E. Miyoshi, Y . Ihara, A. Nishikawa, S. Fujii, Remodeling of cell surface glycoproteins by N-acetylglucosamiiiyltransferase 111 gene transfection: Modulation of metastatic potentials and down regulation of hepatitis B virus replication, G/j~cobioloyj’, 1996. 6. 691-694. 113. M. Yoshimura, Y. Ihara. A. Ohnishi, N. Ijuhin, T. Nishiura. Y. Kanakura, Y. Matsuzawa, N. Taniguchi, Bisecting N-acetylglucosamine on K562 cells suppresses natural killer cytotoxicity and promotes spleen colonization, Cancer Res. 1996, 56, 412- 418. 114. M. Yoshimura, A. Nishikawa, Y. Ihara, S. Taniguchi. N. Taniguchi, Suppression of lung metastasis of B 16 mouse melanoma by N-acetylglucosaminyltransferase I11 gene transfection, Proc Natl Acad Sci U S A , 1995. 92, 8754-8758. 115. M. Yoshimura, Y. Ihara, Y. Matsuzawa, N. Taniguchi, Aberrant glycosylation of E-cadherin enhances cell-cell binding to suppress metastasis, J Biol Chern, 1996, 271, 13811-13815. 116. A. Rebbaa; H. Yamamoto, T. Saito, E. Meuillet, P. Kim, D.S. Kersey, E.G. Bremer, N. Taniguchi, J.R. Moskal, Gene transfection-mediated overexpression of beta 1,4-N-acetylglucosamine bisecting oligosaccharides in glioma cell line U373 M G inhibits epidermal growth factor receptor function, J B i d Chem, 1997, 272, 9275-9279.

168

9 Brunching of N-Glycans: N-Acetyl~lucosuminyltran~eruses

117. Y. Ihara, Y. Sakamoto, M. Mihara, K. Shimizu, N. Taniguchi, Overexpression of N-acetylglucosaminyltransferase 111 disrupts the tyrosine phosphorylation of Trk with resultant signaling dysfunction in PC12 cells treated with nerve growth factor, J Biol Chem, 1997. 272, 9629-9634. 118. M. Tanemura, S. Miyagawa, Y. Ihara, S. Mikata, H. Matsuda, R. Shirakura, N. Taniguchi, Reduction of the major swine xenoantigen GAL alpha( 1,3)GAL by transfection of N-acetylglucosaminyl transferase I11 (GnT-111) gene, Trunsplunt Proc, 1997, 29, 891-892. 119. M. Tanemura, S. Miyagawa, Y. Ihara, H. Matsuda, R. Shirakura, N . Taniguchi, Significant downregulation of the major swine xenoantigen by N-acetylglucosaminyltransferase 111 gene transfection (Vol 235, pg 359, 1997), Biochenz Biophys Res Commun, 1997, 238, 686-686. 120. M. Tanemura, S. Miyagawa, Y. Ihara, H. Matsuda, R. Shirakura, N. Taniguchi, Significant downregulation of the major swine xenoantigen by N-acetylglucosdminyltransferase 111 gene transfection, Biockem Biophys Res Commun, 1997, 235, 359-364. 121. Y. Sheng, M. Yoshimura, S. Inoue, K. Oritani, T. Nishiura, H. Yoshida, M. Ogawa, Y. Okajima, Y. Matsuzawa, N. Taniguchi, Remodeling of glycoconjugates on CD44 enhances cell adhesion to hyaluronate, tumor growth and metastasis in B16 melanoma cells expressing beta 1,4-N-acetylglucosaminyltransferase 111, Int J Cuncer, 1997, 73, 850-858. 122. J.P. Carver, J.-R. Brisson, The three-dimensional structure of N-linked oligosaccharides, In Biology of’ Curhohydrates, Ginsburg, V., and P.W.Robbins (ed), John Wiley and Sons, 1984, New York, N.Y., Vol. 2, pp 289-331. 123. J.P. Carver, The role of three-dimensional structure in the control of N-linked oligosaccharide biosynthesis, Biochem Soc Trans, 1984, 12, 517-519. 124. J.P. Carver, D.A. Cumming, Site-directed processing of N-linked oligosaccharides: the role of three-dimensional structure. Pure Appl Chern, 1987, 59, 1465-1476. 125. A S . Sultan, E. Miyoshi, Y. Ihara, A. Nishikawa, Y. Tsukada, N. Taniguchi, Bisecting GlcNAc structures act as negative sorting signals for cell surface glycoproteins in forskolintreated rat hepatoma cells, J Biol Chem, 1997, 272, 2866-2872. 126. M. Tanemura, S. Miyagawa, Y. Ihara, A. Nishikawa, M. Suzuki, K. Yamamura, H. Matsuda, R. Shirakura, N. Taniguchi, Suppression of the xenoantigen GAL alpha(l,3)Gal by Nacetylglucosaminyltransferase I11 (GnT-111) in transgenic mice, Trunsplunt Proc, 1997, 29, 895-896. 127. M. Yoshimura, Y. Ihara, T. Nishiura. Y. Okajima, M. Ogawa, H. Yoshida, M. Suzuki, K. Yamamura, Y. Kanakura, Y. Matsuzawa, N. Taniguchi, Bisecting GlcNAc structure is implicated in suppression of stroma-dependent haemopoiesis in transgenic mice expressing N-acetylglucosaminyltransferase 111, Biochem J, 1998, 331, 733-742. 128. Y. Ihara, M. Yoshimura, E. Miyoshi, A. Nishikawa, A S . Sultan, S. Toyosawa, A. Ohnishi, M. Suzuki, K. Yamamura, N. Ijuhin, N. Taniguchi, Ectopic expression of N-acetylglucosaminyltransferase 111 in transgenic hepatocytes disrupts apolipoprotein B secretion and induces aberrant cellular morphology with lipid storage, Pruc Nutl Acud Sci U S A , 1998, 95, 25262530. 129. J.J. Priatel, M. Sarkar, H. Schachter, J.D. Marth, Isolation, characterization and inactivation of the mouse Mgat3 gene: The bisecting N-acetylglucosamine in asparagine-linked oligosaccharides appears dispensable for viability and reproduction, Glycobioloqy, 1997, 7, 4556. 130. M. Bhaumik, S. Sundaram, P. Stanley, Mice lacking GlcNAc-TI11 exhibit several biological consequences, Glycohiology, 1996, 6 , 720. 131. S. Narasimhan, H. Schachter, S. Rajalakshmi, Expression of N-acetylglucosaminyltransferase 111 in hepatic nodules during rat liver carcinogenesis promoted by orotic acid, J Biol Chem, 1988,263, 1273-1281. 132. K. Ishibashi, A. Nishikawa, N. Hayashi, A. Kasahara, N. Sato, S. Fujii, T. Kamada, N. Taniguchi, N-acetylglucosaminyltransferase 111 in human serum, and liver and hepatoma tissues: Increased activity in liver cirrhosis and hepatoma patients, Clin Chim Actu, 1989, 185, 325-332. 133. A. Nishikawa, J. Gu, S. Fujii, N. Taniguchi, Determination of N-acetylglucosaminyltransferases 111, IV and V in normal and hepatoma tissues of rats, Biochim Biophys Actu Gen Suhj, 1990, 1035, 313-318.

References

I69

134. E. Miyoshi, A. Nishikawa, Y. Ihara; J. Gu. T. Sugiyama, N. Hayashi, H. Fusamoto. T. Kamada, N , Taniguchi, N-acetylglucosaminyltransferase I11 and V messenger RNA levels in LEC rats during hepatocdrcinogenesis, C'crncer Res, 1993, 53, 3899-3902. 135. M. Yoshimura, A. Nishikawa, Y. Ihara, T. Nishiura, H. Nakao, Y. Kanayama, Y . Matuzawa, N. Taniguchi, High expression of UDP-N-acetylglucosamine: beta-D mannoside beta1,4-N-acetylglucosaminyltransferase111 (GnT-I 11) in chronic myelogenous leukemia in blast crisis; Int J Cancer, 1995, 60, 443-449. 136. M. Yoshimura, Y. Ihara, N. Taniguchi, Changes of beta-l,4-N-acetylglucosaminyltransferase 111 (GnT-111) in patients with leukaemia, Clycwonj J , 1995, 12, 234-240. 137. Y. Aoyagi, S. Mori, A. Naitoh, M. Yanagi, Y. Suzuki, T. Suda, 0. Isokawa, H. Igarashi, T. Takahashi, M. Isemura, H. Asakura, Alpha-fetoprotein-producing renal cell carcinoma with increased activity of N-acetylglucosaminyltransferase 111, Nephron, 1996, 74, 409% 414. 138. S. Mori, Y. Aoyagi, M. Yanagi, Y. Suzuki, H. Asakura, Serum N-acetylglucosaminyltransferase I11 activities in hepatocellular carcinoma, J Gustroenteuol Hepatol, 1998, 13, 610619. 139. M. Bhaumik, T. Harris, S. Sundaram, L. Johnson, J. Guttenplan, C. Rogler, P. Stanley, Progression of hepatic neoplasms is severely retarded in mice lacking the bisecting N-acetylglucosamine on N-glycans: Evidence for a glycoprotein factor that facilitates hepatic tumor progression, Cancer Res, 1998. 58, 288 1-2887. 140. P.A. Gleeson. H. Schachter, Control of Glycoprotein Synthesis. VIII. UDP-GlcNAc:GnCn (GlcNAc to Mancll-3)f34-N-acetylglucosaminyltransferase IV, an enzyme in hen oviduct which adds GlcNAc in 81-4 linkage to the nl-3-linked Man residue of the trimannosyl core of N-glycosyl oligosaccharides to form a triantennary structure, J Biol Chem, 1983, 258, 61 626113. 141. 1. Brockhausen, G. Moller, J.M. Yang, S.H. Khan, K.L. Matta, H. Paulsen, A.A. Grey, R.N. Shah, H. Schachter, Control of glycoprotein synthesis. Characterization of ( 1 -4)-N-acetylbeta -D-glucosaminyltransferases acting on the alpha-D-( 1 1 3 ) - and alpha-D-( 1+h)-linked arms of N-linked oligosaccharides, Curhohydr Res, 1992, 236, 28 1-299. 142. I. Brockhausen, E. Hull, 0. Hindsgaul, H. Schachter, R.N. Shah, S.W. Michnick, J.P. Carver, Control of glycoprotein synthesis. Detection and characterization of a novel branching enzyme from hen oviduct, UDP-N-acetylglucosamine: GlcNAc~1-6(GlcNAc~1-2)Mana-R (GlcNAc to Man) /3-4-N-acetylglucosaminyltransferase V1, J Biol Chenf, 1989, 264(19), 11211-11221. 143. S. Oguri, M.T. Minowa, Y. Ihara, N. Taniguchi, H. Ikenaga, M. Takeuchi, Purification and characterization of UDP-N-acetylglucosamine: alpha 1,3-D-mannoside beta 1,4-N-acetylglucosaminyltransferase (N-acetylglucosaminyltransferase-IV) from bovine small intestine, J Bid Cliem, 1991,272, 22121-22121. 144. M.T. Minowa, S. Oguri, A. Yoshida, T. Hara, A. Iwamatsu, H. Ikenaga, M. Takeuchi, cDNA cloning and expression of bovine UDP-N-acety1glucosamine:alpha 1,3-D-mannoside beta 1,4-N-acetylglucosaminyltransferaseIV. J Biol Chem, 1998, 273. 11 556-1 1562. 145. A. Yoshida, M.T. Minowa, S. Takamatsu, T. Hara, S . Oguri, H. Ikenaga, M. Takeuchi, Tissue specific expression and chromosomal mapping of a human UDP-N-acetylglucosamine :alpha 1,3-D-mannoside beta 1,4-N-acetylglucosaminyltransferase,Glq'cobioloqy, 1999, 9, 303-3 10. 146. M.T. Minowa, A. Yoshida, T. Hara, A. Twamatsu, S. Oguri, H. Ikenaga, M. Takeuchi. (1997) Cloning and expression of a newly purified N-acetylglucosaminyltransferase. In International Symposium on Glycosyltrdnsferases and Cellular Communications, N. Taniguchi. (ed), Ministry of Education, Science, Sports and Culture, Osaka, Japan, p. 125. 141. A.H.L. Koenderman, P.W. Wijermans, D.H. van den Eijnden. Changes in the expression of N-acetylglucosaminyltransferase 111, IV, V associated with the differentiation of HL-60 cells, FEBS Lett, 1987, 222, 42-46. 148. H. Nakao, A. Nishikawa, T. Karasuno. T. Nishiura, M. lida, Y. Kanayama, T. Yonezawa, S. Tarui, N. Taniguchi, Modulation of N-acetylglucosaminyltrdnsferase 111. IV and V activities and alteration of the surface oligosaccharide structure of a myeloma cell line by interleukin 6, Biochern Biophys Res C'ommun; 1990, 172, 1260 ~1266.

170

9 Brunching of N-Glycuns: N-Acetylglucosuminyltrunsferuses

149. E. Ogier-Denis, C. Bauvy, R. Moutsita, M. Aubery, P. Codogno, Increased UDP-GlcNAc: alpha-Mannoside beta( 1 1 4 ) N-acetylglucosaminyltransferase activity during chick embryo development, Biochim Biophys Acta Mol Cell Res, 1990, 1054, 149-153. 150. K. Yamashita, K. Totani, Y. Iwaki, I. Takamisawa, N. Tateishi, T. Higashi, Y. Sakamoto, A. Kobata, Comparative study of the sugar chains of gamma-glutamyltranspeptidases purified from human hepatocellular carcinoma and from human liver, J Biochem (Tokyo), 1989, 105(5), 728-735. 151. T. Mizuochi, R. Nishimura, C. Derappe, T. Taniguchi, T. Hamamoto, M. Mochizuki, A. Kobata, Structures of the asparagine-linked sugar chains of human chorionic gonadotropin produced in choriocdrcinoma, appearance of triantennary sugar chains and unique biantennary sugar chains, J Biol Chem, 1983,258, 14126-14129. 152. N . Zhang, K.C. Peng, L. Chen, D. Puett, M. Pierce, Circular dichroic spectroscopy of Nacetylglucosaminyltransferase V and its substrate interactions, J Biol Chem, 1997, 272, 42254229. 153. M.M. Palcic, J. Ripka, K.J. Kaur, M. Shoreibah, 0. Hindsgad, M. Pierce, Regulation of N-acetylglucosaminyltransferase V activity. Kinetic comparisons of parental, Rous sarcoma virus-transformed BHK, and L-phytohemagglutinin-resistmt BHK cells using synthetic substrates and an inhibitory substrate analog, J Biol Chrm, 1990, 265, 6759-6769. 154. R.D. Cummings, I.S. Trowbridge, S. Kornfeld, A mouse lymphoma cell line resistant to the leukoagglutinating lectin from Phaseolus vulgaris is deficient in UDP-GlcNAc:a-D-mannoside ~l,6N-acetylglucosaminyltransferase, J Biol Chem, 1982, 257, 13421-13427. 155. 0. Hindsgaul, S.H. Tahir, O.P. Srivastava, M. Pierce, The trisaccharide P-D-GlcNAcpNAc(1+2)-a-D-Manp-( 1+6)-P-D-Manp, as its 8-methoxycarbonyloctyl glycoside, is an acceptor selective for N-acetylglucosaminyltransferase V, Carbohydrate Res, 1988, 173, 263-272. 156. 0. Hindsgaul, K.J. Kaur, G. Srivastava, M. Blaszczyk-Thurin, S.C. Crawley, L.D. Heerze, M.M. Palcic, Evaluation of deoxygenated oligosaccharide acceptor analogs as specific inhibitors of glycosyltransferases, J Biol Chem, 1991, 266, 17858-17862. 157. J. Gu, A. Nishikawa, N. Tsuruoka, M. Ohno, N. Yamaguchi, K. Kanagawa, N. Taniguchi, Purification and characterization of UDP-N-acetylglucosamine:alphd-6-D-mannoside beta1 6N-acetylglucosaminyltransferase (N-acetylglucosaminyltransferase V) from a human lung cancer cell line; J Biochem, 1993, 113, 614-619. 158. O.P. Srivastava, 0. Hindsgaul, M. Shoreibah, M. Pierce, Recognition of oligosaccharide substrates by N-acetylglucosaminyltransferase V, Carbohydrate Rex, 1988, 179, 137- 161. 159. S.C. Crawley, 0. Hindsgaul, G. Alton, M. Pierce, M.M. Palcic, An enzyme-linked immunosorbent assay for N-acetylglucosaminyltransferase-V, A n d Biochem, 1990, 185, 112117. 160. I. Brockhausen, J. Carver, H. Schachter, Control of Glycoprotein Synthesis. XIV. The use of oligosaccharide substrates and HPLC to study the sequential pathway for N-acetylglucosaminyltransferases I, 11, 111, IV, V and VI in the biosynthesis of highly branched N-glycans by hen oviduct membranes, Biocheni Cell Biol, 1988, 66, 1134- 1151. 161. I. Brockhausen, F. Reck. W. Kuhns; S. Khan, K.L. Matta, E. Meinjohanns, H. Paulsen, R.N. Shah, M.A. Baker, H. Schachter, Substrate specificity and inhibition of UDP-G1cNAc:GlcNAc beta 1-2Man alpha 1-6R beta 1-6-N-acetylglucosaminyltransferaseV using synthetic substrate analogues, Glycoconj J , 1995, 12, 371-379. 162. 0. Kanie, S.C. Crawley, M.M. Palcic, 0. Hindsgaul, Acceptor-substrate recognition by Nacetylglucosaminyltransferase-V: Critical role of the 4”-hydroxyl group in P-D-GlcpNAc(1 +2)-u-D-Manp-( 1+6)-fl-D-Glcp-OR, Carhohydr Res, 1993, 243, I39-- 164. 163. S.H. Khan, S.C. Crawley, 0. Kanie, 0. Hindsgaul, A trisaccharide acceptor analog for N acetylglucosaminyltransferase V which binds to the enzyme but sterically precludes the transfer reaction, J Biol Chem, 1993,268, 2468-2473. 164. S.H. Khan, K.L. Matta, Synthesis of 4-nitrophenyl 0-(2-acetamid0-2-deoxy-beta-D-glucopyranosy1)-(1-2) -0-(6-0-rnethyl-alpha-D-mannopyranosyl)-(1+6)-beta-D-glucopyranoside and its 4’,6’-di-O-methyl analog. Potential inhibitors of N-acetylglucosaminyl-transferase V (GnT-V), Carbohydr Res, 1993,243, 29-42. 112)165. S.H. Khan, K.L. Matta, Synthesis of 0-(2-acetamido-2-deoxy-beta-D-glucopyranosyl)-( 0-alpha -D-mannopyranosyl-( 1+6)-O-beta-D-glucopyranosyl-( 1+4)-2-acetdmido-2-deoxy-

References

171

D- glucopyranose. A potential acceptor-substrate for N-acetylglucosaminyltransferase-V (GnT-V): J Curhohydr Chem, 1993, 12, 335 348. 166. S.H. Khan, K.L. Matta, Synthesis of 2-acetamido-2-deoxy-beta-D-glucopyranosyl-( I +2)-alpha-D-mannopyranosyl( 1+6)-beta-D-mannopyranosyl-( 1-i4)-2-acetamido-2-deoxy-D-glucopyranose. Acceptor-substrate recognition by N-acetylglucosaminyltransferase-v (GnT-V), Carbohydr Res, 1995,278, 351-362. 167. T. Linker, S.C. Crawley, 0. Hindsgaul, Recognition of the acceptor beta-D-GlcpNAc-( 1 1 2 ) alpha-D-Manp-( 1+6)-beta-D-Glcp-OR by N-acetylglucosaminyltransferase-V: None of the hydroxyl groups on the Glc-residue are important, Curbohydr Res. 1993, 245. 323-33 1. 168. M.G. Shoreibah, 0. Hindsgaul, M. Pierce, Purification and characterization of rat kidney U DP-N-acetylglucosamine:alpha-6-D-Mannosidebeta- 1,6-N-acetyIglucosaminyltransferase, J Biol Chenz, 1992,267, 2920-2927. 169. L. Chen, N. Zhang, B. Adler, J. Browne, N. Freleen, M. Pierce, Preparation of antisera to recombinant, soluble N-acetylglucosaminy~transferdse V and its visualization in situ, Glycoconjugate J , 1995, 12, 813-823. 170. M. Shoreibah, G.S. Perng, B. Adler, J. Weinstein, R. Basu, R. Cupples, D. Wen, J.K. Browne, P. Buckhaults, N. Fregien, M. Pierce, Isolation, characterization, and expression of a cDNA encoding N-acetylglucosaminyltransferase V, J Biol Clzem, 1993, 268, 15381-15385. 171. J. Weinstein, S. Sundaram, X.H. Wang, D. Delgado, R. Basu, P. Stanley, A point mutation causes niistargeting of Golgi GlcNAc-TV in the Lec4A Chinese hamster ovary glycosylation mutant, J Biol Chem, 1996,271, 27462-27469. 172. H. Saito, J.G. Gu, A. Nishikawa, Y. Ihara, J. Fujii, Y. Kohgo, N. Taniguchi, Organization of the human N-acetylglucosaminyltransferase V gene, Eur J Biochem, 1995, 233, 18-26. 173. P. Stanley, T.S. Raju, M. Bhaumik, CHO cells provide access to novel N-glycans and developmentally regulated glycosyltransferases, GIycohiology, 1996, 6, 695-699. 174. W. Chaney. S. Sundaram. N. Friedman, P. Stanley: The Lec4A CHO glycosylation mutant arises from miscompartmentalization of a Golgi glycosyltransferase, J Cell Biol. 1989. IOY, 2089-2096. 175. J.W. Dennis, S. Laferte, C. Waghorne, M.L. Breitman, R.S. Kerbel, pl-6 branching of Asnlinked oligosaccharides is directly associated with metastasis, Science, 1987, 236, 582-585. 176. J.W. Dennis, S. Laferte, Asn-linked oligosaccharides and the metastatic phenotype, In Altered Glycosylation in Tumor Cell.s, Reading, C.L., Hakomori, S.-I., and Marcus, D.M. (ed), Alan R.Liss Inc., 1988, New York, N.Y., pp 257--267. 177. J.W. Dennis, Changes in glycosylation associated with malignant transformation and tumor progression, In Cell Surjace Carbo1iydrate.s and Cell Denelopment, M.Fukuda (ed), CRC Press, Inc., 1992, Boca Raton, Florida. pp 161-194. 178. L.A. Smets, W.P. Van Beek, Carbohydrates of the tumor cell surface, Biochirn Biophvs Actu. 1984, 738,237-249. 179. S. Yagel, R. Feinmesser, C. Waghorne, P.K. Lala, M.L. Breitman, J.W. Dennis, Evidence that betal-6 branched Asn-linked oligosaccharides on metastatic tumor cells facilitate invasion of basement membranes, Znt J Cuncrr, 1989, 44, 685-690. 180. S. Yousefi, E. Higgins, Z. Daoling, A. Pollex-Kruger, 0. Hindsgaul, J.W. Dennis, Increased UDP-GlcNAc:GalbetaI-3GalNAc-R (GlcNAc to GalNAc) beta-1,6-N-acetylglucosaminyltransferase activity in metastatic murine tumor cell lines. Control of polylactosamine synthesis, J Biol Chern, 1991, 266, 1772-1782. 181. B. Fernandes, U. Sagman, M. Auger, M. Demetrio, J.W. Dennis, betal-6 Branched oligosaccharides as a marker of tumor progression in human breast and colon neoplasia, Cancer Res, 1991, 51, 718-723. 182. M. Demetriou, I.R. Nabi, M. Coppolino, S. Dedhar, J.W. Dennis, Reduced contactinhibition and substratum adhesion in epithelial cells expressing GlcNAc-transferase V, J Cell B i d , 1995, 130, 383 -392. 183. M. Asada, K. Furukawa, K. Segawa, T. Endo. A . Kobata, Increased expression of highly branched N-glycans at cell surface is correlated with the malignant phenotypes of mouse tumor cells. Cancer Res, 1997, 57, 1073 1080. 184. J.W. Dennis, K. Kosh, D.-M. Bryce, M.L. Rreitman, Oncogenes conferring metastatic potential induce increased branching of Asn-linked oligosaccharides in rat2 fibroblasts, Oncogene, 1989, 4, 853-860.

172

9 Branching of N-Glyrans: N-Acetylglucosaminyltran~eruses

185. J.W. Dennis, S. Laferte, Oncodevelopmental expression of -GlcNAcp1-6Manal-6Manplbranched asparagine-linked oligosaccharides in murine tissues and human breast carcinomas, Cuncer Res, 1989, 49, 945-9 50. 186. E.W. Easton, I. Blokland, A.A. Geldof, B.R. Rao, D.H. Van den Eijnden, The metastatic potential of rat prostate tumor variant R3327-MatLyLu is correlated with an increased activity of N-acetylglucosaminyltransferase 111 and V, FEBS Lett, 1992, 308, 46-49. 187. W.K.F. Seelentag, W.P. Li, S.F.H. Schmitz, U. Metzger, P. Aeberhard, P.U. Heitz, J. Roth, Prognostic value of beta 1,6-branched oligosaccharides in human colorectal carcinoma, Cuncer Res, 1998, 58, 5559-5564. 188. M. Yao, D.P. Zhou, S.M. Jiang, Q.H. Wang, X.D. Zhou, Z.Y. Tang, J.X. Gu, Elevated activity of N-acetylglucosaminyltransferase V in human hepatocellular carcinoma, J Cancer Res Clin Oncol, 1998, 124, 27-30. 189. K. Yamashita, Y. Tachibana, T. Ohkura, A . Kobata, Enzymatic basis for the structural changes of asparagine-linked sugar chains of membrane glycoproteins of baby hamster kidney cells induced by polyoma transformation, J Biol Chem, 1985,260, 3963-3969. 190. J. Arango, M. Pierce, Comparison of N-acetylglucosaminyltransferase V activities in Rous Sarcoma-transformed baby hamster kidney (RS-BHK) and BHK cells, J Cell Biochem, 1988, 37, 225-231. 191. Y. Lu, W. Chaney. Induction of N-acetylghcosaminykransferase V by elevated expression of activated or proto- Ha-ras oncogenes, Mol Cell Biochem, 1993, 122, 85-92. 192. M. Pierce, J. Arango, Rous sarcoma virus-transformed baby hamster kidney cells express higher levels of asparagine-linked tri- and tetraantennary glycopeptides containing [GlcNAcbeta( 1,6)Man-alpha(1,6)Man] and poly-N-acetyllactosamine sequences than baby hamster kidney cells, J Biol Chem, 1986, 261, 10772-10777. 193. E.W. Easton, J.G.M. Bolscher, D.H. Van den Eijnden, Enzymatic amplification involving glycosyltransferases forms the basis for the increased size of asparagine-linked glycans at the surface of NIH 3T3 cells expressing the N-ras proto-oncogene, J Biol Chenz, 1991, 266, 21 674-21680. 194. E. Miyoshi, A. Nishikawa, Y. Ihara, H. Saito, N. Uozumi. N. Hayashi, H. Fusamoto, T. Kamada, N. Taniguchi. Transforming growth factor beta up-regulates expression of the N-acetylglycosaminyltransferase V gene in mouse melanoma cells, J Biol Chem, 1995, 270, 6216-6220. 195. R.D. Cummings, S. Kornfeld, The distribution of repeating Galpl-4GlcNAcpl-3 sequences in asparagine-linked oligosaccharides of the mouse lymphoma cell line BW5147 and PHAK 2.1, J Biol Chem, 1984,259, 6253-6260. 196. D.H. van den Eijnden, A.H.L. Koenderman, W.E.C.M. Schiphorst, Biosynthesis of blood group i-active polylactosaminoglycans. Partial purification and properties of an UDPG1cNAc:N-acetyllactosaminide 01 +3-N-acetylg~ucosaminy~transferasefrom Novikoff tumor cell ascites fluid, J Biol Chem, 1988: 263, 12461 12471. 197. P. Buckhaults, L. Chen. N. Fregien, M. Pierce, Transcriptional regulation of N-acetylglucosaminyltransferase V by the src oncogene, J Biol Chem. 1997, 272, 19575-19581. 198. B. Wasylyk, S.L. Hahn, A. Giovane. The Ets family of transcription factors, Eur J Biochem, 1993,211. 7-18. 199. L. Chen, W.J. Zhang, N. Fregien, M. Pierce, The her-2/neu oncogene stimulates the transcription of N-acetylglucosaminyltransferase V and expression of its cell surface oligosaccharide products, Oncogene, 1998, 17, 2087-2093. 200. N. Taniguchi, M. Yoshimura, E. Miyoshi, Y. Ihara, A. Nishikawa, R. Kang, Y. Ikeda, Gene expression and regulation of N-acetylglucosaminyltransferases 111 and V in cancer tissues, A h Enz Reg, 1998,38,223-232. 201. R. Kang, H. Saito, Y. Ihara, E. Miyoshi, N. Koyama, Y. Sheng, N. Taniguchi, Transcriptional regulation of the N-acetylglucosaminyltransferase V gene in human bile duct carcinoma cells (HuCC-TI) is mediated by Ets-I, J B i d Chem, 1996, 271, 26706-26712. 202. M. Pierce, P. Buckhaults, L. Chen, N. Fregien, Regulation of N-acetylglucosaminyltransferase V and Asn-linked oligosaccharide beta( 1,6) branching by a growth factor signaling pathway and effects on cell adhesion and metastatic potential, Glycoconj J , 1997, 14, 623630. -

References

113

203. P.J. Seberger, W.G. Chaney, Control of metastasis by Asn-linked. beta 1-6 branched oligosaccharides in mouse mammary cancer cells, Glycohiology, 1999, 9, 235-241. 204. T.-Z. Ju, H.-L. Chen, J.-X. Gu, H. Qin, Regulation of N-acetylglucosaminyltransferase V by protein kinases, Gly'coconj J , 1995, 12, 767-712. 205. Y. Guilloux, S. Lucas, V.G. Brichard, A. Vanpel, C. Viret, E. Deplaen, F. Brasseur. B. Lethe. F. Jotereau, T. Boon, A peptide recognized by human cytolytic T lymphocytes on HLA-A2 melanomas is encoded by an intron sequence of thc N-acetylglucosaminyltransferase V gcne, J E.up Med, 1996, 183, 1173-1 183. 206. T. Taguchi, T. Ogawa, K . Kitajima, S. Inoue. Y. Inoue, Y. Ihara, Y. Sakamoto, K. Nagai, N. Taniguchi, A method for determination of UDP-GlcNAc:GlcNAc beta 1 -6(GlcNAc beta 12)Man alpha 1-R [GlcNAc to Manlbeta 1-4N-acetylglucosaminyltransferase VI activity using a pyridylaminated tetraantennary oligosaccharide as an acceptor substrate, Anal Biochetn, 1998,255, 155-157. 207. T. Taguchi, K. Kitajima, S . Inoue, Y. Inoue, J.M. Yang, H. Schachter, I. Brockhausen, Activity of UDP-GlcNAc:GlcNAc beta I -6(GlcNAc beta l i 2 ) M a n alpha I+R[GlcNAc to Manlbeta 1+4N-acetylglucosaminyltransferase VI (GnT VI) from the ovaries of Oryzias latipes (Medaka fish), Bioclzem Biophys Res Coninizm, 1997, 230, 533-536. 208. C. Campbell, P. Stanley, A dominant mutation to ricin resistance in Chinese hamster ovary cells induces UDP-G1cNAc:glycopeptidc b-4-N-acetylglucosaminyltransferase I11 activity, J B i d Chem, 1984, 259. 13370-13378. 209. T.S. Raju, M.K. Ray, P. Stanley, LEC18, a dominant Chinese hamster ovary glycosylation mutant synthesizes N-linked carbohydrates with a novel core structure, J Biol Chem, 1995. 270, 30294-30302. 210. T.S. Raju, P. Stanley, LEC14, a dominant Chinese hamster ovary glycosylation mutant cxpresses complex N-glycans with a new N-acetylglucosamine residue in the core region, J Biol Chem, 1996,271, 7484-7493. 21 1. T.S. Raju, P. Stanley, Gain-of-function Chinese hamster ovary mutants LECl8 and LEC14 each express a novel N-acetylglucosaminyltransferase activity, J Biol Chenz, 1998,273, 1409014098. 212. H. Bakker, M. Agterberg, A. Van Tetering, C.A.M. Koeleman, D.H. Van den Eijnden, I. Van Die, A Lymnaea stagnalis gene, with sequence similarity to that of mammalian beta 1i4-galactosyltransferases, encodes a novel UDP-GlcNAc:GlcNAc beta-R beta 1 1 4 - N acetylglucosaminyltransferase, J B i d C/ian, 1994, 269, 30326-30333. 213. M. Sarkar, E. Hull, Y. Nishikawa, R.J. Simpson, R.L. Moritz, R. Dunn, H. Schachter, Molecular cloning and expression of cDNA encoding the enzyme that controls conversion of high-mannose to hybrid and complex N-glycans: UDP-N-acetylglucosamine:a-3-D-mannoside ~-l,2-N-acetylglucosaminyltransferase 1, Pvoc Nut1 Acad Sci USA, 1991, 88, 234-238. 214. R. Kumar, J. Yang, R.D. Larsen, P. Stanley, Cloning and expression of N-acetylglucosaminyltransferase I. the medial Golgi transferase that initiates complex N-linked carbohydrate formation, Pvoc Natl Acad Sci U S A , 1990, 87, 9948-9952. 215. S. Narasimhan. R. Yuen, C.J. Fode, J. Ah, S. Rajalakshmi, K. Schappert, J.W. Dennis, H. Schachter, Cloning of cDNA encoding chicken beta-I ,2-N-acetylglucosaminyltransferase I, Glycohiology, 1993, 3, 531. 216. J. Mucha, S. Kappel, H. Schachter, W. Hane, J. Glossl, Molecular cloning and characterization of cDNAs coding for N-acetylglucosaminyltransferases I and I1 from Xvnopus lueais ovary, Glycoconj J , 1995, 12, 413. 217. A. Nishikawa, Y. Ihara, M. Hatakeyama, K . Kangdwa, N. Taniguchi, Purification, cDNA cloning, and expression of UDP-N-acety1glucosamine:beta-D-mannosidebeta-1,4N-acetylglucosaminyltransferase 111 from rat kidney, J Biol Chem, 1992, 267, 18199-18204. 218. M. Bhaumik, M.F. Seldin, P. Stanley, Cloning and chromosomal mapping of the mouse Mgat3 gene encoding N-acetylghcosaminyltransferase 111, Gene, 1995, 164, 295- 300. 219. H. Saito, A. Nishikawa, J.G. Gu, Y. Ihara, H. Soejima, Y . Wada, C. Sekiya, N. Niikawa, N. Taniguchi, CDNA Cloning and Chromosomal Mapping of Human N-Acetylglucosaminyltransferase V, Biochern Bi0phy.s Res Conitnun, 1994. 198; 318-327. 220. C. Breton, A. Imberty, personal communication. 221, J. Marth, personal communication.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

10 The Galactosyltransferases Nancy L. Shaper, Martin Charron, Neny- Wen Lo, Jane R. Scoccu, and Joel H. Shaper

10.1 Introduction The glycosyltransferases are recognized as a functional family of an estimated 300350 distinct, intracellular, membrane-bound enzymes that participate coordinately in the biosynthesis of the carbohydrate moieties on glycoconjugates. The galactosyltransferases are a subset of the glycosyltransferases that transfer galactose (Gal), usually in the presence of a transition metal ion (e.g., Mnf2), from the nucleotide sugar UDP-a-D-galactose (U DP-Gal) to an appropriate hydroxyl group of an acceptor molecule according to the following reaction: UDP-Gal

+ Acceptor 0Gal-Acceptor + UDP

(1)

Transfer of the galactose can take place either with retention of the anomeric configuration to form aGal-linked glycoconjugates or inversion of the anomeric configuration to form PGal-linked glycoconjugates. In vertebrates, a-galactosyltransferase activities have been described that catalyze the formation of Gala3- and Gala4linked glycosidic bonds and the acceptor sugar is frequently, if not exclusively, a Gal residue. With respect to the P-galactosyltransferases, P3 and P4 activities have been described and acceptor sugars include GlcNAc, Glc, Gal, GalNAc and Xyl. In addition, Gal can be transfered directly to the protein procollagen via a 5-hydroxyL-lysine residue or it can be transferred to ceramide (Cer) forming GalPlCer, the major glycolipid of the myelin sheath. Galactose-containing structures are found in N-linked glycans, 0-linked glycans. the core structures of glycosaminoglycans and glycosphingolipids (Table 1). To date, 15 distinct vertebrate genes encoding galactosyltransferase enzymatic activities have been cloned and characterized. Bascd on sequence comparison and available information on the genomic organization, the majority of these cloned genes can be grouped into three separate gene families termed the P4-galactosyltransferase, the

176

10 The GuluctosL.ltrun.Cfevases

Table 1. Examples of galactose-containing glycans that are assembled in part, by the action of specific galactosyltransferases. A. Outer chain/Non-reducing terminal sequences 1. Type 1 chain (03-N-acetyllactosamine) 2. Type 2 chain (P4-N-acetyllactosamine) 3. aGal epitope 4. Blood group B

Galp3GlcNAcP-R Galp4GlcNAcP-R Gala3Galp4GlcNAc!3-R Gala3Galp4GlcNAcfl-R

I 5. Lewis X (Le’)

Fucu2 Galp4GlcNAcp-R

I 6. Lewis A (Lea)

Fucu3 Galp3GlcNAcP-R

I Fuca4

B. Core structures for 0-glycans 1. Core 1 2. Core 2

Galp3GalNAca-Ser/Thr Galp3GalNAca-Ser/Thr

I GlcNAcP6 C. Core structure of glycosaminoglycans D. GIycosphingolipids 1. Galactosykerdmide 2. Lactosylceramide 4. Lactotetraosylceramide; type 1 (Lc4) 5. Lactoneotetraosylceramide; type 2 (nLc4) 6. Globo series 7. Isoglobo series 8. Ganglio series

GlcUAP3Galp3Galp4XylP3Ser-R Galpl Cer CalP4GlcPICer 1Cer Galp3GlcNAcP3Galp4Glc~ Galp4GlcNAc~3Galfi4GIcp1Cer GalNAc~3Gala4Galp4GIcS1 Cer GalNAcS3Galw3Galp4G1cB1Cer G a l ~ 3 G a l N A c ~ 4 G a l ~ 4 GCer 1c~l

Abbreviations used: ceramide (an N-acylsphingosine), (Cer); fucose, (Fuc); galactose (Gal); Nacetylgalactosamine (GalNAc); N-acetylglucosmaine (GlcNAc); xylose (Xyl); glucuronic Acid (GlcUA).

P3-galactosyltransferase and the a3-galactosyltransferase (or ABO blood group) gene family. The goals of this review are to summarize the current available information on these three vertebrate gene families and to acquaint the reader with the various relevant databases that are useful for obtaining information about specific enzymes. As a direct dividend of the different genome projects, information on additional members of the galactosyltransferase gene families has been accumulating at a rapid rate. It is anticipated that this review can be updated by the reader by means of regularly inspecting the latest additions to the databases. Lastly, this review will focus only on the galactosyltransferases from vertebrates, with particular emphasis being given to P4GalT-I as it is one of the most well characterized of all the glycosyltransferases and consequently has served as the “baseline” for comparison with other galactosyltransferases. A summary of the galactosyltransferases that have been identified in prokaryotes and their structural relation-

10 2 Using the Databanks to Obtuiiz Infomiition on the Gulacto,yltranJJerascr

1I7

ship with their eukaryotic counterparts has been recently compiled [ 11. Recent reviews of glycosylation in yeast have been published in a special issue of Biochimica et Biophysica Acta (vol. 1426, 1999), and reviews on glycosylation in bacteria [2] and plants [3] are also available. For a recent general comprehensive review on glycosylation see [4].

10.2 Using the Databanks to Obtain Information on the Galactosyltransferases Research on the galactosyltransferases over the last four decades has resulted in a wealth of information that is now easily accessible in various databases. The reader is encouraged to explore three databases in particular: 1) Entrez; 2) ExPASy; and 3) CAZY. As detailed below, these databases contain information regarding nucleotide sequence, protein sequence and enzyme properties such as reaction catalyzed, K,, metal ion requirements, inhibitors, etc. Entrez (http://www.ncbi.nlm.nih.gov/Entrez/nucleotide.html) offers 23 search fields (e.g., accession number, protein name, author, keyword, EC number). If the accession number is used. the nucleotide sequence, organism and tissue source of sequence and journal references can be obtained. Entrez also provides direct links to other databases such as PubMed. ExPASy (http://expasy.hcuge.ch/) contains an ENZYME database which can be queried by EC number (all galactosyltransferases are 2.4.1 .-) or a descriptive name (i.e., galactosyltransferase). Its important to note that EC numbers are used to identify reactions catalyzed and not the structures of the proteins that carry out the catalysis; thus the same EC number can include more than one protein sequence. Links are provided to the BRENDA database and the Kyoto University Ligand Chemical Database. BRENDA is protein based and gives information on nomenclature, reaction and specificity, enzyme structure, isolation, purification, stability, cross-references to structure databanks and literature references. While BRENDA does not have a direct link to the nucleotide databases, this link is provided if Entrez is queried using the EC number. The Kyoto University Ligand Chemical Database provides a short summary statement regarding a particular enLyme and at the bottom of the search summary a view of all link information including nucleotide and protein databases is provided. CAZY (Carbohydrate Active enzymes) (http://afmb.cnrs-mrs.fr/-Pedro/ CAZY/db.html) provides an complete list of accession numbers (nucleotide and protein) of glycosyltransferases with an assigned EC number.

10.2.1 Nomenclature

A consensus nomenclature for the different galactosyltransferase enzymatic activities has not been established. An additional complication is the fact that several

178

I0 The Guluctosyltrunsferuses

different systems of nomenclature are being used concurrently. Consequently a specific galactosyltransferase (and corresponding gene) may be identified by several different acronyms. The emerging system of nomenclature for the P3- and 84galactosyltransferases is based on the individual gene families. All members of the 83- and P4-galactosyltransferase family (see Tables 2 and 3) are denoted as P3GalTand P4GalT-, respectively, followed by either a roman or arabic numeral to designate each family member, more or less in the order of discovery. This is the system that is used in the Tables compiled in this review; available EC numbers are included along with any descriptive names that may be associated with a particular enzyme. Lastly, the chromosomal assignments are noted for the human and mouse galactosyltransferase orthologs.

10.3 The Dual Role of P4-Galactosyltransferase-I (P4GalT-I) in Oligosaccharide and Lactose Biosynthesis: The Early Days The enzyme P4-galactosyltransferase-I (EC 2.4.1.38; UDP-Gal:GlcNAcP4galactosyltransferase) is a trans Golgi membrane-bound enzyme which is now recognized to carry out two separate biosynthetic reactions in mammals. In essentially all tissues the primary function of P4GalT-I is to catalyze the transfer of galactose (Gal) from UDP-Gal to N-acetylglucosamine (GlcNAcP-R), forming the P4-Nacetyllactosamine (Galp4GlcNAcP-R) structure frequently found in the nonreducing terminal sequences of glycoproteins, glycolipids and proteoglycans. In mammals, P4GalT-I has been recruited for a second biosynthetic function, the tissue specific production of the disaccharide lactose (GalP4Glc), which takes place exclusively in the lactating mammary gland. The synthesis of lactose is carried out by a protein heterodimer assembled from P4GalT-I and a-lactalbumin (lactose synthase; EC 2.4.1.22). a-Lactalbumin is an abundant, noncatalytic milk protein that is expressed exclusively in the epithelial cells of the mammary gland during lactation. The net result of the binding of a-lactalbumin to P4GalT-I is to lower the K, for glucose -3 orders of magnitude. Consequently, glucose becomes an efficient acceptor at physiological concentrations. The demonstration of the dual role of P4GalT-I in oligosaccharide biosynthesis and lactose biosynthesis is a major cornerstone of Glycobiology. By the early 1960s the sequence of the non-reducing terminal trisaccharide (Sia-GalP4GlcNAc-R) had been established for a number of secreted mammalian glycoproteins [ 5 ] . In addition, a number of cell free systems had been developed from liver and it could be shown that Gal could be transfered from UDP-Gal to an asialoagalactoglycoprotein substrate [6].Using mammary gland extracts, Watkins and Hassid [7] demonstrated that the last step in lactose biosynthesis was the transfer of Gal (from UDP-Gal) to Glc. The puzzling observation was that the liver galactosyltranferase activity could not use Glc efficiently as an acceptor substrate. Conversely, the galactosyltransferase in mammary glands, termed lactose synthase, could not use GlcNAc efficiently as a substrate. These observations suggested the presence of two separate galacto-

2.4.1.38 and 2.4.1.22

P4GalT-I1

2.4.1 .-

2.4.1 .-

2.4.1. I33

P4GalT-V

P4GalT-VI (GalT-2) lactosylceramide synthase

P4GalT-VII Galactosyltransferase I

1p32-34

9pl3

Chromosome (Human)

GalP4XylpSer

5q35

Galp4GlcNAc 3q13 I P6 GalP3GalNAcaSer/Thr G a l~ 4G lc N A c ~3G a l~ 4G lc ~ C e r Galp4GlcNAc 20q 13.1.-13.2‘ I P6 Gal(33GalNAcaSer/Thr Galp4GlcpCer 18qll

GalP4GlcNAcP-R

Galp4GlcNAcP-Rb

Galp4GlcNAcp-Rb

Product formed (underlined)

[621

~ 7 ~ 4 [281 ~ 9 ~ 4

1 1

1 1

Reference

“Human (h), mouse (m), bovine (b), rat (r). chicken (c). the presence of a-lactalbumin, 04GalT-I and -11 transfer Gal to Glc forming lactose (GalP4Glc). The dimer formed from a-lactalbumin and P4GalT-I is termed lactose synthase (E.C. No. 2.4.1.22). ‘The correct chromosomal assignment is chromosome 20; not chromosome 11 [65].

2.4.1_ -

P4GalT-IV

AF038663 (h) AB004550 (h) AF142673 (m) AF048687 (r) AF038664 (h) AF142674(m) AB028600 (h) AF142675 (h)

503880 (m) XI4085 (h) X14558 (b) U19890 (c) U19889 (c) Y12509 (h) AF038660 (h) AF142670 (m) Y12510 (h) AF038661 (h) AF142671(m) AF038662 (h) AF022367 (h) AF142672 (m)

2.4.1.38 and 2.4.1.22

P4GalT-I

2.4.1.-

Accession No.”

E.C. No.

Name

Table 2. The vertebrate fi4-galactosyltransferase gene family.

2

5

9

.a

2.4.1.2.4.1.62

2.4.1.2.4.1.149

P3GalT-111

P3GalT-IV (GDl b/GMl/GAl synthase) GalT-3

b3GalT-V

P3GnT

"Human (h). mouse (m), rat (r)

~~

2.4.1.-

P3GalT-I1

~

2.4.1.-

P3GdlT-1

~

E.C. No.

Name E07739 (h) AF029790 (m) Y15014 (h) Y15060 (h) AF029791 (m) Y15062 (h) AF029792 (m) AB003478 (r) Y15061 (h) AF082504 (m) AB020337 (h) AF145784 (h) AF092050 (h) AF092051 (m)

Accession No."

Table 3. The vertebrate P3-galactosyltransferase gene family.

6p21.3

GM1: Gal~3GalNAcp4(NeuAca3)GalD4GlcCer Galb3GlcNAcp3GalNAc-R ~~

GlcNAcp3Galp4GlcNAc-R

3q25

GalP3GlcNAcp-

2 1q22.3

lq31

Chromosome (Human)

GalP3GlcNAcP-

GdlP3GlcNAcp-

Product formed (underlined)

Reference

0

00

-

syltransferase activities and provided the impetus to purify the putative different enzymes for comparative analysis. The purification of lactose synthase proved to be an intractable problem until Brodbeck and Ebner [8] demonstrated that the partially purified synthase activity could be separated into two protein components (the A-protein and B-protein) by gel filtration chromatography. Neither component had the ability to synthesize lactose by itself; however, when the A-and B-proteins were combined, lactose synthase activity was reconstituted. Subsequently, the authors showed that the Bprotein was the abundant milk protein a-lactalbumin. The last piece of the puzzle was provided when Brew et al. [9] demonstrated that the A-protein was a 84galactosyltransferase activity (P4GalT-I) that transfered Gal to GlcNAc (a P4-Nacetyllactosamine synthase). In addition, when a-lactalbumin was added to the liver galactosyltransferase, the acceptor sugar specificity was also changed from GlcN Ac to Glc. Since the P4-galactosyltransferse was the first glycosyltransferase purified it was originally thought that the number of glycosyltransferases may be limited and that acceptor sugar specificity would be determined by interaction of each enzyme with a different “specifier” protein. Interestingly, after more than 30 years, a-lactalbumin is still the only specifier protein known.

10.3.1 P4GalT-I: Isolation and Characterization of cDNA Clones The search for an efficient purification strategy for the isolation of sufficient quantities of active P4GalT-I ultimately resulted in the introduction of sequential affinitychromatography, using affinity-ligands which mimic both the nucleotide donor substrate and the acceptor sugar substrate [70].Using this strategy, a soluble form of P4GalT-I was purified to apparent homogeneity essentially by a two-step procedure. A direct dividend of this technology was the isolation of sufficient quantites of purified enzyme for both polyclonal antibody production and partial amino acid sequence determination. Using these tools, the catalytic domain of bovine P4GalT-I was independently cloned [ 10, 111. With the completion of the coding sequence it was established that the enzyme was a type I1 membrane-bound protein with a short NH2-terminal cytoplasmic domain, a single transmembrane domain and a large, glycosylated COOH-terminal catalytic domain which is positioned within the Golgi lumen [12-151. This protein domain structure and type 11 membrane topology has proven to be characteristic of all Golgi resident glycosyltransferases cloned to date. Curiously, the Golgi compartment appears to be the main subcellular repository for type I1 membrane-bound proteins.

10.3.2 The Murine P4GalT-I Gene: Genomic Organization and Structure of the 5’-End The P4GalT-I gene is organized in six exons that are distributed over -45 kb of genomic sequence. The organization of the 5’-end of the murine P4GalT-I gene is

182

I0 The Guluc.tos~ltransferases

-

unusual in that three transcriptional start sites are contained within a 725-bp contiguous piece of DNA. The most distal start site is used exclusively during the later stages of spermatogenesis, primarily in haploid round spermatids [ 161. Recently, we have shown that an amazingly short 87-bp genomic fragment, that flanks this P4GalT-I male germ cell specific start site, is sufficient to drive correct male germ cell specific expression of a reporter gene in transgenic mice [17]. Transcription of the P4GalT-I gene in somatic tissues takes place primarily from a start site positioned 500 bp downstream of the germ cell start site. The resulting transcript is 4.1 kb in length. The third, most proximal start site is used predominately in the mammary gland during lactation and results in the production of a 3.9 kb mRNA [12]. Since the 4.1 kb start site is positioned upstream of the first inframe ATG and the 3.9 kb start site is positioned between the first two in-frame ATGs (which are 39 bp apart), translation of each mRNA results in the synthesis of two catalytically identical, structurally related protein isoforms that differ only in the length of their respective, short, NH2-terminal cytoplasmic domains (24 vs 11 amino acids). The position of the two start sites also dictates that the 4.1 kb mRNA has a rather long 5’-untranslated region of -200 nt whereas the 3.9 kb mRNA has a very short 5’-untranslated region of -25 nt.

-

10.3.3 P4GalT-I and Lactose Biosynthesis

-

During lactation, P4GalT-I enzyme levels in the mammary gland increase 50-fold. Thus a regulatory problem arose; how could cells maintain the comparatively low level of constitutively expressed enzyme required for glycoconjugate biosynthesis in all somatic tissues yet specifically increase levels in the mammary gland during lactation? To understand how Nature solved this problem a detailed analysis of the structure and regulation of the murine P4GalT-I gene has been carried out. The main insight from this analysis was that mammals have evolved a two-step mechanism to generate the requisite levels of P4GalT-I required for lactose biosynthesis. In step one, steady state P4GalT-I mRNA levels are upregulated as a result of the switch to the 3.9 kb transcriptional start site that is governed by a stronger mammary gland promoter [18]. In step two, the 3.9 kb p4GalT-I mRNA was demonstrated to be translated more efficiently (three- to five-fold), relative to the constitutively expressed 4.1 kb transcript, due to the deletion of the long GC-rich 5’-untranslated region characteristic of the 4.1 kb mRNA [ 191. 10.3.4 P4GalT-I and the Vertebrate P4GalT Gene Family

Prior to 1997, conventional wisdom, based in part on in vitro assays, held that a single P4-galactosyltransferase was responsible for the biosynthesis of all the P4-Nacetyllactosamine structures present in glycoproteins and glycolipids. In retrospect, the indication of a second P4GalT came from a study in which an a-lactalbumin insensitive UDP-Ga1:GlcNAc P4-galactosyltransferase enzymatic activity was identified in pig trachea [20]. Subsequently, two chicken P4GalT cDNA clones were re-

10.3 The Dual Rok of ~~4-Culu~to,s~~ltmtz.~fera.sr-I (P4GulT-I)

183

ported; one clone (CKP4GalT-I) was shown to be the orthologue of the well characterized human P4GalT-I sequence whereas the second clone (CKP4GalT-II), which also encoded an u-lactalbumin sensitive P4-galactosyltransferase activity, proved to be the orthologue of a previously uncharacterized set of EST clones that mapped to human chromosome 1p32-34 [21]. Subsequently, it was shown [22] that the sequence on lp32- 34, as well as a new set of EST sequences on human chromosome 1q21-23, were P4GalTs (Table 2). Using either classical cloning approaches or additional searching of the dbEST database it has now been established that there are seven P4GalT family members (Table 2). The genes are numbered (34GalT-I to -VII, with P4GalT-I referring to the enzyme activity characterized in the late 1960s. At the protein level, p4GalT-I to -V11 are somewhat similar in size (327-398 amino acids) and are predicted to have the same protein domain structure; i.e., each has a short NH2-terminal cytoplasmic domain, a single transmembrane domain and a large COOH-terminal catalytic domain. In addition, each of the family members contains a set of highly conserved “signature” sequences; most notably, these include the FNRA, DVD and GWGGEDDD motifs (Figure lB). The crystal structure of the P4GalT-I catalytic domain has recently been reported [23]and it shows that one face of the molecule contains a deep pocket involved in UDP-Gal binding. The FNRA motif lines one side of this pocket. The DVD motif is at the bottom of the pocket. The WGWGG sequence is located at the periphery of the pocket with the EDDD residues located at the bottom of the pocket. Two disulfide bridges were also noted in the crystal structure; Cysl34-Cys176 and Cys247-Cys266. While these four Cys residues are conserved in P4GalT-I to -VI they are absent in P4GalT-VII (Figure 1B). The enzymatic activity exhibited by each family member, using synthetic acceptors, is summarized in Table 2. Although P4GalT-I and -11 are both a-lactalbumin responsive ~4-galactosyltransferases,its important to note that only p4GalT-1 is upregulated in the lactating mammary gland and consequently is responsible for lactose biosynthesis in vivo [24]. Moreover, p4GalT-I1 exhibits a more restricted pattern of tissue expression [24]. p4GalT-I11 also exhibits a different expression pattern compared to P4GalT-I, in that it is expressed at high levels in human fetal brain [24]. With respect to P4GalT-IV, Ujita et al. [25] report that it is involved in the synthesis of poly-N-acetyllactosamine in core-2 branched 0-glycans whereas Schwientek et al. [26] report that the enzyme is involved in glycosphingolipid biosynthesis. Two differing activities have also been assigned to P4GalT-V; Sat0 et al. [27] report that the enzyme transfers Gal to GlcNAc-R although the efficiency is reduced when compared to P4GalT-I. van Die et al. [28] report that P4GalT-V is involved in the synthesis of poly-N-acetyllactosamine in core-2, -4 or -6 branched 0-glycans. The activities reported for P4GalT-V are interesting in light of the fact that the amino acid sequence of this enzyme and P4GalT-VI are -70% identical (in fact, of all the family members, these two sequences share the greatest identity; Figure 1A). It is clear that P4GalT-VI is a lactosylceramide (GalP4GlcpCer) synthase. The cloning strategy used by Nomura et al. 1291 included purification of the lactosylceramide synthase activity to apparent homogeneity from rat brain followed by partial amino acid sequence determination. Finally, the newest (and most divergent)

184

I0 The Gu1uctosyltrunsfevclse.s -I (9p) -11 (Ip) -111 (lq) -IV(3q) -V(20q) -VI(18q) -VII(5q)

-I(9p)

100

55

-II(lo)

100 -111 (Iq)

50 I

47

41 I

41

38 I

34

20

33 I

34

I

21

100

47

36

36

20

-IV(3q)

100

34

36

22

100

68

20

-VI(18q)

100

20

-V(20q)

I

100

-VII(Sq)

Figure la. Percent identity at the amino acid level between the human P4GalT gene family members.

FNRA DVD GWGXEDDD

P4GalT-I to -VI

f l c c cc FNRA DVD GWGXEDDD

P4GalT-VII

c

ccc

Figure lb. Domain structure showing position of conserved motifs and Cys residues in the lumenal domain of (34GalT-I to -VTI.

family member, P4GalT-VI1, has recently been identified as the UDP-galactose: 0P-D-xylosylprotein 4-P-~-galactosyltransferase(EC 2.4.1 ,133) that functions in the synthesis of the core structure of glycosaminoglycans (Table 1C [30]).

10.3.5 Evolution of the P4-Galactosyltransferase Gene Family It is generally accepted that the increase in the number of functional genes coincident with, and required for the evolution of vertebrates from non-vertebrates resulted from multiple gene duplication events. To account for this increase in genetic complexity within a relatively short evolutionary time period, two different mechanisms of gene duplication have been proposed. The first mechanism, referred to as tandem duplication, involves duplication of a short genomic segment presumably by unequal crossover, resulting in two copies of a given gene. One gene maintains the original function, while the second is free to to adopt a specialized pattern of expression or evolve a new function, through subsequent divergence in the promotor region or coding sequence, respectively. Gene duplication by this mechanism generally results in gene clustering, with homologs mapping to the same chromosomal region. Examples of tandemly duplicated glycosyltransferase genes include:

10.4 The Vertebrate ~ 3 3 ~ l ~ ~ c . t o c ~ l t m n s(b3GulT) f e m s e Gene Fnmily

185

1) the a2-fucosyltransferases FUT-1 and -2 that map to human chromosome and 3) 19q13.3; 2) the 1x3-fucosyltransferasesFUT-3, -4 and -5 that map to 19~13.3; the w3-galactosyltransferase human pseudogene, the AB blood group transferase gene and the Forssman synthase gene that map to human chromosome 9q34. A second mechanism of gene duplication was proposed by Ohno [31]. Based on a comparative analysis of isozyme number and genome size within the chordates, which consists of both invertebrates and vertebrates, he concluded that the increase in genetic complexity required for the vertebrate radiation could not have been achieved solely by tandem duplication. Rather, he suggested that two separate, whole genome duplications (tetraploidization) occurred in an ancestor of vertebrates. As a result, one copy of a given gene would be present in invertebrates and four copies would be present in vertebrates (assuming gene loss did not occur). As with tandem duplication, one gene copy maintains the orginal function; however, three homologs are now available to evolve either a specialized pattern of expression or a new function. This mechanism of gene duplication can be distinguished from tandem duplication since the four homologs (or tetralogs [32]) are not clustered but are located on four different chromosomes (or different arms of the same chromosome). The fact that each P4GalT gene resides on a different human chromosome or different arm of a given chromosome (Table 2) suggests that the family may have arisen via this mechanism.

10.4 The Vertebrate P3Galactosyltransferase (P3GalT) Gene Family A number of different glycan structures have been identified in which Gal is linked through a P3-glycosidic bond to GlcNAc-R, GalNAc-R, or Gal-R (Table 1). By analogy with the P4GalT family, it might be anticipated that there would be a family of p3GalTs that have evolved by duplication of one or more ancestral genes followed by subsequent divergence and the acquisition of new function(s). Through the independent efforts of several groups, the existence of a P3GalT gene family has been confirmed [33-391. As summarized in Table 3, six distinct family members have been identified either by expression cloning or database searches. Based on the fact that P3GalT-I1 to -V have been mapped to different human chromosomes (Table 3), it is interesting to speculate that they also may have arisen as a consequence of genome tetraploidization, as discussed for the P4GalT gene family. Interestingly the sixth member of this gene family (P3GnT in Table 3) is not a P3GalT but rather a P3-N-acetylglucosaminytransferase (P3GnT; E.C. 2.4.1.149), which based on in vitro assay, appears to function in the biosynthesis of poly-Nfound in N - and 0-linked glycans acetyllactosamine (GlcNAc~3GalP4GlcNAc@-R) and glycosphingolipids [39]. While the sequence identity of P3GnT relative to the other P3GalT family members is 17% (Figure 2A), nevertheless the three amino acid sequence motifs that distinguish this gene family are also present, in the same relative positions. Lastly, it is important to note that this list is not complete; there are a number of

-

186

-I

10 The Guluc.to.s~ltran~~f~ruses

100

51

38

33

34

17

-II(lq)

100

36

29

21

17

-111 (3q)

100

32

31

16

-IV(6p)

100

23

18

-V(21q)

100

17

p3GnT

100

Figure 2a. Percent identity at the amino acid level between the human P3GalT gene family members.

AlRXSrrW

c

YVM/LKTD

c

EDVFRVG

cc

Figure 2b. Domain structure showing the approximate position of conserved motifs and Cys residues in the lumenal domain of P3GalT-I to -V.

specific P3GalT enzymatic activities that have not been accounted for by the cloned sequences and: there are at least three additional sequences in the human dbEST databank that share significant sequence similiarity to P3GalT family members.

10.4.1 General Characteristics of the P3-Galactosyltransferase Gene Family Members At the protein structure level, the P3GalT family members are also organized as type I1 transmembrane proteins. At the amino acid level the individual P3GalT family members exhibit sequence identity with P3GalT-I that ranges from 51% (P3GalT-11) to 17%)(P3GnT) (Fig. 2A). The P3GalT family is characterized by three highly conserved signature motifs (AIRXS/TW; YVM/LKTD; EDVF/YVG) located in the catalytic domain (Figure 2B). By analogy with the P4GalT gene Family it is anticipated that these highly conserved sequences will be positioned at the catalytic center of each enzyme. The genes encoding the P3GalT family members appear to be more compact compared to the P4GalT (and a3GalT) family members, in that the coding sequence is contained in a single exon. Based on the available information regarding their respective acceptor sugar substrates, the five family members that are p3-galactosyltransferases can be subdivided into two groups. Group 1 consists of P3GalT-I, -11, -111 and -V and each enzyme shows a preference for GlcNAc-R as the acceptor sugar substrate. Thus, members of this group are UDP-galactose: P-N-acetylglucosamine ~3-galactosyltransferases, and are responsible for the synthesis of the type 1 chains (P3-N-acetyllactosamine) that are positioned at the non-reducing terminus of glycoconjugates and serve as the

10.4 The Vertebrate ~3Galuc~tos~ltran.~feruse (83GulT) Gene Fumily

187

precursor for the Lewis A and sialylated Lewis A antigen (Table 1). Based on in vitro assays with low molecular weight, artificial substrates, P3GalT-I, -11, and -111 also appear to be able to galactosylate Gal-R [34-361. However this substrate is galactosylated much less efficiently relative to GlcNAc-R, and consequently it is not clear if it would be used in vivo. P3GalT-I, -11, and -111 appear to prefer to use glycolipids as their substrate, rather than glycoproteins. Galactosylation of glycans with non-reducing terminal GalNAc (GalNAc-R) could not be detected. P3GalT-V is the latest member of the Group 1 enzymes to be reported and appears to function in mucin glycoprotein biosynthesis [37, 381. P3GalT-V shows a distinct substrate preference for the 0-linked core 3 sequence, GlcNAcp3GalNAcR. Thus P3GalT-V may be the ortholog of the p3-galactosyltransferase previously characterized from pig trachea [40]. What is interesting about the Group 1 family members is that, based on Northern analysis, they appear to exhibit tissue restricted expression. In a panel of human tissues, p3GalT-I was detected at low levels only in the adult brain. P3GalT-11, -111 were most highly expressed in the heart and at somewhat lower levels in the brain [36]. In mouse tissuse, P3GalT-I and -11 are expressed at high levels in the brain whereas P3GalT-I11 was expressed at comparable levels in brain, ovary, uterus and stomach 1351. In contrast P3GalT-V is expressed primarily in gastrointestinal and pancreatic epithelia and in established tumor cell lines derived from these tissues [37].

10.4.2 P3GalT-IV: UDP-ga1actose:GM1 P3-galactosyltransferase(GM1 Synthase; GalT-3) Group 2 consists of a single member P3GalT-IV which is the GM1 synthase and consequently functions in the biosynthesis of the gangliosides. P3GalT-IV transfers Gal to the non-reducing terminal GalNAc-R of the ganglioside GM2 (GalNAcP4[Siaa3]Gal~3Gal~-Cer). Both asialoGM2 (GA 1 ) and GD2 are efficient substrates [33].

10.4.3 Other Vertebrate P-GalactosyltransferaseActivities There are several P-galactosyltransferases that catalyze the transfer of Gal to an acceptor other than a non-reducing terminal monosaccharide. This group includes the hydroxyllysine galactosyltransferase (EC 2.4.1.50) which transfers Gal to hydroxyllysine present in proteins such as procollagen. From this group, only the ceramide U DP-galactosyltransferase has been cloned and characterized at the molecular level (411.

10.4.4 UDP-Ga1actose:Ceramide P-Galactosyltransferase (CGalT; EC 2.4.1.45) The ceramide P-galactosyltransferase (CGalT; 2-hydroxyacylsphinogosine 1 +galactosyltransferase) transfers Gal from UDP-Gal to ceramide to form galactosyl-

188

I0 The Galactosyltransfera.~es

ceramide. Galactosylceramide is a major glycosphingolipid of the lipid bilayer of the myelin sheath and is also present in polarized epithelial cells (reviewed in [42]. CGalT follows none of the general rules established for the other galactosytransferases characterized to date. CGalT is a resident protein of the endoplasmic reticulum and the nuclear envelope as established by immunocytochemistry at the EM level [43]. Second, it is organized as a type I membrane-bound protein with its catalytic domain positioned within the lumen of the endoplasmic reticulum. Third, CGalT exhibits no amino acid sequence identity with any other P-galactosyltransferase. However the amino acid sequence of CGalT does show sequence identity in the range of 25%) with the mammalian UDP-glucuronyltransferases, which are a large gene family of detoxifying enzymes [44]. This conservation of amino acid sequence and a similiar genomic organization suggests a common evolutionary origin of CGalT and the glucuronyltransferases [41, 421. Comparision of CGalT with the ceramide glucosyltransferase (GlcT-1; glucosylceramide synthase; EC 2.4.1.80) is interesting. GlcT-1 transfers glucose from UDP-Glc to ceramide forming glucosylceramide, which is the core structure for hundreds of glycosphingolipids (Table 1). Since both GlcT-1 and CGalT each transfer a monosaccharide directly to ceramide and the only difference between Glc and Gal is the position of the hydroxyl group at the fourth carbon atom (C-4), one might anticipate that these two transferases would be related. However, GlcT-1 exhibits no amino acid sequence identity with CGalT, indicating independent evolutionary pathways for these two enzymes [45].

10.5 The Vertebrate a3-Galactosyltransferase Gene Family In vertebrates, a-galactosyltransferase activities have been described which catalyze the transfer of Gal, a3- or a4- linked, to the acceptor sugar Gal. However, at the present time only genes encoding a3-galactosyltransferase enzymatic activities have been cloned and characterized (reviewed by [46]). As shown in Table 4, the a3GalT gene family is currently comprised of three members; the first two members are a3galactosyltransferases that function in the biosynthesis of the w-Gal epitope and the blood group B determinant, respectively (Table 1). The third member of this gene family is an a3-N-acetylgalactosaminyltransferasethat is involved in the synthesis of the Forssman antigen.

10.5.1 a3-Galactosyltransferase(a3GalT: UDP-Gal:Galp4GlcNAca3Galactosyltransferase;EC 2.4.1 37) a3GalT is a Golgi membrane-bound enzyme that catalyzes the transfer of Gal from UDP-Gal to the P4-N-acetyllactosamine structure (GalP4GlcNAc-R) present at the nonreducing terminus of either glycolipids or glycoproteins [47]. Consequently, the a3-linked Gal occupies the terminal nonreducing position of N-acetyllactosamine-

2.4.1 .87b

2.4.1.37 2.4.1.88

a3GalT

a3GalT (blood group B-transferase) a3GalNAcT (Forssman glycolipid synthetase)

504989 (b) M26925 (m) 505421 (h) M60263 (h) M85153 (m) L36535 (p) AF006673 (h) U66140 (d) AC002319 (h)

Accession No."

"Human(h). bovine (b). mouse (m). pig (p). dog (d). bThis enzyme has two additional EC numbers; 2.4.1.124 and 2.4.1.151.

E.C. No.

Name

Table 4. The vertebrate n3-galactosyltransferase gene family

~[Fuca2]~~4GlcNAc-R GalNAca3GalNAc P3Gala4GalPR

Gala3Gal P4GlcNAc-R

Product formed (underlined)

9q34 (h)

1551 is71

~ 7 1 [69]

2 (m) lq2.1O-q2.11 (p) 9q34 (h)

[511

POI

1661 [681

Reference

9q34 (h) 12q13--14 (h)

Chromosome

2s

'" -+ ,

2

c. -+.

Q

I

P

?

R

rb

E h

rb

z

T w

2

T

2

b

P

k

190

I0 The Galuctos)~ltran~erases

type carbohydrate chains and of lactosaminoglycans, and as such it is a noncharged alternativc to chain termination by sialic acid. a3GalT cannot use lactosylceramide (GalP4GlcpCer) as a substrate nor can it transfer Gal to fucosylated substrates such as the blood group H structure (Fuca2GalP3/4GlcNAc-R), or the Lewis X structure (Gal[34[Fuccr3]GlcNAc-R).In this respect it differs from the blood group B a3GalT as discussed below. a3GalT and the corresponding a3-galactosylated cell surface glycoconjugates show both a tissue and species-specific expression. Although this enzyme is widely expressed in many mammalian species, enzymatic activity has not been detected in Old World monkeys, apes, and man (reviewed in [48, 491). Interestingly, these species also express high levels of antibody (-1% of circulating IgG) to the a-Gal epitope. It has been speculated that antibody production is the result of immune stimulation by gastrointestinal bacteria that express various a-galactosyl antigens (reviewed in [48]). The question of what happened to the human a3GalT gene was answered when it was shown that the human genome contains two homologs that are inactive. The first homolog, located on chromosome 9q34 is a pseudogene that contains a frameshift mutation that results in a truncated protein [50, 511. This sequence is thought to constitute the remnant a3GalT human gene because it contains at least two introns. The second homolog is an intronless pseudogene organized as a retroposon that is located on chromosome 12 [51]. Continued interest in the enzyme results from the fact that the presence of the a-Gal epitope is a major impediment to xenotransplantation from non-primates such as pig or from primates such as New World monkeys, into humans. Although various experimental strategies to prevent xenograft rejection have been investigated, none has yet proven effective. These include: 1) introduction of the a2 fucosyltransferase gene into the donor to produce the fucosylated H structure (a poor substrate for a3GalT) [52]; 2) introduction of an a-galactosidase gene and the a2-fucosyltransferase gene into the donor [53];and 3) overexpression of the a3sialyltransferase gene in the donor; the 1x3-sialyltransferasecompetes with a3GalT for a common acceptor substrate [54].

10.5.2 The Blood Group B a3-Galactosyltransferase (EC 2.4.1.37) The ABO blood group antigens were the first major alloantigens recognized in humans. It was subsequently established that the antigens (A, B and H) were oligosaccharides and were not exclusive to red blood cells but were also present on epithelial cells. A, B and AB individuals express glycosyltransferase activities converting the H antigen into A or B antigens whereas O(H) individuals lack these activities. The blood group A-transferase transfers GalNAc in an al,3-linkage to the Gal residue of the blood group H structure, Fuca2GalP4GlcNAc-R, present on glycoproteins and glycolipids. In contrast, the blood group B-transferase transfers Gal in an al,3-linkage to the Gal residue of the blood group H structure. Its important to note that efficient acceptor substrates for the B-transferase must contain

a fucose residue ul,2-linked to the terminal Gal and it is this feature which distinguishes this enzyme from u3GalT. The genetic basis of the ABO system was clarified when Yamomoto et al. [55] cloned the gene(s) and showed that only a single locus was present. The A- and B-transferases differ by only 4 amino acids. A single base deletion near the NH2terminus results in the 0 phenotype. The shift in reading frame caused by the deletion results in the translation of an entirely different protein. Subsequent mapping has shown that the ABO locus is located on human chromosome 9q34. When the amino acid sequence was compared to other cloned glycosyltransferases the only significant match observed was to u3GalT (EC 2.4.1.87). The identity was -50% when the COOH-terminal two-thirds of each protein were used for comparison.

10.5.3 The Forssman Glycolipid Synthetase (EC 2.4.1.88) This enzyme, which transfers GalNAc in an a3-linkage to GalNAc, is responsible for the synthesis of the Forssman glycolipid (GalNAca3GalNAcp3Gala4Galp4. GlcpCer). Expression of this glycolipid is regulated durine murine development and it has been suggested that it is involved in tissue morphogenesis [56]. The sequence, which has been cloned from a canine cDNA library [57],shows identity at the amino acid level to both the B-transferase (42%) and to u3GalT (35%) if the COOH-terminal two-thirds of each protein is used for comparison. Although it has been reported that the Forssman glycolipid is expressed in certain disease states in humans, a human cDNA sequence has not yet been published. However, the genome sequencing project has deposited a genomic sequence in GenBank that is 8S'% identical to the COOH-terminal half of the canine amino acid sequence. When this sequence and the canine sequence are aligned, the human sequence appears to contain a deletion in the NH2-terminal half of the protein indicating that the gene may be inactivated. The lack of any matching sequences in the human dbEST database (as of 7/99) further supports this idea.

10.5.4 Evolution of the a3GalT Gene Family As shown in Table 4, all three gene family members map to human chromosome 9q34. This suggests that these genes originated by tandem duplication from a single ancestral gene. This hypothesis is consistent with the observed sequence identity between the ABO blood group transferases, a3GalT and the Forssman glycolipid synthetase. In addition, the ABO blood group transferase gene and the u3GalT gene have a similar genomic organization. After duplication of the ancestral gene -400 million years ago, Saitou et al. [58]proposed that one gene copy might have evolved into the ABO genes and the other one into the human a3GalT gene sequence on chromosome 9q34. At some point in time, after the divergence of New World monkeys from the Old World primates, the human u3GalT gene likely gave rise to an mRNA that, after reverse transcription, was inserted into chromosome 12 giving rise to the intronless pseudogene.

192

10 The (;uluctos?iltran~erases

10.6 A UDP-Gal:Galp3GalNAc a4Galactosyltransferase Activity A number of glycans containing an 1x4-linkedGal have been reported in mammals; these include the human blood group PI antigen, Gala4Galp4GlcNAcp3Galp4. GlcpCer; the P antigen GalNAc~3Gala4Galp4GlcpCer; and the Pk antigen, Galw4Galp4Glc~Cer. However, the corresponding a4-galactosyltransferase activity(s) have not been purified or cloned. Recently an u4GalT enzymatic activity has been characterized from the insect Mamestra brassicae cell line. The prefered acceptor substrate for this activity was demonstrated to be Galp3GalNAc-R; Galp3GlcNAc-R was galactosylated about ten-fold less efficiently [59]. Whether this activity will serve as an entre into the isolation and characterization of a vertebrate 1x4-galactosyltransferaseactivity awaits further studies.

Acknowledgments This work was supported in part by a National Institutes of Health Grant CA45799. Martin Charron is a post-doctoral fellow supported by a Human Frontiers Science Program Grant RG-414/94 M. Neng-Wen Lo is a post-doctoral fellow supported by Grant 960188 (to J.H.S.) from the Mizutani Foundation for Glycoscience.

References 1. Breton, C., Bettler, E., Joziasse, D.H.. Geremia. R.A. and Imberty, A. Sequence-function relationships of prokaryotic and eukaryotic galactosyltransferases. J. Biocheni. (Tokyo) 1998. 123, 1000-1009. 2. Messner, P. Bacterial glycoproteins. Gljwconj. J. 1997, 14, 3-1 1. 3. Lerouge, P., Cabanes-Macheteau, M., Rayon. C., Fischette-Laine, A.C., Gomord, V., and Faye, L. N-glycoprotein biosynthesis in plants: recent developments and future trends. Plrrnt Mol. B i d . 1998, 38, 31-48. 4. Sears, P. and Wong, C.-H. Enzyme action in glycoprotein synthesis. Cell. Mol. L(/g Sci, 1998. 54, 223-252. 5. Spiro, R.G. Studies on the Monosaccharide Sequence of the Serum Glycoptotein Fetuin. J. Bid. Clzem. 1962. 237, 646-652. 6. McGuire, E.J., Jourdian, G.W., Carlson, D.M. and Roseman, S. Incorporation of D-Galactose into Glycoproteins. J. Bid. Cliem. 1965, 240, PC4112-4115. 7. Watkins, W. and Hassid, W.Z. The Synthesis of Lactose by Particulate Enzyme Preparations from Guinea Pig and Bovine Mammary Glands. J. Bid. Chem. 1962, 237, 1432-1440. 8. Brodbeck, U. and Ebner, K.E., Resolution of a soluble lactose synthetase into two protein components and solubilization of microsomal lactose synthetase. J. Bid. Chem 1966; 241, 1391-1397. 9. Brew. K., Vanaman, T.C. and Hill, R.L. The Role of a-Lactalbumin and the A Protein in Lactose Synthetase: a Unique Mechanism for the Control of a Biological Reaction. Pror. Nut/ Acad. Sci U.S. A . 1968, 59, 49 1-497. 10. Shaper, N.L., Shaper, J.H., Meuth, J.L., Fox, J.L., Chang, H., Kirsch, I.R., Hollis, G.F. Bovine galactosyltransferase: identification of a clone by direct immunological screening of a cDNA expression library. Proc.. Nut1 Acrid. Sci. U.S.A . 1986, 83, 1573-1577.

11. Narimatsu, H., Sinha, S., Brew. K., Okayama, H., Qasba, P.K. Cloning and sequencing of cDNA of bovine N-acetylglucosamine (131 -4)galactosyltransferase. Proc. Nut/ Acud. Sci. U.S.A.

1986, 83, 4720-4724. 12. Shaper, N.L., Hollis, G.F., Douglas. J.G., Kirsch, I.R. and Shaper, J.H. Characterization of the full length of cDNA for murine P-I ,4-galactosyltransferase: Novel features at the 5’-end predict two translational start sites at two in-frame AUG’s. J. Bid. Chem. 1988, 263, 1042010428. 13. D’Agostaro, G., Bendiak, B. and Tropak, M. Cloning of cDNA encoding the membranebound form of bovine pl,4-galactosyltransferase.Eur. J. Biuchem. 1989, 183, 21 1-217. 14. Russo, R.N., Shaper, N.L. and Shaper, J.H. Bovine pl ,4-galactosyltransferase: two sets of mRNA transcripts encode two forms o f the protein with different amino-terminal domains. In vitro translation experiments demonstrate that both the short and the long forms of the enzyme are type I1 membrane-bound glycoproteins. J. Bio/. Cheni. 1990, 265, 3324-3331. 15. Russo, R.N., Shaper, N.L., Taatjes, D.J. and Shaper, J.H. ~1,4-galactosyltransferase: a short NH2-terminal fragment that includes the cytoplasmic and transmembrane domain is sufficient for Golgi retention. J. Biol. Chem. 1992, 267. 9241-9247. 16. Harduin-Lepers, A., Shaper, N.L., Mahoney, J.A. and Shaper. J.H. Murine [31,4-galactosyltransferase: round spermatid transcripts are ChdrdctKriZed by an extended 5’-untranslated region. G/ycohio/oqy 1992, 2, 361-368. 17. Charron, M., Shaper, N.L., RaJpu, B. and Shaper, J.H. A novel 14-base-pair regulatory element i s essential for in vivo expression of inurine 04-galactosyltransferase-Iin late pachytene spermatocytes and round spermatids. Mol. Cell. Biol. 1999, 19, 5823-5832. 18. Rajput, B., Shaper, N.L. and Shaper. J.H. Transcriptional regulation of muriiie p1,4galactosyltransferase in somatic cells. Analysis of a gene that serves both a housekeeping and a mammary gland-specific function. J. Bid. Chcwi. 1996, 271, 5 13I 5 142. 19. Charron, M., Shaper, J.H. and Shaper, N.L. The increased level of pI,4-galactosyltransferase required for lactose biosynthesis is achieved in part by translational control. Proc. Nut/ Arud. Sci U.S.A. 1998, Y5, 14805-14810. 20. Sheares. B.T. and Carlson, D . M . Two distinct UDP-galactose: 2-acetaniido-2-deoxy-D-glucose 4 P-galactosyltransferases in porcine trachea. J. Biol. Cheni. 1984. 259, 8045-8047. 21. Shaper, N.L., Meurer, J.A., Joziasse, D.H., Chou, T.D.. Smith, E.J.. Schnaar, R.L. and Shaper, J.H. The chicken genome contains two functional nonallelic ~1,4-galactosyltransferase genes. Chromosomal assignment to syntenic regions tracks fate of the two gene lineages in the human genome. J. B i d . Chem. 1997, 272; 31389-31899. 22. Almeida, R., Amado, M.. David, L., Levery. S.B., Holmes, E.H.. Merkx, G.. Van Kessel, A.G., Rygaard, E., Hassan, H., Bennett, E. and Clausen, H . A family of human P4-galactosyltransferases. Cloning and expression of two novel UDP-galactose:P-N-acetylglucosamine D l ,4-galactosyltransferases.P4CaILT2 and p4GaLT3. J. Biol. Chem. 1997, 272. 3 197% 31991. 23. Gastinel, L.N., Cambillau, C. and Bourne, Y. Crystal structures o f the bovine B4galactosyltransferase catalytic domain and its complex with uridine diphosphogalactose. EMBO J. 1999, 18. 3546--3551. 24. Lo. N.W.. Shaper, J.H., Pevsner. J. and Shaper, N.L. The expanding P4-galactosyltransferase gene family: messages from the databanks. G~ycohiolo~qy 1998, 8 , 51 7-526. 25. Ujita, M.. McAuliffe, J., Schwientek. T., Almeida, R., Hindsgaul, O., Clausen, H. and Fukuda, M. Synthesis of poly-N-acetyllactosaminc in core 2 branched 0-glycans. The requirement of novel B-I ,4-galactosyltransferase 1V and p- I ,3-N-acetylglucosaminyltransferase. J. Bid. Chern. 1998,273. 34843-34849. 26. Schwientek, T., Almeida, R., Levery, S.B.. Holmes. E.H., Bennett, E. and Clausen, H. Cloning of a novel member of the UDP-ga1actose:P-N-acetylglucosamine D l ,4-galactosyltransferase family, P4GaLT4, involved in glycosphingolipid biosynthesis. J . Biol. Chem. 1998. 273, 2933 1 29340. 27. Sato, T.: Furukawa, K., Bakker, H.. van den EiJnden, D.H. and van Die. 1. Molecular cloning of a human cDNA encoding P-1,4-galactosyltransferasewith 37% identity to mammalian UDP-Gal:GlcNAc ~-1,4-galactosyltransferase. Proc. Nut/ Acud. Sci. U.S.A . 1998, YS, 472471. ~

-

194

10 The Galactosyltran.sferases

28. van Die, I., van Tetering, A., Schiphorst, W.E., Sato, T., Furukawa, K. and van den Eijnden, D.H. The acceptor substrate specificity of human P4-galactosyltransferase V indicates its potential function in 0-glycosylation. FEBS Lett. 1999, 450, 52-56. 29. Nomura, T., Takizawa, M., Aoki, J., Arai, H . , Inoue, K., Wakisaka, E., Yoshizuka, N., Imokawa, G., Dohmae, N., Takio, K., Hattori, M. and Matsuo, N. Purification, cDNA cloning, and expression of UDP-ga1:glucosylceramide 0- 1,4-galactosyltransferasefrom rat brain. J. Biol. Chem. 1998,273, 13570-13577. 30. Okajima, T., Yoshida, K., Kondo, T. and Furukawa. K. Human Homolog of Cuenorhahditis eleguns sqv-3 Gene is Galactosyltransferase 1 Involved in the Biosythesis of the Glycosaminoglycan-Protein Linkage Region of Proteoglycans. J. Biol. Chem. 1999, 274, 22915-22918. 31, Ohno, S. Evolution by gene duplicution. 1970 Springer-Verlag, New York. 32. Spring, J. Vertebrate evolution by interspecific hybridisation-are we polyploid? FEBS Lett. 1997, 400, 2-8. 33. Miyazaki, H., Fukumoto, S., Okada, M., Hasegawa, T. and Furukawa, K. Expression cloning of rat cDNA encoding UDP-galactose:GD2P1,3-galactosyltransferase that determines the expression of G D ~ ~ / G M I / G J. ABid. I . Chem. 1997, 272, 24194-24799. 34. Kolbinger, F., Streiff, M.B. and Katopodis, A.G. Cloning of a human UDP-galactose:2-acetamido-2-deoxy-D-glucose 3P-galactosyltransferase catalyzing the formation of type 1 chains. J. Bid. Chem. 1998, 273, 433-440. 35. Hennet, T., Dinter, A,, Kuhnert, P., Mattu, T.S., Rudd, P.M. and Berger, E.G. Genomic cloning and expression of three murine UDP-galactose: P-N-acetylglucosamine P I ,3-galactosyltransferase genes. J. Biol. Chem. 1998, 273, 58-65. 36. Amado, M., Almeida. R., Carneiro, F., Levery, S.B., Holmes, E.H., Nomoto, M., Hollingsworth, M.A., Hassan, H., Schwientek, T., Nielsen, P.A., Bennett, E.B. and Clausen, H. A family of human (13-galactosyltransferases.Characterization of four members of a UDPgalactose:P-N-acetyl-glucosamine/P-N-acetyl-galactosamine P-I ,3-galactosyltransferase family. J. Biol. Chem. 1998,273, 12770-12778. 37. Isshiki, S., Togayachi, A,, Kudo, T., Nishihara, S., Watanabe, M., Kubota, T., Kitajima, M., Shiraishi, N., Sasaki, K., Andoh, T., and Narimatsu, H. Cloning, expression, and characterization of a novel UDP-galactose: P-N-acetylglucosamine ~1,3-galactosyltransferase (P3Gal-T5) responsible for synthesis of type 1 chain in colorectal and pancreatic epithelia and tumor cells derived therefrom. J. Biol. Chem. 1999, 274, 12499-12507. 38. Zhou, D., Berger, E.G. and Hennet, T. Molecular cloning of a human UDP-ga1actose:GlcNAc p1,3GalNAcPI ,3galactosyltransferase gene encoding an 0-linked core3-elongation enzyme. Eur. J. Biochem. 1999, 263, 571-576. 39. Zhou, D., Dinter, A,, Gallego, R.G., Kamerling, J.P., Vliegenthart, J.F., Berger, E.G. and Hennet, T.A. ~-1,3-N-acetylglucosaminyltransferase with poly-N-acetyllactosamine synthase activity is structurally related to P-1,3-galactosyltransferases.Proc. Nut/ Acud. Sci. U.S. A . 1999, 96, 406-41 1 . 40. Sheares, B.T. and Carlson, D.M. Characterization of UDP-galactose: 2-acetamido-2-deoxy-~glucose 3P-galactosyltransferases from pig trachea. J. Biol. Chem. 1983,258, 9893-9898. 41. Schulte, S. and Stoffel, W. Ceramide UDPgdlactoSyltrdnSferaSe from myelinating rat brain: purification, cloning, and expression. Proc. Nutl Acud. Sci. U.S.A. 1993, 90, 10265-10269. 42. Coetzee, T., Suzuki, K., and Popko, B. New perspectives on thc function of myelin galactolipids. Trends Neurosci. 1998, 21 126-1 30. 43. Sprong, H., Kruithof, B., Leijendekker, R., Slot, J.W., van Meer, G., van der Sluijs, P. UDPga1actose:ceramide galactosyltransferase is a class I integral membrane protein of the endoplasmic reticulum. J. Bid. Chem. 1998, 273, 25880-25888. 44. Mackenzie, P.I., Owens, I.S., Burchell, B., Bock, K.W., Bairoch, A., Belanger, A., FournelGigleux, S., Green, M., Hum, D.W., Iyanagi, T., Lancet, D., Louisot, P., Magdalou, J., Chowdhury, J.R., Ritter, J.K., Schachter, H., Tephly, T.R., Tipton, K.F., Nebert, D.W. The UDP glycosyltransferase gene superfamily: recommended nomenclature update based on evolutionary divergence. Pharrnucogenetics 1997, 7, 255-269. 45. Ichikawa, S., Sakiyama, H., Suzuki, G., Hidari, K.I., Hirabayashi, Y. Expression cloning of a cDNA for human ceramide glucosyltransferase that catalyzes the first glycosylation step of glycosphingolipid synthesis. Proc. Nutl Acud. Sci. C! S. A . 1996, 93, 4638-4643.

References

195

46. Joziasse, D.H., Shaper, J.H. and Shaper, N.L. The ul,3-Galactosyltransferase Gene, in al,3Galactosyltransferase, a-Gal Epitopes and the Natural Anti-Gdl Antibody”, in the series “Subcellular Biochemistry”, U. Galili and J.L. Avila, eds. Plenum Press, New York, 1999. 47. Blanken, W.M. and van den Eijnden, D.H. Biosynthesis of terminal Galal,3Gal~l,4GlcNAcR oligosaccharide sequences on glycoconjugates. Purification and acceptor specificity of a UDP-Ga1:N-acetyllactosaminide al,3-galactosyltransferasefrom calf thymus. J. Biol. Chem. 1985,260, 12927-12934. 48. Galili U. Evolution of al,3-galactosyltransferase and of the a-Gal epitope. Subcell. Biochern. 1999,32, 1-23. 49. Galili, U., Wang, L., LaTemple, D.C. and Radic, M.Z. The natural anti-Gal antibody. Subcell. Biochem. 1999, 32, 79-106. 50. Larsen, R.D., Rivera-Marrero, C.A., Ernst, L.K.. Cummings, R.D. and Lowe, J.B. Frameshift and nonsense mutations in a human genomic sequence homologous to a murine UDP-Gdl:P&Gal( 1,4)-~-GkNAca( 1,3)-galactosyltransferasecDNA. J. Biol. Chern. 1990, 265, 70557061. 51. Joziasse, D.H., Shaper, J.H., Wang, J.E. and Shaper, N.L. Characterizaiton of an a-143galactosyltransferase homologue on human chromosome 12 that is organized as a processed pseudogene. J. Bid. Chem. 1991,266, 6991 --6998. 52. Sharma, A., Okabe, J., Birch, P., McClellan, S.B., Martin, M.J., Platt, J.L. and Logan, J.S. Reduction in the level of Gal(al,3)Gal in transgenic mice and pigs by the expression of an a(l,2)fucosyltransferase.Proc. Nut1 Acad. Sci. U.S.A. 1996, 93, 7190-7195. 53. Osman, N., McKenzie, I.F., Ostenried, K., Ioannou, Y.A., Desnick, R.J. and Saudrin, M.S. Combined transgenic expression of a-galactosidase and al,2-fucosyltransferase leads to optimal reduction in the major xenoepitope Galu(l,3)Gal. Proc. Nut1 Acud. Sci. U.S.A. 1997, 94, 14677-14682. 54. Tanemura, M., Miyagawa, S., Koyota, S., Koma, M., Matsuda, H., Tsuji, S., Shirakura, R. and Taniguchi, N. Reduction of the major swine xenoantigen, the a-galactosyl epitope by transfection of the a 2,3-sialyltransferase gene. J. Biol. Chern. 1998, 273, 16421-16425. 55. Yamamoto, F., Clausen, H., White, T., Marken, J. and Hakomori, S. Molecular genetic basis of the histo-blood group ABO system. Nuture 1990, 345, 229-233. 56. Willison, K.R. and Stern, P.L. Expression of a Forssman antigenic specificity in the preimplantation mouse embryo. Cell 1978, 14, 785-793. 57. Haslam, D.B. and Baenziger, J.U. Expression cloning of Forssman glycolipid synthetase: a novel member of the histo-blood group ABO gene family. Proc. Nut1 Acad. Sci. U.S.A. 1996, 93, 10697-10702. 58. Saitou, N. and Yamamoto, F., Evolution of primate ABO blood group genes and their homologous genes. Mol. Biol. E d . 1997, 14, 399- 41 1 . 59. Lopez, M., Gazon, M., Juliant, S., Plancke, Y., Leroy, Y., Strecker, G., Cartron, J.P., Bailly, P., Cerutti, M., Verbert, A. and Delannoy, P. Characterization of a UDP-Ga1:Gal P I 3GalNAca1, 4 galactosyltransferase activity in a Mamestru brassicae cell line. J. Biol. Chem. 1998,273, 33644-33651. 60. Masri, K.A., Appert, H.E. and Fukuda, M.N. Identification of the full-length coding sequence for human galactosyltransferase (P-N-acetylglucosaminide: (31,4-galactosyltransferase).Biochern. Biophys. Res. Commun. 1988, 1.57, 657-663. 61. Basu, M. De, T., Das, K.K., Kyle, J.W., Chon, H-C., Schaeper, R.J., and Basu, S. Glycolipids. Met. in Enzymol. 1987, 138, 515-607. 62. Lo et al. Unpublished data. 63. Sasaki, K., Sasaki, E., Kawashima, K., Hanai, N., Nishi, T. and Hasegawa, M. P-1,3galactosyltransferase. Patent: JP 1994181759-A 1 05-Jul-1994. 64. Daniotti J.L., Martina J.A., Zurita A.R. and Maccioni H.J.F. Influence of N-glycosylation and N-glycan trimming on the activity and intracellular traffic of mouse GMl/GDl b/GAl synthase. Deposited in GenBank. 65. Shaper. Unpublished data. 66. Joziasse, D.H., Shaper, J.H., van den Eijnden, D.H., Van Tunen, A.J. and Shaper, N.L. Bovine a1-3-galactosyltransferase: Isolation and characterization of a cDNA clone: Identification of homologous sequences in human genomic DNA. J. Biol. Chern. 1989,264, 14290-14297.

196

I0 The Guluctosyltrunsferuses

67. Joziasse, D.H., Shaper, N.L., Kim, D., van den Eijnden, D.H. and Shaper, J.H. Murine c(-1,3galactosyltransferase: A single gene locus specifies four isoforms of the enzyme by alternative splicing. J. Bid. Chem. 1992, 267, 5534-5541. 68. Larsen, R.D., Rajan, V.P., Ruff, M.M., Kukowska-Latallo, J., Cummings, R.D. and Lowe, J.B. Isolation of a cDNA encoding a murine UDPgalactose:P-D-galactosyl-l,4-N-acetyl-~glucosaminide a- 1,3-galactosyltransferase:Expression cloning by gene transfer. Proc. Nut1 Acud. Sci. U.S.A. 1989, 86, 8227-8231. 69. Dabkowski, P.L., Vaughan, H.A., McKenzie, I.F. and Sandrin, M.S. Isolation of a cDNA clone encoding the pig a1,3 galactosyltransferase. Trunsplunt Proc. 1994, 26, 1335. 70. Barker, R.. Olsen, K.W., Shaper, J.H. and Hill, R.L. Agarose derivatives of uridine diphosphate and N-acetylglucosamine for the purification of a galactosyltransferase. J. Biol. Chern. 1972,247, 7135-7147.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

11 Fucosyltransferases Ernest0 T. A. Marques, Jr.

11.1 Introduction Fucosyltransferases catalyze the transfer of fucose from the nucleotide sugar GDPfucose to a specific hydroxyl group of an acceptor-sugar molecule on an N-linked or 0-linked glycoprotein, or glycolipid, or to an amino acid residue. More than 80 putative fucosyltransferase sequences have been identified in organisms from bacteria to primates, and several of the genes encoding these enzymes have been cloned and characterized. Fucosyltransferases that transfer fucose (Fuc) to Fuc, galactose (Gal), glucose (Glc), N-acetylglucosamine (GlcNAc) and to the amino acids serine (Ser) or threonine (Thr), have all been identified. These enzymes have been classified according to the type of linkage formed with the acceptor molecule, their acceptor substrate specificity, their amino acid sequence homology to other fucosyltransferases and their susceptibility to certain inhibitors. Many fucosyltransferases are expressed in specific tissues, suggesting that they are controlled by specific promoters and are subject to differing types of regulation. Several of the fucosylated structures made by these enzymes are known to act as specific ligands for other molecules. Fucosylated glycoconjugates are involved in important physiological functions, including fertilization, development, cell differentiation and the immune response and they also form the major human blood group antigens. Fucosyltransferases have been found to be abnormally expressed in neoplasic tissues, and fucosylated glycoconjugates appear to play a role in tumor metastasis. In addition, mimicking of mammalian fucosylated structures by human parasites (bacteria, protozoa and helminths) allows then to interact with the host and to evade the host’s immune response; in some cases, the host immune response to these carbohydrates can lead to autoimmune disease. This Chapter will discuss some common aspects of these enzymes, then introduce a few features unique to each group of enzymes. Finally, it will highlight some of the biological processes in which each type of fucosyltransferase has been implicated. This Chapter is intended to give a general overview and is by no means all-

198

11 Fucosyltransferases

inclusive. Those readers who are interested in a particular fucosyltransferase will find it useful to investigate the cited literature.

11.2 General Characteristics Fucosyltransferases catalyze the transfer of L-fucose from the donor nucleotide sugar guanosine 5'-diphospho-~-~-fucose, (GDP-fucose) to a hydroxyl group of an acceptor molecule as follows: GDP-fucose

+ HO-R + fucoseal-0-R + GDP

(1)

where R can be carbon-2 of Fuc or Gal; 3 of Glu; 3, 4 or 6 of GlcNAc or the hydroxy amino acids Ser or Thr.

11.2.1 Nomenclature The original nomenclature for the various isoforms of fucosyltransferases, based upon the order in which the genes were isolated. was established by Pamela Stanley [ l ] and further extended by John Lowe. FucT I and I1 were used to designate fucosyltransferases activities found in the lectin-resistant CHO cell lines LEC 11 and LECl2; the next five fucosyltransferases identified were human a3,4-fucosyltransferases whose genes have been cloned (FucT 111to VII) [2-8]. Although the original nomenclature has been used to refer to the gene as well as the enzyme, the currently prefered system is to reserve the original nomenclature for the enzymes thenselves and use capital letters and Arabic numbers to designate the genes. For example, FucT 111 is the enzyme encoded by the FUT3 gene [9, 101. Since the original FucT I and I1 referred to a3 enzymes that were actually the products of other genes [ 111, the preferred nomenclature is to use FUTl to refer to the a2 enzyme related to the H blood group and FUT2 to the Secretor (Se) blood type enzyme. In addition, the FucT VIII product of the FUT8 gene is an a6-fucosyltranferase [12]. The most recently defined isoform that has been cloned is a mouse a3-fucosyltransferase, FucT IX [ 131. Newly identified fucosyltransferase genes from different species are compared to the kown isoforms and are usually given a designation corresponding to the isoform with which they share highest homology and similar enzymatic characteristics. When the gene and its product show appreciable similarity to more than one isoform, the gene is often given a special designation (for example, bovine Fuc-Tb is similar in its characteristics to FucT 111, V and VI and the FucTh gene is also thought to be most similar to the common ancestor gene that give raise to the FUT3, FUTS and FUT6 genes [14]). On other occasions the gene may be defined as a new isoform.

11.2 General Charucteristics

199

11.2.2 Gene Structure Most of the fucosyltransferases in several organisms are products of single-exon genes. Among the exceptions, however, are the human and mouse FUT7 and the human FUT8 genes. Phylogenetic analysis of the DNA sequences of the putative fucosyltransferases genes has suggested that the various fucosyltransferases present in higher organisms have evolved from two or three ancestor genes by consecutive duplication and subsequent divergence. A remarkable example of this duplication phenomenon is the presence of 18 different putative fucosyltransferases genes in Caenorhabditis elegnns, a simple multicellular organism.

11.2.3 Sequence Peptide Motifs A G D P binding domain is common to all fucosyltransferases. Although conserved within families of fucosyltransferases, this domain differs in sequence between families. The G D P binding domain is localized between the acceptor binding region and the catalytic site. In adition, a conserved peptide sequence motif, YXFX(L/ V)XFEN(S/T)XXXDYXTEK, has been identified for the putative catalytic site of the a3-fucosyltransferases, on the basis of sequence alignment of the deduced amino acid sequences of fucosyltransferase genes deposited in the Gene Bank. The conserved motif (V/T)G(V/I)H(V/I)R(R/H)(G/H) is found in both the a2 and a6-fucosyltransferases [ 151. 11.2.4 Specificity Fucosyltransferases show extremely high specificity for their acceptor and donor substrates. As expected, the acceptor binding region of each fucosyltransferase has evolved to interact with a distinct acceptor molecule. As a result, the acceptor binding region is the least conserved portion of these enzymes. The acceptorsubstrate binding region has often been referred to as the hypervariable region. Subtle changes in this region, however, can dramatically affect substrate specificity. For example, studies using site-directed mutagenesis in the hypervariable region have demonstrated that substitution of only two amino acid residues (NT86.87) in the FucT V with the amino acids in the equivalent positions in the FucT 111 (HT) produces a mutant FucT V with an acceptor substrate specificity more similar to that expected for a FucT I11 [16]. The acceptor-substrate binding region can recognize specific hydroxyl groups at certain positions in the acceptor molecule. Studies with acceptor substrate analogs for FucT 111, which transfers Fuc to the OH-4 group of the GlcNAc, have demonstrated that recognition of the OH group at position six of the terminal Gal residue in the acceptor substrate (Galpl,3GlcNAc) is critical for binding of this substrate to the enzyme [ 171. In some cases, amino acid residues near the carboxy terminus have also been implicated in substrate recognition and catalysis [ 18, 191. The threedimensional folding of the enzymes appears to bring the carboxy terminus to close

200

I 1 Fucosyltransferases

Figure 1. Schematic diagram of a common structural organization of a fucosyltransferase. The domains are not to scale. The carboxy terminus is depicted to represent the possible interaction of this region with the acceptor binding and catalytic domains. Scissors indicate the site of an in vivo proteolytic clevage that results in release of soluble fucosyltransferases.

proximity with the hypervariable region. The presence of conserved Cys residues near the amino and carboxyl termini of these enzymes suggests that disulfide bonds may be involved in mediating their proper folding. 11.2.5 Protein Structure and Topology

Golgi fucosyltransferases are type I1 transmembrane proteins, with a cytoplasmic domain, a transmembrane region and a catalytic site that resides in the secretory pathway. A schematic representation of the most common structural organization of the fucosyltransferases is presented in Figure 1. These enzymes usually fucosylate N- and 0- linked glycoproteins and glycolipids; they are also themselves glycoproteins, with two or more glycosylation sites. Frequently, fucosyltransferases have a stem region that is very susceptible to endoproteolysis, and a significant fraction of the total active enzyme can be secreted. Most of the enzyme is retained in the Golgi compartment, but small quantities have also been found in the endoplasmic reticulum and cytoplasmic membranes. A distinct group of fucosyltransferases, as well as an increasing number of other glycosyltransferases, has been identified in the cytosol. These enzymes have been

11.2 General Churucteristics

201

shown to fucosylate cytosolic and nuclear glycoproteins, but much less is known about their structures or biological roles. 11.2.6 Enzymatic Reaction Mechanism The enzymatic mechanism of the human a3-fucosyltransferase V was determined to be an ordered sequential bi-bi reaction, and this model can apparently also be applied to other fucosyltransferases. The K , of the fucosyltransferases for the donor substrate GDP-fucose is generally 5-20 pM, while that for the appropriate acceptor is 1.5-15 mM. The transfer of fucose from the donor nucleotide sugar to the acceptor molecule is carried out by the inversion of the configuration at the anomeric center of the L-fucose [4].The enzymatic reaction appears to be mediated by a general base mechanism; this conclusion is supported by the presence of a catalytic residuc with a pKa of 4.1 [20]. The presence of highly conserved lysine residues in several fucosyltransferases suggests that these residues may be also involved in the catalytic process. Three of these basic residues are conserved at equivalent positions in FucT 111, IV, V and VI, but only two are conserved in FucT VII. A site-directed mutagenesis study yielded the interesting observation that substitution of Arg223, which is located within the conserved peptide sequence motif, for the third Lys in FucT VII produced an enzyme with twice the specific activity of the original FucTVII. Conversely, substitution of an Arg for the corresponding conserved Lys255in FucT V produced an enzyme with much lower specific activity than that of the native form [21]. Inhibitor studies and studies of primary and secondary solvent isotope effects, suggest that a proton transfer leads to the formation of a charged sp2-hybridized transition state, subsequent cleavage of the glycosidic bond of the nucleotide sugar, and the formation of the linkage to the acceptor [22]. The proposed structure of the transition state (Figure 2) is analogous to that in glycosidase reactions, the sugar forming a flattened half-chair conformation. with the formation of a oxocarbenium ion character at the anomeric position. More studies are necessary to confirm this model. Arguing against this proposed reaction mechanism is the fact that deoxyfuconojirimiycin, a potent inhibitor of the a-fucosidases and a transition state analog, has very little effect on fucosyltransferases. The first evidence identifying the residues involved in substrate binding and catalysis was obtained using inhibitors that target specific amino acids, such as N ethylmaleimide (NEM), which forms a covalent linkage with sulfhydryl groups of cysteine residues. Inhibition of some fucosyltransferases by NEM could be prevented by addition of GDP-Fuc, suggesting that cysteine residues are involved in the GDP-binding site. These early studies with NEM were used to differentiate the 1x3-fucosyltransferaseactivities present in tissue homogenates, where more then one fucosyltransferase isoform is often present. The enzymes were classified as NEMsensitive and NEM-insensitive. The availability of the amino acid sequence of several a3-fucosyltransferase and the use of sequence alignment and site-directed mutagenesis has made it possible to map the relevant cysteine residues for the NEMsensitive enzymes, FucT I11 C Y S ' ~FucTV ~ , C Y S 'and ~ ~FucT VI Cys14*[23,241. The

202

I 1 Fucosyltrunsferases

+ b

I * o=a-o

I

0

I < o=a-o I ?

11.3 Specgc Fucosyltransjhuases

203

NEM-insensitive enzymes, FucT IV and FucT VII, have a Ser17*or Thr'27 at these respective positions.

11.2.7 Inhibitors It has been postulated that fucosyltransferase inhibitors can have important pharmacological uses as anti-neoplasic, anti-inflammatory, anti-helminthic, and antimicrobial agents. Although investigations into fucosyltransferase inhibitors have been carried on in academic institutions and pharmaceutical companies, only few reports of such compounds are available. GDP (K, 30 pM), GMP (K, 94 pM) and GTP (K, 170 pM) have been clearly established as non-specific inhibitors of the fucosyltransferases; in contrast. ATP, CTP and TTP do not inhibit enzymes. Recently, a donor nucleotide sugar mimetic, GDP-2F-fucose, was shown to be among the most potent inhibitors (K, 4.2 pM) [22]. Although these fluoro-nucleotide sugars have no therapeutic relevance as fucosyltransferase inhibitors because of their lack of specificity, they have frequently been used in vitro. Potentially specific fucosyltransferase inhibitors have been designed by mimicking the acceptor molecules. One of these compounds, an azatrisaccharide similar to the Lewis trisaccharide, has been tested for its ability to inhibit Fuc-T V. Although this inhibitor is not very potent (K, 2.4 mM) when used alone, it produces a synergistic effect when mixed with GDP [25].

11.3 Specific Fucosyltransferases 11.3.1 GDP-Fucose: Fucal(Fucal,2Fuc)a2-fucosyltransferase The first evidence of a fucosyltransferase that can transfer fucose to position two of another fucose emerged from the characterization of a Schistosoma mansoni carbohydrate epitope recognized by a murine monoclonal antibody, mAb128. This mAb recognizes carbohydrate structures present in glycolipids and glycoproteins. and glycoconjugates containing this epitope have been shown to be involved in the pathogenesis of schistosomiasis. Early mass spectrometry studies of affinity-purified S. mansoni egg glycolipids produced two possible structures for the epitope [26],and Ann Dell's group has defined the complete structure. They identified the fucoseal ,2fucose in schistosome egg glycolipids [27] and cercarial glycoproteins [28]. Recently, Hokke and collaborators demonstrated the enzymatic activity of this fucosyltransferase, by showing that cercarial extracts of the schistosome Tricohilharzia occelata could fucosylate a fucose present in an acceptor containing the Lewis X structure [29]. Efforts from several groups to clone the gene encoding this Fucal,2Fuc fucosyltransferase are underway. One fucosyltransferase from gene S. mansoni has been cloned, but this enzyme did not possess such activity (301.

204

I 1 Fucosyltrunsferuses

11.3.2 GDP-Fucose: Gal~l(Fucal,2Gal)a2-fucosyltransferase The oligosaccaride products of the reactions catalyzed by these enzymes are the H blood group antigens. In humans these structures are precursors of the ABO blood group determinants. To date, two of these enzymes that have been characterized in humans. The H blood group producing fucosyltransferases are located in the chromosome 19q13.3 cluster. The FUCl, type H enzyme is present in hematopoetic tissues, and FUC2. the Secretor (Se) enzyme. is present in secretory tissues and fluids. There is a third isoforn of this enzyme that has been cloned and characterized in rabbits [31], but this isoform has yet to be identified in humans. The H-type fucosyltransferase appears to preferentially fucosylate N-linked acceptor oligosaccharides in glycoproteins containing polylactosamine sequences, such as the lysosomal associated membrane proteins, LAMP-1 and LAMP-2 [32], although the mechanism for this selectivity is still unclear. The expression of the a2fucosyltransferases and the H carbohydrate determinant has been studied in rabbit neurons, and both have been found to be developmentally regulated [33]. In addition, expression of the rabbit homolog of FUCl , RFT-I, in neuroblastoma cells has been shown to inhibit axonal outgrowth [34]. These reports suggest roles for these enzymes in the regulation of neural development. However, humans with the null allele of the type H enzyme, the Bombay phenotype, or with null alleles of the Se enzyme, the so-called “non-secretor phenotype”, show no detectable abnormalities in neural development. The human a2-fucosyltransferases have also been used in transgenic animals in many attempts to remodel the carbohydrate structures on the surfaces of tissues as a means of reducing xenograft rejection [35].This strategy has been shown to be particularly promising when the enzyme is associated with other glycosyltransferases. 11.3.3 GDP-Fucose: Gal~1,4/3GlcNAc(Fuca1,3/4GlcNAc)a3/4-fucosyltransferases This group of fucosyltransferases is perhaps the best characterized, to date. They are involved in the synthesis of molecules that play important roles in cell-cell interactions and adhesion, including Led, sLe”. LeX and sLeX.These fucosylated structures are ligands for the selectins and have been implicated in many biological functions (reviewed by 1361 see also chapter ). Six fucosyltransferase isoforms have been identified in mammals; these isoforms are expressed in different tissues and have distinct acceptor substrate specificities in vitro (Figure 3). The specificity has also been investigated in vivo side by side in BHK-21 cells and the expression ratios of sLex/Lexin these cells have been determined for FucT 111 (14 : 1), FucT IV (1 : 7), FucT V (3: l), FucT VI (1.1: 1) and FucT VII (1 :0) [37]. Blood group Lewis: FucT 111, V and VI

The Lewis group contains three enzymes, FucT 111, (the “Lewis enzyme”) [2]; V [4] and VI [3] (the “Plusmu-type enzymes”). In humans these enzymes are closely located in the same locus on chromosome 19~13.3.These enzymes share more than

11.3 Spec$c Fuco.~yyltrunsferases

205

Figure 3. Substrate specificities of the various vertebrate a3/4-fucosyltransferases. This graph shows the capacity of each a3-fucosyltransferase isoform to catalyze the synthesis of the various Lewis determinants.

85% sequence identity and are believed to have evolved from a common ancestor gene. FucT 111 is present in liver, gall-bladder, kidney and milk but is not found in the plasma. This enzyme catalyzes the creation of the Fucul,4GlcNAc linkage and produces Lea and sLea structures [38]. These carbohydrate epitopes are also often found to be overexpressed in cancers of the digestive tract, such as colon cancers. Furthermore, the expression of sLe” by colon cancer cell lines has been correlated with increased metastatic capacity, and this structure is a potential therapeutic target. In fact, expression of sLea could be blocked in a colon cancer cell line by transfection of this fucosyltransferase in the antisense orientation [ 39, 401. The FucT V and VI, are found in the plasma, and they catalyze the formation of Fucal,3GlcNAc linkages and produce LeX and sLeX structures. High expression levels of these enzymes have also been identified in neoplastic tissues, especially in cancers of the lung [41]. The levels of a3-fucosyltransferase activity in plasma are also increased in chronic liver diseases, such as hepatitis, cirrhosis and hepatocellular carcinoma. Indeed, quantitation of this fucosyltransferase activity in plasma can be used as indicator of these diseases [42]. Myeloid enzyme: FucT IV

The myeloid enzyme, FucT IV. is also known as ELAM-1 ligand fucosyltransferase (ELFT) [7, 43, 441. The human FucT IV is located on chromosome llq21 and is expressed in cells of the myeloid lineage. Even though some of the protein is secreted, the secreted molecules represent only a small fraction of the fucosyltransferase activity in plasma. This enzyme can fuco-

206

I1 Fucos2,ltransfevases

sylate GlcNAc in an a3linkage in type-2 neutral acceptor sugars to produce the LeX structure. Although it can fucosylate internal GlcNAc residues of sialylated acceptors to formVIM-2-type structures, it cannot fucosylate the terminal GlcNAc near a sialic acid. Therefore, FucT IV is unable to produce the sLe' determinant. FucT IV is higly expressed in immature myelocytes, and its expression appears to decrease as the myeloid cells develop and mature [45, 461. High levels of expresion of FucT IV in lung tumors are associated with more undifferentiated tumors and indicate a poor prognosis for the patient [47]. FucT IV has also been implicated in the etiology of atherosclerosis. The expression of this enzyme appears to be down regulated in the foam cells that are derived from monocytes [48] and which are directily involved in the formation of sclerotic lessions. Leukocyte enzyme: FucT VII This enzyme plays a critical role in the immune system by regulating the trafficking of leukocytes to the sites of infection. Human FucT VII [6, 81 is located on chromosome 9q34.3. This enzyme can fucosylate GlcNAc in an a3 linkage in sialylated type-2 acceptor sugars to produce sLeX,but it can not fucosylate non-sialylated sugars. FucT VII acivity is complementary to that of FucT IV. A great deal of information about the biological role of this enzyme has been gained from studies of FucT VII-deficient mice. Leukocytes from these mice are less able to adhere, as a result of their lack of sufficient E- and P-selectin ligand activity. FucT VII-deficient mice also have much a lower L-selectin ligand activity in their high endothelial venules (HEV). These effects appear to cause blood leukocytosis and a reduced capacity of the leukocytes to leave the intravascular space in response to immunological stimuli [49]. The synthesis of P-selectin or P- and E-selectin ligands is regulated according to the level of FucT VII expression in the T lymphocytes [50]. Thl lymphocytes have higher levels of FucT VII expression than do Th2 lymphocytes, an observation which correlates with a greater capacity of Thl lymphocytes to bind to P-selectin [51]. As would be expected, the expression of FucT VII appears to be controlled by cytokines. It is enhanced by IL12, a Thl driving cytokine, and is inhibited by a Th2 cytokine, IL4 [52]. Abnormal levels of FucT VII expression have been identified in several types of tumors, where its expression correlates with increased metastatic potential. Neuronal enzyme: FucT IX This fucosyltransferase was recently isolated by expression cloning from a mouse brain library; the human analog has not been identified. This enzyme catalyzes the synthesis of the Le' determinant and has a substrate specificity similar to that of the FucT IV. The mRNA for FucT IX is mainly found in kidney and brain, whereas FucT IV mRNA is not found in these tissues. The LeXepitope is expressed in neurons and glial cells, and it appears likely that this enzyme is responsible for the synthesis of these structures in such cells. FucT IX has been found by in situ hybridisation in neurons but not in glial cells; the enzymes responsible for the LeX structure in glial cells have yet to be identified [ 131.

11.3 Specific FucosyltransJerases

201

11.3.4 GDP-Fucose: Gal~l,3GlcNAc(Fucal,3GlcNAc) bacterial (Helicobactev pylovi) a3-fucosyltransferase Helicobacter pylori is a pathogenic bacterium that lives in the acidic environment of the stomach, attached to the gastric mucosa. H. pylori infection is one of the most important agents causing gastric and duodenal ulcers and is also associated with gastric adenocarcinomas. This H. pylori enzyme has been shown to contain the conserved peptide motif common to all the a3-fucosyltransferases. It is distinguished from the eukaryotic fucosyltransferases by its larger size, lack of a transmembrane domain, and presence of ten seven-amino acid repeats which contain a putative leucine zipper motif [53, 541.

11.3.5 GDP-Fucose: GlcNAc-N(Fucal,6GlcNAc)a6fucosyltransferases One isoform of fucosyltransferase with this activity, FucT VIII, has been identified in human and pig. The two enzymes have been purified, and the FUT8 genes have been cloned and found to share 95% homology. It has been suggested that other isoforms with this activity may exist. FucT VIII has been found elevated in patients with hepatocellular carcinoma present with high levels of fetoprotein containing ctl,6-fucosylation. The levels of FUT8 mRNA and al,6-fucosyltranferase activity are also elevated in several chronic liver pathologies [12, 551.

11.3.6 GDP-Fucose: 0-Ser(Fuca1-0-Ser)GlcNAc polypeptide fucosyltransferases Fucose has been found to be linked by 0-glycosidic bond to Ser and Thr residues of epidermal growth factor (EGF),domains of many blood proteins involved in coagulation and fibrinolysis, including the tissue plasminogen activator factor, t-PA. In most cases, 0-linked fucose is found to occur within the sequence CXXGG(S/T)C of the E G F domains; although the Gly residues appear not be essential for the substrate, the proper folding of the EGF domains seems to be required. An enzyme that catalyzes the fucosylation of EGF domains has recently been purified from CHO cells. It appears to be a transmembrane protein, and the fact that it bears high mannose type oligosaccharides suggests that this molecule may be localized to the endoplasmic reticulum or cis-Golgi [ 561.

11.3.7 Unconventional Types of Fucosylation: FucPl -P-Ser and cytoplasmic Fucal,2-GalP1,3-GlcNAc-Pro(Dictyostelium discoideum) FucPl -P-Ser Synthesis of Fuc fl1-P-Ser represents a new type of fucosylation that has been identified in Dictyostelium discoideum, an amoeba often used for biological studies.

208

11 FucosyltrunsJiruses

This structure was identified through the characterization of the epitope recognized by monoclonal antibody mAb83.5. mAb83.5 recognizes this fucosylated structure on Ser-rich proteins that are developmentally regulated. The enzymatic activity was demonstrated by incubating peptides consisting of Ser and Gly with microsomal preparations from D.discoideum [57].By definition this enzyme may not be classified as a fucosyltransferase, since it appears to catalyze the transfer of phospho-fucose, not fucose, from the nucleotide; thus it has been designated Fuc-phosphotransferase. Fucal,2-Galpl,3-GlcNAc-Pro

Characterization of the post-translational modifications of a cytosolic glycoprotein, FP21, from D. discoideum, has led to the identification of this type of fucosylation in the cytosol. At the same time several cytosolic forms of glycosylation were also discovered. This D. discoideum enzyme has been purified and characterized by Christopher West and colleages. The enzyme has a molecular mass about twice that its Golgi counterpart and has substrate specificity similar to that of the human Secretor-type enzyme. Recently, a cytosolic protein involved in ubiquitination, termed SKP1, has also been shown to contain a glycosylation site attached to a hydroxyproline (Galal,6Gala1,L-Fucal,2-Gal~1,3-GlcNAc-Pro). It is interesting that the hydroxylation of the Pro residue appears to be an important regulatory step, since only a fraction of the Pro residues were found to be hydroxylated, but all hydroxylproline residues were glycosylated [ 581.

Acknowledgments We thank Dr. Deborah McClellan for superb editorial assistance and Dr. Brian Collins, Dr. Yoshitaka Ichikawa, Dr. Ronald Schnaar and Dr. Gerald Hart for their comments and suggestions that greatly contributed to this chapter. Dr. E. T. A. Marques Jr. is supported by CAPES grant 1219/94-2 and NIH grant R37AI19217.

References 1. D. R. Howard, M. Fukuda, M. N. Fukuda, P. Stanley, The GDP-fucose:N-acetylglucosaminide 3-alpha-L-fucosyltransferasesof LECl1 and LEC12 Chinese hamster ovary mutants exhibit novel specificities for glycolipid substrates, J Biol Chem, (1Y87), 262, 16830-7. 2. J. F. Kukowska-Latallo, R. D. Larsen, R. P. Nair, J. B. Lowe, A cloned human cDNA determines expression of a mouse stage-specific embryonic antigen and the Lewis blood group alpha( 1,3/1,4)fucosyltransferase,Genes Deu, (lYYO), 4, 1288-303. 3. B. W. Weston, P. L. Smith, R. J. Kelly, J. B. Lowe, Molecular cloning of a fourth member of a human alpha (1,3)fucosyltransferase gene family. Multiple homologous sequences that determine expression of the Lewis x, sialyl Lewis x, and difucosyl sialyl Lewis x epitopes [ published erratum appears in J Biol Chem 1993 Aug 25;268(24):18398], J Biol Chem, (lYYZ), 267, 2457584. 4. B. W. Weston, R. P. Nair, R. D. Larsen, J. B. Lowe, Isolation of a novel human alpha (1,3)fucosyltransferase gene and molecular comparison to the human Lewis blood group alpha

References

209

(1,3/1,4)fucosyltransferasegene. Syntenic, homologous, nonallelic genes encoding enzymes with distinct acceptor substrate specificities, J Biol Chem, (1992), 267, 41 52-60. 5 . S. Natsuka, K. M. Gersten, K. Zenita, R. Kannagi, J. B. Lowe, Molecular cloning of a cDNA encoding a novel human leukocyte alpha-l,3-fucosyltransferasecapable of synthesizing the sialyl Lewis x determinant, J B i d Cheni, (1994). 269, 20806. 6. S. Natsuka, K. M. Gersten, K. Zenita, R. Kannagi, J. B. Lowe, Molecular cloning of a cDNA cncoding a novel human leukocyte alpha-l,3-fucosyltransferasecapable of synthesizing the sialyl Lewis x determinant [published erratum appears in J Biol Chem 1994 Aug 12;269(32):20806], J Biol Chem, (1994), 269, 16789-94. 7. J. B. Lowe, J. F. Kukowska-Latallo, R. P. Nair, R. D. Larsen, R. M. Marks, B. A. Macher, R. J. Kelly, L. K. Ernst, Molecular cloning of a human fucosyltransferase gene that determines expression of the Lewis x and VIM-2 epitopes but not ELAM-1-dependent cell adhesion, J Bid Chem, (1991), 266, 17467-77. 8. K. Sasaki, K. Kurata, K. Funayama, M. Nagata, E. Watanabe, S. Ohta, N. Hanai, T. Nishi, Expression cloning of a novel alpha 1,3-fucosyltransferase that is involved in biosynthesis of the sialyl Lewis x carbohydrate determinants in leukocytes, J B i d Chem, (1994), 269, 14730-7. 9. M. Costache, A. Cailleau, P. Fernandez-Mateos, R. Oriol, R. Mollicone, Advances in molecular genetics of alpha-2- and alpha-3/4-fucosyltransferases, Trun@s Clin Bid, (1997), 4 , 36782. 10. M. Costache, P. A. Apoil, A. Cailleau, A. Elmgren, G. Larson, S. Henry, A. Blancher, D. Iordachescu, R. Oriol, R. Mollicone, Evolution of fucosyltransferase genes in vertebrates, J Biol Chem, (1997), 272, 29721-8. 1 I . S. Goelz, R. Kumar, €3. Potvin, S. Sundaram, M. Brickelmaier, P. Stanley, Differential expression of an E-selectin ligand (SLex) by two Chinese hamster ovary cell lines transfected with the same alpha (l,3)- fucosyltransferase gene (ELFT), J Biol Chem. (1994). 269, 1033-40. 12. S. Yanagidani, N. Uozumi, Y. Ihara, E. Miyoshi, N. Yamaguchi, N. Taniguchi, Purification and cDNA cloning of GDP-L-Fuc:N-acetyl-beta-D-glucosaminide:alphal-6 fucosyltransferase (alphal-6 FucT) from human gastric cancer MKN45 cells, J Biochem (Tokyo), (1997), 121, 626-32. 13. T. Kudo, Y. Ikehara, A. Togayachi, M. Kaneko, T. Hiraga, K. Sasaki, H. Narimatsu, Expression cloning and characterization of a novel murine alphal, 3-fucosyltransferase, mFucTIX, that synthesizes the lewis x (CD15) epitopc in brain and kidney, J Bid Chem, (1998), 273, 26729-38. 14. A. Oulmouden, A. Wierinckx, J. M. Petit, M. Costache, M. M. Palcic, R. Mollicone, R. Oriol, R. Julien, Molecular cloning and expression of a bovine alpha( 1,3)- fucosyltransferase gene homologous to a putative ancestor gene of the human FUT3-FUTS-FUT6 cluster, J Biol Chem, (1997), 272, 8764-73. 15. C. Breton, R. Oriol, A. Imberty, Conserved structural features in eukaryotic and prokaryotic fucosyltransferases, Glycohiology, (1998), 8, 87T94. 16. A. T. Nguyen, E. H. Holmes, J. M. Whitaker, S. Ho, S. Shetterly, B. A. Macher, Human alphal,3/4-fucosyltransferases.I . Identification of amino acids involvcd in acceptor substrate binding by site-directed mutagenesis, J Bid Chon, (1998), 273, 25244-9. 17. M. Du, 0. Hindsgaul, Recognition of beta-D-Gal p-( 1+3)-beta-D-Glc pNAc-OR acceptor analogues by the Lewis alpha-( 1+3/4)-fucosyltransferase from human milk, Curhohyrir Res, (1996), 286. 87-105. 18. Z. Xu, L. Vo, B. A. Macher, Structure-function analysis of human alphal,3-fucosyltransferase. Amino acids involved in acceptor substrate specificity, J Bid Chem, (1996), 271, 8818-23. 19. L. Vo, S. Lee, M. C. Marcinko, E. H. Holmes, B. A. Macher, Human alphal,3/4-fucosyltransferases. 11. A single amino acid at the COOH terminus of FucT I11 and V alters their kinetic properties, J Biol Clzem, (1998), 273, 25250-5. 20. B. W. Murray, S. Takayama, J. Schultz, C. H. Wong, Mechanism and specificity of human alpha-l,3-fucosyltransferase V, Biochemi.~try,(1996), 35, 11 183--95. 21. A. L. Sherwood. A. T. Nguyen, J. M. Whitaker, B. A. Macher, M. R. Stroud, E. H. Holmes, Human alphal,3/4-fucosyltransferases.Ill. A Lys/Arg residue locatcd within the alphal ;3FucT motif is required for activity but not substrate binding, J Bid Chem, (1998), 273, 2525660.

2 10

I I Fucosyltransjerases

22. B. W. Murray, V. Wittmann, M. D. Burkart, S. C. Hung, C. H. Wong, Mechanism of human alpha- 1,3-fucosyltransferase V: glycosidic cleavage occurs prior to nucleophilic attack, Biochemistry, (1997), 36, 823-31. 23. E. H. Holmes, Z. Xu, A. L. Shenvood, B. A. Macher, Structure-function analysis of human alpha 1+3fucosyltransferases. A GDP-fucose-protected, N-ethylmaleimide-sensitive site in FucT-I11 and FucT-V corresponds to Ser178 in FucT-IV, J Biol Chem, (1995), 270, 8145-51. 24. C. J. Britten, M. I. Bird, Chemical modification of an alpha 3-fucosyltransferase; definition of amino acid residues essential for enzyme activity, Biochim Biophys Acta, (1997), 1334, 5764. 25. L. Qiao, B. Murray, M. Shimazaki, S. J., W. C-H., Synergistic Inhibition of human a1,3fucosyltransferase V, Journal oJ the American Chemical Society, (1996), 118, 7653-7662. 26. S. B. Levery, J. B. Weiss, M. E. Salyan, C. E. Roberts, S. Hakomori, J. L. Magnani, M. Strand, Characterization of a series of novel fucose-containing glycosphingolipid immunogens from eggs of Schistosoma mansoni, J Biol Chem, (1992), 267. 5542-51. 27. Khoo Kh, Chatterjee D, Caulfield JP, Morris HR, Dell A, Structural characterization of glycosphingolipids from the eggs of Schistosoma mansoni and Schistosoma japonicum, Glycohioloyy, (1997), 7, 653-661. 28. K. H. Khoo, S. Sarda, X. Xu, J. P. Caulfield, M. R. McNeil, S. W. Homans, H. R. Morris, A. Dell, A unique multifucosylated -3GalNAc beta lh4GlcNAc beta l h 3 G a l alpha 1- motif constitutes the repeating unit of the complex 0-glycans derived from the cercarial glycocalyx of Schistosoma mansoni, J Biol Chem, (1995); 270, 17114-23. 29. C. H. Hokke, A. P. Neeleman, C. A. Koeleman, D. H. van den Eijnden, Identification of an alpha3-fucosyltransferase and a novel alpha2- fucosyltransferase activity in cercariae of the schistosome Trichobilharzia ocellata: biosynthesis of the Fucalphal+2Fucalphal + 3[Gal(NAc)betal+4]GlcNAc sequence, Glycobiology, (1998), 8, 393-406. 30. E. T. Marques, Jr., J. B. Weiss, M. Strand, Molecular characterization of a fucosyltransferase encoded by Schistosoma mansoni, Mol Biochem Parasitol, (1998), 93, 237-50. 31. S. Hitoshi, S. Kusunoki, I. Kanazawa, S. Tsuji, Molecular cloning and expression of a third type of rabbit GDP-L- fucose:beta-D-galactoside 2-alpha-L-fucosyltransferase, J Biol Chem, (1996), 271, 16975-81. 32. P. A. Prieto, R. D. Larsen, M. Cho, H. N. Rivera, A. Shilatifard, J. B. Lowe, R. D. Cummings, D. F. Smith, Expression of human H-type alphal,2-fucosyltransferase encoding for blood group H ( 0 ) antigen in Chinese hamster ovary cells. Evidence for preferential fucosylation and truncation of polylactosamine sequences, J Biol Chem, (1997), 272, 2089-97. 33. S. Hitoshi, S. Kusunoki, 1. Kanazawa, S. Tsuji, Dorsal root ganglia-specific expression of the beta-galactoside alphal,2-fucosyltransferasegenes in rabbits, J Neurochem, (1998), 70, 2174-8. 34. S. Hitoshi, N. Koijima, S. Kusunoki, J. Inokuchi, I. Kanazawa, S. Tsuji, Expression of the beta-galactoside alpha 1,2-fucosyltransferase gene suppresses axonal outgrowth of neuro2a neuroblastoma cells, J Neurochem, (1996), 66, 1633-40. 35. C. G. Chen, E. J. Salvaris, M. Romanella, A. Aminian, M. Katerelos, N . Fisicaro, A. J. d’Apice, M. J. Pearse, Transgenic expression of human alphal,2-fucosyltransferase(H- transferase) prolongs mouse heart survival in an ex vivo model of xenograft rejection, Trunsplantation, (1998), 65, 832-7. 36. S. D. Rosen, C. R. Bertozzi, The selectins and their ligands, Curr Upin Cell Biol, (1994), 6, 663-73. 37. E. Grabenhorst, M. Nimtz, J. Costa, H . S. Conradt, In vivo specificity of human alpha1,3/4fucosyltransferases 111-VIi in the biosynthesis of LewisX and sialyl LewisX motifs on complextype N- glycans. Coexpression studies from Bhk-21 cells together with human beta-trace protein, J Biol Chem, (1998), 273, 30985-94. 38. J. Costa, E. Grabenhorst, M. Nimtz, H. S. Conradt, Stable expression of the Golgi form and secretory variants of human fucosyltransferase I11 from BHK-21 cells. Purification and characterization of an engineered truncated form from the culture medium, J Biol Chern, (1997), 272, 11613-21. 39. N. Yamada, Y. S. Chung, S. Takatsuka, Y. Arimoto, T. Sawada, T. Dohi, M. Sowa, Increased sialyl Lewis A expression and fucosyltransferase activity with acquisition of a high metastatic capacity in a colon cancer cell line, Br J Cancer, (1997), 76, 582-7.

Rgferences

21 1

40. A. G. Klopocki. A. Laskowska, J. Antoniewicz-Papis, M. Duk, E. Lisowska, M. Ugorski. Role of sialosyl Lewis(a) in adhesion of colon cancer cells-the antisense RNA approach, Euv J Bioclzem, (1998), 253, 309-18. 41. M. Martin-Satue, R. Marrugat, J. A. Cancelas, J. Blanco, Enhanced expression of alpha(l,3)fucosyltransferase genes correlates with E-selectin-mediated adhesion and metastatic potential of human lung adenocarcinoma cells, Cunc,er Re.v, (1998), 58, 1544-50. 42. T. Hada, K. Fukui, M. Ohno, S. Akamatsu, S. Yazawa. K. Enomoto, K. Yamaguchi, Y. Matsuda, Y. Amuro, N. Yamanaka, Increased plasma alpha (1 +3)-L-fucosyltransferase activities in patients with hepatocellular carcinoma, Glycoconj J , (1995), 12, 627-31. 43. S. E. Goelz, C. Hession, D. Goff, B. Griffiths, R. Tizard, B. Newman, G. Chi-Rosso, R. Lobb, ELFT: a gene that directs the expression of an ELAM-1 ligand, Cell, (1990), 63, 1349-56. 44. R. Kumar, B. Potvin, W. A. Muller, P. Stanley, Cloning of a human alpha(l,3)-fucosyltransferase gene that encodes ELFT but does not confer ELAM-1 recognition on Chinese hamster ovary cell transfectants, J Biol Chem, (1991), 266, 21777-83. 45. N. Le Marer, M. M. Palcic, J. L. Clarke, D. Davies, P. 0. Skacel, Developmental regulation of alpha 1,3-fucosyltransferase expression in CD34 positive progenitors and maturing myeloid cells isolated from normal human bone marrow, Glycohiology, (1997). 7, 357-65. 46. J. L. Clarke, W. Watkins, Alpha1 33-L-fucosyltransferase expression in developing human myeloid cells. Antigenic, enzymatic, and mRNA analyses, J Biol Clzem, (1996), 271, 10317 28. 47. J. Ogawa, H. Inoue, S. Koide, Expression of alpha-l,3-fucosyltransferasetype IV and VII genes is related to poor prognosis in lung cancer, Cuncer Res, (1996), 56, 325-9. 48. P. Cullen, S. Mohr, B. Brennhausen. A. Cignarella, G. Assmann, Downregulation of the selectin ligand-producing fucosyltransferases Fuc- TIV and Fuc-TVII during foam cell formation in monocyte-derived macrophages, Arrevioscler Thvornb Vase B i d , (1997), 17, 159 1-8. 49. P. Maly, A. Thall, B. Petryniak, C. E. Rogers, P. L. Smith, R. M. Marks, R. J. Kelly, K. M. Gersten, G. Cheng, T. L. Saunders, S. A. Camper, R. T. Camphausen, F. X. Sullivan, Y. Isogai, 0. Hindsgad, U. H. von Andrian, J. B. Lowe, The alpha(l,3)fucosyltransferaseFuc-TVII controls leukocyte trafficking through an essential role in L-, E-, and P-selectin ligand biosynthesis, Cell, (1996), 86, 643-53. 50. R . N. Knibbs, R. A. Craig, P. Maly, P. L. Smith, F. M. Wolber, N. E. Faulkner, J. B. Lowe, L. M. Stoolman, Alpha( 1,3)-fucosyltransferase VII-dependent synthesis of P- and E- selectin ligands on cultured T lymphoblasts, J Immunol, (1998), 161, 6305-15. 51. C. A. van Wely, A. D. Blanchard, C. J. Brittcn, Differential expression of alpha3 fucosyltransferases in Thl and Th2 cells correlates with their ability to bind P-selectin, Biochern Biophys Res Comrnun, (1998), 247, 307-1 1. 52. A. J. Wagers, C. M. Waters, L. M. Stoolman, G. S. Kansas, Interleukin 12 and interleukin 4 control T cell adhesion to endothelial selectins through opposite effects on alphal,3-fucosyltransferase VII gene expression, J Exp Med, (1998), I88, 2225-223 1. 53. Z . Ce, N. W. Chan, M. M. Palcic, D. E. Taylor, Cloning and heterologous expression of an alphal,3-fucosyltransferasegene from the gastric pathogen Helicobacter pylori, J B i d Chem, (1997), 272, 21357-63. 54. S. L. Martin, M. R. Edbrooke, T. C. Hodgman, D. H. van den Eijnden, M. I. Bird, Lewis X biosynthesis in Helicobacter pylori. Molecular cloning of an alpha( 1,3)-fucosyltransferase gene, J Biol Chern, (1997), 272, 21349-56. 55. N. Uozumi, S. Yanagidani, E. Miyoshi,'Y. Ihara, T. Sakuma, C. X. Ciao, T. Teshima, S. Fuji, T. Shiba, N. Taniguchi, Purification and cDNA cloning of porcine brain GDP-L-Fuc:N-acetylbeta- D-glucosaminide alpha1 i6fucosyltransferase, J Biol Clzem, (1996), 271, 27810- 7. 56. Y. Wang, M. W. Spellman, Purification and characterization of a GDP-fucose:polypeptide fucosyltransferase from Chinese hamster ovary cells, J Biol Chem, (1998), 273, 81 12-8. 57. G. Srikrishna, L. Wang, H. H. Freeze, Fucosebeta-1-P-Ser is a new type of glycosylation: using antibodies to identify a novel structure in Dictyostelium discoideum and study multiple types of fucosylation during growth and development, Glycohiology, (1998), 8, 799-8 1 1. 58. P. Teng-umnuay, H. R. Morris, A. Dell, M. Panico, T. Paxton, C. M. West, The cytoplasmic F-box binding protein SKPl contains a novel pentasaccharide linked to hydroxyproline in Dictyostelium, J Biol Chem, (1998), 273, 18242-9.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

12 Sialyltransferases Joseph T. Y. Lau and Sherry A . Wuensch

12.1 Introduction Sialic acids, ubiquitous components of mammalian glycoproteins and glycolipids, are a group of closely-related carboxylated sugars found at the non-reducing termini of glycan chains. It has long been appreciated that the extent of sialylation is a prime factor governing the circulatory half-life of serum glycoproteins [3]. More recent is the appreciation that sialylated glycans participate in physiologic processes involving cell to cell and cell to matrix interactions. Sialylated glycans serve as ligands to a diverse array of cellular receptors, notable among these receptors are members of the Selectin and the Siglec families. Expression of sialylated glycans changes with development, differentiation, and during oncogenic transformation. Cell-surface sialyl epitopes are also recognized by a number of pathogenic organisms and toxins. These issues are addressed in a number of reviews and will not be exhaustively discussed here [ 12, 82, 90, 93-95]. The attachment of sialic acids is mediated by the sialyltransferases. Fifteen or more mammalian sialyltransferases are believed to exist to account for the complete complement of sialic acid linkages in mammals. Of these, 13 distinct sialyltransferase species have been cloned [31, 91, 961. These are summarized in Table 1. Sialic acids and sialyltransferases are not unique to mammals; they have also been documented in birds [91], fish [16, 21, 971, amphibians [ 5 , 24, 531, echinoderms [37, 461, and bacterial sources [9, 22, 1061. Sialyltransferases in yeast and in insects [83] have also been reported but their existence remains somewhat controversial.

12.2 General Features of Sialyltransferases Sialyltransferases mediate the general reaction: CMP-sialic acid + acceptor

--f

CMP + sialyl-acceptor

PST- 1 SAT-V/SAT-I11

glycoprotein Siau2-3Galpl-4GlcNAc GD3 (b-series) glycoprotein GDla, GTI b, GD3

a

Shown in bold is the sugar residue to which the sialic acid is attached. Sources from which sialyltranferases have been clone: p = pig, m = mouse, c = chicken, h = human, r f = frog

STXSia IV STXSia V

STXSia I1 ST8Sia 111

GM3 (a-series)

SAT-I1 (SAT-III), GD3 synthase STX

2,8-STs STXSia I

Siau2-3Galpl-3GalNAca-Ser/Thr Siau2-3Gal~l-3GalNAca-Ser/Thr

GalPl-4GlcNac GalNAca-Ser/Thr Gale 1-3GalNAca-Ser/Thr Siaa2-3Galp1-3GalNAca-Ser/Thr GalPl-3GalNAcu-Ser/Thr

GalPl,3GalNAc Galpl-3GalNAc GalPl,3GlcNAc Galpl-4GlcNAc Gal-Glc-ceramide (GA3)

Preferred accepter(s)”

GMlb (Siaa2-3Galp1-3)GlcNacp-R

STY, ST60-I1

ST-6N, SiaT-1, ST-I ST30-I

ST-30,ST3GalA.I ,ST-2 ST3GalA.2, SAT-IV ST-3N, ST-3 ST-Z, SAT-3, ST-4 GM3 synthase

Other names

ST6GlcNAc

ST6GalNAc 111

ST6GalNAc 11

2,6-STs ST6Gal I ST6GalNAc I

2,3-STs ST3Gal I ST3Gal I1 ST3Gal 111 ST3Gdl I v

Name

Table 1. The sialyltransferase gene family.

= rat,

b = bacteria, ham = hamster, and

ham [IS], m [ 1091, h [75],r [Xl],b [9] m [521, h 1411

r [65],m [49, 501, x [53] m [lox]

h [29, 77, 851, m [108],r [ l o ] ; 1111, c [13]

r [ 1021, h [25],m [26],c 1571, b [ 1061 c [551

P ~ 3 1m, ~ 3 1c, W I ,h PI m (621, r [62],h [42],b [22] r [103],h [43],m [51] h [45, 861, m [51j

Cloned inb

j;

i? ‘U

a

4

2

2

S’ G-

t v

GI

.r

P

E

12.3 Cloning and Identificrrtion Strutegiesfor Sialyltransjcrases

21 5

Sialyltransferases, as common to all glycosyltransferases, are type I1 transmembrane proteins. A generalized sialyltransferase is a protein containing a short 3-11 amino acid NHz-teminal cytosolic domain, a 16-20 amino acid transmembrane domain, a 30-200 amino acid stem region, followed by a extensive 300350 residue COOH-terminal catalytic domain. Sialyltransferases generally reside within the Golgi network [6, 671, although catalytically active soluble forms can be generated in vivo by proteolytic cleavage at the stem domain. For example, a significant level of soluble ST6Gal I is present in serum that is elevated as part of the systemic response to trauma [38, 1051. The serum soluble ST6Gal I is thought to be derived from the liver by proteolytic liberation from its membrane anchor [102]. Shared among the 13 cloned species of higher vertebrate sialyltransferases are two conserved regions, termed sialylmotifs L and S of 48 and 23 amino acids, respectively, that are localized within the catalytically active domain. Single point mutation studies using ST6Gal I as model indicate that the larger “L” motif contributes to the binding of the donor substrate [ 141, while the shorter and more carboxy terminal-situated “S” motif contributes to binding of both donor and acceptor substrates [ 151. Outside of the sialylmotif regions, there is little noticeable sequence similarity among the sialyltransferase species. However, it is not known the sialylmotif consensus regions are present in all sialyltransferases of higher vetebrates. At least among bacteria, presence of the recognizable sialylmotif domains is not universal. For example, a sialyltransferase exhibiting the higher vertebrate ST6Gal I activity has been isolated from the marine bacterium, Plzotobacteriurn damsel, which does not possess the sialylmotif regions [ 1061.

12.3 Cloning and Identification Strategies for Sialyltransferases Sequence elucidation of glycosyltransferases is arguably the major turning point in understanding the regulation and functionality of glycans. It permits generation of pure recombinant proteins for structural and catalytic analysis and for the production of specific antibody probes. Primary sequence information also permits the isolation and characterization of the cognate genes, thereby allowing studies on their regulation and the introduction of mutations into the germline by targeted mutagenesis to address functional issues. ST6Gal I is the second glycosyltransferase [102], after the bovine milk p1,4 galactosyltransferase [87], to be cloned. This end was achieved by the traditional approach which involved purification of native ST6Gal I from rat liver. Antisera raised against the native protein was used to screen a lambda g tll library. This conventional approach proved to be impractical for most sialyltransferases, given the low steady state levels of these enzymes in vivo. More innovative approaches have been adopted. One method is expression cloning (reviewed in [20]),introduced into the glycobiology field by John Lowe. Briefly, a mixture of cDNA sequences from a source known to express a sialyltransferase activity is introduced into recipient cells that do not express the enzyme. Transfected cells are screened for the

2 16

12 Sialyltransferases

presence of an introduced sialyltransferase cDNA by their ability to display the cognate sialyl epitope. In an innovative approach adopted from cloning of members of steroid hormone superfamilies, Paulson’s group and then Tsuji’s group took advantage of conserved sialylmotifs to synthesize redundant oligonucleotides. The degenerate oligonucleotides were used as primers in RT-PCR to generate cDNAs to additional sialyltransferase species [911. This procedure has been wildly successful, and the majority of the known sialyltransferase cDNAs were identified by this approach.

12.4 Sialyltransferase Classification and Nomenclature Sialyltransferases can be divided into three broad classes according to the manner of sialic acid attachment to the acceptor substrate. Under the nomenclature proposed by [92], these broad classes are designated ST3, ST6, and ST8 to denote enzymes that transfers sialic acids to the 3rd, 6th, or gth hydroxyl positions of the underlying sugar, respectively. A summary of known and hypothetical sialyltransferases is summarized in Table 1. Thirteen distinct sialyltransferase species are named. Two additional activities, the GM3 synthase and the a2,6 to GlcNAc sialyltransferase are also listed although their primary structural information remain elusive.

12.5 The a2,3-ST Family All known ST3 enzymes transfers sialic acids to the third position of the galactose in acceptor glycans. Four species of the ST3 subfamily have been cloned, ST3Gal I, ST3Gal 11, ST3Gal 111, and ST3Gal IV. These four enzymes exhibit overlapping acceptor specificities but distinct profiles of tissue and developmental specificities [5 11. The sialyla2,3 to GalPl,3GalNAc-R linkage, present on both O-linked glycans and gangliosides, can be elaborated by ST3Gal I and ST3Gal 11. Recombinant ST3Gal I and I1 exhibit similar acceptor preference of GalPl,3GalNAc- >>> GalPl,3GlcNAc-, and negligible activity towards Galpl,4GlcNAc- [94]. Between the two enzymes, ST3Gal I1 differs in that it exhibits a distinct preference for glycolipid acceptors. Glycolipid acceptors are also used by ST3Gal I but at comparatively lower efficiency. While most tissues express both ST3Gal I and I1 to some degree, ST3Gal I1 mRNA levels are dramatically elevated level in brain and spleen, organs normally associated with higher ganglioside biosynthesis [8]. In contrast, ST3Gal I mRNA is more moderately expressed in most tissues, with more enrichment in mucin producing tissues such as submandibular glands and with lower levels in brain. Taken together, these observations suggest that ST3Gal I is the dominant enzyme utilized in vivo for sialylation of O-linked chains, while ST3Gal

12.6 The ct2,6-ST Fumily

217

I1 is dominant enzyme in ganglioside biosynthesis. Siaa2,3Galpl,3GalNAc- containing gangliosides (GDla, GTl b, etc) are ligands for MAG (Myelin-Associated Glycoprotein), a member of the Siglec family of carbohydrate-binding receptors [ 1 1, 391. Mutant mice unable to express MAG show age-related degeneration of the myelin sheath and increased neurofilament density within the axon [ 19, 64, 71. 1071. ST3Gal 11, because of its elevated expression in neuronal tissues and preference for glycolipid acceptors, is likely the enzyme utilized in vivo for the synthesis of MAG ligands. However, ST3Gal I is also present, albeit at reduced levels. It remains to be determined whether ST3Gal I and ST3Gal I1 can function interchangeably in vivo in the biosynthesis of ligands for MAG. In contrast to ST3Gal I and 11, GalPl,3GalNAc- is not a preferred acceptor for ST3Gal I11 and ST3Gal IV. ST3Gal 111, formerly known as ST3N, selectively acts on Gal-GlcNAc- acceptors and exhibits a distinct preference of GalPl,3GlcNAc >> Galpl,4GlcNAc. ST3Gal IV, in contrast, prefers GalP1,4GlcNAc >> GalPl,3GlcNAc. Furthermore, both ST3Gal 111 and ST3Gal JV exhibit negligible activity towards glycolipids, i.e. nLc4Cer and LacCer. Taken together, this data suggest the existence of additional a2,3-STs that modify nLc4Cer and LacCer in vivo. An additional a2,3-ST is the CM3 synthase that elaborates the a2,3-sialyl linkage to Gal-Glc-ceramide (GA3). Neither ST3Gal I, 11, 111, nor IV exhibits GM3 synthase activity. Thus, this linkage must be mediated by a distinct sialyltransferase [4,71.

12.6 The a2,6-ST Family Biosynthesis of the a2,6-sialyl linkage to Gal, GalNAc, or GlcNAc are mediated by the a2,6-STs. The Sia(a2,6) to Galp1,4GlcNAc- linkage is unique in that it can be synthesized by only one sialyltransferase, the ST6Gal I. By virtue of being the first sialyltransferase cloned [ 1021, the largest body of literature exists for ST6Gal I, formerly known as ST6N, SiaT-1, or ST1. Expression of ST6Gal I is highest in liver [78, 981, where its expression is mediated by circulatory glucocortoids and IL-6 [35, 61, 99, 104, 105, 1141. Elevated hepatic ST6Cal I expression is part of the acute phase reaction, a hepatic response to systemic stress or localized injury [35].ST6Gal I is also elevated in mature B cells [loo], where its catalytic product. Sia(a2,6) Gal(P1,4)GlcNAc-, serves as ligand for the B cell surface receptor, CD22 [lo, 401. Mice unable to elaborate ST6Gal I exhibit impaired B cell proliferative response, reduced serum 1gM levels, and attenuated antibody production to T-dependent and T-independent antigens [32]. ST6Gal I and its cognate sialyl product in endothelial cells has also been implicated in lymphocyte trafficking [27, 281. Expression of ST6Gal I gene is transcriptionally regulated by a number of physically distinct promoters, each of them are operative with their distinct subset of cellular specificities. For example, promoters P1 and P2 are responsible for expression in liver and B cells, respectively [66, 1001.

2 18

12 Siulyltrunsferuses

Three distinct enzymes are known that mediate the synthesis of the Sia(a2,6) to core GalNAc residue of O-linked oligosaccharides. These three enzymes, ST6GalNAc I, 11, and 111 differ in their range of acceptor specificities. ST6GalNAc I can mediate Sia(a2,6) transfer to the GalNAc in Siaa2,3Galpl,3GalNAc-R (Sialyl-T antigen), Galpl,3GalNAc-R (T antigen), and GulNAc-R (Tn antigen) [55]. ST6GalNAc I1 can mediate sialic acid transfer to sialyl-T and T antigens [58]. ST6GalNAc 111’s activity is even more restricted and only acts on the sialyl T structure. SialylGalNAc I11 can also act on the ganglioside GMlb, which also contains the Siaa2,6Galp1,3GalNAc-R structure [88]. It is interesting to note that among the three enzymes, only ST6GalNAc I can utilize the Tn structure to produce the sialyl-Tn epitope. a2,6-Sialylation of Tn blocks all further modification on the glycan chain. The sialyl-Tn epitope, while expressed heavily in embryonal chick brains [55], is mostly absent in adult human tissues. In humans, expression of sialyTn is limited to adenocarcinomas and certain malignant tumors [34]. A third class of a2,6-STs are enzyme(s) that mediate the transfer of Sia(a2,6) to GlcNAc in N-glycans. Though the activity has been described from a number of species [ 17, 801, the enzyme(s) responsible for this activity have not been identified.

12.7 The a2,S-ST Family 1x2,s sialyl are present in both glycolipid and glycoproteins. In glycoproteins, a2,8 sialic acids exist as long homopolymers (polysialic acids) in a small set of molecules, most notably in embryonic forms of NCAM [60]. Polysialylation alters the homophilic binding properties of NCAM and has been thought to influence normal changes in cellular adhesiveness during embryonic brain development [84]. The presence of polysialic acids on NCAM correlates with expression of polysialyltransferases and promotes neoblastoma cell growth [ 331. Polysialic acids have also been found in the a subunit of the sodium channel protein of the developing brain [ 1131. Presence of polysialic structures on O-glycans has also been reported in breast carcinoma and leukemia cells [69].At least two distinct sialyltransferases, ST8Sia I1 and ST8Sia IV mediate the elaboration of polysialyl chains on glycoproteins. ST8Sia I1 and IV differ mostly in tissue distribution [2, 791. ST8Sia 11, formerly also known as STX, is highly enriched in fetal brain and marginally in most adult tissues 12, 591. In contrast, ST8Sia IV, also known as PST-1, is strongly expressed in adult lung, spleen, and heart, but only marginally in embryonic and adult brain [33, 791. There are also catalytic differences between ST8Sia I1 and IV. Under in vitro conditions, longer a2,8 sialyl polymers are formed by ST8Sia IV than by ST8Sia I1 [47]. At least three sialyltransferases mediate the a2,8 sialyl attachment in glycolipids. These enzymes, ST8Sia I, 111, and V are differentially expressed during mouse brain development [52, 91, 108, 1101. ST8Sia I and I11 are strongly expressed in day 20 fetal brain and declining to minimal levels in adults; expression of ST8Sia V is highest in adult brain. ST8Sia I, 111, and V exhibit highly similar and overlapping acceptor specificities 1521. ST8Sia I strongly prefers GM3 as substrate. ST8Sia 111,

12.8 Regulation und Functionality qf Sialyltransferases

2 19

acts on both GM3 but can also utilize GD3. ST8Sia 111, in vitro, also acts on glycoprotein acceptors such as fetuin, and to a lesser extent, al-acid glycoprotein. This is in contrast to ST8Sia I, which exhibits minimal activity towards glycoprotein acceptors [ 1081. It has been proposed that ST8Sia I11 serves as the polysialic “initiator” by attaching the first a2,8-linked sialic acid onto Siaa2,3Galpl,4GlcNAc on glycoproteins. However, both cloned polysialyl-transferases (ST8Sia I1 and IV) are able to elaborate a2,8 sialic acid homopolymers without requiring an initiator a2,Ssialyltransferase for the first a2,S-sialyl residue [48, 72, 741. ST8Sia V preferentially modifies G D l a , GTlb, and GD3 but not GM3. The significance and functional specialization among these glycolipid a2,S-STs remain not well understood at the present time.

12.8 Regulation and Functionality of Sialyltransferases The set of biologically relevant sialic acid structures is elaborated by the family of sialyltransferases acting in concert. With only a few exceptions, a single sialic acid structure can be synthesized, at least under in vitro conditions by more than one sialyltransferase. What remains unclear is the in vivo contribution of the individual sialyltransferase species. A tantalizing indication of functional specialization among the sialyltransferases is that strict development and tissue-specific programs govern the expression of the individual sialyltransferase genes. Understanding the mechanistic basis governing sialyltransferase gene expression should be paramount, although to date only a few sialyltransferase genes have been mapped and their structures elucidated in detail. These are summarized in Table 2 and include only the ST6Gal I, ST3Gal I, ST3Gal IV, and a number of the a2,S-STs.

Table 2. Mapped sialyltransferase genes Sialyltransferase

ST6Gal I ST3Gal I ST3Gal I1 ST3Gal IV ST8Sia I

STXSia I1 STRSia IV

* N A = not available

Gene name/chromosome location Human

Mouse

SIATllchr 3 (q21Lq28) [loo] SIAT4AIChr 8 IS] SIAT4BIChr 1 ( ~ 2 1 ~ ~[8] 34) SIAT4CIChr 11 (q23.3-qter) [R] /Chr 1 1 (q23-24) [44] SIAT8lChr 12 (~12.1-p11.2) [70,761 NA*/Chr 15 (q26) [2] NA*/Chr 5 (p21) 121

Siat-l/Chr 16 [36]

Siat-S/Chr 6 [70]

220

12 Siulyltransferases

Figure 1. ST6Gal 1 gene organization and tissue specificity of Sintl mRNA isoforms. The murine ST6Gal I gene, Siut-1, is schematically depicted on top. A number of mRNA isoforms are generated by alternate usage of promoters PI, P2, and P3 (lower left). Shaded box regions represent protein coding domains that are shared among known mRNA isoforms. Probes specific toward exons Q, H, X I , or the common coding regions were used to probe a panel of R N A from various mouse tissues. (Lanes 1-8: testis, kidney, muscle, liver, lung, spleen, brain, and heart, respectively)

In the one well-documented exception to the “one linkage, multiple sialyltransferase” generality, the Siaa2,6 to GalPl,4GlcNAc structure is synthesized only by ST6Gal I. Compelling evidence suggests that ST6Gal I and its cognate sialyl products contribute to different biological processes in different tissues. As mentioned previously, ST6Gal I product serves as ligand for CD22 and participates in B cell differentiation and activation. In liver, ST6Gal I expression is closely associated with the acute phase reaction, a hepatic response to systemic trauma. Significantly, multiple and physically distinct promoters, each exhibiting individual programs of tissue and developmental specificities, govern the expression of the ST6Gal I gene [Figure 11. At least three distinct ST6Gal I transcriptional promoters exist. Promoter P1 mediates transcription in hepatic tissues; P2 transcription is restricted only to mature B cells; P3, in contrast, appears to be a non-tissue specific promoter that is active in most cells and tissues albeit at a low level. P1 transcription is modulated by glucocorticoids and 11-6, and is believed to be responsible for increased ST6Gal I expression during the acute phase response. ST6Gal I expression by P2, on the other hand, may contribute to the synthesis of ligands for CD22 during B-differentiation. Existence of multiple promoters is likely not unique among the sialyltransferases. Because sialyltransferases are expressed in most cells and tissues where their functionality may differ, evolution of multiple promoters provide additional pathways to modulate their expression. Evidence for multiple promoter exists for at least one other sialyltransferase, the ST3Gal IV [44]. Regulation of sialyltransferase expression by post-transcriptional mechanisms remains largely unexplored. For ST6Gal I, differential promoter usage results in divergent 5’-untranslated regions in the mRNAs. In vitro translation experiments

12.8 Regulation a i d Functionality of Sialyltvansferuses

22 1

suggest that the divergent 5’-UTs confer different translational efficiencies of the ST6Gal I mRNA isotypes [I]. Post-translational modification of sialyltransferase proteins may also modulate enzymatic activity. Naturally occurring ST6Gal I protein disulfide dimers, are shown to be catalytically inactive although their capacity to bind acceptor remains undiminished [ 681. The physiologic significance of these naturally occurring ST6Gal I dimers remains unknown. Allosteric modifiers of ST6Gal I enzymatic activity have also been sporadically reported. Rat al-macrogloblobulin has been reported as an inhibitor [30]while a rat colon cytosolic factor has been reported to be a stimulator of ST6Gal I activity [73]. As a final caveat, much of our current understanding of sialyltransferase acceptor specificities, and our extrapolation of biologic role based on these specificities, are based on behavior of recombinant enzymes in in vitro assays. Only the soluble, catalytic domain, and often times the catalytic domain fused to a heterologous sequence, is expressed. Taken out of their physiologic environment (i.e. lumen of the Golgi), observed sialyltransferase acceptor specificities may be deceptive. Indeed a recent report demonstrated substantial differences in catalytic specificities between the soluble and complete forms of two glycosyltransferases, a p1,4-GalNAc transferase and ST6Gal I [112]. When both forms are expressed in cells, the complete GalNAc transferase can efficiently mediate GM2 production, while little GM2 is synthesized in the soluble GalNAc transferase transfected cells. Similarly, cells transfected with intact ST6Gal I sequence is at least twice as efficient in sialylation as the counterpart transfected with soluble form of ST6Gal I. It is clear that much work remains before a thorough understanding of the complexities of the sialyltransferases and their in vivo functional contributions. With more complete understanding of the genes that encode the distinct sialyltransferases, specific targeted mutations may be introduced by the “knock out” approach whereby in vivo functional questions can be addressed.

References 1 . Aasheim HC, Aas-Eng DA, Deggerdal A , Blomhoff HK, Funderud S, et al. Cell-specific expression of human beta-galactoside 2,6-sialyltransferase transcripts differing in the 5’ untranslated region. Eur. J. Biochcm. 1993; 213: 461-415. 2. Angata K, Nakayama J , Fredette B, Chong K, Ranscht B, et al. Human STX polysialyltransferase forms the embryonic form of the neural cell adhesion molecule. Tissue-specific expression, neurite outgrowth, and chromosomal localization in comparison with another polysialyltransferase, PST. J. Bid. Chem. 1997; 272: 7182-7 190. 3. Ashwell G, Morel1 AG. The Role of Surface Carbohydrates in the Hepatic Recognition and Transport of Circulating Glycoproteins. A ~ LEnzymol. ?. 1974; 41: 99- 128. 4. Basu M, De T, Das KK, Kyle JW, Chon HC, et al. Glycolipids. Meth. Enzjwol. 1987; 138: 575-607. 5. Becker CG, Becker T, Roth G. Distribution of NCAM-I80 and polysialic acid in the developing tectum mesencephali of the frog Discoglossus pictus and the salamander Pleurodeles waltl. Cell & Tissue Research 1993; 272: 289-301. 6. Burger PC, Lotscher M, Streiff M, Kleene R , Kaissling B, et al. Immunocytochemical localization of 2,3(N)-sialyltransferase (ST3Gal 111) in cell lines and rat kidney tissue sections: evidence for golgi and post-golgi localization. G!ycobiology 1998; 8: 245-257.

222

12 Sialyltransferases

7. Busam K, Decker K. Ganglioside biosynthesis in rat liver. Characterization of three sialyltransferases. Eur. J. Biochent 1986; 160: 23-30. 8. Chang ML, Eddy RL, Shows TB, Lau JTY. Three genes that encode human beta-galactoside 2,3-sialyltransferases. Structural analysis and chromosomal mapping studies. Glycohioloyy 1995; 5: 319-325. 9. Claus H, Vogel U, Muhlenhoff M, Gerardy-Schahn R, Frosch M. Molecular divergence of the sia locus in different serogroups of Neisseria meningitidis expressing polysialic acid capsules. Mol. Gen. Genet. 1997; 257 28-34. 10. Collins BE, Kiso M, Hasegawa A, Tropak MB, Roder JC, et al. Binding specificities of the sialoadhesin family of I-type lectins. Sialic acid linkage and substructure requirements for binding of myelin-associated glycoprotein, Schwann cell myelin protein, and sialoadhesin. J. Biol. Chern. 1997; 272: 16889-16895. 1I . Collins BE, Yang LJ, Mukhopadhyay G, Filbin MT, Kiso M, et al. Sialic acid specificity of myelin-associated glycoprotein binding. J. Biol. Chern. 1997: 272: 1248-1255. 12. Crocker PR, Clark EA, Filbin M, Gordon S, Jones Y , et al. Siglecs: a family of sialic-acid binding lectins [letter]. Glycohiology 1998; 8: v 13. Daniotti JL, Rosales F, V, Kunda P, Nishi T, Maccioni HJ. Cloning, characterization and developmental expression of 2,8 sialyltransferase (GD3 synthase, ST8Sia I) gene in chick brain and retina. lntl J. Dev. Neurosci. 1997; 15: 767-776. 14. Datta AK, Paulson JC. The sialyltransferase “sialylmotif” participates in binding the donor substrate CMP-NeuAc. J. Biol. Chem. 1995; 270: 1497-1500. 15. Datta AK, Sinha A, Paulson JC. Mutation of the sialyltransferase S-sialylmotif alters the kinetics of the donor and acceptor substrates. J. Biol. Chern. 1998; 273: 9608-9614. 16. Datta S, Datta SC. Purification and characterization of fish surface mucin. ltul. J. Biochrm. 1987; 36: 143 152. 17. de Heij HT, Koppen PL, Van den Eijnden DH. Biosynthesis of disialylated beta-Dgalactopyranosyl-( 1-3)-2-acetamido-2-deoxy-beta-D-glucopyran osyl oligosaccharide chains. Identification of a beta-D-galactoside -(2-3)- and a 2-acetamido-2-deoxy-beta-D-glucoside -(2-6)- sialyltransferase in regenerating rat liver and other tissues. Curbolzycir. Rex 1986; 149: 85-99. 18. Eckhardt M, Muhlenhoff M, Bethe A. Koopman J, Frosch M, et al. Molecular characterization of eukaryotic polysialyltransferase-1. Nuture 1995; 373: 71 5-718. 19. Fruttiger M, Montdg D, Schachner M, Martini R. Crucial role for the myelin-associated glycoprotein in the maintenance of axon-myelin integrity. Eur. J. Neurosci. 1995; 7: 511515. 20. Fukuda MN, Bierhuizen MF, Nakayama J. Expression cloning of glycosyltransferases. G/ycohioloyy 1996; 6: 683-689. 21. Ghosh R, Sikder SK, Das A. Structure of the carbohydrate portion of the sialo-glycopeptide isolated from the skin of the fish Mastacembalus armatus. Curbohydr.. Res. 1987; 162: 257269. 22. Gilbert M, Watson DC, Cunningham AM, Jennings MP, Young NM, et al. Cloning of the lipooligosaccharide -2,3-sialyltransferase from the bacterial pathogens Neisseria meningitidis and Neisseria gonorrhoeae. J. Biol. Chem. 1996; 271: 28271-28276. 23. Gillespie W, Kelm S, Paulson JC. Cloning and expression of the Galpl,3GalNAc a2,3-sialyltransferase. J. Biol. Chrrn. 1992; 267: 21004-21010. 24. Gornati R, Basu S, Bernardini G, Rizzo AM, Rossi F, et al. Activities of glycolipid glycosyltransferases and sialidases during the early development of Xenopus laevis. Mol. Cell. Biothem. 1997; 166: 117-124. 25. Grundmann U; Nerlich C, Rein T. Zettlmeissl G. Complete cDNA sequence encoding human a-gdlactoside w2,6-sialyltransferase. Nucltlic Acids Res. 1990; 18: 667 26. Hamamoto T, Kawasaki M, Kurosawa N, Nakaoka T, Lee YC, et al. Two step single primer mediated polymerase chain reaction. Application to cloning of putative mouse, beta-gdlactoside alpha2,6-sialyltransferase cDNA. Bioory. Med. Chem. 1993; 1: 141-145. 27. Hanasaki K, Varki A, Powell LD. CD22-mediated cell adhesion to cytokine-activated human endothelial cells. Positive and negative regulation by u2-6-sialylation of cellular glycoproteins. J. Bid. Chern. 1995; 270: 7533-7542.

References

223

28. Hanasaki K, Varki A, Stamenkovic I, Bevilacqua MP. Cytokine-Induced beta-Galactoside u2,6-Sialyltransferase in Human Endothelial Cells Mediates 2,6-Sialylation of Adhesion Molecules and Cd22 Ligands. J. Bid. Clicm. 1994; 269: 10637-10643. 29. Haraguchi M, Yamashiro S, Yamamoto A, Furukawa K, Takamiya K, et al. Isolation of GD3 synthase gene by expression cloning of GM3 -2,8-sialyltransferase cDNA using antiGD2 monoclonal antibody. Proc. Natl Acad. Sci. U S A 1994; 91: 10455-10459. 30. Harder PG, Jamieson JC. Identification of rat 1 macroglobulin as an inhibitor of rat Galbeta1 -4GlcNAc alpha2-6 sialyltransferase. Glycohioloyy 1997; 7: 791 -801, 3 I . Harduin-Lepers A, Recchi MA, Delannoy P. 1994, the year of sialyltransferases. Glycohioloyy 1995; 5: 741-758. 32. Hennet T, Chui D, Paulson JC, Marth JD. Immune regulation by the ST6Gal sialyltransferase. Proc. Natl Acad. Sci. U S A 1998; 95: 450444509, 33. Hildebrandt H, Becker C, Gluer S, Rosner H, Gerardy-Schahn R, et al. Polysialic acid on the neural cell adhesion molecule correlates with expression of polysialyltransferases and promotes neuroblastoma cell growth. Cancer Res. 1998; 58: 779-784. 34. Itzkowitz S, Kjeldsen T, Friera A, Hakomori S, Yang US, et al. Expression of Tn, Sialosyl Tn, and T-antigens in human pancreas. GListroenteroloyy 1991; 100: 1691- 1700. 35. Jamieson JC, Mccaffrey G, Harder PG. Sialyltransferase: a novel acute-phase reactant. Comp. Biochem. Physiol. B: Cornp. Biocheni. 1993; 105: 29-33. 36. Kalcheva I, Elliott RW, Dalziel M, Lau JTY, JTY. The gene encoding beta-galactoside 2,6sialyltransferase maps to mouse chromosome 16. Mammalian Genome 1997; 8: 61 9-620. 37. Kamerling JP, Schauer R, Vliegenthart JF, Hotta K. Identification of the sialic acids from the egg jelly coat of the sea urchin Pseudocentrotus depressus (Okayama). Hoppe-Seylers Zeitschrifi,fur Physiologische Chernie 1980; 361: 1511-1516. 38. Kaplan HA, Woloski BMRNJ, Hellman M, Jamieson JC. Studies on the effect of inflammation on rat liver and serum sialyltransferase: Evidence that inflammation causes release of Gal pl-4 GlcNAc u2-6 sialyltransferase from liver. J. Biol. Chern. 1983; 258: 11 505-1 1509. 39. Kelm, S., Brossmer, R., Isecke, R., Gross, H. J.. Strenge, K., and Schauer, R. Functional groups of sialic acids involved in binding to siglecs (sialoadhesins) deduced from interactions with synthetic analogues. Eur. J. Bioc,herri. 1998; 255: 663-672 40. Kelm S, Pelz A, Schauer R, Filbin MT, Tang S, et al. Sialoadhesin, MAG, and CD22 define a new family of sialic acid-dependent adhesion molecules of the immunoglobulin superfamily. Curr. Bid. 1994; 4: 965-972. 41. Kim YJ, Kim KS, Do S, Kim CH, Kim SK, et al. Molecular cloning and expression of human 2,8-sialyltransferase (hSTXSia V). Biochenz. Biophysic. Res. Commun. 1997; 235: 327330. 42. Kim YJ. Kim KS, Kim SH, Kim CH, KO JH, et al. Molecular cloning and expression of human Gal beta 1,3GalNAc alpha 2,3-sialytransferase (hST3Gal 11). Biochem. Biophy.sic. Rex Commun. 1996; 228: 324--327. 43. Kitagawa H, Livingston BD, Paulson JC. Cloning of novel genes related to the sialyltransferase gene family. Glycoconj. J. 1993; 10: 233-234.(Abstract) 44. Kitagawa H, Mattei MG, Paulson JC. Genoniic organization and hromosomal mapping of the Gal beta 1,3GalNAc/Gal beta 1,4GlcNAc alpha 2,3-sialyltransferase. J. Biol. Chern. 1996; 271: 931-938. 45. Kitagawa H, Paulson JC. Cloning of a Novel 2,3-sialyltransferase That Sialylates Glycoprotein and Glycolipid Carbohydrate Groups. J. Bid. Chen7. 1994; 269: 1394-1401. 46. Kitazume S, Kitajima K, Inoue S, Troy FA, Cho JW, et al. Identification of polysialic acidcontaining glycoprotein in the jelly coat of sea urchin eggs. Occurrence of a novel type of polysialic acid structure. J. Biol. Chem. 1994; 269: 22712-22718. 47. Kojima N, Tachida Y. Tsuji S. Two polysialic acid synthases, mouse STXSia 11 and IV, synthesize different degrees of polysialic acids on different substrate glycoproteins in mouse neuroblastoma Neuro2a cells. J. Biochenz. 1997; 122: 1265-1273. 48. Kojima N, Tachida Y; Yoshida Y, Tsuji S. Characterization of mouse ST8Sia TI (STX) as a neural cell adhesion molecule-specific polysialic acid synthase. Requirement of core 1,6linked fucose and a polypeptide chain for polysialylation. J. Bid. Chern. 1996; 271: 1945719463.

224

12 Sialyltrclnsferases

49. Kojima N, Yoshida Y, Kurosawa N, Lee YC, Tsuji S. Enzymatic activity of a developmentally regulated member of the sialyltransferase family (STX): evidence for u2,8-sialyltransferase activity toward N-linked oligosaccharides. FEBS Lett. 1995; 360: 1-4. 50. Kojima N, Yoshida Y, Tsuji S. A developmentally regulated member of the sialyltransferase family (ST8Sia 11, STX) is a polysialic acid synthase. FEBS Lett. 1995; 373: 119-122. 51. Kono M, Ohyama Y, Lee YC, Hamamoto T, Kojima N , et al. Mouse beta-galactoside alpha 2,3-sialyltransferases: comparison of in vitro substrate specificities and tissue specific expression. Glycobioloyy 1997; 7 469-479. 52. Kono M, Yoshida Y, Kojima N, Tsuji S. Molecular cloning and expression of a fifth type of a2,8-sialyltrdnSferaSe (ST8Sia V). Its substrate specificity is similar to that of SAT-V/Ill, which synthesize GDlc. GTla. G Q l b and GT3. J. B i d . Chem. 1996; 271: 29366-29371. 53. Kudo, M., Takdyamd, E., Tashiro, K., Fukamachi, H., Nakata, T., Tadakuma, T., Kitajima, K., Inoue, Y., and Shiokawa, K. Cloning and expression of an U2,8-polySialyltranSfe~dSe (STX) from Xenopus laevis. Glycobioloyy 1998; 8: 771-777 54. Kurosawa N, Hamdmoto T, Inoue M, Tsuji S. Molecular cloning and expression of chick Gal beta 1,3GalNAc alpha 2,3-sialyltransferase. Biochim. Biophys. Acfa 1995; 1244: 21 6-222. 55. Kurosawa N, Hamamoto T, Lee YC, Nakaoka T, Kojima N, et al. Molecular Cloning and Expression of GalNAc u2,6-sialyltransferase. J. Biol. Chem. 1994; 269: 1402-1409. 56. Kurosawa N, lnoue M, Yoshida Y, Tsuji S. Molecular cloning and genomic analysis of mouse Gal beta 1, 3GalNAc- specific GalNAc 2,6-sialyltransferase. J. Biol. Chem. 1996; 271: 1510915116. 57. Kurosawa N, Kawasaki M, Hamamoto T, Nakaoka T, Lee YC, et al. Molecular Cloning and Expression of Chick Embryo Gal beta I , 4Glcnac alpha 2,6-SiaiykrdnsferaseComparison with the Mammalian Enzyme. Euv. J. Biochem. 1994; 219: 375-381. 58. Kurosawa N, Kojimd N, Inoue T, Hamamoto T, Tsuji S. Cloning and expression of Galpl,3GalNAc- specific GalNAca2,6-sialyltransferase. J. Bid. Chem. 1994; 269: 1904819053. 59. Kurosawa N, Yoshida Y, Kojima N, Tsuji S. Polysialic acid synthase (ST8Sia IIjSTX) mRNA expression in the developing mouse central nervous system. J. Neurochem. 1997; 69: 494-503. 60. Lackie PM, Zuber C, Roth J. Polysialic acid of the neural cell adhesion molecule (N- CAM) is widely expressed during organogenesis in mesodermal and endodermal derivatives. Dij’ierentlulion 1994; 57: 119- 131. 61. Lammers G, Jamieson JC. Studies on the effect of experimental inflammation on sialyltransferase in the mouse and guinea pig. Comp. Bioclwm. Physiol. B: Comp. Biockem. 1986; 84: 181-187. 62. Lee YC, Kojima N, Wada E, Kurosawa N, Nakaoka T, et al. Cloning and Expression of cDNA for a New Type of Gal beta 1JGalNAc 2,3-sialyltransferase. J. B i d . Clien7. 1994; 269: 10028- 10033. 63. Lee YC, Kurosawa N, Hamamoto T Nakaoka T, Tsuji S. Molecular Cloning and Expression of Gal-beta-l,3GalNAc- alpha-2.34 lytransferase from Mouse Brain. Euv. J. Bioclienz. 1993; 216: 377-385. 64. Li C, Tropak MB, Gerlai R, Clapoff S, Abramow-Newerly W, et al. Myelination in the absence of myelin-associated glycoprotein. Nutiire 1994; 369: 747-750. 65. Livingston BD, Paulson JC. Polymerase chain reaction cloning of a developmentally regulated member of the sialyltransferase gene family. J. Biol. Chenz. 1993: 268: 11504-1 1507. 66. Lo NW. Lau JTY. Transcription of the beta-galactoside 2,6-sialyltransferdse gene in B lymphocytes is directed by a separate and distinct promoter. Glycobiolo~~y 1996; 6: 271-279. 67. Lovelock, C. and Lucocq, J. Quantitative immunoelectron microscopy reveals a2,6 sialyltransferase is concentrated in the central cisternae of rat hepatocyte Golgi apparatus. Eur. J Cell Biol. 1998; 76: 18-24 68. Ma J , Colley KJ. A disulfide-bonded dimer of the Golgi beta-galactoside u2,6-sialyltransferase is catalytically inactive yet still retains the ability to bind galactose. J. Bid. Chem. 1996; 271: 7758~-7766. 69. Martersteck CM, Kedersha NL, Drapp DA, Tsui TG, Colley KJ. Unique u2,8-polysialylated glycoproteins in breast cancer and leukemia cells. GI-vcobiology. 1996; 6: 289-301,

Rqfevences

225

70. Matsuda Y, Nara K, Watanabe Y, Saito T, Sanai Y. Chromosome mapping of the G D 3 synthase gene (SIAT8) in human and mouse. Gmonzics 1996; 32: 137-139. 71. Montag D, Giese KP, Bartsch U, Martini R, Lang Y, et al. Mice deficient for the myelinassociated glycoprotein show subtle abnormalities in myelin. Neuron 1994; 13: 229-246. 72. Muhlenhoff M, Eckhardt M, Bethe A, Frosch M, Gerardy-Schahn R. Polysialylation of NCAM by a single enzyme. Curr. Bid. 1996; 6: 1188-1 191. 73. Nagpurkar, A,, Hunt, D., and Mookerjea, S. Specific stimulation of n2-6 sialyltransferase activity by a novel cytosolic factor from rat colon. In/. J. Biochem. CeII Bid. 1996; 28: 1337-1348 74. Nakayama J, Fukuda MN. A human polysialyltransferasc directs in vitro synthesis o f polysialic acid. J. Bid. Cllem. 1996; 271: 1829--1832. 75. Nakayama J; Fukuda MN, Fredettc B, Ranscht B. Expression cloning of a human polysialyltransferase that forms the polysialylated neural cell adhesion molecule present in embryonic brain. Proc. Natl Acud. Sci. U S A 1995; 92: 7031-7035. 76. Nakayama J, Fukuda M N , Hirabayashi Y, Kanamori A, Sasaki K, et al. Expression cloning o f a human GT3 synthase. G D 3 A N D GT3 are synthesized by a single enzyme. J. B i d . Chem. 1996; 271: 3684-3691. 77. Nara K, Watanabe Y, Maruyama K , Kasahara K, Nagai Y, et al. Expression cloning of a CMP-NeuAc:NeuAc 2-3Gal beta I -4Glc beta 1-1 ’Cer 2,8-sialyltransferase (GD3 synthase) from human melanoma cells. Proc. Nail Acud. Sci. U S A 1994; 91: 7952-7956. 78. O’Hanlon TP, Lau K M , Wang XC, Lau JTY. Tissue-specific expression of P-galactoside a2,6sialyltransferase: transcript heterogeneity predicts a divergent polypeptide. J. Biol. Chem. 1989; 264: 17389-17394. 79. Ong E, Nakayama J, Angata K , Reyes L, Katsuyama T. et al. Developmental regulation of polysialic acid synthesis in mouse directed by two polysialyltransferases, PST and STX. Glycobioloyy 1998; 8: 41 5-424. 80. Paulson JC, Weinstein J, de S. Biosynthesis of a disialylated sequence in N-linked oligosaccharides: identification of an N-acetylglucosaminide (2-6)-sialyltransferase in Golgi apparatus from rat liver. Eur. J. Biochenz. 1984; 1411: 523-530. 81. Phillips GR, Krushel LA. Crossin KL. Developmental expression of two rat sialyltransferases that modify the neural cell adhesion molecule, N-CAM. Mol. Bruin Res. 1997; Dev.Brain Rcs. 102: 143-155. 82. Pilatte Y, Bignon J, Lambre CR. Sialic acids as important molecules in the regulation of the immune system: patholophysiological implications of sialidases in immunity. Glycobiolm~y 1993; 3: 201-217. 83. Roth J, Kempf A. Reuter G , Schauer R. Gehring WJ. Occurrence of sialic acids in Drosophila melanogaster. Science 1992; 256: 673-675. 84. Rothbard JB, Brackenbury R, Cunningham BA. Edelman G M . Differences in the carbohydrate structures of neural cell-adhesion molecules from adult and embryonic chicken brains. J. Biol. Clzcm. 1982; 257: I 1064-1 1069. 85. Sasaki K, Kurata K, Kojima N, Kurosawa N, Ohta S, et al. Expression cloning of a G(M3)spccific-a2,8-sialyltraiisfcrasc (G(D3) synthase). J. Biol. Cheni. 1994; 269: 15950--15956. 86. Sasaki K, Watanabe E, Kawashima K. Sekine S, Dohi T, et al. Expression Cloning of a Novel Gal beta( 1 -3/1-4)GlcNAc alpha 2,3-Sialyltransferase Using Lectin Resistance Selection. J. Biol. Cliem. 1993; 268: 22182- 22787. 87. Shaper NL, Shaper JH, Neuth JL, Fox JL, Chang H, et al. Bovine galactosyltransferase: Identification of a clone by direct immunological screening of a cDNA expression library. Proc. Nut1 Acad. Sci. U S A 1986; 83: 1573-1 577. 88. Sjoberg ER, Kitagawa H , Glushka J, van Halbeek H, Paulson JC. Molecular cloning o f a developmentally regulated N-acetylgalactosamine n2,6-~ialyltrdnsferasespecific for sialylated glycoconjugates. J. Bid. Clieni. 1996; 271: 7450-7459. 89. Soutiropoulou, G., Anisowicz, A., and Sager, R . Isolation and cloning from human mammary epithelial cells. 1994. GenBank/EMBL Data Bank under accession # U14550. 90. Thomas P. Cell surface sialic acid as a mediator of metastatic potential in colorectal cancer. Cuncer J. 1996; 9: 32-36. 91. Tsuji, S. Molecular cloning and functional analysis of sialyltransferases. J. Biockem. (Tokyo) 1996; 120: 1-13

226

12 Siulyltransferuses

92. Tsuji S, Datta AK, Paulson JC. Systematic nomenclature for sialyltransferases [letter]. G/ycobiology 1996; 6: v--vii. 93. Varki A. Biological roles of oligosaccharides. Glycobiology 1993; 3: 97-1 30. 94. Varki A. Selectin ligands: will the real ones please standup?J. Clin. Inoestig. 1997; 100: S31-S35. 95. Varki A. Sialic acids as ligands in recognition phenomena. FASEB J. 1997; 11: 248-255. 96. Wainwright LJ, Field MC. Molecular cloning of eukaryotic glycoprotein and glycolipid glycosyltransferases; A survey. Glycobiology 1995; 5: 463-472. 97. Waki H; Murata A, Kon K, Maruyama K, Kimura S, et al. Isolation and characterization of a trisialyllactosylceramide, GT3, containing an 0-acetylated sialic acid in cod fish brain. J. Biochem. 1993; 113: 502-507. 98. Wang XC, O’Hanlon TP, Lau JTY. Regulation of P-galactoside a2,6-sialyltransferase gene expression by dexamethasone. J. Bid. Chem. 1989; 264: 1854-1859. 99. Wang XC, Smith TJ, Lau JTY. Transcriptional regulation of the liver P-galactoside a2,6sialyltransferase by glucocorticoids. J. Biol. Chem. 1990; 265: 17849-1 7853. 100. Wang XC, Vertino A, Eddy RL, Byers MG, Jani-Sait SN, et al. Chromosome mapping and organization of the human beta-galactoside alpha 2,6-sialyltransferase gene. Differential and cell-type specific usage of upstream exon sequences in B-lymphoblastoid cells. J. Biol. Chem. 1993; 268: 4355-4361. 101. Watanabe Y, Nara K, Takahashi H, Nagai Y, Sanai Y. The molecular cloning and expression of 2,8-sialyltransferase (GD3 synthase) in a rat brain. J. Biochem. 1996; 120: 1020-1027. 102. Weinstein J, Lee EU, McEntee K, Lai PH, Paulson JC. Primary structure of (3-galactoside a2,6-sialyltransferase. Conversion of membrane-bound enzyme to soluble forms by cleavage of the NH2-terminal signal anchor. J. Biol. Chem. 1987; 262: 17735-17743. 103. Wen DX, Livingston BD, Medzihradszky KF, Kelm S, Burlingame AL, et al. Primary Structure of Galbeta1,3(4)GlcNAc 2,3-Sialyltransferase Determined by Mass Spectrometry Sequence Analysis and Molecular Cloning-Evidence for a Protein Motif in the Sialyltransferase Gene Family. J. Bid. Chrm. 1992; 267: 21011-21019. 104. Woloski BM, Fuller GM, Jamieson JC, Gospodarek E. Studies on the effect of the hepatocyte-stimulating factor on galactose-beta 1-4-N-acetylglucosamine 2-6-sialyltransferase in cultured hepatocytes. Biochim. Biophys. Acta 1986; 885: 185-191. 105. Woloski BMRNJ, Gospodarek E, Jamieson JC. Studies on monokines as mediators of acute phase response: Effects on sialyltransferases, albumin, -1-acid glycoprotein and (3-Nacetylhexosaminidase. Biochem. Biophys. Rex Cornrnun. 1985; 130: 30-36. 106. Yamamoto T, Nakashizuka M, Terada I. Cloning and expression of a marine bacterial betagalactoside 2,6-sialyltransferase gene from Photobacterium damsela JTOl60. J. Biochem. 1998; 123: 94-100. 107. Yin X, Crawford TO, Griffin JW, Tu P, Lee VM, et al. Myelin-associated glycoprotein is a myelin signal that modulates the caliber of myelinated axons. J. Neurosci. 1998; 18: 19531962. 108. Yoshida Y, Kojima N, Kurosawa N, Hamamoto T, Tsuji S. Molecular cloning of Sia 2,3Gal beta 1,4GlcNAc 2,8-sialyltransferase from mouse brain. J. Bid. Chem. 1995; 270: 1462814633. 109. Yoshida Y, Kojima N, Tsuji S. Molecular cloning and characterization of a third type of Nglycan 2,8-sialyltransferase from mouse lung. J. Biochem. (Tokyo) 1995; 118: 658-664. 110. Yoshida Y, Kurosawa N; Kanematsu T, Taguchi A, Arita M, et al. Unique genomic structure and expression of the mouse a2,8-sialyltransferase (ST8Sia 111) gene. Glycobiology 1996; 6: 573-580. 1 11. Zeng G, Gao L, Ariga T, Yu RK. Molecular cloning of cDNA for rat brain GD3-synthase. Biochem. Biophys. Res. Commun. 1996; 226: 319-323. 112. Zhu, GF, Allende, ML, Jaskiewicz, E: Qian, R, Darling, DS, Worth, CA, Colley, KJ, and Young, W. W. Two soluble glycosyltransferases glycosylate less efficiently in vivo than their membrane bound counterparts. Glycobiology 1998; 8: 83 1-840 113. Zuber C, Lackie PM, Catterall WA, and Roth J. Polysialic acid is associated with sodium channels and the neural cell adhesion molecule N-CAM in adult rat brain. J. Bid. Chem. 1992; 267: 9965-9971. 114. Dalziel and Yau. Unpublished observations.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

13 Biochemistry of Sialic Acid Diversity Roland Schuuev

13.1 Introduction A great variety of sialic acids (Sia) is known to occur in animals and some microorganisms. Sia significantly contribute to the enormous structural diversity of complex carbohydrates. We are trying to understand the multifacetted metabolic and biological effects of these acidic monosaccharides in general and of the specific roles of their diRerent substituents on physiological and pathological processes. In this respect, the main topics presently studied are cell protection, fertilization, differentiation, cell adhesion, immunology, inflammation and tumour growth. Although it is the present opinion that the structures of most Sia have been elucidated, the enzymes involved in their metabolism need more attention. The properties and structures of the catabolic enzymes sialidase, sialate lyase and sialate esterases have been studied, but on the anabolic site, only the CMP-N-acetylneuraminic acid hydroxylase and sialate-8-0-methyltransferase have been purified to homogeneity, and only the gene structure of the hydroxylase was investigated. It is challenging to focus on these enzymes, especially also on the widespread sialate 0-acetyltransferases, in order to better understand the regulation of the expression of diverse Sia under variable biological conditions.

13.2 Occurrence and Biosynthesis Sialic acids (Sia) are a family of monosaccharides comprising about 40 members which can be considered as derivatives of 2-keto-3-deoxy-nononic acid (Kdn), the 5amino derivative representing the long-known neuraminic acid [ 11. However, Kdn formally is a 5-desamino-5-hydroxy-neuraminicacid. Thus, all sialic acids are derivatives of neuraminic acid with different substituents at the amino or hydroxyl

228

13 Biochemistry of Sialic Acid Diversity

groups, or they are otherwise modified [ 1-41. In [ 11 most natural Sia found so far are listed, together with their main sources and abbreviations. The amino group of neuraminic acid is acetylated or glycolylated, while at all non-glycosidic hydroxyl residues one or various acetyl groups may occur. Usually, there is only one 0-acetyl group, mostly at 0-9. but di- and tri-0-acetylated Sia are known. especially in mucins from bovine submandibular gland and human colon [4]. At 0-9 lactyl or phosphoryl residues and at 0-8 methyl or sulfate groups can occur. All these different substituents may be combined, e.g. 8-0-methyl with 9-0-acetyl and Nglycolyl, which is typical for the starfish Asferias rubens, yielding the manifold Sia types [1-4]. The most recently discovered Sia is 5-0-methylated Kdn (KdnSMe), a component of the lipopolysaccharide of the nitrogen-fixing bacterium Sinorhizobium fredii [ 51. These Sia, except N-acetylneuraminic acid-9-phosphate (NeuSAc9P), found only as free sugar as an intermediate during Sia biosynthesis [6], usually occur in glycosidic linkages of oligosaccharides, glycoproteins, gangliosides and lipopolysaccharides [ 1-41. In contrast, the 2,3-unsaturated sialic acids, usually Neu2en5Ac but also found with various N - and 0-substituents [ 11, lack a glycosidic hydroxyl residue and therefore exist only as free monosaccharides in very low concentrations in the body fluids of man and animals. Another ncuraminic acid derivative incapable of binding glycosidically to other sugars, is 2,7-anhydro-N-acetylneuraminicacid (Neu2,7an5Ac) [ 11, which is the product of an intramolecular transglycosidation reaction during the hydrolytic release of Neu5Ac from glycosidic linkage by the action of a sialidase from the leech [ 71. N-Acetylneuraminic acid (NeuSAc), N-glycolylneuraminic acid (Neu5Gc) and N acetyl-9-O-acetylneuraminic acid (Neu5,9Acz) are the three most frequently occurring Sia. Only Neu5Ac is ubiquitous, while the others are not found in all species. The best investigated example after Neu5Ac is NeuSGc, which occurs often in the animal kingdom, but not in healthy human tissues (only in some tumours in small amounts) and not in bacteria. Also Kdn [8] is increasingly detected in microorganisms and animals. Never have all kinds of Sia been found in one cell or organism. The distribution depends on the animal and cell species as well as on the function of a cell and seems to be strongly regulated on the gene level. The source with the highest amount of different Sia is the cow. In its submandibular gland mucin 15 species were detected, most of them being 0-acetylated in the Sia side-chain, but also 9-0-lactylated as well as N-acetylated or N-glycolylated [ l , 91. In man the number of Sia is much smaller, Neu5Ac prevailing and followed by Neu5,9Acl, Neu5,7Ac~,Neu5,7,9Aci, Neu5Ac9Lt, Neu2en5Ac, and Neu2,7anSAc [ 11. The latter two Sia occur only in traces. Remarkably, from echinoderms, e.g. A . rubens, many Sia types are known, some of them representing the well-known monosaccharides of higher animals, such as NeuSAc, Neu5Gc and 0-acetylated species, as well as dehydro derivatives thereof in traces. However, a large percentage of these Sia are additionally methylated at 0-8 and this kind of modification, as well as 0-sulfation in other echinoderms, seems to be unique to this group of animals [ 11. In contrast, the formation of Neu5Gc and 0-acetylated Sia has been conserved in animals evolutionarily higher than the echinoderms. This means that the corresponding enzymes leading to these

13.3 General Biological Functions

229

Sia types have first been expressed at least 500 million years ago, when echinoderms evolved. NeuSAc also may originate from this time, since Sia are not regularly found in animals lower than echinoderms [1-4, lo]. Remarkably, Neu5Ac in polymeric form also occurs during a short period in Drosophilu rnelunogaster embryogenesis [ 111 and in larvae of the cicada Plziluerzus spurnurius [64]. The sporadic occurrence of Sia in microorganisms gives rise to the question whether these cells have started to produce Sia independently from animals or are even precursors. However, microorganisms have developed different strategies to acquire Sia. While some synthesize Neu5Ac in a way similar to animals [6], it was reported that Escherichiu coli K1 can achieve this with the aid of sialate lyase from pyruvate and N-acetylmannosamine [ 121. In contrast, Neisseriu rncningitidis acquires Sia from the host’s CMP-Neu5Ac but produces the sialyltransferase required itself [ 131. It was shown that both African and American trypanosomes capture glycosidically bound Sia from the host by the activity of trans-sialidases [ 141. It is still under debate whether these microorganisms have acquired the enzymes of sialic acid metabolism from host cells by a natural gene transfer, as evidence is accumulating for sialidases, the primary structures of which show similarities between animals and (often pathogenic) microorganisms [15, 161. The biosynthesis of Kdn was reported in cultured mammalian cells to proceed similarly to NeuSAc, from phosphoenolpyruvic acid and mannose [65]. Although free neuraminic acid cannot exist, this sugar has been found in glycosidic linkage [ 1-41 and seems to originate from Neu5Ac by enzymatic de-N-acetylation. It has been found to cyclicize in sialyl-6sulfo Lewis’, which cannot bind to selectins, but may be reopened and give an active ligand [17]. The enzymes involved in the metabolism of the diverse Sia, characterized or still putative, are summarized in Figure 1.

13.3 General Biological Functions One of the answers to the question why so many different Sia species exist may be the observation that they are mainly located on the outer cell membrane, which interacts with the components of other cell surfaces, extracellular substances and effector molecules. At this site, Sia much contribute to the enormous chemical diversity of glycoproteins and glycolipids. Their composition seems to be cell-specific and represents the “make-up” of a cell. The diversity may even increase by further modification of the Sia molecules and add to the biological variety of sialoglycoconjugates. In this way a complicated “writing” or “language” may be formed for cellular communication and Sia modification may sensitively tune the behavior of cells. As the glycan structures are often branched, in contrast to amino acid and nucleotide sequences in proteins and nucleic acids, respectively, these carbohydratebased “words” have been compared to the complex Chinese characters [ 181. The biological effects caused by the different substituents on the neuraminic acid molecule can best be understood in the light of the basic functions of Sia, as described in detail in [ 1-3, 16, 19-20]. These are due to their bulky structures and the

230

13 Biochemistry of Siulic Acid Diversity SialateS-0-methyl-transferase

Sialate-9-0-lactyl-transferase ( 7 )

Sialate-7(97)-0-acelyl-transferase (Golgi, microorganisms)

Sialate-9-0-acetylesterase (lysosomes, cytosol, microorganisms)

Sialate-7,8,9-O-acetyl-migrase(7)

HOHS

S

H3C 1 NI

t

11

H

OH

3

0

(cytosol, microsornes)

Sialate4-0-acetyl-esterase (lysosomes (?), virus)

Figure 1. Enzymes modifying the structure of NeuSAc and of its 0-acetylated derivatives. The wellcharacterized enzymes are shown in bold print, whereas the existence of those with a question mark is likely but has not yet unequivocally been demonstrated. For more details and references see the text.

relatively strong negative charge, which have repulsive effects on molecules in the cell membrane and on whole cells, and are involved in the binding and transport of cationic compounds. Furthermore, Sia influence the structure of macromolecules, protect them from enzymatic, mainly proteolytic attack and are involved in the specificity of antigens. However, Sia seem to play their most important role as a mask to prevent biological recognition, or, just the opposite, to themselves represent recognition sites [ l , 19, 201. With regard to the masking effect, one should distinguish between a more general masking of biological recognition sites such as hormone receptors or antigens, in which Sia may cooperate with other members of the glycan chain, and a more specific protection of subterminal galactose or N-acetylgalactosamine residues. Enzymatic removal of Sia in the first case may induce better access of molecules to cell surfaces or increase antigenicity, while in the second case the exposed galactose can lead to the interaction with galactose-recognizing receptors, followed by binding of molecules and cells [l-3, 10, 19, 201. For the second aspect, the discovery of the binding of glycoproteins after desialylation to a galactose-recognizing receptor of hepatocytes [21] as well as the attachment and in some cases phagocytosis of erythrocytes, lymphocytes and thrombocytes to a similar receptor of macrophages after sialidase treatment [ 1, 19, 20, 22, 231 was pacemaking. It was delineated from these obse;vations that the adhesion of molecules and cells via galactose may be regulated by sialylation of these residues [ l , 19, 201. Such a reversible interaction was

13.4 N-Glycolylneuraminic Acid

231

observed with lymphocytes, which bind to macrophages after sialidase treatment, but disintegrate from these cells after a few hours, probably due to resialylation [23]. Such a reflexible system may especially be required during embryogenesis and other growth processes including the spreading of tumour cells. Modifications of Sia can interfere with this mode of cell interaction, because 0-acetylation or N-acetyl hydroxylation of Sia hinder the action of sialidases and correspondingly the demasking of subterminal galactose residues required for the interaction with a galactose receptor. This, for example, leads to a longer life-time of rat erythrocytes [24] and possibly also of serum glycoproteins. The number of examples for sialic acids as recognition sites, ligands of receptor proteins or “counter-receptors” is rapidly growing and summarized in 11-3, 10, 19, 201. While it has been known for long that some viruses, bacteria and proteins from lower animals bind to sialic acids, such receptors, sialoadhesins, were detected in vertebrates only recently, and termed “siglecs” [19, 25, 261. A fifth member of these transmembrane molecules belonging to the immunoglobulin superfamily has been discovered on myeloid cells and is related to the CD33 antigen 1261. The properties as well as their investigation on the gene level and the amino acids involved in Sia binding of the sialoadhesins have been reviewed 1191. In this counter-receptor function of Sia, modulation of their structure is of great influence. 0-Acetylation and N-acetyl hydroxylation of NeuSAc not only renders such modified ligands less vulnerable towards degrading enzymes, but they can create new specificities of interaction, and can impair or abolish the binding strength of NeuSAc to its receptors. Examples will be given below, together with observations that such Sia modifications modulate the binding of viruses.

13.4 N-Glycolylneuraminic Acid N-Glycolylneuraminic acid was found in echinoderms up to primates. The animals with the highest proportion of NeuSGc in the Sia fraction and investigated in detail are the starfish A. ruhens and the pig. Regarding its biosynthesis, after the formation of Neu5Ac from glucose in a multi-step reaction in the cytosol [6], this monosaccharide is linked to CMP in the cell nucleus, followed by hydroxylation to CMPNeuSGc in the cytosol. Together with CMP-NeuSAc, this nucleotide is transported into the Golgi compartment and transferred by sialyltransferases, which in most cases do not much discriminate between NeuSAc and NeuSGc, onto growing glycoprotein and ganglioside molecules [ I , 41. The mammalian CMP-NeuSAc hydroxylase (EC 1.14.99.18) (Figure 1) is a soluble enzyme which requires several cofactors for activity. From NAD(P)H the electrons are transferred to cytochrome b5 by NAD(P)H cytochrome bs reductase. Reduced cytochrome b5 is a cofactor of the specific hydroxylase, the action of which results in CMP-NeuSGc with the aid of molecular oxygen. The hydroxylase has been isolated from mouse liver and pig submandibular gland, as summarized in [ 1 , 31. Both enzymes are monomers with a molecular weight of about 65 kDa. They reveal a high substrate affinity (pmolar

232

13 Biochemistry of Siulic Acid Diversity

range) and pronounced substrate specificity. The hydroxylase has binding sites for cytochrome b5 and for CMP-NeuSAc. Another stretch of amino acids is typical for Rieske iron-sulfur proteins. On electron spin resonance studies, the enzyme from pig exhibited a signal characteristic for such a 2Fe-2S centre. These conserved motifs have been identified in the primary structures of the pig and mouse enzymes, obtained by gene cloning and exhibiting 93% homology. Although the CMPNeuSAc monooxygenase seems to be soluble in the cytosol of the mammals studied, it is believed that cytochrome b5 and the corresponding reductase are bound to the membrane of the endoplasmic reticulum. In the starfish A . ruhens, however, the components of this hydroxylation system are more tightly associated, since the CMP-NeuSAc hydroxylase was found to be firmly bound to subcellular membranes [27].Although this is a marked difference to the mammalian hydroxylase, the kinetic and other properties of the starfish enzyme including its dependency on cytochrome bs, even from mammalian sources, are similar. This may be interpreted that the CMP-NeuSAc hydroxylase has strongly been conserved during the long evolution time from echinoderms to mammals. Although the great apes, like chimpanzees, bonobos, gorillas and orangutans, produce NeuSGc in significant amounts, in healthy human tissues only traces of this sialic acid can be identified by sensitive techniques [1-3, 281. It is assumed that these minute quantities (below 1% of the Sia fraction) are derived from food stuffs, as was observed with mice and rats when feeding NeuSGc-containing substances [29]. This means that the biosynthesis of NeuSGc has been suppressed sometime in evolution after the divergence of the great apes from hominids. It was found that cDNA coding for the human hydroxylase lacks a 92 bp region, which results in the expression of a truncated, inactive protein lacking the N-terminus with the Rieske centre in human tissues [30, 311. The sialylation of glycoconjugates with NeuSGc seems to depend on the amount of hydroxylase protein expressed [32]. A positive correlation between the activity of the hydroxylase and the molar ratio of NeuSAc and NeuSGc was found in all tissues investigated. Also the hydroxylase activity corresponded well to the amount of enzyme protein present, as estimated with various anti-hydroxylase antibodies. Exceptions, however, were heart and lung, in which disproportionately large amounts of immunoreactive protein were found. An explanation for this phenomenon is not yet available, although the presence of inhibitors or phosphorylation of the enzyme may be responsible. Since Sia in glycoconjugates in general play many important roles in the functioning of an organism, it is reasonable to assume that the formation and enzymatic modification of NeuSAc characteristic for particular cell types in mature and developing tissues is under strict control by intra- and extracellular factors [I]. It is conceivable that in vivo besides the gene level several factors can influence this multicomponent system of NeuSAc hydroxylation. It took a long time until we started to recognize the manyfold biological roles of NeuSGc. These are summarized in Figure 2 and [ 1-3, 10, 19, 201, and here only the latest observations will be discussed. Interestingly, some of the effects are similar to the O-acetylation of Sia, discussed below, like the negative influence on sialoglycoconjugate degradation, which may extend the life-time of sialoglycoconjugates. Similar to O-acetylated Sia, NeuSGc is a strong modulator of receptor biology, which may be of great significance in cell differentiation, embryogenesis, immunol-

13.4 N-Glycolylnrururninic Acid

233

Decreasing hydrophilicity Reduction of sialidase. trans-sialidase and sialate-pyruvate lyase activities Hindrance of sialic acid receptor interactions (viruses, mammals) Epitope for virus binding Epitope for invertebrate lectin binding Modulation of colonization and virulence of microorganisms Role in morphogenesis Tumor-associated antigen Reduction of sialidase action Influence on complement reactivity Influence on glycosylation. e g polysialylation, Modulation of immune reactions or branching

-

,coo-

OCH,

ACOH,C OH HOHG 1 C

I

oAc

Increasing hydrophilicity Reduction of sialidase and sialate-pyruvate lyase actions Binding of viruses, bacteria, plant, invertebrate and mammalian lectins Hindering recognition by mammalian receptors Modulation of immune response Blood group determinant Tumor-associated antigen Differentiation marker Involvement in ester formation and glycosylation Specific role in echinoderms

3

Decreasing hydrophilicity Providing resistance towards the action of sialate-pyruvate lyases and most sialidases Counter-receptor for mouse hepatitis virus Inhibitor of influenza viruses

Figure 2. Survey of the biological significance of the sialic acid substituents.

ogy and infection biology. These topics have been discussed in a very stimulative manner [28] considering the differences in NeuSGc expression between man, great apes and other animals and the interaction of these different Sia with siglecs. While CD22 on B-lymphocytes, which plays a critical role in humoral immune response, in mouse strongly prefers NeuSGc, sialoadhesin of the same animal binds best to NeuSAc [19, 281. As expected, in man these two siglecs, as well as the myelinassociated glycoprotein (MAG) prefer NeuSAc. It is a challenging task to study whether the preferred linkage of human siglecs to NeuSAc is of evolutionary importance, especially in brain development and function [28]. Also the species specificity of microbial infections should be investigated in more detail with regard to the Sia species. The infectiosity of influenza A and B viruses is influenced by the presence of NeuSAc or NeuSGc, humans are not infected by E. coli K99 or pig gastroenteritis virus, both recognizing NeuSGc, and in case of the uptake of the malaria parasite Plusmodium ,falciparum into erythrocytes from primates and man, respectively, a difference due to NeuSAc or NeuSGc is assumed [28]. The origin of glycoconjugate-bound NeuSGc in several human tumours, shown in various laboratories by conventional analytical methods, modern mass spectrometry, and antibodies (summarized in [1-4, 281) needs elucidation. Neither the activity of CMPNeuSAc hydroxylase in tumour tissues nor the re-expression of a corresponding gene could be demonstrated. One possible explanation is the nutritional origin of NeuSGc, possibly by preferred uptake of Neu5Gc or the degradation product N glycolylmannosamine released from foodstuffs in the intestine (see above) by tumour

234

13 Biochemistry of Siulic Acid Diversity

cells. A further role of N-acetyl hydroxylation is to supply the molecule with another hydroxyl group required for esterification yielding N-(0-methyl)glycolylneuraminic acid in starfish, and for glycosylation resulting in Neu5Gc(a2-0,1,,,1,1)Neu5Gc- or Ne~SGc8Me(a2-O~l~~~~~l)Neu5Gc8Meoligosaccharide structures in sea urchin and starfish [ 1, 101.

13.5 0-Acetylated Sialic Acids Two enzymes are known for Sia esterification, an acetyl-coenzyme A:sialate-4-0acetyltransferase (EC 2.3.1.44) responsible for esterification at the pyranose ring, and a corresponding 7(9)-0-acetyltransferase (EC 2.3.1.45), modifying the glycerol side chain. The first enzyme seems to be relatively rare, since 4-0-acetylation was found only in horse, donkey, guinea pig, echidna (Tuchyglossus uculeatus), the Japanese dace “ugui” (Triholodon hukonensis) and a South American pit viper, as summarized in 11-4, 331. The second enzyme is believed to be more frequent, since 0-acetylation at the side-chain has been found in many animal species, from echinoderms to man, and also in some bacterial glycoconjugates [ 1-4, 20, 341. It can be predicted that more sources of these esterified, labile Sia will be detected, since the analytical techniques are improving [35]. Loss of the ester groups may occur by the action of Sia-specific 4- and 9-0-acetylesterases frequently present in tissues [ 1-3, 361, during the purification process of glycoproteins and gangliosides, during the release of Sia from their glycosidic linkages by mild acids or sialidase, and during the purification procedure of the liberated Sia before analysis. These difficulties may be overcome at least for 9-0-acetylated Sia by the direct and very sensitive detection of such Sia in tissues or isolated glycoconjugates, which is possible with the aid of intact influenza C viruses 135, 371 or of chimeras composed of the hemagglutinin and esterase portion of the surface protein from these viruses and the Fc portion of human IgG [ 381. The viruses or chimeras bound to 0-acetylated sialoglycoconjugates are visualized by their esterase activity or by antibodies. In thin layer chromatography overlay and microtiter plate binding assays, pmol amounts of glycoprotein- or ganglioside-bound Neu5,9Ac2 can be detected [66]. While hydroxylation occurs in the cytosol at the CMP-NeuSAc level before the transport and transfer, as described above, both 4- and side-chain-0-acetylation (Figure 1) seem to take place predominantly at glycosidically linked Sia in the Golgi apparatus, as was studied in rat liver 11-3, 391, bovine submandibular gland 1341, and guinea-pig liver 1331. The 0-acetyltransferases are membrane-bound and colocalized with sialyltransferases in the tissues investigated. The 4-0-acetyltransferase can be solubilized with detergents and is active with NeuSAc either free or bound to oligosaccharides and glycoconjugates [ 671. With regard to side-chain 0-acetylation, it is the present view that only one enzyme is responsible for this modification. There is experimental evidencc that the primary insertion site of 0-acetyl groups from AcCoA is at C-7 [34]. From this position the ester group migrates to C-8 and C-9. The whole glycerol side-chain may become 0-acetylated by a second acetyl

13.5 O-Acetyluted Sicllic Acids

235

transfer to 0-7, migration to 0-8, and finally by the incorporation of a third acetyl to 0-7. A relatively high amount of Sia with partly or fully esterified side-chains exist in mucins from bovine submandibular gland and human colon [ 1-4, 341. This isomerization appears to be an enzymatic process, catalyzed by a hypothetical “sialate O-acetyl migrase or isomerase”, as studied in microsomes from bovine submandibular gland. It is about 160 times faster than the mere physical process under the same conditions of e.g. pH and temperature, and it can be inhibited by protein denaturation [34]. The Golgi-located O-acetylation seems to require the cooperation by an acetylCoA transporter, according to experiments with rat liver [40] and bovine submandibular gland [34], and by expression cloning in COS-1 cells [41]. However, at least some O-acetylation before Sia transfer to glycoconjugates cannot be excluded, since CMP-Neu5,9Acz was isolated from bovine submandibular glands, and the highest specific radioactivity was found in the fraction of free Sia from this tissue in metabolic labelling studies [l]. These two possible pathways of Sia O-acetylation are shown in Figure 3. The 4-O-acetylation renders the corresponding sialoglycoconjugates completely resistant towards the action of almost all sialidases and of sialate-lyases. Only influenza virus sialidases, which in contrast to other sialidases possess a pocket in their active centre into which the 4-O-acetyl group fits, are able to slowly hydrolyze the glycosidic bond of N-acetyl-4-O-acetylneuraminic acid (Neu4,5Acz). The structural feature of virus sialidase was a basis for the development of potent anti-influenza drugs like 4-amino- or 4-guanidino Neu2en5Ac [42]. It is assumed that Sia 4-0acetylation protects glycoconjugates from the attack of catabolic enzymes and thus extends the lifetime of molecules and cells. In virology, such Sia gain increasing importance. First evidence is available that mouse hepatitis virus binds to 4-0acetylated sialoglycoconjugates [43]. Most remarkably, it possesses sialate 4-0acetylesterase to destroy its receptor (Figure 4); this enzyme has so far been only found in horse liver [ l , 361. It represents a third receptor-destroying enzyme of viruses, after sialidase and sialate-9-O-acetylesterase on influenza and other viruses [ 1-31. Neu4,5Acz-containing glycoconjugates, e.g. glycoproteins from horse and guinea-pig serum, inhibit the binding of influenza viruses, because Neu4,5Ac2 is recognized by their hemagglutinin, but is (almost) resistant to their sialidase action. However, these viruses can become resistant to such inhibitors, because their affinity to them decreased by a mutation in their hemagglutinin [44]. It was shown during an equine influenza outbreak caused by an avian virus that an isolate of the avian virus lost its capability to bind Neu4,5Ac2 and thus behaved like classical equine influenza strains not recognizing Neu4,5A~-containingglycoproteins. This means an increase of virulence in animals expressing such Sia. The studies also show that the kind of Sia plays a role in viral evolution during inter-species transmission. Since it is difficult to isolate glycoconjugates with only Neu5,7Ac2, due to the isomerization tendency of this Sia, the specific role of this modification is not easy to establish. However, the ganglioside GD3 containing only Neu5,7Ac2 was isolated from human B lymphocytes, which has other immunological properties than the corresponding Neu5,9Ac2-containing isomer [45]. 9-O-Acetylation seems to enhance the signal transduction mediated by GD3 [46]. Furthermore, Neu5,7Ac2

AcCoA

A

B

Figure 3. Two models for the 0-acetylation of Sia in Golgi-membranes of bovine submandibular glands. A, as CMPglycoside, before the transfer onto nascent glycoconjugates (mucin) and B, after this transfer. OAT, sialate-7(9)-0acetyltransferase; ST, sialyltransferase; M, 0-acetylisomerase. For further details see the text.

CMPNeu5Ac

13.5 O-Acetyluted Sialir Acid.s

231

Neu4,5Ac,

8 AcO

Sialate-4-O-acetylestetaase Mouse Hepatitis Virus S

0

5

10 15 20 time of elution [minl

25

0

5

10 15 20 time of elution [min]

25

Figure 4. Viral sialate-O-acetylesterases as “Receptor-Destroying Enzymes”. The chromatograms (HPLC, fluorescent Sia derivatives [35])show the conversion of Neu4,5Ac2 and Neu5,9Ac2 into NeuSAc under the influence of mouse hepatitis and influenza C viruses, respectively.

may serve as a reservoir for 8- and 9-O-acetylated Sia. Whether Neu5,8Ac2 plays a biological role could not yet be established because it is so labile [ 11. Most functions of Sia side-chain O-acetylation are therefore ascribed to the more stabile isomer Neu5,9Ac2. These are summarized in the references cited and in Figure 2. It is conceivable that a higher grade of O-acetylation makes a glycoprotein or ganglioside molecule more hydrophobic in its carbohydrate part, which may be important for the architecture of cell membranes or for cell communication. The retardation of sialic acid degradation leads to a longer life-time of glycoconjugates and cells with side-chain O-acetylated Sia [24]. Correspondingly, Sia O-acetyl groups are believed to protect colonic much from the action of bacterial sialidases and other degrading enzymes [47]. This has a great (patho-)physiological impact, since the mucus barrier is responsible for lubrication, control of microbial colonization and protection of intestinal epithelia against enzymes, toxins, as well as osmotical and mechanical injuries. There is evidence that Sia O-acetylation is a potent regulator of molecular, viral and cellular adhesion. The siglecs, e.g. MAG, macrophage sialoadhesin, and CD22 are all hindered to bind to Sia if these are 9-O-acetylated [19, 251. It is not yet known which consequences this observation, made by in vitro experiments, has in a living organism. Since lymphocytes contain Neu5,9Ac2, this may influence the immune response by modulation of the interaction between B- and T-lymphocytes via the CD22 receptor. The core epitope of CDwGO, a marker on activated Blymphocytes, consists of a 9-O-acetylated disialosyl sequence bound to the ganglioside GD3, the disialosylparagloboside DSPG or possibly also to surface glycoproteins, and seems also to be involved in adhesion contacts of these cells [48]. A

238

13 Biochemistry of Siulic Acid Diversity

further anti-adhesion effect by sialic acid O-acetylation was observed on binding of influenza A and B viruses to sialylated glycans, which recognize only unsubstituted Sia (summarized in [ l , 2, 191). This is also known for malaria parasites and mouse erythrocytes [49]. In contrast, influenza C viruses, as well as corona- and rotaviruses, recognize only Neu5,9Ac2, which is believed to play a crucial role in the infection mechanism by influenza C virus in the human respiratory tract. The occurrence of this sialic acid e.g. in human nasal mucin was demonstrated [ 11. The binding sites of both the receptor and the sialate-9-O-acetylesterase were studied in crystals of the HEF envelope glycoprotein of the influenza C virus [50]. Although the receptorbinding domain is structurally similar to the Sia-binding domain of influenza A virus hemagglutinin, it only binds Neu5,9Ac2. The esterase domain is similar to esterases from bovine brain and Streptomyces scabies. The expression of O-acetylated Sia also varies during differentiation and tumorigenesis [1-4, 101. For example, the ganglioside GD3 with a terminal Neu5,9Ac2 residue was found to stimulate neurite growth and branching (511 and the developmentally regulated expression of O-acetylated sialoglycans in fetal murine cerebral cortex was studied with the aid of a monoclonal antibody [52].An increased content of this ganglioside was found in the human malignant neuroectodennal tissues melanoma [53], basalioma [54, 681, and brain tumor [55]. In contrast, the degree of O-acetylation decreases in the much from human colon tumor concommitantly with the grade of malignancy and metastatic potential (561. Interestingly, more Sia Le" ligands were expressed by this loss of Sia O-acetylation, which may enhance the metastatic potential of colon cancer [57]. Downregulation of 9-O-acetyl GD3 (CDw60) was reported in poorly differentiated, invasive human breast carcinomas [ 581, thus rendering this ganglioside a prognostic marker.

13.6 O-Methylated and O-Sulfated Sialic Acids Sialic acids, mostly NeuSGc, methylated at 0-8, have been found only in gangliosides and glycoproteins of a few echinoderms, the best studied example being the starfish A . rubens [I-3, 27, 591. From subcellular membranes of gonads of (proposed EC this animal S-adenosyl-~-methionine:sialate-8-O-methyltransferase 1.14.99.18) (Figure 1) was solubilized and purified 22,000-fold by affinity chromatography on adenosine-homocysteine-Sepharose [ 591. The molecular mass of the Mn2+-dependent enzyme is about 60 kDa. Remarkably, it has a temperature optimum of 37"C, although this starfish species lives in a relatively cold environment. Both free and glycosidically linked Neu5Ac and Neu5Gc are methylated, bound Sia exhibiting higher affinity. With regard to the significance of these methyl ether groups, only a hindrance of sialidase action on these Sia from glycoconjugates has been observed [59]. It is speculated that this modification may influence the biosynthesis of unusual structures found in these animals, i. e. branched glycan chains of gangliosides and oligosaccharides in which Neu5Gc or Neu5Gc8Me residues are glycosidically linked to

13.6 0-Mrthyluted and 0-Suljiited Sialic Acids

239

each other via the N-glycolyl residue, as reviewed [ 1, 3, 591 and mentioned above in the chapter on NeuSGc. This type of oligomerization of Sia detected in various starfish and sea urchins appears to be an alternative to the formation of the wellknown a2,8-linkages, which are prevented by 8-0-methylation. However, also this method of elongation may be regulated by methylation of the N-glycolyl hydroxyl, leading to N-( 0-methy1)glycolylneuraminic acid, a stop signal found in glycolipids of the starfish A . arnurensis [60]. Sialic acid 8-0-sulfation was also only identified in echinoderms, e.g. in sea urchins [l-41. Nothing is known about its enzymatic origin and a specific 8-0-sulfotransferase is waiting to be discovered. Two functions of this sialic acid variation have been described: Firstly, the incorporation of N-acetyl- or N-glycolyl-8-0sulfoneuraminic acid (NeuSAc8S, NeuSGc8S) into growing oligosialyl chains of glycoconjugates in sea urchin spermatocytes or egg cells seems to terminate further elongation of the glycan moieties [61]. Secondly, this modification facilitates the formation of lactones between Sia residues [62]. Furthermore, the anionic character of Sia is much increased and thus indicates a functional role. The sulfation at C-9 of the non-reducing NeuSGc in the poly-Sia chains [ N ~ U ~ G ~ ~ S ( ~ ~ - ~ , ~ ~ , , I ~ I ) N ~ U on ova of the sea urchin Hemicentrotus pulcherrinius has been found to be involved in binding of spermatozoa [63]. This finding adds a new Sia (NeuSGc9S) to the long list of these monosaccharides. It may also prevent further growth of the oligosaccharide chain and protect it against degradation, since Sia-0-sulfation was reported to hinder the action of sialidases [63]. Thus, already echinoderms exhibit an astonishing variety of Sia, more so when 90-acetyl groups, found in a number of such species [l-4, 591, are included. One may hypothesize that the “invention” of Sia in general, attributed to this group of animals, and of the different forms of this monosaccharide, has enhanced the diversification of the structures of glycoconjugates and in this way supported the development of higher and more specialized animals. Since methylated and sulfated Sia have never unequivocally been found in vertebrates, they possibly were not required for further evolution.

Acknowledgments Thanks are due to Matthias Iwersen, Guido Kohla, and Yanina N. Malykh for valuable suggestions and discussions.

References I . Schauer, R., and Kamerling, J.P. Chemistry, biochemistry and biology of sialic acids. In: Glycoproteins ZZ (Montreuil, J., Vliegenthart, J.F.G., and Schachter, H., eds.) (1997) Elsevier, Amsterdam, pp. 243-402. 2. Varki, A. Diversity in the sialic acids. Glycobiology (1992) 2, 25-40. 3. Schauer, R., Kelm, S., Reuter, G., Roggentin, P., and Shaw, L. Biochemistry and role of sialic acids. In: Biology of’Siulic Acids (Rosenberg, A., ed.) (1995) Plenum Press, New York, pp. 667.

240

13 Biochemistry of Sialic Acid Diversity

4. Corfield, A.P., and Schauer, R. Occurrence of sialic acids. In: Sialic Acids-Chemistry, Metuholism and Function (Schauer, R., ed.) (1982) Springer Wien, New York, pp. 5-50. 5. Gil-Serrano, A.M., Rodriguez-Carvajal, M.A., Tejero-Mateo, P., Espartero, J.L., ThomasOates, J., Ruiz-Sainz, J.E., and Buendia-Claveria, A.M. Structural determination of a 5 - 0 methyl-deaminated neuraminic acid (Kdn)-containing polysaccharide isolated from Sinorhizohium,fredii. Biochem. J. (1998) 334, 585-594. 6. Corfield, A.P., and Schauer, R. Metabolism of sialic acids. In: Sialic Acids-Chemistry, Mefaholism and Function (Schauer, R., ed.) (1982) Springer Wien, New York, pp. 195-261. 7. Luo, Y., Li, S.-C., Chou, M.-Y., Li, Y.-T., and Luo, M. The crystal structure of an intramolecular trans-sialidase with a NeuAcu2i3Gal specificity. Structure (1998) 6, 521-530. 8. Nadano, D., Iwasaki, M., Endo, S., Kitajima, K., Inoue, S., and Inoue, Y. A naturally occuracid (KDN). J. ring deaminated neuraminic acid, 3-deoxy-D-glycero-D-galactononulosonic Biol. Chem. (1986) 261, 11550-11557. 9. Klein, A,, Diaz, S.: Ferreira, I., Lamblin, G.: Roussel, P., and Manzi, A.E. (1997) New sialic acids from biological sources identified by a comprehensive and sensitive approach:liquid chromatography-electrospray ionization-mass spectrometry (LC-ESI-MS) of SILA quinoxalinones. Glycohiology (1997) 7, 421-432. 10. Schauer, R. Origin and biological role of the great chemical diversity of natural sialic acids. Trend~sGlycosci. Glycotechnol. (1997) 9, 3 15-330. 11. Roth, J., Kemp, A,, Reuter, G., Schauer, R., and Gehring, W.J. Occurrence of sialic acids in Drosophila melanoguster. Science (1992) 256, 673-675. 12. Ferrero, M.A., Reglero, A., Fernandez-Lopez, M., Ordas, R., and Rodriguez-Aparicio, L.H. N-Acetyl-D-neuraminic acid lyase generates the sialic acid for colominic acid biosynthesis in Esclierichia coli K1. Biochem. J. (1996) 317, 157-165. 13. Mandrell, R.E., Smith, H., Jarvis, G.A., Griffis, J.M., and Cole, J.A. Detection and some properties of the sialyltransferase implicated in the sialylation of lipopolysaccharide of Neixseriu gonourhoeae. Microbial Pathogen (1993) 14, 307-313. 14. Engstler, M., Schauer, R., and Brun, R. Distribution and developmentally regulated transsialidases in the Kinetoplastida and characterization of a shed trans-sialidase activity from procyclic Trypanosoma congolense. Acfu Tvopicu (1995) 59, 1 17-129. 15. Roggentin, P., Schauer, R., Hoyer, L.L., and Vimr, E.R. The sialidase superfamily and its spread by horizontal gene transfer. Mol. Microhiol. (1993) 9, 915-921. 16. Traving, C., and Schauer, R. Structure, function and metabolism of sialic acids. Cell. Mol. Lije Sci. (1998) 54, 1330-1349. 17. Mitsuoka, C., Ohmori, K., Kimura, N., Kanamori, A,, Komba, S., Ishida, H., Kiso, M., and Kannagi, R. Regulation of selectin binding activity by cyclization of sialic acid moiety of carbohydrate ligands on human leukocytes. Proc. Nut1 Acad. Sci. USA (1999) 96, 15971602. 18. Kasei, K . 4 , and Hirabayashi, J. Galectins: a family of animal lectins that decipher glycocodes. J. Biochem. (1996) 119, 1-8. 19. Kelm, S., and Schauer, R. Sialic acids in molecular and cellular interactions. lnf. Rev. Cytology (1997) 175, 137-240. 20. Schauer, R. Sialic acids and their roles as biological masks. Trends Biocheni. Sci. (1985) 10, 357-360. 21. Harford, J., Klausner, R.D., and Ashwell, G. Inhibition of the endocytic pathway for asialoglycoprotein catabolism. Biol. Cell. (1984) 51, 173-179. 22. Jancik, J.M., Schauer, R., Andres, K.H., and von During, M. Sequestration of neuraminidasetreated erythrocytes. Studies on its topographic, morphologic and immunologic aspects. Cell Tissue Res. (1978) 186, 209-226. 23. Fischer, C., Kelm, S.; Ruch, B., and Schauer, R. Reversible binding of sialidase-treated rat lymphocytes by homologous peritoneal macrophages. Curbohydr. Res. (1991) 213, 263-273. 24. Kiehne, K., and Schauer, R. The influence of a- and P-galactose residues and sialic acid 0acetyl groups of rat erythrocytes on the interaction with peritoneal macrophages. Biol. Cheni. Hoppe-Seyler (1992) 373, 1 1 17-1 123. 25. Kelm, S., Pelz, A., Schauer, R., Filbin, M.T., Tang, S., de Bellard, M.-E., Schnaar, R.L., Mahoney, J.A., Hartnell, A,, Bradfield, P., and Crocker, P.R. Sialoadhesin, myelin-associated

References

24 1

glycoprotein and CD22 define a new family of sialic acid-dependent adhesion molecules of the immunoglobulin superfamily. Curr. Bid. (1994) 4, 965-972. 26. Cornish, A.L., Freeman, S., Forbes, G., Ni, J., Zhang, M., Cepeda, M., Gentz, R., Augustus, M., Carter, K.C., and Crocker. P.R. Characterization of siglec-5, a novel glycoprotein expressed on myeloid cells related to CD33. Blood(1998) Y2, 2123-2132. 27. Gollub, M., Schauer, R., and Shaw, L. Cytidine monophosphate-N-acetylneuraminate hydroxylase in the starfish Asterius ruhens and other echinoderms. Comp. Biorhem. Physiol. Purr B (1998) 120, 605-615. 28. Muchmore, E.A., Diaz, S., and Varki, A. A structural difference between the cell surfaces of humans and the great apes. Am. J. Pl7j:r.. Anthropol. (1998) 107, 187--198. 29. Nohle, U., and Schauer, R. Metabolism of sialic acids from exogeneously administered sialyllactose and mucin in mouse and rat. Hoppe-Seyler's Z. Physiol. Chem. (1984) 365. 14571467. 30. Irie. A., Koyama, S., Kozutsumi, Y., Kawasaki, T., and Suzuki, A. The molecular basis for the absence of N-glycolylneuraminic acid in humans. J. Bid. C/iem. (1998) 273, 15866-15871. 31. Chou. H.-H., Takematsu, H., Diaz, S., Iber, J.: Nickerson, E., Wright, K.L., Muchmore, E.A., Nelson, D.L., Warren, S.T.; and Varki, A. A mutation in human CMP-sialic acid hydroxylase occurred after the Homo-Purr divergence. Proc. Nut/ Acud. Sci. USA (1998) 95, 11751-11756. 32. Malykh. Y.N., Shaw, L., and Schauer, R. The role of CMP-N-acetyheuraminic acid hydroxylase in determining the level of N-glycolylneuraminic acid in porcine tissues. G/yeoconj. J. (1998) 15, 885 -893. 33. Iwersen, M., Vandamme-Feldhaus, V., and Schauer, R. Enzymatic 4-0-acetylation of Nacetylneuraminic acid in guinea-pig liver. G1ycoconj. J , (1998) 15, 895-904. 34. Vandamme-Feldhaus, V.. and Schauer, K. Characterization of the enzymatic 7-0-acetylation of sialic acids and evidence for enzymatic 0-acetyl migration from C-7 to C-9 in bovine submandibular glands. J. Biochem. (1998) 124, 11 1-121. 35. Reuter, G., and Schauer, R. Determination of sialic acids. Me~hodsEnzymol. (1994) 230, 168199. 36. Schauer, R., Reuter, G., and Stoll, S. Sialate-0-acetylesterases--key enzymes in sialic acid catabolism. Biochimie (1988) 70, 151 1 -1519. 37. Harms, G., Reuter, G., Corfield, A.P., and Scliauer, R. Binding specificity of influenza C-virus to variably 0-acetylated glycoconjugates and its use for histochemical detection of N-acetyl-90-acetylneuraminic acid in mammalian tissues. Gl.vcoconj. J. (1996) 13, 621-630. 38. Klein, A,, Krishna, M., and Varki, A. 9-0-Acetylated sialic acids have widespread but selective expression: analysis using a chimeric dual-function probe derived from influenza C hemagglutinin-esterase. Proc. Ntrtl Acud Sci. U S A (1994) 91, 7782-7786. 39. Diaz, S., Higa, H.H.. Hayes, B.K., and Varki, A. 0-Acetylation and de-0-acetylation of sialic acids. 7- And 9-0-acetylation of a-2,6-linked sialic acids on endogenous N-linked glycans in rat liver Golgi vesicles. J. Biol. Chern. (1989) 264, 19416-19426. 40. Higa, H.H., Butor, C.. Diaz, S., and Varki,A. 0-Acetylation and de-0-acetylation of sialic acids. 0-Acetylation of sialic acids in rat liver Golgi apparatus involves an acetyl intermediate and essential histidine and lysine residues: A transmembrane reaction. J. Biol. Cherui. (1989) 264, 19427- 19434. 41. Kanamori, A,. Nakayama, J., Fukuda, M.N., Stallcup, W.B., Sasaki, K., Fukuda. M., and Hirabayashi, Y. Expression cloning and characterization of a cDNA encoding a novel membrane protein required for the formation of 0-acetylated ganglioside: A putative acetyl-CoA transporter. Proc. Nu11 Acud. Sri. USA (1997) 94, 2897-2902. 42. von Itzstein. M., and Kiefel, M.J. Sialic acid analogues as potential antimicrobial agents. In: Curbohydrutes 0 7 Drug Design (Witczak, Z.J.. and Nieforth, K.A.. eds.) (1997) Marcel Dekker, New York; Basel: Hong Kong, pp. 39-82. 43. Regl, G., Kaser, A,, Iwersen, M., Schmid, H., Kohla, G., Strobl, B., Vilas, U.; Schauer, R., and Vlasak, R. The hemagglutinin-esterase of mouse hepatitis virus strain i s a sialate-4-0acetylesterase. J. Virol. (1999), in press. 44. Matrosovich, M., Gao. P.. and Kawaoka, Y . Molecular mechanisms of serum resistance of human influenza H3N2 virus and their involvement in virus adaptation in a new host. J. Virol. (1998) 72, 6373-6380.

242

13 Biochemistry of Siulic Acid Diversity

45. Kniep, B., Claus, C., Peter-Katalinic, J., Monner, D.A., Dippold, W., and Nimtz, M. 7-0Acetyl-GD3 in human T-lymphocytes is detected by a specific T-cell-activating monoclonal antibody. J. Biol. Chem. (1995) 270, 30173--30180. 46. Reivinen, J., Holthofer, H., and Miettinen, A. Tyrosine phosphorylation of p72syk induced by anti-9-0-acetyl GD3 antibodies in human peripheral blood mononuclear cells. Scand. J. Immunol. (1998) 48, 615--622. 47. Roberton, A.M., and Corfield, A.P. Mu ch degradation and its significance in inflammatory conditions of the gastrointestinal tract. In: Medical Importance of' the Normal Microfloru (Tannock, G.W., ed.) (1998) Kluwer Academic Publishers, Dordrecht, Boston, London, pp. 222-261. 48. Vater, M., Kniep, B., GroB, H.-J., Claus, C., Dippold, W., and Schwartz-Albiez, R. The 9-0acetylated disialosyl carbohydrate sequence of CDw60 is a marker on activated human B lymphocytes. Immunol. Lett. (1997) 59, 151-157. 49. Klotz, F.W., Orlandi, P.A., Reuter, G., Cohen, S.J.. Haynes, J.D., Schauer, R., Howard, R.J., Palese, P., and Miller, L.H. Binding of Plasmodium falcipurum 175 kilodalton erythrocyte binding antigen and invasion of murine erythrocytes requires N-acetylneuraminic acid but not its 0-acetylated form. Mol. Biochem. Purusifol. (1992) 51, 49-54. 50. Rosenthal, P.B., Zang, X., Formanowski, F., Fitz, W., Wong, C.-H., Meier-Ewert, H., Skehel, J.J., and Wiley, D.C. Structure of the haemagglutinin-esterase-fusion glycoprotein of influenza C virus. Nature (1998) 396, 92-96. 51. Araujo, H., Menezes, M., and Mendez-Otero, R. Blockage of 9-0-acetyl gangliosides induces microtubule depolymerization in growth cones and neurites. Eur. J. Cell Biol. (1997) 72, 20221 3. 52. Zhang, G., Ji, L., Kurono, S., Fujita, S.C., Furuya, S., and Hirabayashi, Y. Developmentally regulated 0-acetylated sialoglycans in the central nervous system revealed by a new monoclonal antibody 493D4 recognizing a wide range of 0-acetylated glycoconjugates. Glycoconj. J. (1997) 14, 847-857. 53. Sjoberg, E.R., Manzi, A.E., Khoo, K.H., Dell; A,, and Varki, A. Structural and immunological characterization of 0-acetylated GD2. Evidence that GD2 is an acceptor for ganglioside 0-acetyltransferase in human melanoma cells. J. Biol. Chern. (1992) 267, 16200-1 621 1. 54. Paller, AS., Arnsmeier, S.L., Robinson, J.K., and Bremer, E.G. Alteration in keratinocyte ganglioside content in basal cell carcinomas. J. Invest. Dermatol. (1992) 98, 226-232. 55. Ariga, T., Blaine, G.M., Yoshino, H., Dawson, G., Kanda, T., Zeng, G.C., Kasama, T., Kushi, Y., and Yu, R.K. Glycosphingolipid composition of murine neuroblastoma cells: 0Acetylesterase gene downregulates the expression of 0-acetylated GD3. Biochemistry (1995) 34, 11 500-1 1507. 56. Corfield, A.P., Myerscough, N., Warren, B.F., Durdey, P., Paraskeva, C., and Schauer, R. Reduction of sialic acid 0-acetylation in human colonic mucins in the adenoma-carcinoma sequence. Glycoconj. J. (1999) 16, 307-317. 57. Mann, B., Klussmann, E., Vandamme-Feldhaus, V., Iwersen, M., Hanski, M.-L., Riecken, E.-O., Buhr, H.J., Schauer, R., and Hanski, C . Low 0-acetylation of sialyl-Le' contributes to its overexpression in colon carcinoma metastases. Int. J. Cancer (1997) 72, 258-264. 58. Gocht, A., Rutter, G., and Kniep. B. Changed expression of 9-0-acetyl GD3 (CDw60) in benign and atypical proliferative lesions and carcinomas of the human breast. Histochem. Cell Biol. (1998) 110, 217-229. 59. Kelm, A., Shaw, L., Schauer, R., and Reuter, G. The biosynthesis of 8-0-methylated sialic acids in the starfish Asterias ruhens: Isolation and characterisation of S-adenosyl-L-methionine: sialate-8-0-methyltransferase. Eur. J. Biochenz. (1998) 251, 874-884. 60. Higuchi, R., Inukai, K., Jhou, J.X., Honda, M., Komori, T., Tsuji, S., and Nagai, Y. Structure and biological activity of ganglioside molecular species. Liebigs Ann. Chem. (1993) 359366. 61. Kitazume, S., Kitajima, K., Inoue, S., Haslam, S.M., Morris, H.R., Dell, A,, Lennarz, W.J., and Inoue, Y. The occurrence of novel 9-0-sulfated N-glycolylneuraminic acid-capped ~ 2 1 5 O,~,,~,~-linkedoligo/polyNeuSGc chains in sea urchin egg cell surface glycoprotein. Identification of a new chain termination signal for polysialyltransferase. J. Biol. Chem. (1996) 271, 6694-6701.

References

243

62. Ijuin, T., Kitajima, K., Song, Y., Kitazume, S., Inoue, S., Haslam, S.M., Morris. H.R.; Dell, A,, and Inoue, Y. Isolation and identification of novel sulfated and nonsulfated oligosialyl glycosphingolipids from sea urchin sperm. Glycoconj. J. (1996) 13, 401-413. 63. Kitazume-Kawaguchi, S. Polysialylated glycoproteins found in sea urchin eggs. Trc,nd.s Glycosci. Glycotechnol. (1998) 10. 383- 392. 64. Malykh, Y.N., Krisch, B., Gerardy-Schahn, R.. Lapina, E.B., Shaw, L., and Schauer, R. The presence of N-acetylneuraminic acid in Malpighian tubules of larvae of the cicada Philaenus spurnuriiis. Glyeoconj. J. (2000) 16, in press. 65. Angata, T., Nakata, D., Matsuda, T., Kitajima. K., and Troy, F.A. 11. Biosynthesis of KDN (2-Keto-3-deoxy-D-glycero-D-galacto-nononic acid). J. Biol. Chern. (1999), 274. 22949-22956. 66. Hubl, U., Ishida, H., Kiso, M., Hasegawa. A,, and Schauer, R . Studies on the sensitivity and specificity of the influenza C virus binding assay of 9-0-acetylated sialic acids and its application to human melanoma. J. Biocheni., submitted. 67. Iwersen, M., and Schauer, R., unpublished. 68. Fahr, C., and Schauer, R. Detection of gangliosides with 9-0-acetylated sialic acids in basaliomas and normal human skin by influenza C virus assay. J. Drrmatol. submitted.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

14 Carb0hydrate Sulfotransferases Steven D. Rosen, Annette Bistrup, and Stefun Hemmerich

14.1 Introduction All eukaryotic cells carry out sulfation (i.e., sulfonation) reactions through the action of the class of enzymes known as sulfotransferases [l, 21. Two general classes of sulfotransferases exist: 1) cytosolic sulfotransferases, whose substrates include small molecules such as steroids, neurotransmitters, and various end products of metabolism [ 3 ] ; and 2) the Golgi-localized, membrane-bound sulfotransferases which modify protein tyrosine residues and carbohydrates [2]. The activities of the cytosolic sulfotransferases frequently lead to the inactivation or elimination of their substrates by increasing water solubility and decreasing biological activity [ 31. In contrast, the set of biological activities conferred onto their respective substrates by the Golgi-resident sulfotransferases is much more diverse as the activities frequently involve the creation of specific epitopes that can be recognized by soluble proteins, extracellular matrix proteins and membrane-bound receptors. The focus of this review will be the Golgi sulfotransferases, with emphasis on the carbohydrate sulfotransferases.

14.2 Basic Features of Sulfotransferase Reactions Sulfotransferases catalyze the transfer of a sulfate (i.e., sulfonate, SO3) group from an activated donor onto the hydroxyl or less commonly the amino group of the acceptor molecule. The activated donor is invariably 3’ phosphoadenosine 5’ phosphosulfate (PAPS) [4]. PAPS is synthesized from ATP and S 0 4 2 - by the sequential action of ATP sulfurylase, which generates adenosine 5’ phosphosulfate (APS), and APS kinase, which transfers a phosphate from ATP to APS. Chlorate, a drug which has been extremely useful for implicating the importance of sulfation

246

14 Carbohydrate Sulfotransfrruses

modifications in a variety of systems, is a chemical analogue of sulfate and serves as a competitive inhibitor of sulfate binding to ATP sulfurylase [5, 61. In lower species (plants, yeast, fungi, and bacteria) the ATP sulfurylase and APS kinase are encoded by separate genes. In animal cells, however, both activities are contained within a PAPS synthetase polypeptide which contains domains for the two activities [7, 81. The enzyme is widely distributed in tissues of the body and may serve as the source of PAPS for all sulfotransferase reactions-whether cytosolic or Golgi-associated. PAPS is synthesized within the cytosol of cells. Sulfate enters cells through an active transport process [4]. A sulfate transporter has been identified which is mutated in diastrophic dysplasia, a disease characterized by undersulfation of proteoglycans in cartilage [9]. For utilization by Golgi sulfotransferases, PAPS must cross a lipid bilayer. A PAPS translocase, which functions through an antiport mechanism has been characterized at the biochemical level [ 101.

14.3 Tyrosine Sulfation Golgi-associated sulfotransferases catalyze tyrosine sulfation and carbohydrate sulfation reactions. Tyrosine O-sulfation is a common post-translational modification that is found in both membrane and secretory proteins of eucaryotic cells [ 1 I]. The transfer of sulfate from PAPS to tyrosine residues of acceptor proteins is catalyzed by tyrosylprotein sulfotransferases, which recognize a motif consisting of several acidic amino acids in the vicinity of the tyrosine [ 11I. cDNAs encoding two highly related enzymes of this class have recently been identified [12, 131. The predicted proteins are type I1 membrane glycoproteins with very short amino terminal cytoplasmic tails. Although the functional roles for tyrosine sulfation are not well understood, recent findings have shown that tyrosine sulfation contributes to protein-protein interactions in several systems. Notable among these are the binding of PSGL-1 by P-selectin [ 141; the interaction of GPIba with von Willebrand factor [15] and binding between chemokine receptors and chemokines [ 161. Transcripts for both tyrosylsulfotransferases are broadly distributed in human tissues, but the relative levels differ among tissues [ 12, 131. A plausible explanation for this is that the two tyrosylsulfotransferases exhibit differential specificity for protein acceptors.

14.4 Diversity of Carbohydrate Sulfation Carbohydrates exhibit enormous diversity because of variation in monosaccharide composition, glycosidic linkage positions, and chain branching [ 171. Further diversity is achieved by covalent modification of carbohydrates via acetylation, methylation, phosphorylation, and sulfation [ 11. Sulfation can be incorporated into the carbohydrates of glycolipids and glycoproteins. Modifications of both N-linked and O-linked chains are found.

14.4 Divrrsity

ef

Curhohydrute Sulfntion

247

Table 1. Examples of sulfated monosaccharides found in glycoproteins or glycolipids. Sulfated monosaccharide

Examples of glycoprotein/glycolipid

Sulfatide [ 561 Ovarian cystadenoma glycoprotein 1571 Cystic fibrosis respiratory mucins [58] Colon carcinoma mucins [ 591 Various nervous system glycolipids and glycoproteins(NCAM. MAG. L1, PO), essential feature of HNK-I epitope I601 Lutenizing hormone, Thyroid stimulating hormone [ 6 11 Urokinase [62] Chondroitin 4-sulfite [ 561 Chondroitin 6-sulfate 1561 Keratan sulfate [63] GlyCAM-I, CD34 ligands for L-selectin [64] HIV gp120 [65] Porcine zona pellucida [ 661 Cystic fibrosis respiratory mucins [58] Colon carcinoma mucins 1591 Gal-6-sulfate

Carrageenan [ 561 GlyCAM-1, CD34 ligands for L-selectin [64] Keratan sulfate [ 631 Fucan from sea urchin cgg jelly coat [67] Chondroitin sulfate from sea cucumber [68] Rhizobiuni NGR234 nodulation factor [69]

Fucose-4-SOd

Fucoidan from Fiicus ~~~.siculosus [ 701 Fucan from sea urchin egg jelly coat [67]

GlcNAc-6-SO4 (reducing terminal)

Rhizobium nzrliloti nodulation factor [44]

Man-6304

D. discoidrunt secreted glycoproteins [71] Heparan sulfate froin breast cancer 1721

GlcNS IduA-2-SOd GlcA-2-SOd GlcNS-6-SOd GlcNS-3-SO4

Heparan sulfate, heparin [ 18, 191

Sulfated carbohydrates are present from bacteria to human [ l ] . A wide range of monosaccharide modifications have been reported, some of which are listed in Table 1. As sulfate modifications are extensive and tend to occur in clusters in glycoproteins, they significantly affect the physicochemical character of carbohydrate chains and protein scaffolds, particularly by imparting a net negative charge at physiological pH. In addition, sulfation provides a further diversification mecha-

248

14 Carbohydrate Sulfotrunsferuses

Table 2. Examples of specific sulfated carbohydrates involved in biological recognition phenomena. Sulfated carbohydrate/occurrence

Receptor and biological process

Siaa2,3[S04-61GalP1,4[Fuca I ,3]GlcNAc Siaa2,3GalPl,4[Fuca 1,3][S04-61GlcNAc High endothelial cell ligands

L-selectin on lymphocyte: lymphocyte homing [64, 731

(S04-4)GalNAc[ll,4GlcNAc~I ,2Man N-linked chains of LH and TSH

Macrophage/endothelial mannose receptor: serum clearing of LH and TSH j61, 741

GlcNH(R)PI, ~ G ~ c N A ~ P ~ , ~ G ~ ,4GlcNAc-6-S04 cNAcPI R = COC15:2 Nod factor produced by Rhizohiurn rneliloti

Receptor on roots of alfalfa [44, 451

[IduA(2-S04)al,4GlcNS (6-S04)a1,4], smooth muscle heparan sulfate

Platelet derived growth factor A chain [75]

nism to carbohydrate chains which can potentially augment information transfer. This potential is realized, as it is well established that sulfated carbohydrates comprise recognition determinants for many proteins and play important roles in a diverse set of biological systems. Recent reviews have surveyed these systems [ 1, 2, 181. A few representative examples are provided in Table 2. The most complete structural information about sulfate-dependent protein interactions is available from extensive studies on the many interactions of heparan sulfate proteoglycans (HSPGs). Via their covalently attached heparan sulfate chains, HSPGs bind to a large array of growth factors, cytokines, chemokines, extracellular matrix proteins, cell-cell adhesion molecules, viral attachment receptors, blood coagulation components, and lipid carrier molecules [ 18-21]. Among the functions served by these interactions are: 1) sequestration of soluble growth/differentiation factors for presentation to specific cell surface receptors; 2) modulation of growth factor-receptor activation; 3 ) cell-extracellular matrix adhesion; 4) cell-cell adhesion and viral infection of host; 5 ) neutralization of coagulation proteases; and 6) binding of lipoprotein lipase and its presentation to triglyceride-rich lipoproteins.

Heparan sulfate (HS) chains are synthesized by extending a core tetrasaccharide with alternate addition of glucuronic acid (GlcA) and GlcNAc moieties to produce linear polymers with many disaccharide units. Chain diversity is based on the degree of epimerization of GlcA to iduronic acid (IduA) and extensive variation in five different sulfation modifications along the chains [18, 191 (Table 1). The fine structure of the HS chains, determined by these modifications, confers preferential binding to particular proteins and receptors. High affinity binding generally depends on discrete regions of the chains (called blocks) which are extensively modified and are interspersed among regions of limited modification. For example,

14.5 Biochemical Denionstration qf Carbohydrate Su!$otransf>rasrs

249

crystallographic studies demonstrate that FGF-2 (bFGF) interacts with both IduA2-O-sulfate and GlcNS residues of heparin/HS chains within a hexasaccharide sequence [22]. This interaction is likely to function in the presentation of the growth factor to cells bearing FGFR1, its specific transmembrane receptor. To promote signaling through this receptor, a dodecasaccharide sequence of heparin/HS, containing both IduA-2-O-sulfate and GlcNS 6-O-sulfate residues, is required [23, 241. Another example is seen in the natural anticoagulation mechanisms of blood vessels in which heparin/HS potentiate the activity of antithrombin. This interaction depends on 6-O-sulfation of GlcN(Ac/S) and the 3-O-sulfation of a separate GlcNS in a hexasaccharide sequence [ 181. An analogous mechanism may be operative in the binding of L-selectin to a mucin-like physiological ligand. Here, the Gal-6-0- and GlcNAc-6-O-sulfation modifications appear to contribute synergistically to ligand activity [25].

14.5 Biochemical Demonstration of Carbohydrate Sulfotransferases Rather than providing an exhaustive list, this section will highlight representative examples of sulfotransferase assays. Similarly, only a few examples of enzyme purifications will be mentioned. The occurrence in nature of a tremendous diversity of carbohydrate sulfation modifications implies that multiple sulfotransferase enzymes exist. Numerous sulfotransferase activities have been identified in cell extracts. Typically, the sulfotransferase assays are based on the transfer of labeled sulfate from PAPS to acceptors, which are either model carbohydrates. oligosaccharide chains or glycoproteins. Two recent examples, pertinent to lymphocyte homing (Table 2), concern the identification of GlcNAc-6-O-sulfotransferase [26] and Gal6-O-sulfotransferase [27] activities in extracts of porcine lymph nodes. In both cases, disaccharide acceptors, incorporating key features of L-selectin ligands, were used. Chemical analysis of the sulfated products verified the regiochemistry of the sulfation modifications. Since these activities were enriched in lymph nodes, where sulfdte-dependent L-selectin ligands are elaborated, these activities are excellent candidates for the enzymes that actually participate in ligand synthesis. In the case of the GlcNAc-6-O-sulfotransferase, the activity was in fact shown to be highly enriched in isolated high endothelial cells 1261, the actual cell type in lymph nodes where ligand activity is expressed. Biochemical fractionation of extracts has led to the purification to homogeneity of several carbohydrate sulfotransferases, a few illustrative examples of which are considered below [27-341. In some cases, the starting material was conditioned medium from cultured cells and in others detergent extracts of tissues. Invariably, one of the purification steps employs the use of 3'5' ADP-agarose as an affinity matrix. Other commonly used steps employ matrices based on heparin, WGA, and blue-Sepharose. Typically, the various sulfotransferases are relatively minor proteins in the starting extracts, since purification factors range from 2,000-fold for conditioned medium [34] to 65,000-fold for a crude detergent extract [28]. One method for verifying the final product is to perform photoaffinity labeling on the product with [3'P]3'5'-ADP (32) or 8-azido[''PP]-PAPS(27).

250

I4 Curbohydru te Sulfo trunsfivmes

14.6 Molecular Cloning of Carbohydrate Sulfotransferases A list of the cloned Golgi-associated sulfotransferases from vertebrate sources is given in Table 3. Four basic strategies have been used to obtain cDNAs that encode carbohydrate sulfotransferases: 1) biochemical purification; 2) expression cloning; 3) cloning by homology; and 4) genetic screens. A recent example of the biochemical approach is provided by the cloning of chondroitin 6-sulfotransferase from chicken chondrocytes. The enzyme was purified as a soluble activity from conditioned medium of cultured chicken chondrocytes [35]. Amino acid sequences obtained from peptide fragments of the purified protein allowed the generation of a PCR product, which was used as a probe to clone a full-length cDNA [33]. When expressed recombinantly, the enzyme was shown to have both chondroitin-6-sulfotransferase activity (GalNAc-6-O-) and keratan sulfate sulfotransferase activity (Gal-6-0-) [33]. Expression cloning has been successfully applied in obtaining cDNAs for the rat [36] and human [37] GlcA-3-O-sulfotransferase. This enzyme is required for the formation of the HNK-1 epitope (Table l), which is present on a variety of neural cell adhesion molecules. Expression cloning [ 381 was carried out with antibodies that recognize the HNK-1 epitope in a sulfate-dependent manner. The approach required cotransfection with a cDNA for the glucuronyltransferase to provide the terminal GlcA residue to which sulfate can be added. Various homology cloning techniques have been applied to obtain cDNAs related to ones already cloned. One recent example is provided by the hepardn sulfate glucosaminyl 3-O-sulfotransferases (3-OSTs). This enzyme is limiting for the production of anticoagulant forms of heparan sulfate [18]. Murine 3-OST-1 was first cloned by a biochemical purification approach 1391. Subsequently, the human ortholog of this enzyme and several other homologous human OSTs were obtained by identifying related ESTs and using these in library screens to obtain full-length cDNAs [40]. Similarly, a human homolog of the chicken chondroitin-6-sulfotransferase (C6/KSST) mentioned above was identified in a human brain library by crosshybridization with the chick cDNA [41]. The human enzyme (KSGal6ST) obtained exhibited Gal-6-O-sulfotransferase (keratan sulfate) activity [41]. Subsequently, a more closely related human homologue (C6ST) of the chicken gene was cloned which showed chrondroitin 6-O-sulfotransferase activity [42]. A few sulfotransferases have been identified through genetic approaches. In Drosophila, one of the genes required for the formation of embryonic dorsalventral polarity is known as pipe [43]. The molecular cloning of pipe revealed two isoforms which are closely related to the IduA-2-O-sulfotransferase cloned from hamster (28% amino acid identity). It is speculated that the spatially restricted sulfation of a proteoglycan catalyzed by one of these enzymes is pivotal to the establishment of a dorsal-ventral morphogen gradient in the embryo. Sulfotransferase activity has yet to be demonstrated explicitly for the pipe gene. Nodulating bacteria such as Rhizobium meliloti elaborate secreted nod factors, which are required for the infection and nodulation of appropriate legume plants. These factors are lipooligosaccharides based on p1,4 linked GlcNAc backbone [44]. In the case of R. meliloti, one of the nod genes involved in the elaboration of its nod

GlcNAc Gal

GalNAc GalNAc or Gal GlcA & IduA in DS & CS GlcN(Ac) GlcNS

IduA in HS GlcNAc

GlcA Gal

High Endothelial Cell GlcNAc-6-0

Keratan/Gal-h-O

Chondroitin-6-0

ChondroitinIKeratan Sulfate 6-0 Dermatan/Chondroitin Sulfate 2 - 0 Heparan Sulfate glucosaminyl 6 - 0 Heparan Sulfate glucosaminyl 3 - 0

Heparan Sulfate glucuronosyl 2 - 0 Heparan Sulfate GlcNAc N-deacetylase/N-

HNK-I/GlcA-3-0

GalCer/Gal-3-O Tyrosylprotein TYr

AB014679 ABOl I451 AF083066 AF13 1325 AF13 I326 AB003791 AF090 137 ABOl2192 AB008937 D499 15 AB0203 16 AB006179 AF019386 AF019385 AF 105 374?? A F 105375'? AF105376? D8881 I U18918 AF049894 M92042 U36601 U02304 X75885 AF074924 AF033827 AF022129 D88667 AF038009 AF038008 AF06 1254

Human Mouse Human Human Mouse Human Human Human Mouse Chicken Human Human Human Mouse Human Human Human Hamster Human Mouse Rat Human Mouse Mouse Human Human Rat Human Human Mouse Human GlcNAc6ST

GlcNAc

GlcNAc-6-0

TPST-2

GalCer3ST TPST- 1

NST3 HNKGlcA3ST

NST2

3-OST-2 3-OST-3A 3-OST-3B 2-OST NSTl

C6/KSST D/CS-2ST 6-OST 3-OST-1

KSGa16ST CHSTl C6ST

CHST2 HEC-GICNAC~ST

Accession No.

Species

Abbreviation

~

Residue

~~

Activity

_______

Table 3. List of cloned carbohydrate and tyrosylprotein sulfotransferases References

.c

a

o\

252

14 Carbohydrate Sulfotransjerases

factor is NodH. The sequence of NodH is homologous to sulfotransferases and the protein encoded by NodH has been confirmed to catalyze the addition of sulfate to the 6-position of GlcNAc at the reducing terminus of a lipooligosaccharide [45,46]. When sulfated, the nod factor elicits responses from alfalfa, whereas the unsulfated nod factor causes a conversion in host range to a different legume [45].

14.7 Primary Structures of Carbohydrate Sulfotransferases The carbohydrate sulfotransferases and the tyrosine sulfotransferases, like the glycosyltransferases [47], have a type I1 membrane organization with a typically short cytoplasmic tail at the C-terminus, a transmembrane domain, and a large Cterminal domain which is predicted to be intraluminal. In contrast to this general pattern, 3-OST- 1 lacks the first two domains and apparently resides intraluminally in the Golgi [39]. Again, as is the case for the glycosyltransferases, the intraluminal domain of the carbohydrate sulfotransferases can be divided into a membraneproximal stem domain and a globular domain of about 260-280 residues which extends to the C-terminus and is the putative sulfotransferase domain [40].The stem region varies in length from 10 to 100 residues and frequently shows a high content of Ser, Pro, Leu, Ala, and Gly (SPLAG domain) [40]. Regions of this character are predicted to have very limited secondary structure and therefore to be quite flexible [401. The Carbohydrate sulfotransferases that have been molecularly cloned segregate into four families within which isoforms can be defined based on similar catalytic activities and sequence homology in the putative catalytic domains (Figure 1). The families are: 1) the heparan sulfate N-deacetylase/N-sulfotransferases(NSTs, three isoforms in human); 2) heparan sulfate D-glucosaminyl 3-O-sulfotransferases (3OSTs, four isoforms in human); 3 ) 2-O-sulfotransferases acting on heparan-, dermatan- and/or chondroitin sulfates; and 4) the 6-O-sulfotransferases which add sulfate to the 6-position of Gal, GalNAc, or GlcNAc (GSTs. four isoforms in human). Several other carbohydrate sulfotransferases have been cloned but do not as yet belong to a family, as only individual members are presently known (Table 3, Figure 1). Significant sequence homology is observed only in the putative sulfotransferase catalytic domains (Figure 1). Even within members of a given family, there is no meaningful homology among cytoplasmic, transmembrane, or stem regions. A dendrogram showing the relatedness of the putative catalytic domains of the human carbohydrate sulfotransferases is presented in Figure 1. Also included in the comparison are the two tyrosylsulfotransferases. Of the four carbohydrate sulfotransferase families, the heparan sulfate NSTs and the heparan sulfate 3-OSTs are most closely related, with homology scores of >60% between the aligned sequences within a family and -40% across the families. The heparan sulfate 6-0sulfotransferase, on the other hand, exhibits only low homology to the 2-0sulfotransferases (-30Yo) and even less to the other sulfotransferase families (-25%). The HNK- 1 sulfotransferase and the galactosyl ceramide 3-O-sulfotransferase,

14.7 Primary Structures of’ Carbohydrate Su!fotransjerast>s

253

3-0sT-3~ (148-406) 3-OST-3s (133499) 3-OST-2 (110-367) 3-OST-1 (49-307) NST-1 (599882) NST-3 (590873)

]

NST

NST-2 (596-884) D/CS-PST (105406) 2-OST (66-356)

I

]20ST

6-OST (79-410) GalCer3ST (72423) HNKlGlcA3ST (79-256) C6ST (131-479)

I

KSGal6ST (59411) GlcNAcGST (163-530)

I

HEC-GICNAC~ST (41-386) TPST-1 (63-370) TPST-2 (62-377)

25

50

GST

75

]TPST

100

Homology (#)

Figure 1. Comparison of Sulfotransferases. The amino acid sequences of the putative sulfotransferase domains were aligned pairwise using the ClustalW algorithm 1901 at default settings. The homology scores were calculated as the sum of the identity score and one half of the high similarity score. The resulting scores were rounded towards the next integer, imported into a matrix and the phylogenctic tree was assembled using the G C G program “Growtree”. All sequences correspond to human clones. The accession numbers are (top to bottom): AF105375, AF105377, AF105374, AF019386, U36600. AF074924, U36601, AB020316. AB007917, AB006179, D88667. AF033827, AB012192, AB003791, AF083066. AF131325, AF038009, and AF049891. Human 2-OST is annotated in Genbank (accession number AB007917) as KIAA 0448 protein with unknown function. We identified this protein by BLAST screening of the human sequences in Genbank using the hamster enzyme (accession number D8881 I ) as the query sequence [92]. The human KIAA 0448 protein was found to be 97.5% identical to the hamster enzyme and was thus assigned as human 2OST. Scores of 25% indicate no significant overall homology. Comparisons that yield homology scores of -25‘%1(actual range 23.5 -25) are assigned the value of 25‘% on the dendrogram.

-

show no significant overall homology to each other or to other sulfotransferases (-25%). The Gal/GalNAc/GlcNAc 6-0-sulfotransferases (GSTs) and the protein tyrosylsulfotransferases (TPSTs) constitute two distinct families. The GSTs are greater then 40% homologous to each other and appear to fall into two subfamilies, those that sulfate Gal and/or GalNAc and those that sulfate GlcNAc. Within each subfamily, homologies are 2 55%. Detailed inspection of the putative catalytic domains reveals regions that are conserved among all sulfotransferases including the cytosolic enzymes, as well as others that are highly conserved within families. The structural and functional

254

14 Carbohydrate Sulfotruns$erusrs

constraints imposed by the common sulfate donor PAPS are reflected in the presence of sequence motifs that are found in all sulfotransferases identified to date. Considerable information about the probable configuration of the active site has been obtained from two recent reports on the structure of estrogen sulfotransferase (EstST) co-crystallized with the inactive co-factor adenosine 3 : 5‘-diphosphate and either estrogen [48] or vanadate [49]. Residues interacting with the 5’-phosphate in PAPS are in a characteristic P-loop motif (consensus sequence TYPKSGTTWL) [48, 501 similar to the one found in ATP- and GTP- binding proteins [51, 521. Within this loop of EstST is a conserved lysine residue (Lys 48 in EstST; Lys 614 is the corresponding residue in human heparan sulfate N-deacetylase/N-sulfotransferase- 1) which has been shown by mutational analysis to be critically involved in catalysis [53, 541. The lysine residue is conserved in all the sulfotransferases represented in Figure 1, with the exception of the GST family, in which the corresponding residue is an arginine and the sequence immediately following is (S/T)GSSF (Figure 2). Interestingly, within both the 3OST family and the GST family of sulfotransferases, the candidate sequence for

HECGlcNAc6ST GlcNAc6ST KSGal6ST C6ST

HECGlcNAc6ST GlcNAc6ST KSGal6ST C6ST

HECGlcNAc6ST GlcNAc6ST KSGa16ST C6ST Figure 2. Alignment of regions of high homology among the four members of the GST family of carbohydrate sulfotransferases that modify ti-hydroxyl residues. Protein sequences were aligned using the ClustalW algorithm [901. Sequences that conform to the consensus sequences for binding to PAPS are shaded in gray. The 5’-phosphate and 3’-phosphate binding motifs are indicated. Sequences that are unique to the GST family are shaded in black. The amino acid numbers refer to the amino acid sequence for HEC-GlcNAc6ST [25].

14.7 Prirnury Structurrs o f Curhohydrute Sulfotrunsjerrrses

255

the P-loop is highly conserved. For example, among the five members of GST family, five of the ten residues are identical and all but one residue exhibit similarity (Figure 2). The region which may be involved in binding the 3'-phosphate of PAPS is less well conserved, but among all sulfotransferases, sequence in this region conforms to the consensus sequence (K/R/H)(hydrophobic),R(N/D)(P/G)(X),SX [50]. Again, within the GST family, this motif is very highly conserved, with a central L(V/F)RDPR(A/G)(V/I)XXSRsequence. In addition to the PAPS binding motifs found in all sulfotransferases, each family of sulfotransferases contains sequences that appear to be unique within the family. For example, within the GST family, the sequence (L/V)RYED(L/V)(A/V) (aa residues 273-280 in the HEC-GlcNAc6ST sequence, Figure 2) is very highly conserved. Although a similar motif appears to be present in the majority of other sulfotransferases, there is a high degree of divergence outside the GST family. Similarly, the highly conserved sequence F(Y/F)L(F/Y/M)EP(L/V/A)(W/Y)H(V/I) is unique within the GST family. It is possible that these two elements contribute to a binding pocket that interacts with the 6-hydroxyl group of the appropriate oligosaccharide acceptor (Gal, GalNAc, or GlcNAc) to bring it into apposition with the donor phosphosulfate group. Rapid progress in the molecular cloning of carbohydrate sulfotransferases has yielded a large number of full length cDNAs in the past few years. A striking feature is the emergence of several families, each comprised of members with related catalytic domains. This multiplicity is probably attributable to the functional requirements for carbohydrate chains with complex and reproducible patterns of sulfation. The heparan sulfate chains provide the best understood example. Here it is established that precise regulation of carbohydrate fine structure dictates specificity of protein binding. The diversity of structures is critically dependent on the placement of sulfates along the GAG chains and thus on the availability of the appropriate sulfotransferases in the cells and tissues where these structures are synthesized. A useful example is provided by the 3-OST enzymes, which catalyze the addition of sulfate to the 3-OH position of glucosamine in heparan sulfate chains. 3OST-1, the rate limiting enzyme in production of anticoagulant forms of heparan sulfate, transfers sulfatc to glucosamine in GlcA-GlcNS ? 6s. 3-OST-2 transfers sulfate to glucosamine in GlcA2S-GlcNS and IduA2S-GlcNS and 3-OST-3A transfers sulfate to glucosamine in IduA2S-GlcNS [ 551. Importantly, these latter two enzymes are not capable of generating anticoagulant forms of HS. Taken together with the fact that three enzymes exhibit different patterns of tissue expression, these results argue that the individual 3-OST isoforms are specialized to add the rare 3-0-sulfate modification in specific sequence contexts. The resultant HS chains would thus acquire different biological functions. In the case of 3-OST-2 and 3-OST-3A, the biological functions of the specifically modified HS are not yet known [55]. The roles of the nonconserved domains of the enzyme family in determining acceptor specificity or in other aspects of regulation are completely unexplored at this point. The above speculations are based on limited information. With the availability of an increasing number of cDNAs encoding carbohydrate sulfotransferases. rapid

256

14 Carbohydrate Suljotransjkrases

progress is anticipated in elucidating the biological consequences of specific sulfation modifications within carbohydrates and the contributions of individual enzymes to these processes.

Acknowledgments

Sulfotransferase research in the Rosen laboratory is supported by grants from the NIH (R37 GM23547, GM57411) and from Roche Bioscience.

References I . L. V. Hooper, J. U. Baenziger, Sulfotransferase and glycosyltransferase analyses using a 96well filtration plate. Anal. Biockem. 1993 212, 128-33. 2. K. G. Bowman, C. R. Bertozzi, Carbohydrate sulfotransferases: mediators of extracellular communication. Chem. Bid. 1999 6, R9-R22. 3. C. N. Falany, Enzymology of human cytosolic sulfotransferases. FASEB J. 1997 11, 206-16. 4. C. D. Klaassen, J. W. Boles, Sulfation and sulfotransferases 5: the importance of 3'-phosphoadenosine 5'-phosphosulfate (PAPS) in the regulation of sulfation. FASEB J. 1997 11. 40418. 5. P. A. Baeuerle, W. B. Huttner, Chlorate-a potent inhibitor of protein sulfation in intact cells. Biochen?. Biophys. Rex Commun. 1986 141, 870-877. 6. A. C. Rapraeger, A. Krufia, B. B. Olwin, Requirement of heparan sulfate for bFGF-mediated fibroblast growth and myoblast differentiation. Science 1991 252, 1705-1708. 7. H. Li, A. Deyrup, J. R. Mensch, Jr., M. Domowicz, A. K. Konstantinidis, N. B. Schwartz, The isolation and characterization of cDNA encoding the mouse bifunctional ATP sulfurylaseadenosine 5'-phosphosulfate kinase. J. Biol. Chem. 1995 270, 29453-9. 8. J . P. Girard. E. S. Baekkevold, F. Amalric, Sulfation in high endothelial venules: cloning and expression of the human PAPS synthetase. FASEB J. 1998 12, 603-612. 9. J. Hastbacka, A. de la Chapelle, M. M. Mahtani, G . Clines, M. P. Reeve-Daly, M. Daly, B. A. Hamilton, K. Kusumi. B. Trivedi, A. Weaver. et al., The diastrophic dysplasia gene encodes a novel sulfate transporter: positional cloning by fine-structure linkage disequilibrium mapping. Cell 1994 78, 1073-87. 10. J. D. Ozeran, J. Westley, N. B. Schwartz, Identification and partial purification of PAPS translocase. Bioclwnistrj~1996 35, 3695-703. 11. W. B. Huttner, Tyrosine sulfation and the secretory pathway. Annu. Reu. Physiol. 1988 50, 363- 376. 12. Y. Ouyang, W. S. Lane, K. L. Moore, Tyrosylprotein sulfotransferase: purification and molecular cloning of an enzyme that catalyzes tyrosine 0-sulfation, a common posttranslational modification of eukaryotic proteins. Proc. Nut1 Acud. Sci. U S A 1998 95, 2896-901. 13. R. Beisswanger, D. Corbeil, C. Vannier, C. Thiele, U. Dohrmann, R. Kellner, K. Ashman. C. Niehrs, W. B. Huttner. Existence of distinct tyrosylprotein sulfotransferase genes: molecular characterization of tyrosylprotein sulfotransferase-2. Proc. Nut1 Acud. Sci. U S A 1998 95, 11 134-9. 14. D. Sako, K . M. Comess, K. M . Barone, R. T. Camphausen, D. A. Cumming, G. D. Shaw, A sulfated peptide segment at the amino terminus of PSGL-1 is critical for P-selectin binding. Cell 1995 83, 323-331. 15. J. Dong, C. Q. Li, J. A. Lopez, Tyrosine sulfation of the glycoprotein Ib-IX complex: Identification of sulfated residues and effect on ligand binding. Biochemistry 1994 33, 13946-13953. 16. M. Farzan, T. Mirzabekov, Peter Kolchinsky, Richard Wyatt, M. Cayabyab, N. P. Gerard, Craig Gerard, J. Sodroski, H. Choe, Tyrosine Sulfation of the Amino Terminus of CCR5 Facilitates HIV-I Entry. Cell 1999 96, 667-676.

References

257

17. R. M. Nelson. A. Venot, M. P. Bevilacqua, R. J. Linhardt, I. Stamenkovic, Carbohydrateprotein interactions in vascular biology. Awnu. Rev. C M . D ~ PBiol. . I995 1 1 , 601-3 I . 18. R. D. Rosenberg. N. W. Shworak, J . Liu, J . J. Schwartz, L. Zhang. Hcparan sulfate proteoglycans of the cardiovascular system. Specific structures emerge but how i s synthesis regulated? J. Clin. Invest. 1997 100, S67--75. 19. M. Salmivirta, K. Lidholt, U. Lindahl. Heparan sulfate: a piece of information. FASEB J. 1996 10, 1270-9. 20. D. J. Carey, Syndecans: multifunctional cell-surface co-receptors. Biochenz. J. 1997 327, 1 16. 21. R . E. Hileman, J. R . Fromm. J. M. Weiler, R. J. Linhardt, Glycosaminoglycan-protein interactions: definition of consensus sites in glycosaminoglycan binding proteins. Biorssuys 1998 20, 156-67. 22. S. Faham, R . E. Hileman, J. R . Fromm, R. J. Linhardt, D. C. Rees, Heparin structure and interactions with basic fibroblast growth factor. Science 1996 271, 11 16-20. 23. S. Guimond, M. Maccarana, B. B. Olwin, U. Lindahl, A. C. Rapraegcr, Activating and inhibitory heparin sequences for FGF-2 (basic FGF). Distinct requirements for FGF-1. FGF2. and FGF-4. J. Bid. Chmni. 1993 268. 23906 23914. 24. A. C. Rapraeger, In the clutches of proleoglycans: how docs heparan sulfate rcgulate F G F binding? Client Bid. 1995 2, 645-9. 25. A. Bistrup, S. Bhakta, J. K. Lee: Y. C. Belov, M. D. Gunn, F.-R. Zuo, C.-C. Huang, R . Kannagi, S. D. Rosen, S. Hemmerich, Sulfotransferases of Two Specificities Function in the Reconstitution of High-Endothelial-Cell Ligands for L-Selectin. J. Cell Biol. 1999 (in press). 26. K. G. Bowman, S. Hemmerich, S. Bakhta. M. S. Singer, A. Bistrup, S. D. Rosen, C. R. Bertozzi, Identification of an N-acetylglucamine-6-0-sulfotransferase activity specific to lymphoid tissue: an enzyme with a possible role in lymphocyte homing. Cheni. Biol. 1998 5, 447-460. 27. K. Shailubhai, Q. K . Huynh, H. Boddupalli. H. H. Yu, G . S. Jacob. Purification and characterization of a lymph node sulfotransferase for 6-0-sulfation of the galactose residues and other sialyl lewis x-related sugars. Biochtw. Biopliy,s. Rex Cornm. 1999 256, 170- 176. 28. E. Brandan, C. B. Hirschberg, Purification of rat liver N-heparan-sulfate sulfotransferase. J. Bid. Clzem. 1988 263, 2417-2422. 29. H. Wlad, M . Maccarana, I. Eriksson, L. Kjcllen, U. Lindahl, Biosynthesis of heparin. Different molecular forms of 0-sulfotransferases. J. Biol. Chtw. 1994 269. 24538-41, 30. 1. Pettersson. M. Kusche, E. Unger, H. Wlad. L. Nylund. U. Lindahl, L. Kjellen, Biosynthesis of heparin. Purification of a 110-kDa mouse mastocytoma protein required for both glucosaminyl N-deacetylation and N-sulfation. J. Biol. Cliern. 199 I 266, 8044-9. 31. 0. Habuchi, Y. Matsui, Y. Kotoya, Y. Aoyama, Y . Yasuda, M . Noda, Purification if chondrotin-6-sulfotransferase secreted from cultured chick embryo chondrocytes. J. Biol. Chenz. 1993 268, 21968-21974. 32. L. V. Hooper, 0. Hindsgaul, J. U . Baenziger, Purification and characterization of the GalNAc4-sulfotransferase responsible for sulfation of GalNAcb1,4GlcNAc-bearing oligosaccharides. J. Bid. Cllrwz. 1995 270, 16327-16332. 33. M. Fukuta, K. Uchimura, K . Nakashima, M. Kato, K. Kimata, T. Shinomura, 0. Habuchi. Molecular cloning and expression of chick chondrocyte chondroitin 6-sulfotransferase. J. Biol. Cliem. 1995 270, 18575-18580. 34. S. Yamauchi, Y. Hirahara, H. Usui, Y. Takeda, M. Hoshino, M. Fukuta, J . H. Kimura, 0. Habuchi, Purification and characterization of chondroitin 4-sulfotransferase from the culture medium of a rat chondrosarconia cell line. J. B i d Cheni. 1999 274, 2456 ~ 6 3 . 35. 0. Habuchi. Y. Matsui, Y. Kotoya. Y. Aoyama, Y. Yasuda, M. Noda, Purification of chondroitin 6-sulfotransferase secreted from culturcd chick embryo chondrocytes. J. Biol. Chern. 1993 268, 21968-21974. 36. H. Bakker, I. Friedmann, S. Oka, T. Kawasaki, N. Nifant'ev, M. Schachner, N. Mantei, Expression cloning of a cDNA encoding a sulfotransferase involved in the biosynthesis of the HNK-1 carbohydrate epitopc. J. Bid. Cliem. I997 272, 29942-6. 37. E. Ong, J. C. Yeh, Y. Ding, 0. Hindsgaul, M. Fukuda, Expression cloning of a human sulfotransferase that directs the synthesis of the HNK-1 glycan on the neural cell adhesion molecule and glycolipids. J. Biol. Chem. 1998 273, 5 1 9 0 ~5. ~

258

14 Carbohydrate Sulfbtrunsferuses

38. A. Aruffo, B. Seed. Molecular cloning of a CD28 cDNA by a high efficiency COS expression system. Proc. Nut1 Acud. Sci. U S A 1987 84, 8753-8577. 39. N . W. Shworak, J. Liu, L. M. S. Fritze, J. J. Schwartz, L. Zhang, D. Logeart, R. D. Rosenberg, Molecular cloning and expression of mouse and human cDNAs encoding heparan sulfate D-glucosaminyl 3-0-sulfotransferase. J. Biol. Chcnt 1997 272, 28008-28019. 40. N. W. Shworak, J. Liu, L. M. Petros, L. Zhang. M. Kobayashi, N. G. Copeland, N. A. Jenkins, R. D. Rosenberg, Multiple isoforms of heparan sulfate D-glucosaminyl 3-0-sulfotransferase. Isolation, characterization, and expression of human cdnas and identification of distinct genomic loci. J. Bid. Cheni. 1999 274, 5170-84. 41. M. Fukuta, J. Inazawa, T. Torii, K. Tsuzuki, E. Shimada, 0. Habuchi, Molecular cloning and characterization of human kerdtan sulfate Gal-6-sulfotransferase. J. Biol. Chern. 1997 272: 32321-8. 42. M. Fukuta, Y. Kobayashi, K. Uchimura, K. Kimata. 0. Habuchi, Molecular cloning and expression of human chondroitin 6-sulfotransferase. Biochim. Biophys Aria 1998 1399, 5761. 43. J. Sen, J. S. Goltz, L. Stevens, D. Stein, Spatially restricted expression of pipe in the Drosophila egg chamber defines embryonic dorsal-ventral polarity. Cell 1998 95, 471-81. 44. P. Lerouge, P. Roche, C. Faucher, F. Maillet, G. Truchet, J. C. Prome, J. Denarie, Symbiotic host-specificity of Rhizobium meliloti is determined by a sulphated and acylated glucosamine oligosaccharide signal. Nature 1990 344, 781-784. 45. P. Roche, F. Debelle, F. Maillet, P. Lerouge, C. Faucher, G. Truchet, J. Denarie, J. C. Prome, Molecular basis of symbiotic host specificity in rhizobium meliloti: nodH and nodPQ genes encode the sulfation of lipo-oligosaccharide signals. Cell 199I 67, 1 131 -I 143. 46. D. W. Ehrhardt, E. M. Atkinson, K. F. Faull, D. 1. Freedberg, D. P. Sutherlin, R. Armstrong, S. R. Long, In vitro sulfotransferase activity of NodH, a nodulation protein of Rhizobium meliloti required for host-specific nodulation. J. Bacteriol. 1995 177, 6237-6245. 47. K. J. Colley, Golgi localization of glycosyltransferases: more questions than answers. Glycohioloyy 1997 7, 1-13. 48. Y. Kakuta, L. G. Pedersen, C. W. Carter, M. Negishi, L. C. Pedersen, Crystal structure of estrogen sulphotrdnsferase Nat. Struct. Bid. I997 4, 904-8. 49. Y. Kakuta, E. V. Petrotchenko, L. C. Pedersen, M. Negishi, The sulfuryl transfer mechanism. Crystal structure of a vanadate complex of estrogen sulfotransferase and mutational analysis. J. Biol. Chem. 1998 273, 27325-30. 50. Y. Kakuta, L. G. Pedersen, L. C. Pedersen, M. Negishi, Conserved structural motifs in the sulfotransferase family Trendy Biochern. Sci. 1998 23, 129%30. 51. H. Chiba, K. Komatsu, Y. C. Lee, T. Tomizuka, C. A. Strott, The 3’-terminal exon of the family of steroid and phenol sulfotransferase genes is spliced at the N-terminal glycine of the universally conserved GXXGXXK motif that forms the sulfonate donor binding site. Proc. Natl Acud. Sci. U S A 1995 92, 8176-9. 52. K. Komatsu, W. J. Driscoll, Y. C. Koh, C. A. Strott, A P-loop related motif (GxxGxxK) highly conserved in sulfotransferases is required for binding the activated sulfate donor. Biochem. Biophys. Rex Comnzun. 1994 204, 1178-85. 53. Y. Kakuta, L. C. Pedersen, K. Chae, W. C. Song. D. Leblanc, R. London, C. W. Carter, M. Negishi, Mouse steroid sulfotransferases: substrate specificity and preliminary X-ray crystallographic analysis. Biochrm. Phurmacol. 1998 55, 3 13-7. 54. T. Sueyoshi, Y. Kakuta, L. C. Pedersen, F. E. Wall, L. G. Pedersen, M. Negishi, A role of Lys614 in the sulfotransferase activity of human heparan sulfate N-deacetylase/Nsulfotransferase. FEBS Lett. 1998 433, 21 1-4. 55. J. Liu, N. W. Shworak, P. Sinay, J. J. Schwartz, L. Zhang, L. M. Fritze, R. D. Rosenberg, Expression of heparan sulpate D-glucosaminyl 3-0-sulfotransferase isoforms reveals novel substrate specificities. J. Bid. Chem. 1999 274, 5185-92. 56. CarbBank, University of Georgia CCRC. http:// 128.192.9.29/carbbank/CarbBank.htm. 57. C. T. Yuen, A. M. Lawson, W. Chai, M. Larkin, M. S. Stoll, A. C. Stuart, F. X. Sullivan, T. J. Ahern, T. Feizi, Novel sulfated ligands for the cell adhesion molecule E-selectin revealed by the neoglycolipid technology among 0-linked oligosaccharides on an ovarian cystadenoma glycoprotein. Biochemistry 1992 31, 9126-9131.

References

259

58. J. M. Lo-Guidice, J. M. Wieruszeski, J. Lemoine, A. Verbert, P. Roussel, G. Lamblin, Sialylation and sulfation of the carbohydrate chains in respiratory mucins from a patient with cystic fibrosis. J. Bid. Chem. 1994 269, 18794-18813. 59. C. Capon, J. M. Wieruszeski, J. Lemoine, J. C. Byrd, H. Leffler, Y. S. Kim, Sulfated lewis X determinants as a major structural motif in glycans from LS174T-HM7 human colon carcinoma mucin. J. Biol. Chem. 1997 272, 31957-68. 60. M. Schachner, R. Martini, Glycans and the modulation of neural-recognition molecule function. Trends Neurosci. 1995 18, 183-91. 61. L. V. Hooper, S. M. Manzella, J. U. Baenziger, From legumes to leukocytes: Biological roles for sulfated carbohydrates. FASEB J. 1996 10, 1137-1 146. 62. A. A. Bergwerff, J. Van Oostrum, J. P. Kamerling, J. F. Vliegenthart, The major N-linked carbohydrate chains from human urokinase. The occurrence of 4-O-sulfated, (alpha 2-6)sialylated or (alpha 1-3)-fucosylated N-acetylgalactosamine(beta 1-4)-N-acetylglucosamine elements. Eur. J. Biochem. 1995 228, 1009-19. 63. G . M. Brown, T. N. Huckerby, H. G. Morris, B. L. Abram, I. A. Nieduszynski, Oligosaccharides derived from bovine articular cartilage keratan sulfates after keratanase I1 digestion: Implications for keratan sulfate structural fingerprinting. Biochemistry 1994 33, 48364846. 64. S. Hemmerich, H. Leffler, S. D. Rosen, Structure of the 0-glycans in GlyCAM-I, an endothelial-derived ligand for L-selectin. J. Biol. C'hem. 1995 270, 12035.-12047. 65. A. Shilatifard, R. K. Merkle, D. E. Helland, J. L. Welles, W. A. Haseltine, R. D. Cummings, Complex-type N-linked oligosaccharides of gpl20 from human immunodeficiency virus type 1 contain sulfated N-acetylglucosamine. J. Virol. 1993 67, 943-952. 66. C. H. Hokke, J. B. L. Damm, J. P. Kamerling, J. F. G. Vliegenthart, Structure of three acidic 0-linked carbohydrate chains of porcine zona pellucida glycoproteins. FEBS Lett. 1993 329, 29-34. 67. A. P. Alves, B. Mulloy, J. A. Diniz, P. A. Mourao, Sulfated polysaccharides from the egg jelly layer are species-specificinducers of acrosomal reaction in sperms of sea urchins. J. Bid. Chem. 1997 272,6965-71. 68. R . P. Viera, P. A. S. Mouriio, Occurence of a unique fucose-branched chondroitin sulfate in the body wall of a sea cucumber. J. Biol. Chem. 1988 263, 18176-18183. 69. N. P. Price, F. Talmont, J. M. Wieruszeski, D. Prome, J. C. Prome, Structural determination of symbiotic nodulation factors from the broad host-range Rhizobium species NGR234. Curhohyd. Res. 1996 289, 115-36. 70. M. S. Patankar, S. Oehninger, T. Barnett, R . L. Williams, G. F. Clark, A revised structure for fucoidan may explain some of its biological activities. J. Biol. Chem. 1993 268, 21770-21776. 71. H. H. Freeze, D. Wolgast, Structural analysis of N-linked oligosaccharides from glycoproteins secreted by Dictyostelium discoideum. Identification of mannose 6-sulfate. J. Biol. Chem. 1986 261. 127-34. 72. N. Parthasarathy, L. F. Gotow, J. D. Bottoms, T. E. Kute, W. D. Wagner, B. Mulloy, Oligosaccharide sequence of human breast cancer cell heparan sulfate with high affinity for laminin. J. Biol. Chem. 1998 273, 21 I 1 1-4. 73. S. D. Rosen, C. B. Bertozzi, Leukocyte adhesion: two selectins converge on sulphate. Curr. B i d 1996 6, 261-264. 74. D. Fiete, M. C. Beranek, J. U. Baenziger, The macrophage/endothelial cell mannose receptor cDNA encodes a protein that binds oligosaccharides terminating with S04-4-GalNAcbeta1,4GlcNAcbeta or Man at independent sites. Proc. Nut1 Acud Sci. U S A 1997 94. 1125661. 75. E. Feyzi, F. Lustig, G. Fager, D. Spillmann, U. Lindahl, M. Salmivirta. Characterization of heparin and heparan sulfate domains binding to the long splice variant of platelet-derived growth factor A chain. J. Biol. Chem. 1997 212, 5518-24. 76. K. Uchimura, H. Muramatsu, T. Kaname, H. Ogawa, T. Yamakawa, Q. W. Fan, C. Mitsuoka, R. Kannagi, 0. Habuchi, I. Yokoyama, K. Yamamura, T. Ozaki, A. Nakagawara, K. Kadomatsu, T. Muramatsu, Human N-acetylglucosamine-6-0-sulfotransferase involved in the biosynthesis of 6-sulfo sialyl Lewis X: molecular cloning, chromosomal mapping, and expression in various organs and tumor cells. J. Biochem. (Tokyo) 1998 124, 670--8.

260

14 Curhohydrute Sulfotrunsferuses

77. K. Uchimura, H. Muramatsu, K. Kadomatsu, Q. W. Fan, N. Kurosawa, C. Mitsuoka, R. Kannagi, 0. Habuchi, T. Muramatsu, Molecular cloning and characterization of an Nacetylglucosamine-6-O-sulfotransferase. J. Bid. Chem. 1998 273, 22577-83. 78. X. Li, T. F. Tedder, CHSTl and CHST2 sulfotransferases expressed by human vascular endothelial cells: cDNA cloning, expression, and chromosomal localization. Genomics 1999 55. 34557. 79. K. Uchimura, K. Kadomatsu, Q. W. Fan, H. Muramatsu, N . Kurosawa, T. Kaname, K. Yamamura, M. Fukuta, 0. Habuchi, T. Muramatsu, Mouse chondroitin 6-sulfotransferase: molecular cloning, characterization and chromosomal mapping. Glycohioloyy 1998 8, 489-96. 80. H. Habuchi, M. Kobayashi, K. Kimata, Molecular characterization and expression of heparan-sulfate 6-sulfotransferase. Complete cDNA cloning in human and partial cloning in Chinese hamster ovary cells. J. Bid. Chem. 1998 273, 9208-13. 81. M. Kobayashi, H. Habuchi, M. Yoneda, 0. Habuchi, K. Kimata, Molecular cloning and expression of Chinese hamster ovary cell heparan-sulfate 2-sulfotransferase. J. Bid. Chem. 1997 272, 13980-5. 82. J. Dixon, S. K. Loftus, A. J. Gladwin, P. J. Scambler, J. J. Wasmuth, M. J . Dixon, Cloning of the human heparan sulfate-N-deacetylase/N-sulfotransferase gene from the Treacher Collins syndrome candidate region at 5q32-q33.1. Genomics 1995 26, 239-244. 83. M. Kusche-Gullberg, I. Eriksson, D. S. Pikas, L. Kjellen, Identification and expression in mouse of two heparan sulfate glucosaminyl N-deacetylase/N-sulfotransferase genes. J. Bid. Cheni. 1998 273, 11902-7. 84. Y. Hashimoto, A. Orellana, G. Gil, C. B. Hirschberg, Molecular cloning and expression of rat liver N-heparan sulfate sulfotransferase. J. Bid. Chern. 1992 267, 15744-1 5750. 85. D. E. Humphries, J. Lanciotti, J. B. Karlinsky, cDNA cloning, genomic organization and chromosomal localization of human heparan glucosaminyl N-deacetylase/N-sulphotransferase2. Biochem. J. 1998 332, 303 -7. 86. A. Orellana, C. B. Hirschberg, Z. Wei, S. J. Swiedler, M. Ishihara, Molecular cloning and expression of a glycosaminoglycan N-acetylglucosaminyl N-deacetylase/N-sulfotransferase from a heparin-producing cell line. J . Biol. Chem. 1994 269, 2270-6. 87. 1. Eriksson, D. Sandback. B. Ek, U. Lindahl, L. Kjellen, cDNA cloning and sequencing of mouse mastocytoma glucosaminyl N-deacet ylase/N-sulfotransferase, an enzyme involved in the biosynthesis of heparin. J. Biol. C h n . 1994 269, 10438-43. 88. J. Aikawa, J. D. Esko, Molecular cloning and expression of a third member of the heparan sulfate/heparin GlcNAc N-deacetylase/N-sulfotransferase family. J. Bid. Cl7em. 1999 274, 2690.- 5. 89. K. Honke, M. Tsuda, Y. Hirahara, A. Ishii, A . Makita, Y. Wada, Molecular cloning and expression of cDNA encoding human 3’-phosphoadenylylsuIfate:galactosylceramide3’-sulfotransferase. J. Bid. Chern. 1997 272, 4864-4868. 90. J. D. Thompson, D. G. Higgins, T. J. Gibson, CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, positions-specific gap penalties and weight matrix choice. Nucleic Acids Reseurch 1994 22, 4373-80. 91. M. Kobayashi, G. Sugumaran, J. Liu, N. W. Shworak, J. E. Silber, and R. D. Rosenberg, Molecular Cloning and Characterization of a Human Uronyl 2-Sulfotransferase That Sulfates Iduronyl and Glucuronyl Residues in Dermatan/Chondroitin Sulhte. J. Biol. Chem. 1999 274 10474- 10480. 92. S. Hemmerich, unpublished.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

15 Novel Variant Pathways in Complex-type Oligosaccharide Synthesis Dirk H. van den Eijnden

15.1 Introduction Complex-type oligosaccharide chains commonly contain Gal(31+4GlcNAc ( N acetyllactosamine, lacNAc) units as building blocks of their so called backbone regions. These backbones connect the core structure, which is directly linked to the protein or lipid aglycon, with the terminal sugars structures. The number of units in these backbones may vary from one per branch, as in most serum glycoproteins, to several, forming polylactosaminoglycan extensions as occur on a number of cellular glycoproteins and glycolipids. The terminal structures carried, such as Lewis’ and sialyl-Lewisx, confer specific biological properties (e.g. blood group activity; mediation of cell adhesion through ligand/lectin interactions) on the molecules and cells exposing them. Originally IacNAc based oligosaccharides were defined to be of the complex-type (as opposed to oligomannosidic type and hybrid type) when they were attached to an N-protein-linked core [ 11. However, IacNAc based structures also frequently occur on 0-linked chains and are found on sphingoglycolipids (lacto-series) as well. Moreover, to date many oligosaccharide structures have been described that are based on a variant backbone building block: the GalNAcPl i 4 G l c N A c ( N , N ’ diacetyllactosediamine, 1acdiNAc) unit (see below). It is proposed here to use the term ‘complex-type’ jointly for structures based on IacNAc, lacdiNAc and other variants, whether linked to N- or 0-proteins, or lipids.

15.2 The lacNAc Pathway of Complex-type Oligosaccharide Synthesis The biosynthesis of lacNAc units of complex-type oligosaccharides is controlled by the common UDP-Gal:GlcNAcp-RP 144-galactosyltransferase ((34-GalT) that can

262

15 Novel Variant Pathways in Complex-type Oligosaccharide Synthesis

act on almost any non-reducing, terminal P-linked GlcNAc residue ([2] and references cited therein). These GlcNAc residues may be introduced to N-linked chains by the action of N-acetylglucosaminyltransferase (GlcNAcT) I, 11, IV, V and VI [3], to 0-linked chains by core 2, 3, 4, and 6 synthases [4, 51, to glycolipids by a lactosylceramide specific P3-GlcNAcT [6], and by aglycon non-specific P3-GlcNAcTs (i-enzymes) [7-91 and several P6-GlcNAcTs acting on linear polylactosaminoglycans [ 10- 131. Subsequent action of enzymes like sialyltransferases, fucosyltransferases, a-galactosyltransferase and sulfotransferases can yield a vast variety of terminal structures (reviewed in [ 141).

15.3 Occurrence and Biology of 1acdiNAc-based Complex-type Oligosaccharides The lacdiNAc variant backbone has now been found in an vast number of complextype N - and 0-protein-, and lipid-linked oligosaccharides (reviewed in [ 151). The group of glycoconjugates carrying such oligosaccharides is diverse and comprises glycoprotein hormones, transport proteins, enzymes, protective glycoproteins, membrane sphingoglycolipids, etc. Glycoconjugates with lacdiNAc based chains occur throughout the animal kingdom and have been described in vertebrates such as mammals and reptiles, but are particularly found in invertebrates such as schistosomes and insects [15] (Table 1). The lacdiNAc backbone may carry terminal sialic acid or fucose residues in structures that are analogous to those occurring on lacNAc chains (Table 1). Interestingly, termination by a2i6-linked NeuAc or sulfate at OH-4 (of GalNAc in 1acdiNAc) is confined to the glycans of mammalian glycoconjugates. By contrast, terminal a1i3-linked Fuc (to form the lacdiNAc analog of the LewisX epitope) occurs in vertebrate as well as in invertebrate glycans. On the other hand, several other terminal motifs are only found in invertebrate glycoconjugates (Table 1). While these invertebrate structures often seem to be highly antigenic in man, some of the terminal 1acdiNAc-based structures on mammalian glycoconjugates have been suggested to be high affinity ligands for animal lectins thus serving a specific biological function. Thus asialo-glycoproteins with an exposed GalNAc (in 1acdiNAc) will be rapidly cleared by the hepatic asialoglycoprotein receptor, in particular since this receptor is known to have a much higher affinity for GalNAc than for Gal [ 161. Furthermore, the 4’-0-sulfatedlacdiNAc epitope, occurring on the glycans of lutropin, is recognized by a receptor on hepatic endothelial and Kupffer cells resulting in a functional rapid half-life of this glycohormone in the circulation [ 171. The a3-fucosylated form of IacdiNAc present on the glycans of recombinant protein C has been suggested to confer antiinflammatory properties on this anticoagulant factor by being a ligand for Eselectin on vascular endothelium [ 181. Similarly, the immunosuppressive effect of glycodelin A has been proposed to be mediated by the recognition of a6-sialylatedlacdiNAc termini by CD22 on B cells, whereas the same epitope and/or a3-

15.4 Biosynthesis

of

lacdiNAc Backbone Units

263

Table 1. Occurrence in different species of various terminal substitution motifs on N . N’-diacetyllactosediamine (1acdiNAc) units of protein- and lipid-linked complex-type glycans (based on references compiled in [ 151). Terminal structure on IacdiNAc-type glycan

Occurrence in

NeuAca2+3GalNAcfi1+4GlcNAc. . NeuAca2i6GalNAcpl i 4 G l c N A c . . S0~--4GalNAcfi1+4GlcNAc.. GalNAcfi1+4[Fucal+3]GlcNAc. .

reptiles mammals mammals mammals, reptiles, amphibians, lower fishes, insects, schistosomes nematodes molluscs insects schistosomes schistosomes

Tyvl+3GalNAcfi1+4~Fucal+3]GlcNAc.. *Gal~l---3GalNAc~l i4GlcNAc. . *GalNAcal+4GalNAcfi1+4GlcNAc. .

*Galal~3GalNAcfi1~4[*Fucal+3]GlcNAc. . *Fucal+3GalNAc~l+41*Fucal +3]GlcNAc. .

The sugars marked by

* may be further substituted.

fucosylated-1acdiNAc on this glycoprotein is believed to be implicated in the contraceptive properties of glycodelin A, through its interaction with a lectin at the surface of spermatocytes or oocytes [ 191.

15.4 Biosynthesis of IacdiNAc Backbone Units As lacdiNAc based chains occur relatively abundant in invertebrate glycoconjugates the first studies on the biosynthesis of this unit were in lower animals. In cercariae of the schistosome Tvichohillzrrrzia ocelluta [20], adult worms of the schistosome Schistosoma mansoni [21], the albumen gland [22] and the prostate gland of the snail Lymnueu stagnalis, and in several insect cell lines [23] a UDPGa1NAc:GlcNAcP-R P 1 +4-N-acetylgalactosaminyltransferase (P4-GalNAcT) has been found with acceptor properties strongly resembling those of mammalian p4GalT and catalyzing a very similar reaction:

-

P4-N-acetylgalactosaminyltransferase UDP-GalNAc

+ GlcNAc-R

Mn’+

-

GalNAcPl

4

4GlcNAc-R + UD P

(1)

P4-galactosyltransferase UDP-Gal + GlcNAc-R

Mn’

’

GalPl

4

4GlcNAc-R

+ UDP

(2)

It should be noted that this enzyme does not share acceptor properties with other described P4-GalNAcTs, such as the enzymes involved in the synthesis of the Sd” blood group determinant, ganglioside GM2 and chondroitin sulfate [20], which

264

15 Novel Variant Pathways in Complex-type Oligosaccharide Synthesis

distinguishes the ‘1acdiNAc synthase’ from these other enzymes. Previously, a (34GalNAcT has been described in pituitary glands that was proposed to serve the specific function of catalyzing the synthesis of lacdiNAc units on the glycans of glycoprotein hormones such as lutropin and thyrotropin, by specifically recognizing a Pro-Xaa-Arg/Lys tripeptide motif neighbouring the glycosylation site of the hormone [24]. Substrate specificity studies on the invertebrate P4-GalNAcTs, and kinetic analysis of the reaction catalyzed, however, indicated that the invertebrate enzymes are also clearly different from the pituitary enzyme and are hormone nonspecific 120, 221. A mammalian version of the hormone non-specific P4-GalNAcT has recently been described in bovine mammary gland [25]. The activity found was demonstrated to be due to a genuine P4-GalNAcT by competition studies, inhibition by a specific antibody and physical separation techniques, and the possibility of a P4-GalT promiscuously utilizing UDP-GalNAc 126, 271 could be discounted [25] (I.M. Van den Nieuwenhof, W.E.C.M. Schiphorst and D.H. van den Eijnden). Also this P4-GalNAcT shows acceptor properties strongly resembling those of P4-GalT [ 531.

15.5 The IacdiNAc Pathway of Complex-type Oligosaccharide Synthesis Studies on the hormone non-specific P4-GalNAcTs have revealed a novel variant pathway of complex-type oligosaccharide synthesis which is denoted as the ‘lacdiNAc pathway’ (Figure 1). Also in this pathway the backbone can be further acted upon by other glycosyltransferases to form terminal structures (Table 2). A common reaction in the bovine mammary gland is the sialylation by u6-sialyltransferase (u6-NeuAcT) to form NeuAcu2+6GalNAcp 1-+4GlcNAcp, which terminus is abundantly present on bovine milk glycoproteins (reviewed in [ 151). Both colostrum [28]and rat liver [29]u6-NeuAcT can catalyze this reaction in vitro very efficiently. In pituitary glands a sulfotransferase has been described which catalyzis the formation of SO4 --4GalNAcP1+4GlcNAcP on pituitary hormones [30].Furthermore, human milk u3-fucosyltransferase (u3-FucT) can catalyze the introduction of Fuc to lacdiNAc to form the lacdiNAc analog of Lewis’ (GalNAcP1-+4[ Fucu1+3]GlcNAc~) [31] which is found on mammalian and schistosomal glycoconjugates [ 151. Recently, schistosomal 1x3-FucTs have been described in adult worms of S. mansoni [32] and cercariae of T. ocellatu 1331, which catalyze the same reaction. In addition to the u3-FucT, cercariae of T ocellatu also contain an u2-FucT which can convert the lacdiNAc analog of LewisX into the difucosylated structure GalNAcp144 [ Fucal-+2Fucul+3]GlcNAc~. This carbohydrate epitope, that is found on schistosomal glycoconjugates [34, 351, is highly antigenic in man. Acceptor properties distinguishes the schistosomal 1x2-FucT from the blood group H and secretor u2-FucTs. In the snail L. staynalis a 03-GalT has been described which catalyzes the formation of Gal~l+3GalNAcpl+4GlcNAcP [36], which structural element occurs on snail haemocyanin glycans 1371. This enzyme is clearly different from the mammalian P3-GalT involved in the synthesis of O-linked core 1

15.5 The ImdiNAc Pathway o j Complex-type Oligosaccharide Synthesis

/

265

\

p4-N-acefy/ga/acfosaminy/fransferase

p4-galacfosylIransferase

\

I/ Galpl->4GlcNAc

GalNAcpl->4GlcNAc

(N-acetyllactosamine)

(N,N-diacetyllactosediamine)

I

prevalent in mammals

prevalent in invertebrates

,E4-N-acety/g/ucosaminy/fransferase

further additions of NeuAca2->3 and 2->6, Galal->3, Fucal->2, S0,-3 and GlcNAcb1->3 to Gal; Fucal->3 to GlcNAc.

.I GICNAC!~I->~G~CNAC (N,N-diacetylchitobiose)

1-

further additions of NeuAcu2->6 (not 2->3), Galpl->3, SO,-4 and (slow) GlcNAcP1->3 to GalNAc; Fucal->3 to GlcNAc.

demonstrated in invertebrates

Figure 1. Three variant pathways of complex-type oligosaccharide synthesis. Terminal P-linked GlcNAc residues on protein- and lipid-linked glycans that may be added by the action of various GlcNAcTs (N-linked chains: GlcNAcT I, 11, IV, V, and VI; 0-linked chains: core 2-, 3-, 4-, and 6synthase; lipid-linked chains: kactosylceramide P4-GlcNAcT; all chains: elongating P3-GlcNAcTs and P6-GlcNAcTs acting on linear chains) are at the start of three different pathways in complextype oligosaccharide synthesis (upper center). Competition for these acceptor sites can take place between glycosyltransferases that each control a separate pathway of complex-type oligosaccharide synthesis. 04-Galactosylation. as catalyzed by P4-GalT, is the first committed step in the IacNAc pathway leading to the formation of the common GalPl i 4 G l c N A c unit commonly occurring on mammalian protein- and lipid-linked oligosaccharides. In the first step of the lacdiNAc pathway, which is controlled by P4-GalNAcT and is prevalent in invertebrates, a GalNAcP1 i 4 G l c N A c backbone unit is formed. Further additions of NeuAc, Fuc, Gal and SOd--moieties can take place in both these pathways. The chitobio pathway is controlled by a P4-GlcNAcT and has been demonstrated in invertebrates. The unit formed by that enzyme (GlcNAcPl i4GlcNAc) might be acted upon by the P4-GalNAcT and the resulting product structure is subsequently channelled into the lacdiNAc pathway.

(GalP1+3GalNAca). Furthermore, in the snail an 1x2-FucT has been described that can convert GalPl+3GalNAcPI 44GlcNAcP into Fucal+2GalP1+3GalNAcPI +4GlcNAcP [38]. Also this enzyme can be distinguished from the other a2-FucTs mentioned before. It is of interest to note that P3-GlcNAcT (i-enzyme) [7] is amongst the mammalian enzymes that can act on lacdiNAc-type chains, be it with a very low efficiency (unpublished results). Consequently, the concerted action of this enzyme and P4-GalNAcT may lead to the formation of repeats of GalNAcP1+ 4GlcNAcP 1'3, which structures, by analogy with polylactosaminoglycans (repeats of GalP 1'4GlcNAcP 1 1 3 ) ; should be named polylactosediaminoglycans. So far, however, such structures have not been found in nature. Other mammalian enzymes

266

I5 Novel Vuriunt Putlzivuys in Complex-type Oligosuccharide Synthesis

like 1x3-NeuAcT and a3-GalT have been found to be incapable of acting on GalNAcPl+4GlcNAcP.

15.6 Other Shared Properties of P4-GalT and P4-GalNAcT A unique property of mammalian P4-GalT is that its specificity can be modified from acting on GlcNAcP-R acceptors to acting on Glc to yield G a l p l i 4 G l c (lactose) by the milk protein a-lactalbumin (a-LA) [39, 401. This typically happens during lactation in the mammary gland. Surprisingly, also the P4-GalNAcT of the snail albumen gland has appeared to be responsive to a-LA [41]. In vitro the enzyme is stimulated by this modifier protein to act on Glc to yield the lactose analog GalNAcPl+4Glc. Similarly, the P4-GalNAcT of bovine mammary gland is induced by a-LA to catalyze this reaction [25].This shared property, along with the very similar acceptor specificities of P4-GalT and P4-GalNAcT suggests that both enzymes show similarity at a molecular level. It has to be pointed out that, in view of the analogous case of the blood group A- (a3-GalNAcT) and B- (a3-GalT) enzyme, which are known to differ only in four amino acids in their respective polypeptide chains [42], the differences between P4-GalT and P4-GalNAcT may be very minor.

15.7 Cloning of a snail UDP-G1cNAc:GlcNAcP-R P4-N-acetylglucosaminyltransferase Based on the anticipated molecular similarity between P4-GalT and 04-GalNAcT it has been attempted to clone the P4-GalNAcT from snail DNA libraries by screening them with a cDNA probe derived from bovine P4-GalT [43]. From a snail prostate library a cDNA was obtained that appeared to code for a protein with the typical domain structure of a mammalian glycosyltransferase, and showing stretches in the deduced amino acid sequence with considerable similarity to the sequence of P4-GalT. Expression of this cDNA, however, resulted in the production of an enzyme that appeared to catalyze the transfer of GlcNAc, rather than Gal (<1%) or GalNAc(-7%), in PI +4-linkage to P-N-acetylglucosaminides [43, 441. This novel P4-N-acetylglucosaminyltransferase(P4-GlcNAcT) had not been described before and its existence could not have been predicted from known oligosaccharide structural data. Notwithstanding the evolutionary distance between snails and mammals the genomic organization of the snail gene encoding the novel P4-GlcNAcT strongly resembles that of the murine gene coding for P4-GalT [45]. For all exons of the murine P4-GalT gene equivalents are present in the snail gene. Particularly the exons that are believed to encode the catalytic domain of P4-GalT [46,47] show a high degree of sequence similarity after translation into amino acids.

15.9 Competition Between Puthwuys

261

Interestingly, the snail gene contains two additional exons that are repeats of exon 6. Deletion of these exons have been shown to result in a P4-GlcNAcT showing a strongly elevated kinetic efficiency for both donor and acceptor substrate at the expense of a decreased specificity for the donor sugar [45]. The similarity in sequences and genomic organization of P4-GalT and P4-GlcNAcT has led to the suggestion that these enzymes are members of a glycosyltransferase family that occurs widely spread in the animal kingdom [15, 43, 481. To date eight members of this P4galactosyltransferase family have been identified (reviewed in [ 14]),but so far no 84GalNAcT has been unequivocally found to be a member through its cloning yet.

15.8 The Chitobio Pathway of Complex-type Oligosaccharide Synthesis The acceptor specificity of the cloned snail P4-GlcNAcT is more restricted than that of mammalian P4-GalT [44]. Also the enzyme is not induced by a-LA to act on Glc. Oligomers of P4-GlcNAc are relatively poor acceptors indicating that this enzyme is not involved in the synthesis of large, chitin-like molecules, but short oligomers of GlcNAc may be formed by the enzyme. Furthermore, both the polypeptide structure and acceptor specificity of P4-GlcNAcT reveal that it neither is implicated in the synthesis of the chitobiose core of N-linked glycans. Preferred substrates appear to be those that contain a P-GlcNAc linked to the OH-6 of Gal or GalNAc residues, as is found in vertebrate blood-group I-active and 0-linked core 2- and 4based oligosaccharides, respectively. Based on these properties it has been proposed that the enzyme functions in a novel, variant pathway of complex-type oligosaccharide synthesis, the ‘chitobio pathway’, which might be common in the snail [43,44] (Figure 1). For example, as snail P4-GalNAcT efficiently acts on GlcNAcPl +4GlcNAc [22], the concerted action of the two glycosyltransferases will lead to the formation of a GalNAc~l+4GlcNAc~1+4GlcNAcstructural element that may occur on snail glycans. Interestingly, this element has been reported as a minor constituent on the glycans of the snake venom serine protease batroxobin [49]. It also is probably present on N-glycans of filarial parasites [50]. In turn the terminal lacdiNAc unit of this element may be further modified as has been described above and is shown in Table 2.

15.9 Competition Between Pathways The three variant pathways of complex-type oligosaccharide synthesis may not operate independently. For instance, when the enzymes (P4-GalT, P4-GalNAcT and P4-GlcNAcT) controlling each of these pathways are expressed simultaneously, they can potentially compete for common terminal, non-reducing GlcNAc acceptor sites (Figure 1). In lactating bovine mammary gland the activity of P4-GalNAcT is

GalNAcPli4GlcNAc

GalNAcP1+4GlcNAc

GalNAc~1+4GlcNAc

GalNAcPl i4GlcNA c

GalNAcPli4GlcNAc

PJ-CIIINAIT ,>rr, 4"U

p3-GalT --

a3-FucT

_i

GalNAcfi1+4[ Fucal-3]GlcNAcP-R

SO4- -4GalNAcP1+4GlcNAcP-R

GlcNAcrjl+3GalNAcPI -4GlcNAc

GalNAcPl+4[ Fucal+2Fucul-3]GlcNAcP-R

GalNAcPl+4GlcNAc~l-3GalNAcPl~4GlcNAc

Fucali2GalPl-3GalNAcNAc~l+4GlcNAcP-R

p4-GalhAcT

a2-FucT

a2-FucT

-

NeuAca2-6GalNAcPl-4GlcNAcP-R

GalP1-3GalNAcP1+4GlcNAcP-R

-

4-SulfoT

uh-NeuAcT

-

Table 2. Conversions in the IacdiNAc pathway of complex-type oligosaccharide synthesis.

m

N

9'

2

9 E'

-7

00

References

269

only about 1-5‘)/0 of that of p4-GalT. Because of the presence of a-LA, however, the activity of the p4-GalT on GlcNAc termini is suppressed [39, 401. By contrast, the activity of p4-GalNAcT on GlcNAc is stimulated by a-LA rather than inhibited [41], which forms the enzymatic basis for the occurrence of lacdiNAc units (in addition to IacNAc units) on bovine milk glycoprotein glycans. The competition between the IacNAc and lacdiNAc pathways in bovine mammary gland thus may be controlled by the concentration of a-LA. In invertebrates the activity of p4-GalT is generally very low or absent [20, 22, 23, 511. Instead, in the snail there may be competition between the lacdiNAc and the chitobio pathways. However, because of the specificity differences of the two controlling enzymes, the competition will be mainly confined to accepting GlcNAcs in Pl i6-linkage. Furthermore, action of the P4-GlcNAcT yields a GlcNAc~l+4GlcNAc product structure on which the p4GalNAcT can act. So far no other invertebrate enzymes have been described that can act on this structure. Consequently, in this instance competition will merely result in glycans that contain an additional GlcNAc residue, whereas the terminal structures will be the same.

15.10 Concluding Remarks In addition to the three pathways discussed other routes may exist. One may be initiated by the addition of a Glc residue to GlcNAc to form GlcP1+4GlcNAcp, a reaction that has been described in the snail [52]. The novel pathways described in this chapter were discovered by studies in invertebrates. Notwithstanding the evolutionary distance they have shown to be eye-opening for our understanding of glycosylation in mammals. This should encourage further studies in invertebrates. References I , R. Kornfeld and S. Kornfeld, Assembly of asparagine-linked oligosaccharides. Annu. Ren. Biochem., 1985, 54, 631-664. 2. I. van Die, A. van Tetering, W.E.C.M. Schiphorst, T. Sato, K. Furukawa and D.H. Van den Eijnden, The acceptor substrate specificity of human P4-galactosyltransferase V indicates its potential function in 0-glycosylation. FEBS Lett., 1999, 450,52-56. 3. H.Schachter, The ‘yellow brick road’ to branched complex N-glycans. GIycobioloqy, 1991, I , 453-461. 4. F. Vavasseur, K. Dole, J. Yang, K.L. Matta, N. Myerscough, A. Corfield, C. Paraskeva and I. Brockhauscn, 0-glycan biosynthesis in human colorectal adenoma cells during progression to cancer. Eur. J. Biochenz., 1994,222, 415-424. 5. T. Schwientek, M. Nomoto, S.B. Levery, G. Merkx, A.G. van Kessel, E.P. Bennett, M.A. Hollingsworth and H. Clausen, Control of 0-glycan branch formation-Molecular cloning of human cDNA encoding a novel p 1,6-N-acetylglucosaminyltransferaseforming core 2 and core 4.J. Biol. Chem., 1999, 274, 4504-4512. 6. D.K. Chou and F.B. Jungalwala, N-acetylglucosaminyltransferase regulates the expression of neolactoglycolipids including sulfoglucuronylglycolipids in the developing nervous system. J. Biol. Chem.: 1993,268, 21727-21733.

270

15 Novel Variant Pathways in Complex-type Oligosaccharide Synthesis

7. D.H. Van den Eijnden, A.H.L. Koenderman and W.E.C.M. Schiphorst, Biosynthesis of blood group i-active polylactosaminoglycans. Partial purification and properties of an UDP-GlcNAc: N-acetyllactosaminide P 143-N-acetylglucosaminyl transferase from Novikoff tumor cell ascites fluid. J. Biol. Chem., 1988,263, 12461- 12471. 8. K. Sasaki, K. Kuratamiura, M. Ujita, K. Angata, S. Nakagawa, S. Sekine, T. Nishi and M. Fukuda, Expression cloning of cDNA encoding a human ~-1,3-N-acetylglucosaminyItransferase that is essential for poly-N-acetyllactosamine synthesis. Proc. Nut1 Acad. Sci. USA, 1997, 94, 14294-14299. 9. D.P. Zhou, A. Dinter, R.G. Gallego, J.P. Kamerling, J.F.G. Vliegenthart, E.G. Berger and T. Hennet, A ~-1,3-N-acetylglucosaminyltransferasewith poly-N-acetyllactosamine synthase activity is structurally related to P-1,3-galactosyltransferases.Proc. Nut1 Acad. Sci. USA, 1999, 96,406-4 1 1. 10. D.H. Van den Eijnden, H. Winterwerp, P. Smeeman and W.E.C.M. Schiphorst, Novikoff ascites tumor cells contain N-acetyllactosaminide 81+3 and PI -6 N-acetylglucosaminyltransferase activity. J. Biol. Chem., 1983, 258, 3435-3437. 11. F. Piller, J.P. Cartron, A. Maranduba, A. Veyrieres, Y. Leroy and B. Fournet, Biosynthesis of blood group I antigens. Identification of a UDP-GlcNAc:GlcNAc Pl-3Gal(-R) pl-6(GlcNAc to Gal) N-acetylglucosaminyltransferase in hog gastric mucosa. J. Biol. Chem., 1984, 259, 13385-13390. 12. Y. Sakamoto, T. Taguchi, Y. Tano, T. Ogawa, A. Leppanen, M. Kinnunen, 0. Aitio, P. Parmanne, 0. Renkonen and N. Taniguchi, Purification and characterization of UDPGlcNAc:Gal~-l-4GlcNAc~l-3*Gal~l-4Glc(NAc)-R (GlcNAc to *Gal) p-l,6N-acetylgIucosaminyltransferase from hog small intestine. J. Bid. Chem., 1998, 273, 27625-27632. 13. J.C. Yeh, E. Ong and M. Fukuda, Molecular cloning and expression of a novel 0-1,6-Nacetylglucosaminyltransferase that forms core 2, core 4, and I branches. J. Biol. Chem., 1999, 274, 3215-3221. 14. D.H. van den Eijnden, On the origin of oligosaccharide species. Glycosyltransferases in action, in: Carbohydrate in Chemistry and Biology (B. Emst, G . Hart and P. Sinay, eds.) Vol. 1, Wiley/ VCH, Weinheim, Germany, 2000, Wiley-VCH. 15. D.H. Van den Eijnden, H. Bakker, A.P. Neeleman, I.M. Van den Nieuwenhof and I. van Die, Novel pathways in complex-type oligosaccharide synthesis-New vistas opened by studies in invertebrates. Biochem. Soc. Trans., 1997, 25, 887-893. 16. Y.C. Lee, Biochemistry of carbohydrate-protein interaction. FASEB J., 1992, 6 , 3193-3200. 17. D. Fiete, V. Srivastava, 0. Hindsgaul and J.U. Baenziger, A hepatic reticuloendothelial cell that mediates rapid clearance of receptor specific for S0~-4GalNAc~1,4GlcNAc~1,2Mana lutropin. CeN, 1991, 67, 1103-1 110. 18. B.W. Grinnell, R.B. Hermann and S.B. Yan, Human protein C inhibits selectin-mediated cell adhesion: Role of unique fucosylated oligosaccharide. Glycobiology, 1994, 4, 221-225. 19. A. Dell, H.R. Morris, R.L. Easton, M. Panico, M. Patankar, S. Oehninger, R. Koistinen, H. Koistinen, M. Seppala and G.F. Clark, Structural analysis of the oligosaccharides derived from glycodelin, a human glycoprotein with potent immunosuppressive and contraceptive activities. J. Bid. Chem., 1995, 270, 24116-24126. 20. A.P. Neeleman, W.P.W. Van der Knaap and D.H. Van den Eijnden, Identification and characterization of a UDP-Ga1NAc:GlcNAcP-R ~1+4-N-acetylgalactosarninyltransferase from cercariae of the schistosome Trichobilharzia ocellata. Catalysis of a key step in the synthesis of N , N '-diacetyllactosediamino (1acdiNAc)-typeglycans. Glycobiology, 1994, 4 , 641-65 1. 21. J. Srivatsan, D.F. Smith and R.D. Cummings, Demonstration of a novel UDP-GalNAc: GlcNAc PI-4-N-acetylgalactosaminyltransferasein extracts of Schistosoma mansoni. J. Parasitol., 1994, 80, 884-890. 22. H. Mulder, B.A. Spronk, H. Schachter, A.P. Neeleman, D.H. Van den Eijnden, M. De JongBrink, J.P. Kamerling and J.F.G. Vliegenthart, Identification of a novel UDP-GalNAc: from the albumen gland and connective GlcNAcP-R PI-4-N-acetylgalactosaminyltransferase tissue of the snail Lymnaeu staynalis. Eur. J. Biochem., 1995, 227, 175-185. 23. I. van Die, A. van Tetering, H. Bakker, D.H. Van den Eijnden and D.H. Joziasse, Glycosylation in lepidopteran insect cells- Identification of a Pl-4-N-acetylgalactosaminyltransferase involved in the synthesis of complex-type oligosaccharide chains. Glycobiology, 1996, 6, 157-164.

References

271

24. P.L. Smith and J.U. Baenziger, A pituitary N-acetylgalactosamine transferase that specifically recognizes glycoprotein hormones. Science, 1988, 242, 930-933. 25. I.M. Van den Nieuwenhof, W.E.C.M. Schiphorst, I. van Die and D.H. Van den Eijnden, Bovine mammary gland UDP-Ga1NAc:GlcNAcP-R 81+4-N-acetylgalactosaminyltransferase is glycoprotein hormone nonspecific and shows interaction with a-lactalbumin. Glycobiology, 1999, 9, 115-123. 26. M.M. Palcic and 0. Hindsgaul, Flexibility in the donor substrate specificity of P1,4galactosyltransferase: application in the synthesis of complex carbohydrates. Glycobiology, 1991, 1, 205-209. 27. K.Y. Do, S.I. Do and R.D. Cummings, a-Lactalbumin induces bovine milk P-1,4-galactosyltransferase to utilize UDP-GalNAc. J. Biol. Chem., 1995, 270, 18447-1 845I . 28. M. Nemansky and D.H. Van den Eijnden, Bovine colostrum CMP-NeuAc:Galp(1+4) GlcNAc-R a(2i6)-sialyltransferase is involved in the synthesis of the terminal NeuAca(2+6) GalNAcp(lh4)GlcNAc sequence occurring on N-linked glycans of bovine milk glycoproteins. Biochem. J., 1992,287, 31 1-316. 29. C.H. Hokke, J.G. Van der Ven, J.P. Kamerling and J.F.G. Vliegenthart, Action of rat liver Gal p1-4GlcNAc a(2-6)-sialyltransferase on Man pl-4GlcNAcp-OMe, GalNAcpl-4GlcNAc~OMe, Glc~1-4GlcNAcp-OMeand GlcNAcpl-4GlcNAc~-OMeas synthetic substrates. Glycoconj. J . , 1993, 10, 82-90. 30. T.P. Skelton, L.V. Hooper, V. Srivastava, 0. Hindsgaul and J.U. Baenziger, Characterization of a sulfotransferase responsible for the 4-0-sulfation of terminal P-N-acetyl-D-galactosamine on asparagine-linked oligosaccharides of glycoprotein hormones. J. Biol. Chem., 1991, 266, 17142-17150. 31. A.A. Bergwerff, J.A. Van Kuik, W.E.C.M. Schiphorst, C.A.M. Koeleman, D.H. Van den Eijnden, J.P. Kamerling and J.F.G. Vliegenthart, Conversion of GalNAcp(1-4)GlcNAcP-OMe into GalNAcb(1-4)[Fuca( 1 -3)]GlcNAcp-OMe using human milk u3/4-fucosyltransferase. Synthesis of a novel terminal element in glycoprotein glycans. FEBS Lett., 1993, 334, 133138. 32. R. DeBose-Boyd, A.K. Nyame and R.D. Cummings, Schistosomu munsoni-Characterization of an a-1-3 fucosyltransferase in adult parasites. Exp. Purusitol., 1996, 82, 1-10, 33. C.H. Hokke, A.P. Neeleman, C.A.M. Koeleman and D.H. Van den Eijnden, Identification of an 1x3-fucosyltransferaseand a novel a2-fucosyltransferase activity in cercariae of the schistosome Trirhobilhurziu ocellatu~Biosynthesis of the Fucal+2Fucwl+3[Gal(NAc)PI+4]GlcNAc sequence. Glycobiology, 1998, 8, 393-406. 34. K.H. Khoo, S. Sarda, X.F. Xu, J.P. Caulfield, M.R. McNeil, S.W. Homans, H.R. Morris and A. Dell, A unique multifucosylated-3GaI”c~-l+4GlcNAc~-It 3 G a l a l - motif constitutes the repeating unit of the complex 0-glycans derived from the cercarial glycocalyx of Schisfosomu munsoni. J. Biol. Chem., 1995, 270, 17114- 17123. 35. K.H. Khoo, D. Chatterjee, J.P. Caulfield, H.R. Morris and A. Dell, Structural characterization of glycosphingolipids from the eggs of Sdiistosomu mansoni and Schistosoma juponicunz. Glycohiology, 1997, 7, 653-661. 36. H. Mulder, H. Schachter, M. de Jong Brink, J.G. Van der Ven, J.P. Kamerling and J.F.G. Vliegenthart, Identification of a novel UDP-Gal:GalNAc~1-4GlcNAc-R~1-3-Galactosyltransferase in the connective tissue of the snail L J V I ~ L W sfugnulis. U Eur. J. Biochem., 1991, 201, 459%465. 37. J.A. Van Kuik, R.P. Sijbesma, J.P. Kamerling, J.F.G. Vliegenthart and E.J. Wood, Primary structure determination of seven novel N-linked carbohydrate chains derived from hemocyanin of Lymnaea stagnalis. 3-0-methyl-D-galactose and N-acetyl-D-galactosamine as constituents of xylose-containing N-linked oligosaccharides in an animal glycoprotein. Eur. J. Biochem., 1987, 169, 399-41 1. 38. H. Mulder, H. Schachter, J.R. Thomas, K.M. Halkes, J.P. Kamerling and F.G. Vliegenthart, Identifieation of a GDP-Fuc:Galpl-3GalNAc-R (Fuc to Gal) al-2-fucosyltransferase and a GDP-Fuc:GalPl-4GlcNAc (Fuc to GlcNAc) al-3-fucosyltransferase in connective tissue of thc snail Lymnaeu stagnalis. Glycoconj. J . , 1996, 13, 107-1 13. 39. U. Brodbeck, W.L. Denton, N. Tanahashi and K.E. Ebner, The isolation and identification of the B protein of lactose synthetase as a-lactalbumin. J. Biol. Chem., 1967, 242, 13911397.

272

15 Novel Variant Pathways in Complex-type Oligosacchuride Synthesis

40. K. Brew, T.C. Vanaman and R.L. Hill, The role of a-lactalbumin and the A protein in lactose synthetase: a unique mechanism for the control of a biological reaction. Proc. Nut1 Acad. Sci. USA, 1968,59,491-497. 41. A.P. Neeleman and D.H. Van den Eijnden, a-Lactalbumin affects the acceptor specificity of Lymnaea stagnalis albumen gland UDP-Ga1NAc:GlcNAcp-R ~1+4-N-acetylgalactosaminyltransferase-Synthesis of GalNAc~1+4Glc. Proc. Nut/ Acud. Sci. USA, 1996, 93, 1011110116. 42. F. Yamamoto, H. Clausen, T. White, J. Marken and S. Hakomori, Molecular genetic basis of the histo-blood group ABO system. Nuture, 1990, 345, 229-233. 43. H. Bakker, M. Agterberg, A. van Tetering, C.A.M. Koeleman, D.H. Van den Eijnden and I. van Die, A Lymnaea stagnalis gene, with sequence similarity to that of mammalian p1+4galactosyltransferases; encodes a novel UDP-G1cNAc:GlcNAcP-R Pl+4-N-acetylglucosaminyltransferase. J. Biol. Chem., 1994, 269, 30326-30333. 44. H. Bakker, P.S. Schoenmakers, C.A.M. Koeleman, D.H. Joziasse, I. van Die and D.H. Van den Eijnden, The substrate specificity of the snail Lymnaea stagnalis UDP-GlcNAcGlcNAcpR j34-N-acetylglucosaminyltransferase reveals a novel variant pathway of complex-type oligosaccharide synthesis. Glycobiology, 1997, 7, 539 -548. 45. H. Bakker, A. van Tetering, M. Agterberg, A.B. Smit, D.H. Van den Eijnden and I. van Die, Deletion of two exons from the Lymnaeu stagnalis p 1 +4-N-acetylglucosaminyltransferase gene elevates the kinetic efficiency of the encoded enzyme for both UDP-sugar donor and acceptor substrates. J. Bid. Chem., 1997, 272, 18580-18585. 46. E.E. Boeggeman, P.V. Balaji and P.K. Qasba, Functional domains of bovine P-1,4-galactosyltransferase. Glycoconj. J., 1995, 12, 865-878. 47. H.Y. Zu, M.N. Fukuda, S.S. Wong, Y. Wang, Z.D. Liu, Q.S. Tang and H.E. Appert, Use of site-directed mutagenesis to identify the galactosyltransferase binding sites for UDP-galactose. Biochem. Biophys. Res. Commun., 1995, 206, 362-369. 48. I. van Die, H. Bakker and D.H. Van den Eijnden, Identification of conserved amino acid motifs in members of the 81+4-galactosyltransferase gene family. Glycobiology, 1997, 7 (number 8), R 5-R 8. 49. G. Lochnit and R. Geyer, Carbohydrate structure analysis of batroxobin, a thrombin-like serine protease from Bothrops moojenivenom. Eur. J. Biochem., 1995, 228, 805-816. 50. S.M. Haslam, K.M. Houston, W. Harnett, A.J. Reason, H.R. Morris and A. Dell, Structural studies of N-glycans of filarial parasites. Conservation of phosphorylcholine-substituted glycans among species and discovery of novel chito-oligomers. J. Biol. Chem., 1999,274, 20953-20960. 51. C.A. Rivera Marrero and R.D. Cummings, Schistosomu munsoni contains a galactosyltransferase activity distinct from that typically found in mammalian cells. Mol. Biochem. Purasitol., 1990, 43, 59-67. 52. I. Van Die, R.D. Cummings, A. Van Tetering, C.A.M. Koeleman and D.H. Van den Eijnden, Identification of a novel UDP-G1c:GlcNAcp-R glucosyltransferase in Lymnuea stugnulis. Glycoconj. J., 1997, 14, 29. 53. The lactose analog GalNAcP1-4Glc is present in bovine colostrum. Enzymatic basis for its occurrence: see FEBS lett., 1999, 459, 377-380.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

16 Control of Mucin-Type O-Glycosylation: O-Glycan Occupancy is Directed by Substrate Specificities of Polypeptide GalNAc-Transferases He& Hassan, Eric P. Bennett, Ulla Mandel, Michael A . Hollingsworth, and Henrik Clausen

16.1 Introduction Mucin-type O-glycosylation is one of the most abundant forms of protein glycosylation and found in all eukaryotic cells. O-Glycans are widely distributed on mucins, in glycoproteins with mucin-like domains, and at single or clustered sites in glycoproteins in general. O-glycosylation exerts a number of biological effects including direct structural and functional properties for the carrier protein and involvement in molecular adhesion events [ 1, 21. The biosynthetic pathways of 0glycosylation and the structural diversity of O-glycans have been extensively characterized [ 3 ] .However, this type of protein glycosylation is poorly defined with respect to where O-glycans are specifically attached in proteins. The rules that govern selection of sites for O-glycan attachment and the particular structures of O-glycans at specific sites are also not well described. Factors contributing to this are i) the analytical techniques for identification and characterization of O-glycan sites have only recently advanced to allow analysis of mucins (for review see [4-71); and ii) analysis of the sequence context of identified O-glycan sites did not reveal a clear sequence motif that can be used for prediction of O-glycosylation (8-10). Furthermore, analysis of the substrate specificity of the polypeptide GalNAc-transferase activity catalyzing the transfer of GalNAc to Ser/Thr amino acids did not yield conclusive information regarding the importance of primary sequence of acceptor sites (for review see [ l l , 121). One major conceptional limitation regarding interpretation of past studies of regulation of O-glycan attachment sites has been the dogma that O-glycan initiation was ruled by a simple unified mechanism in which one polypeptide GalNAc-transferase activity (possibly two or three) transferred GalNAc to appropriate peptide acceptor sequences. Here, we discuss advances made in the last five years which show that the initiation step of O-glycosylation is a far more complex system than previously recognized. Contrary to past predictions, O-glycan initiation is a highly regulated process, where the substrate specificities and kinetic properties of mem-

214

16 Control of Mucin-Type 0-Glycosylution

bers of a large family of polypeptide GalNAc-transferases are predicted to be the major factor determining 0-glycan occupancy.

16.2 The Mammalian UDP-GalNAc: Polypeptide GalNAc-Transferase Gene Family Mucin-type 0-glycosylation involves an initiation step in which polypeptide GalNAc-transferase activity transfers GalNAc from UDP-GalNAc to the hydroxyl amino acids serine and threonine in proteins [ 131. This generally begins in the cis-Golgi, but may occur throughout the Golgi stacks [14, 151. The processing step where glycosylated sites, GalNAccll-0-SerlThr, are further processed occur throughout the Golgi apparatus and are mediated by numerous glycosyltransferases that form various mucin-type core structures and terminal modifications [3]. The initiation step is controlled by a large family of distinct polypeptide GalNActransferases [ 121; at least eight distinct mammalian isoforms have been cloned and characterized in some detail [ 16-22, 54, 64, 681. Several additional putative GalNAc-transferase genes have been identified, but the enzymatic functions of the encoded proteins have not been established. A summary of the structural features of seven human and one rat GalNAc-transferases are presented in Figure 1. The GalNAc-transferase genes are highly conserved during evolution. since bovine, rat, mouse, or pig orthologues of these (when available) exhibit 90-98% amino acid sequence identity. GalNAc-transferases are predicted to be type I1 transmembrane proteins similar to other resident Golgi glycosyltransferases 1231. GalNAc-transferases exhibit relatively high sequence similarities in the central regions containing the putative catalytic domains, but differ markedly outside in the C-terminal regions and especially in the N-terminal regions. Ten to twelve cysteine residues found throughout the putative catalytic domains and C-terminal regions of all GalNActransferases are conserved throughout evolution, suggesting that the overall structures of GalNAc-transferases are similar. One major structural variation found among isoforms is the length of the stem regions which is predicted to affect the extent of protrusion of the catalytic domain into the Golgi lumen provided by (Figure 1). Structural information on the catalytic domains of GalNAc-transferases is not yet available. A number of residues in the central putative catalytic domain that are conserved among isoforms are critical for enzyme function as determined by site directed mutagenesis 124,251. Hagen et al. 1251 proposed that the catalytic unit (central part of the luminal region) consists of two domains with a common structural fold resembling the lac repressor protein. One domain (GT1 motif, glycosyltransferasemotif-1 ) contains a sequence motif shared among most glycosyltransferases including the DxD-like (DxH in polypeptide GalNAc-transferases) sequon [26], and the other domain (Gal/GalNAc-T motif) contains a sequence motif shared with a large family of P4galactosyltransferases [25]. This conserved motif was shown to consti-

Y

Y

Y

Y

Y

29

p

Y y

Y-+

28

++

Y

I

C C

'c C C

C

kc'c

catalytic unit

C

k k

t 'c

Y

Y

c k

5

C

'c C

C

'c k C

'c C

;

c

609

6>8

622

930

bs79

ricin-like lectin motif

I a l B I ? ]

C C

'c

c c c c c c

by;'

cYC c

s7i

;YYss9

;'> ;

;; ;; 2 ; ;; ;

;

t c'c 'k

;

;;.

c;

GaUGaiNAc-motif

;;; ;

b b Y 'c'c

c

; c E t c c

N-terminal cytoplasmic tallitransmembrane anchor/ stem

'+

37 41

-

14

192

+++

kY ; ;c

+;:+

fG -

= 3 5kb

245kb

* 50kb

e44kb

= 60kb

6

?.

10

1

10

= 36kb

11 16

34kb 4 Ikb

= 45kb

ii

1~

7q36

'q'l

12ql3

12q21 3-q22

2q24-q31

lq4i-q42

18qi2-qZI

65,637

71,129

64,936

63,274

67,044

60,028

59,914

Y12435

A5002744

YO8565

AF049344

YO8564

X92689

X85019

X850iR u41514

IJH8

UH7

UH6

UH5

UH4

UH3

Transcnpt Coding Chromosome M w Accession MAb size exons Predicted no.

unpub

54

22

21

20

18,34

17,34

17,34 33

Ref

Figure 1. The mammalian polypeptide GalNAc-transferase gene family Depicted structures of eight GalNAc-trdnsferdse proteins dre shown with alignment of the central conserved region containing the putative catalytic domains GalNAc-transferases are type I1 transmembrane proteins with a short N-terminal cytoplasmic domain (4-19 aa), a putative trdnSmembrdne signal sequence (17-23 dd) a stem region of highly variable length (from 55 to 416), a large lumindl domain containing the catalytic unit that includes the G T l dnd Gal/GalNAc-T sequence motifs (approx 340 aa), and the putative lectin domain (approx 130 aa) The arrows indicate the N-terminus of functionally active forms of the soluble recombinant GalNAc-transferases including residue numbers C = cysteine residues in the protein sequences where 10-12 Cys are conserved among the GalNAc-transferase family (indicated by dotted lines) The transmembrane regions are indicated by black boxes with numbers of flanking residues Indicated by Y are potential Nglycosylation sites and the position5 of the G T l and Gdl/GdlNAc-T motifs are indicated by solid lines.

I

hCalNAc-T8

hCalNAc-T7

hGalNAc-Th

rCalNAc-TS

hCalNAc-T4

hCalNAc-T3

hCalNAc-T2

hCalNAc-T1

7 4i

6

8 P

9

3

g

$

3 n

b

5,

276

16 Control of' Mucin-Type O-Glycosylution

tute part of a groove containing the catalytic site of (34Gal-TI in x-crystallography studies. Several conserved aspartic acid and glutamic acid residues in this motif are shown to be important for the catalytic activity of polypeptide GalNAc-T1 as well [25]. The C-terminal region exhibits similarity to a ricin-like lectin motif [27, 281, and recently, Hagen et al. [25] demonstrated that mutations believed to be disruptive of a lectin function in this region of GalNAc-T1 did not affect enzyme activity. Thus, the C-terminal domain is apparently not important for simple catalysis with peptide acceptor substrates, and demonstration of a putative lectin function of this domain awaits further investigation. The high number of polypeptide GalNAc-transferases in mammals leads us to consider several scenarios as to the functions of these. The two extremes hypothesis are: i) the isoforms have similar functions and provide extensive genetic redundancy (back-up); or ii) the isoforms have distinct functions by virtue of their kinetic properties and/or their expression patterns. As will be discussed in the following sections, emerging evidence supports the latter hypothesis including the structural and evolutionary features of GalNAc-transferase genes, the distinct kinetic properties of isoforms, and the regulation of expression cell and tissue of isoforms.

16.3 The GalNAc-Transferase Gene Family is Evolutionarily Old Hagen et al. [29, 301 deinonstrated at least nine distinct genes of the GalNActransferase family in Cuenorlzuhditis eleguns. Three genes have GalNAc-transferases activity with an acceptor substrate, EA2; derived from rat submandibular mucin [ 3 11. More detailed characterization of the kinetic properties and substrate specificities of the C. eleguns isoforms is required. Sequence conservation between C. elegans and mammalian genes is high in the catalytic domains. The sequence divergence between mammalian genes and among C. eleguns genes are greater than that found between the species among some of the genes, as shown in Figure 2 by a dendrogram based on a ClustalW multiple sequence alignment. The full size of the mammalian GalNAc-transferase gene family is still unknown, but it is likely to be considerably larger than the C. eleguns family. If individual genes and associated functions have been conserved throughout evolution from C. eleyuns to mammals, this strongly supports the hypothesis that there are distinct functions of members of the GalNAc-transferase family. Currently, Gly3 is grouped with GalNAc-T1, Gly4 is grouped with GalNAc-T2, and Gly5(u,b,c)are grouped with GalNAc-T4. The C. eleguns genes have GalNAc-transferase activities with the EA2 peptide substrate [ 301, and the corresponding mammalian GalNAc-transferases also exhibit activity with this substrate [ 19, 20, 321. Gly7 which is grouped with GalNAc-T7, did not exhibit activity with the EA2 substrate, but the human GalNAc-T7 acts on EA2 only after initial GalNAc-glycosylation by other GalNActransferases [54]. Thus, a more complete comparison of the kinetic parameters of these enzymes is required. Information on the genomic organization of several mammalian and C. elegnns

16.3 The GulNAc-Transferuse Gene Fumily is Evolutionuvil~yOld Gly7 hT7 GlylO Glyll Gly8

277

9 I

hT8

hT3 hT6 hT4 Gly5A hT 1 Gly3 Gly4 hT2 Gly6A rT5 Figure 2. Phylogenetic tree of mammalian and C'. eleguns GalNAc-transferase genes. The mammalian genes are the same as used in Figure 1. The nine C. eleguns genes represent seven distinct genes reported in [30], and two novel genes designated GlylO and Glyll (GenBank accession numbers AL021493 and AL033514, respectively).

genes is available [12, 20, 22, 33, 341. The coding regions of five C. elegans genes Gly3, Gly6, Gly8, GlylO, and G l y l l , are placed in 6, 11, 9, 8, and 7 exons, respectively. Most mammalian GalNAc-transferases have similar complex organizations, and the coding regions of GalNAc-T1, -T2, -T3, -T6, -T7, and -T8, are contained in 11, 16, 10, 5 , and 6 exons, respectively (Figure 1). Analyses of positions of intron/ exon boundaries among genes suggests that some boundaries are shared among different mammalian genes and with C. elegans genes [ 12, 341. The most remarkable example is the organization of GalNAc-T3 and -T6, which both have the coding regions placed in 10 exons [22]. Conservation of intron/exon boundaries indicate evolutionary relationships by gene duplication and subsequent divergence. When shared intron/exon boundaries among genes is combined with analysis of sequence similarity, the strength of predictions of evolutionary relationships increases. In the case of GalNAc-T3 and -T6, the genes exhibit high sequence similarity throughout the coding regions, while similarities among other GalNAc-transferase genes are limited to the central putative catalytic domains. Thus, GalNAc-T3 and -T6 appear to represent a late gene duplication event (in evolutionary terms). GalNAc-T4 is unique in that the coding region is placed in a single exon [20]. None of the C. elegans genes have so far been found to have such a simple genomic organization. If the GalNAc-transferase gene family developed through gene duplication events, GalNAc-T4 may represent an ancestral gene. Another hypothesis is that GalNAc-T4 originated from a retroposon event and was derived from another GalNAc-transferase. Such events often lead to pseudogenes, and one such processed pseudogene predicted to have originated from GalN Ac-T1 was cloned and characterized [33]. GalNAc-T4 is expressed and encodes a functional protein

278

16 Control o j Mucin-Type O-Glycosylution

with unique properties, and it does not show particularly high sequence similarity to any other reported genes. However, a novel gene encoded by multiple exons shows high sequence similarity to GalNAc-T4 [65], and could potentially have given rise to GalNAc-T4 in a retroposon event. The mammalian GalNAc-transferase gene family contains several high-similarity gene pairs, which are defined by sequence similarity throughout the coding regions and conservation of all intron/exon boundaries. A high-similarity gene for GalNAc-T2 has been cloned, but the function of this gene is not established yet. As will be discussed later the kinetic properties of GalNAc-T3 and -T6 are very similar, but their expression patterns differ. Thus, while these clearly establish genetic redundancy for some polypeptide GalNAc-transferases, they do not provide full functional redundancy. High-similarity gene pairs have not been found in C. eleguns. It is probable that the expansion of the mammalian GalNAc-transferase gene family involved gene duplications of distinct isoforms which lead to a series of subfamilies. In order to evaluate functions of specific isoforms, it is therefore important to consider potential genetic and functional redundancies provided by members of such subfamilies of GalNAc-transferases. The first attempted knock-out experiment of a GalNAc-transferase, GalNAc-T1 [35], resulted in the targeting of what appear to be a high-similarity copy gene. The targeted gene has not been fully cloned or characterized functionally, and it may represent a pseudogene as attempts to clone the 5’ region have failed. The genomic organization of the available coding region of the targeted gene is identical to that of GalNAc-T1 [65]. Disruption of this gene in mice did not lead to a phenotype; however, it is not possible to evaluate the significance of this result until the integrity and functionality of the targeted gene have been characterized.

16.4 The Kinetic Properties of GalNAc-Transferase Isoforms are Different Extensive studies on the substrate specificity of polypeptide GalNAc-transferase activity in extracts of cells and organs has been carried out in the past (for comprehensive review of these data refer to [ l l ] and references herein). These studies provided insight into the sequence of acceptor substrates for GalNAc-transferases, and indicated that the enzyme activities measured have preference for acceptor sites with clusters of Ser/Thr residues and proline residues in proximity. Furthermore, it was found that Ser sites were not used or only poorly used [16, 36-38]. Importantly, these studies did not reveal the existence of multiple GalNAc-transferases or that there were differences in the GalNAc-transferase activities from different cells or organs. During the purification of a GalNAc-transferase activity (measured with a peptide acceptor substrate derived from the tandem repeat of human MUC2) found in human placenta, a distinct activity measured with a peptide substrate derived from HIVrrrs gp120 was separated [39]. The purified enzyme was shown to represent GalNAc-T2 [ 171, while the separated activity may have originated from GalNAc-T3 or -T6 [ 18, 221. Subsequently, it was possible to demonstrate differences in

16.4 The Kinetic Properties of GulNAc-Trcrrzsfiruse Isofiwns are Dijerent

279

substrate specificities of GalNAc-transferase activities from different organs with selected peptide substrates [ 17, 391. Given our present knowledge of the number of distinct mammalian GalNActransferases, it is clear that detailed analysis of the kinetic properties of the individual isoforms in appropriate systems is required to fully understand the factors that determine the initiation of 0-glycosylation. As discussed in detail below, several factors influence such analysis, and it is not straightforward to reach unequivocal understanding of the in vivo functions of individual GalNAc-transferase isoforms. It is, however, clear that there are differences in functions and these are found both by in vitro and in vivo analysis. Several novel clever approaches to study GalNActransferase activities in vivo have been developed and future analysis with these should provide a better understanding. The following discussion is limited to the functions of individual isoforms as information derived from total GalNActransferase activities in vitro or in vivo has been reviewed in the past.

16.4.1 Lessons from in vivo Analysis of GalNAc-transferase Substrate Specificities For obvious reasons it is desirable to study the kinetic properties of GalNActransferase isoforms in situ in the appropriate Golgi environment. Several factors may contribute to their biological functions, including potential interactions with other GalNAc-transferases or Golgi-resident proteins, modulation by co-factors, and the relative kinetic properties compared to other glycosyltransferases using the same substrates. Nehrke et al. [40] have designed an in vivo system where a small cDNA construct, encoding a secretion signal sequence, a short variable 0-glycan acceptor sequence, a labeling site for tyrosine phosphorylation and an antibody tag for immunoprecipitation, can be expressed in different cell types and the secreted product isolated and characterized. The system is convenient in allowing direct analysis of 0-glycosylation of the inserted acceptor sequences by SDS-PAGE. In one study, this system was used to compare the in vitro activity of purified bovine GalNAc-T1 with that of the in vivo capacity of COS7 cells. An in vivo acceptor sequence from von Willebrand factor (PHMAQVTVGPGL) was initially examined with systematic substitutions by Ile, Ala, Pro and Glu residues, and marked negative effects of adjacent charged residues and positive effects of adjacent Pro residues were found for in vitro 0-glycosylation with purified GalNAc-T1 [41]. Analysis of the same acceptor sequence in COS-7 cells in vivo, using the secreted reporter construct, revealed that the capacity for 0-glycosylation in these cells was much broader than that of GalNAc-T1 [40]. Specifically, adjacent charged residues or Ser substitution for Thr at the acceptor site had little or no effect on levels of glycosylation. Charged residues at positions - l and +3 reduced 0-glycosylation the most. Although, this may indicate that in vitro assessment of substrate specificities of GalNAc-transferases fail to reflect the in vivo capacity as previously suggested (91, it is clear that the in vitro activities of GalNAc-TI is compared with the entire repertoire of GalNAc-transferases expressed in the host cell used for expression. The repertoire and kinetic properties of GalNAc-transferases expressed in COS7 cells used in the above experiments are not known at present, but multiple genes are expressed [42]. In a subsequent study, Nehrke et al. [42] demonstrated that the

280

16 Control of Mucin-Type O-Glycosylution

capacity for O-glycosylation of six reporter sequences was different in three host cells (COS-7, L6, lO(3) cells), thus providing in vivo evidence for cell-type specific O-glycosylation. Analysis of the in vivo functions of individual GalNAc-transferase isoforms is more complex. In situations where a unique substrate has been identified for a GalNAc-transferase isoform analysis is possible with the system designed by Nehrke et al. [40]. One unique substrate for GalNAc-T3 and -T6 is the v3-100~ sequence of HIV gp120 [18, 221. Using this model substrate, Nehrke et al. [43] showed that O-glycosylation of the HIV substrate was dependent on co-expression of GalNAc-T3 in the host cell. The unique requirement for GalNAc-T3 was abolished if the positively charged Lys residue at position + 3 (GRAFVTIGK) in the acceptor sequence was substituted for Pro. Thus, a single amino acid substitution in the acceptor sequence resulted in O-glycosylation by endogenous GalNActransferases of the host cell COS-7. COS-7 cells express GalNAc-TI and -T2, and in vitro analysis of GalNAc-TI showed that this enzyme was capable of utilizing the Pro substituted peptide substrate. These data demonstrate one example where in vitro and in vivo functions of a GalNAc-transferase isoform show correlation. It should be noted that this in vivo system may be biased by the fact that it is short acceptor sequences and not native proteins are serving as substrate. Rottger et al. [ 151 have developed a different in vivo system, which is particularly suitable for analyses of the functions of individual GalNAc-transferase isoforms. This in situ system is based on relocation of chimeric GalNAc-transferases and chimeric substrates to the ER in cells. de Haan et al. [44] used this system and showed that OST7- 1 cells exhibited a broad O-glycosylation capacity for sequence variations at a single O-glycosylation site in the N-terminal sequence of mouse hepatitis membrane protein (NH2-MSSTTQAP-). However, when individual isoforms of GalNAc-transferases were analysed in situ by relocation to ER, distinct differences were found. Three GalNAc-transferases (GalNAc-T1, -T2, and -T3) were studied and distinct differences in their capacity for O-glycosylation of CD8 and sequence variations of mouse hepatitis membrane protein were demonstrated [15, 441. Interestingly, these studies showed that Ser substitution of the Thr acceptor site in -MSSTTQAP- was only accepted by GalNAc-T3 and not GalNAc-T1 or -T2. In contrast, substitution of the + 3 Pro did not affect GalNAc-T1, but GalNAc-T2 and -T3 did not glycosylated this substrate [44]. This in vivo system allows for analysis of individual isoforms; however, one counts that the enzymes are functioning similarly in an environment distinct from Golgi. The available data from in vivo assessment of the functions of GalNActransferase isoforms is fragmentary. Further correlation of data obtained in vitro and in vivo are needed. Undoubtfully, application of these and other in vivo systems are critical for further progress in the field. 16.4.2 Lessons from in vitro Analysis of the Acceptor Substrate Specificities of GalNAc-transferase Isoforms

In vitro analysis of GalNAc-transferases involve a number of potential obstacles. The first obstacle is the enzyme source. Assays are generally done with recombinant,

16.4 The Kinetic Properties of’ GulNAc-Trunsferuse Isoforms are Diflerent

28 1

soluble, secreted forms expressed in insect cells or COS-7 cells. Only two native GalNAc-transferases, GalNAc-T 1 and -T2, have been purified to apparent homogeneity as soluble forms and their kinetic properties studied 116, 17, 32, 36, 38, 391. The majority of information in the literature is derived from recombinant enzymes. The recombinant forms were designed to resemble the purified forms in the case of GalNAc-T1 and -T2, and the kinetic properties have been comparable. Other recombinant GalNAc-transferases were constructed to include the entire stem regions, and information of potential processing of these has not been obtained. In most cases, expressed GalNAc-transferases were assayed directly with the culture medium of infected/transfected cells. Interestingly, this has proven to be a problem with GalNAc-T4 and -T7, two enzymes that show preference for partially glycosylated GalNAc-peptide substrates [20, 54, 681. The activities of these enLymes are only clearly demonstrable after some purification of the proteins. The second obstacle is the acceptor substrate. The design and length of acceptor substrate peptides are important for the in vitro assessment of O-glycosylation capacity of GalNAc-transferases. The number of potential peptide acceptor substrates is practically unlimited. Most studies have focussed on peptide sequences derived from known 0glycosylation sites, and variations in length and sequence in such model systems have been analysed. The length of acceptor peptides is clearly important and kinetic properties generally increase with increased flanking sequences of acceptor sites 145-471. Thr residues placed in the N-terminus can be glycosylated, but with poor kinetics [45, 48, 491. Detailed studies of GalNAc-transferase isoform activities with overlapping peptides derived from the tandem repeat sequences of M UC 1 have clearly demonstrated the significance of the flanking sequences and the peptide design 146, 471. A third obstacle is potential co-operative effects of GdNAc-transferase isoforms. A direct co-operative effect of different GalNAc-transferases has not been identified to our knowledge, however, this may simply be due to present experimental limitations. For example, the finding that GalNAc-transferase extracts show no or only poor activity for Ser glycosylation sites, may be related to adverse effects of multiple GalNAc-transferase isoforms in extracts, as several isolated GalNActransferases show good activities with these substrates 139, 491. Finally, recent findings reveal that some GalN Ac-transferase activities require partially GalNAc glycosylated peptide substrates for activity 120, 54, 681. Despite the apparent obstacles of in vitro assays, analysis of the kinetic properties by in vitro assays have been highly informative, demonstrating that GalNActransferase isoforms have different kinetic properties, and that the primary sequence context of acceptor sites plays a major role for the functions of these enzymes. General features of kinetic properties of GalNAc-transferase isoforms with (glyco)peptide acceptor substrates include the following: Isoforms may have distinct acceptor substrate specificities

Unique acceptor substrate sites have been identified for several GalNAc-trdnsferase isoforms or subfamilies (Table 1). GalNAc-T2 is the only GalNAc-transferase found to catalyse glycosylation of a peptide sequence from human choriogonadotropin p-chain [ 181, and significantly GalNAc-T2 transfers GalNAc to the same three Ser residues found to carry O-glycans in the native glycoprotein [ S O ] . The

282

16 Control of’ Mucin-Type 0-Glycosylution

Table 1. Unique acceptor substrate specificity of human recombinant GalNAc-transferases (K, (mM)).

npp

Peptide

Sequence

h”’-TI h-T2 h-T3 h-T4 h-T6 h-TI

VTHPGY HIVllrbgpl20 hCG-a GalNAc4TAP24 Prion-a PSGL- I b OSM fragment

Ac-PFVTHPGY D Ac-GRAFVTIGK PRFQDSSSSKAPPPSLPSPSRLPG TAPPAHGVTSAPDTRPAPGSJAPP‘’ KQHTVTTTTKGEN QATEYEYLDYDFLPETEPPEM LSESTTQLPGGGPGCA

NAb’ NA NA NA NDd’ ND 0.30

NA NA 1.20 NA NA ND NA

3.23 0.61 NA NA 0.58 ND 1.61

NA NA NA 0.09 NA 0.02 NA

2.46 0.58 NA NA 0.63 ND 1.60

NA NA NA NA NA NA NA

h; Human recombinant GalNAc-transferases. h’NA; not applicable, indicates that no incorporation is observed with the substrate even after prolonged incubations (24 h). ‘) GalNAc4TAP24 represents the TAP24 peptide terminally glycosylated with GalNAc-T2, and GalNAc attachment site are underlined [49]. d’ND, not determined, incorporation of GalNAc into the peptide monitored by CE and MALDITOF was observed. but apparent K, could not be obtained due to required quantities of peptide. VTHPYG is derived from fibronectin, HIVlllhgpl20 is dcrived from the V3 loop of HIVgpI[rB120, hCG-P is derived from the subunit of human chorionic gonadotropin, TAP24 is derived from the tandem repeat of MUCI. Prion-a is derived from the human Prion protein, PSGL-lb is derived from the N-terminus of human PSGL-I . OSM fragment is derived from ovine submaxillary mucin.

subfamily, GalNAc-T3/T6, both can glycosylate a single Thr acceptor site in fibronectin that is only utilised in fetal and cancer tissues [51]. Since normal plasma fibronectin is not 0-glycosylated at this site, it was possible to analyze the activity of GalNAc-T3 with the mature glycoprotein. Using an antibody defining the presence of an 0-glycan at this particular site, it was demonstrated that GalNAc-T3, and not GalNAc-TI or -T2, could transfer to the unique acceptor site [49]. Thus, the substrate specificity with GalNAc-T3 determined in vitro with short synthetic peptides was maintained in the mature folded, N-and 0-glycosylated, and dimerized glycoprotein. GalNAc-T3 and -T6 also exhibit unique specificity for a single Thr site in the HIV~IIB gp120 V3-100~sequence [ 18, 221. Although 0-glycosylation has been identified on gp120, the actual positions of 0-glycans have not been characterized [52]. Interestingly, the 0-glycosylation site in the V3-100~is at a known T-cell epitope, and it was recently shown that a GalNAc residue at the putative acceptor site in this peptide influenced the conformation of a 15-mer v 3 -1 0 0 ~peptide, and inhibited the activity of cytotoxic T-cells directed to this epitope [53].GalNAc-T3 and -T6 also show distinct substrate specificity for a cluster of Thr residues in the prion protein by in vitro analysis with synthetic peptides [22]. This cluster of acceptor sites is located in a 36-aa disulphide loop similar to the V3-100~in gp120 of HIV, and it is flanked by two utilized N-linked glycosylation sites. The prion loop region is exposed and it would be predicted that 0-glycosylation would occur in the region if the appropriate GalNAc-transferases were expressed. GalNAc-T4 has unique substrate specificities for two sites in the MUCl tandem repeat sequence, and the activity of GalNAc-T4 for these sites is dependent on prior 0-glycosylation

16.4 The Kinetic Properties

of' GulNAc-Transjevuse Isojovms m e Different

283

at other sites in the repeat (this is discussed in more detail below) [64]. GalNAc-T4 also exhibits nearly exclusive specificity for a single utilized Thr glycosylation in the P-selectin-ligand-glycoprotein-1(PSGL-1) [20, 551. It was predicted in early studies that different GalNAc-transferase isoforms would catalyse transfer to Ser and Thr acceptor sites 1361. Acceptor substrates in which Ser was substituted for Thr have failed to be 0-glycosylated in vitro using purified GalNAc-transferases or crude homogenate from various sources [ 36, 371, however, virtually all characterized GalNAc-transferases have been shown to transfer to both Ser and Thr in one or the other peptide sequence contexts. Nevertheless, the kinetic properties for Ser sites by in vitro assays with purified recombinant enzymes generally appear to be worse than for Thr sites. lsoforms may have overlapping substrate specificities

Overlapping substrate specificities are found with many peptide substrates, especially those derived from mucin tandem repeats with multiple acceptor sites in clusters. Although, overlap in usage of acceptor substrate sites exists among GalN Ac-transferases, of the multiple acceptor substrates used, the actual sites of incorporation have not been defined. Furthermore, it may be important to consider the relative kinetic properties for the substrate sites analysed. Detailed analysis of the activities of GalNAc-transferase isoforms with peptides derived from the tandem repeat of MUC 1, clearly revealed that individual isoforms exhibited quite distinct and with some sites unique kinetic properties for the five different acceptor sites in the repeat (Figure 3 ) , in spite of apparent overlapping activity with the tandem repeat 120, 491. Furthermore, the activities of isoforms with common acceptor sites on the MUCl tandem repeat were differentially influenced by 0-glycans (GalNAc or Gal0 I-3GalNAc) attached adjacent and distant to the acceptor sites [561. Isoforms may act in different order on substrates with multiple acceptor sites

One example of this is how GalNAc-TI and -T3 initiates 0-glycosylation of the M UC1 tandem repeat at sites different from GalNAc-T2, yet they all ultimatively transfer to the same three out of five potential acceptor sites in the repeat (Figure 3 ) 1491. Another example stems from one of the most ubiquitous acceptor substrates identified so far and derived from the tandem repeat of MUC2 (PTTTPISK). Analysis of the activities of recombinant GalNAc-T1, -T2, and -T3 revealed clear differences in order of incorporation and the final sites utilized [57]. lsoforms may require prior (GalNAc) glycosylation

Recently, it has become clear that the activities of some GalNAc-transferase isoforms show strict dependence for prior GalNAc glycosylation at adjacent or distant sites 120, 54, 681. Thus, the unique activity of GalNAc-T4 for two sites in the MUCl tandem repeat sequence is dependent on prior attachment of GalNAc residues at some of the three other acceptor sites. The kinetics of this reaction has been studied in more detail [64], and it is apparent that GalNAc-T4 has overlapping specificity

284

16 Control of Mucin-Type Q-Glycosylution PDTR, G,

2% 9 l ? !

1 PDTR, Q

59

I

GalNAc-TI, GalNAc-T3 or GalNAc-T6

/GalNAc-T2

GalNAc-T1 GalNAC-T2:

I

GalNAc-TG

Ga1NAc-T4

PDTR,

PDTR

Qgii

4

@

GalNAc-T4

0

4

First GalNAc to be incorporated by either GalNAc-TI, -T2. -T3 or -T6

0 GalNAc incorporated by GalNAc-TI, -T2. -T3 or -T6 @

GalNAc incorporated by GalNAc-T4

Figure 3. In vitro 0-glycosylation of MUCl tandem repeat with human recombinant GalNActransferases. The tandem repeat sequence of MUCl has five potential acceptor sites for O-glycosylation. Three of these sites are utilised by GalNAc-TI, -T2, -T3 and -T6, however, the order with which sites are glycosylated differs. GalNAc-T1, -T3 and -T6 initiates glycosylation at T in -VTSA-, whereas, GalNAc-T2 initiates at T in -GSTA-. The last site utilized by these enzymes is S in -GSTA-. GalNAc-T4 competes with other GalNAc-transferases for S in -GSTA-, and may have better kinetic properties for this site. GalNAc-T4 complements other GalNAc-transferases in utilizing the last two sites in the MUCl tandem repeat, S in -VTSA- and T in -PDTR-. The activity of GalNAc-T4 is, however, only measurable with substrates having GalNAc residues attached at T in -VTSA- and/or T in -GSTA- [64].

with other GalNAc-transferases for one Ser site (refer to Figure 3 for detailed description). GalNAc-T4 also utilized acceptor sites on naked peptides including the EA2 peptide, but a similar dependence for prior GalNAc glycosylation are found with peptides derived from the MUC2 tandem repeat (PTTTPISMVTPTPTPTC) and MUCSAC (Ac-SAPTTSTTSAPT) [64]. Interestingly, the activity of GalNAcT4 with the EA2 peptide was inhibited by prior incorporation of only one mole of GalNAc to the EA2 peptide. Another GalNAc-transferase isoform, GalNAc-T7,

16 5 Exprevsion o j the GuINAc-TrunsferuJe Genes are Differentially Reguluted

285

showed unique specificity for GalNAc glycosylated peptide acceptor substrates [ 541. The peptide substrate specificities of GalNAc-T4 and T7 are, however, different. Most notably GalNAc-T7 does not utilise glycosylated MUCl peptides, and efficiently utilizes the EA2 peptide when GalNAc is attached [54]. It is not clear what directs the apparent GalNAc-glycopeptide substrate specificities of these GalNActransferase isoforms. Preceding O-glycosylation at adjacent or distant sites may have positive as well as negative effects for activities of some GalNAc-transferases, and these effects are likely to be exerted by conformational changes in the acceptor substrate [56]. In the case of GalNAc-T4 and -T7, the absolute requirement for preceding GalNAc-glycosylation of different substrates, may indicate that other factors than conformation of peptide backbone may be significant. A detailed analysis of the activity of GalNAc-T4 with different GalNAc glycoforms of the MUCl tandem repeat peptide indicate that GalNAc-T4 activity is activated by prior GalNAc attachments at different sites [64]. This suggests that it is not a particular conformation of the substrate that triggers GalNAc-T4. One possibility is that the putative lectin domain in the C-terminal regions of GalNAc-transferase interact with preceding GalNAc residues, and that this event triggers enzyme activity. Further studies are clearly needed to address these issues. In summary, available data indicate that the kinetic properties of GalNActransferase isoforms are different, and that the primary sequence context of acceptor sites is the major factor determining the substrate specificities. Although, there clearly is a need for more studies, the in vitro assessed properties of GalNActransferase isoforms do not appear to be in conflict with their properties assessed in in vivo systems. We envision that characteristic acceptor sequence motifs for GalNAc-transferase isoforms will eventually emerge. These may be used for prediction of the O-glycosylation capacity of a cell with due consideration of the repertoire of GalNAc-transferase isoforms expressed.

16.5 Expression of the GalNAc-Transferase Genes are Differentially Regulated Presently, there is no information on the regulatory elements of any of the GalNActransferase genes. Northern analysis with rodent or human organs have clearly demonstrated that the GalNAc-transferase isoforms have different expression patterns (Table 2). GalNAc-TI and -T2 are widely expressed [16, 17, 321, while GalNAc-T3, -T4, and -T5 have more restricted expression patterns [ 18-21]. GalNAcT6 has a very restricted expression pattern in normal tissues, but is more widely expressed in carcinomas [22]. The expression pattern of the high-similarity pair GalNAc-T3/T6 is therefore different and co-expression is only found in pancreas. GalNAc-T7 is widely expressed in normal human organs, but strong expression was found in a gastric carcinoma cell line, MKN45 [54]. GalNAc-T8 expression is predominantly been detected in kidney [66].The Northern analysis does not provide

~.

-

+

-

-

-.

-

-

-

-

+

-

-

-

-

-

-

-

++ ++ + + + +

Brain lung Liver

++ + +

+++

-

-

++ +++ +

-

-

++ + ++

-

+

+++ +++ +++ +

Kidney Placental Pancreas'

-

-

-

-

-

++ + + -

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

+ + + +c2 + + + +++ +++ + ++z + +++ +++ +++ + + + ++ -

+ ++ +++ +++ +++

Ovary Intestine Colon Stomach Muscle Salivary glands

++ ++ ++ ++ + + + ++ +++ + + + +-+ +++ ++

Spleen Testis

++.

The data are derived from northern blot analysis with multiple human tissue northern blots from CLONTECH, or blots prepared with mRNA from rat and mouse organs. No essential difference in expression pattern of rodent and human isoforms has been found so far [ 12, 20, 22, 60, 19, 21, 54, 681. ')Analyzed only with human GalNAc-transferases Analyzed only with rat/murine GalNAc-transferdses Intensity grading - > negative; +. weakly positive; moderately positive; + + +, strongly positive; blank, no data.

GalNAc-TI + + + GalNAc-T2 + GalNAc-T3 GalNAc-T4 k GalNAc-T5 GalNAc-T6 GalNAc-T7 GdlNAc-TS +

Heart

Table 2. Expression patterns of GalNAc-transferases genes by northern blot analysis.

s

6'

G E-

0

B 0 6 %-

2

$

% c

h

b'

n

o\

k

co m

IQ

16.5 Expression of the GulNAc-Trunsferase Genes are Differentially Regulated

287

information of cell-type expression, and the ubiquitous expression patterns found for GalNAc-T1 and -T2 are to some extent due to expression in fibroblasts [%I. Northern analysis of cell lines have revealed cell type specificities [22, 591. Northern analysis of the expression of GalNAc-T3 in adenocarcinoma cell lines revealed that well-differentiated lines expressed high levels and moderately differentiated cell lines expressed lower levels of GalNAc-T3 [59]. In contrast, cell lines classified as poorly differentiated failed to express detectable GalNAc-T3 mRNA. We have initiated development of a panel of monoclonal antibodies, which can be used for more direct evaluation of the repertoire of GalNAc-transferases in cells and the subcellular localization (Figure 1) [58]. Immunohistological analysis with the available antibodies clearly confirmed the differential expression patterns of isoforms indicated by the Northern analysis, and demonstrated that the repertoire of GalNAc-transferase isoforms in cells vary significantly with biological processes and malignant transformation. Stratified squamous epithelium of the buccal mucosa provides one excellent model for study of differentiation related changes in expression. Immunohistological analysis with monoclonal antibodies specific for different GalNAc-transferase isoforms, shows differential expression according to the differentiation stages of cells [58]. Thus, GalNAc-T2 is expressed in immature cells in the basal cell layers, GalNAc-TI in the mature superficial layers, and GalNAc-T3 throughout the epithelium. Further studies of the expression of GalNAc-T4, -T6, and -T7, have revealed that none are expressed in normal stratified squamous epithelium [67]. GalNAc-T6 is expressed in some oral squamous carcinomas. GalNAc-T2 and -T3 were expressed in most oral carcinomas, and GalNAc-T1 expression was lost in the majority of cases studied [%]. The repertoire of GalNAc-transferases varies markedly in different cell types. Acinar cells of small labial salivary glands express a high number of GalNActransferase isoforms; expression of GalNAc-TI. -T2, -T3, -T4, and -T6 has been detected [20, 22, 581. Northern analysis confirm that the first four genes and GalNAc-TS are expressed in rodent sublingual glands [19, 21, 32, 601. Of the isoforms studied so far, only GalNAc-T8 does not appear to be expressed in acinar cells [66]. The opposite picture is found in sperm cells where among all tested GalNActransferases (GalNAc-T1, -T2, -T3, -T4, -T6, and -T8) only GalNAc-T3 is expressed [58] (Figure 3). Analysis of expression of GalNAc-transferase isoforms in carcinoma cells is of great interest. and is particularly germane to the findings that mucins produced by cancer cells often are aberrantly glycosylated. Cancer-associated changes in glycosylation include immature O-glycan processing and also lower numbers of attached O-glycans [61]. These changes may lead to exposure of the mucin peptide backbone and antigenic epitopes not found in normal cells [62]. Studies aimed at correlating glycosylation changes with expression of GalNAc-transferases are in progress, but no direct correlations between specific GalNAc-transferase isoforms and their functions with O-glycosylation changes have been reported. One example is the highly cancer-associated site-specific O-glycosylation of the IIICS domain of fibronectin [51], found in relation to invading carcinomas [58, 631. As discussed above only GalNAc-T3 and -T6 have this function, and immunohistology show

288

16 Control of’ Mucin-Type O-Glycosylution

that GalNAc-T3 is expressed only in epithelial cells [58].GalNAc-T6, in contrast, is expressed in a tumor fibroblast cell line, WI38, which produces the oncofetal fibronectin glycoform [22]. GalNAc-T6 exhibits a cancer-associated expression pattern in several tissues, but immunohistological analysis of tissues have shown reactivity mainly in the epithelial compartment. Further studies are needed, but this represents an excellent model system since monoclonal antibodies that can distinguish the O-glycosylated protein are available. Similar studies with mucins have been hampered by the difficulties in structural analysis and lack of availability of antibody reagents defined specific glycoforms. Recent advances in both areas holds promise for future studies.

16.6 Predictive Value of in vitro O-glycosylation? Available data on the in vitro O-glycosylation capacity of GalNAc-transferase isoforms is not in disagreement with the O-glycosylation pattern found on glycoproteins. Within the limited amount of data available, no clear examples have been reported of specific acceptor sites that are utilized in vitro by one or more GalNActransferase isoforms, but that are not utilized in vivo by cells expressing the corresponding enzyme isoforms. Surely examples of the reverse have been reported, where in vivo utilized sites do not function as acceptors by in vitro analysis. The latter situation may relate to experimental problems with assaying some GalNAc-transferase activities in extracts, assay conditions, substrate design, and failure to consider if a particular order of action of GalNAc-transferase isoforms are required for demonstration of activity. Furthermore, until all the players (GalNAc-trdnsferases) are available it is not possible to fully assess the accumulated acceptor repertoire.

16.7 Conclusions and Future Perspectives Within the last few years it has become apparent that formation of many glycosidic linkages are catalyzed by entire families of homologous glycosyltransferases. Analysis of the kinetic properties and functions of members of such gene families have shown that these are different. The GalNAc-transferase gene family is no exception. The GalNAc-transferase isoforms have different functions and this can be evaluated by analysis of in vitro activities as discussed in the summation of data. The in vivo properties of isoforms are likely to be similar to the in vitro properties, but future studies need to address specific functions of isoforms in cells where they are expressed. We envision that more extensive characterization of the acceptor sequence specificities of isoforms will continue to reveal that the sequence context play a major role for the activity. Potentially the future will bring more defined sequence motifs for each isoform that can be used for predictive purposes.

References

289

Now, that we have gained some insight into the complex regulation of the first step in mucin-type 0-glycosylation, it is possible to begin addressing potential conditions and diseases associated with variation or loss of individual GalNAc-transferase genes. Studies have already shown that the GalNAc-transferase repertoire may change markedly in cancer cells, but the effect these changes may have on 0glycosylation still needs to be addressed. Genetic polymorphisms in GalNActransferase genes may underly hereditary diseases. Although no examples of such have been reported, it is clear that this possibility has not been addressed. Analysis of individual variation in 0-glycan occupancy in specific glycoproteins have only recently become possible, and identification of loss of GalNAc-transferase activity will require assays for specific isoforms. It is anticipated that loss of one GalNActransferase activity would result in limited changes in 0-glycosylation in general, but it may result in changes in the capacity for 0-glycosylation at specific sites in specific proteins. Application of the “knock-out strategy” to individual GalNActransferase genes may begin to answer some of these questions.

References 1 . Jentoft, N. Why are proteins 0-glycosylated? Trends. Biochem. Sci. I S : 291-294, 1990. 2. Varki, A. Biological roles of oligosaccharides: all of the theories are correct. Glycobiology 3: 97 130, 1993. 3. Schachter, H. and Brockhausen, 1. The biosynthesis of branched 0-glycans. Symp. Soc. Exp. Biol. 43: 1L26, 1989. 4. Gooley, A.A. and Williams, K.L. How to find, identify and quantitate the sugars on proteins. Nature 385: 557-559, 1997. 5. Gerken, T.A.. Owens, C.L., and Pasuinarthy, M. Determination of the site-specific 0glycosylation pattern of the porcine submaxillary mucin tandem repeat glycopeptide. Model proposed for the po1ypeptide:galnac transferase peptide binding site. J , Biol. Chem. 272: 97099719, 1997. 6. Muller, S., Goletz, S., Packer, N., Gooley, A.A., Lawson, A.M., and Hanisch, F.G. Localization of 0-glycosylation sites on glycopeptide fragments from lactation-associated M U C l . J. Biol. Chem. 272: 24780-24793, 1997. 7. Gerken, T.A., Owens, C.L.. and Pasumarthy. M. Site-specific core 1 0-glycosylation pattern of the porcine submaxillary gland m u c h tandem repeat. Evidence for the modulation of glycan length by peptide sequence. J. Biol. Chem. 273: 2658G26588, 1998. 8 . Elhammer, A.P., Poorman, R.A., Brown, E., Maggiora, L.L., Hoogerheide, J.G., Kezdy, and FJ. The specificity of UDP-Ga1NAc:polypeptidc N- acetylgalactosaminyltransferase as inferred from a database of in vivo substrates and from the in vitro glycosylation of proteins and peptides. J. Biol. Chem. 268: 10029-10038, 1993. 9. Hansen, J.E., Lund, O., Tolstrup. N., Gooley, A.A., Williams, K.L., and Brunak, S. NetOglyc: Prediction of mucin type 0-glycosylation sites based on the sequence context and surface accessibility. Glyconjugate J. 15: 115-130, 1998. 10. Cai, Y.D. and Chou, K.C. Artificial neural network model for predicting the specificity of GalNAc-transferase. Analyt. Biochem. 243: 284 285, 1996. 1 1. Gooley, A.A. and Williams, K.L. Towards characterizing 0-glycans: the relative merits of in vivo and in vitro approaches in seeking peptide motifs specifying 0-glycosylation sites. Glycobiology 4: 413-417, 1994. 12. Clausen, H. and Bennett, E.P. A family of UDP-GalNAc: polypeptide N-acetylgdlactosaminyltransferases control the initiation of mucin-type 0-linked glycosylation. Glycobiology 6 : 635646, 1996.

290

16 Control q f Mucin-Type 0-Glycosylution

13. McGuire, E.J. and Roseman, S. Enzymatic synthesis of the protein-hexosamine linkage in sheep submaxilliary mucin. J. Biol. Chem. 242: 3745-3755, 1967. 14. Roth, J., Wang, Y . , Eckhardt, A.E., and Hill, R.L. Subcellular localization of the UDP-Nacetyl-D-galactosamine: polypeptide N-acetylgalactosaminyltransferase-mediated O-glycosylation reaction in the submaxillary gland. Proc. Natl. Acad. Sci. USA 91: 8935-8939, 1994. 15. Rottger, S., White, R., Wandall, H.H., Bennett, E.P., Stark, A., Olivo, J.-C., Whitehouse, C., BERGER, E.G., Clausen, H., and Nilsson, T. Differential localization of three human GalNAc-transferases in HeLa cells suggests initiation of 0-linked glycosylation occur throughout the Golgi apparatus. J. Cell Sci. I l l : 45-60, 1998. 16. Homa, F.L., Hollander, T., Lehman, D.J., Thomsen, D.R., and Elhammer, A.P. Isolation and expression of a cDNA clone encoding a bovine UDP- Ga1NAc:polypeptide N-acetylgalactosaminyltransferase. J. Biol. Chem. 268: 12609-12616, 1993. 17. White, T., Bennett, E.P., Takio, K., Sorensen, T., Bonding, N., and Clausen, H. Purification and cDNA cloning of a human UDP-N-acetyl-alpha-D- ga1actosamine:polypeptide N-acetylgalactosaminyltransferase. J. Biol. Chem. 270: 24156-24165, 1995. 18. Bennett, E.P., Hassan, H., and Clausen, H. cDNA cloning and expression of a novel human UDP-N-acetyl-alpha- D-galactosamine. Polypeptide N-acetylgalactosaminyltransferase, GalNAc-T3. J. Biol. Chem. 271: 17006-17012, 1996. 19. Hagen, F.K., Hagen, T.K.G., Beres, T.M., Balys, M.M., VanWuyckhuyse, B.C., and Tabak, L.A. cDNA cloning and expression of a novel UDP-N-acetyl-D-galactosamine: polypeptide Nacetylgalactosaminyltransferase. J. Biol. Chem. 272: 13843- 13848, 1997. 20. Bennett, E.P., Hassan, H., Mandel, U., Mirgorodskaya, E., Roepstorff, P., Burchell, J., TaylorPapadamitriou, J., Hollingsworth, M.A., Merkx, G., Ceurts van Kessel, A,, Eiberg, H., Steffensen, R., and Clausen, H. Cloning of a human UDP-N-acetyl-a-D-galactosamine: polypeptide N-acetylgalactosaminyltransferase that complements other GalNAc-transferases in complete 0-glycosylation of the MUCl tandem repeat. J. Biol. Chem. 273: 30472-30481, 1998. 21. Hagen, T.K.G., Hagen, F.K., Balys, M.M., Beres. T.M., Van Wuyckhuyse, B., and Tabak, L.A. Cloning and Expression of a Novel, Tissue Specifically Expressed Member of the UDPGa1NAc:Polypeptide N-Acetylgalactosaminyltransferase Family. J. Biol. Chem. 273: 2774927754, 1998. 22. Bennett, E.P., Hassan, H., Mandel, U., Hollingsworth, M.A., Akisawa, N., Ikematsu, Y., Merkx, G., Geurts van Kessel, A,, Olofsson, S., and Clausen, H. Cloning and Characterization of a Close Homologue of Human UDP-N-acetyl-a-D-galactosamine: Polypeptide N-Acetylgalactosaminyltransferase T3, designated GalNAc-T6: Evidence for Genetic but not Functional Redundancy. J. Biol. Chem. 274: 25362-25370, 1999. 23. Paulson, J.C. and Colley, K.J. Glycosyltransferases. Structure, localization, and control of cell type-specific glycosylation. J. Biol. Chem. 264: 17615-17618, 1989. 24. Wragg, S., Hagen, F.K., and Tabak. L.A. Identification of essential histidine residues in UDPN-acetyl-D- ga1actosamine:polypeptide N-acetylgalactosaminyltransferase-T1. Biochem. J. 328: 193-197, 1997. 25. Hagen, F.K., Hazes, B., Raffo, R., deSa, D., and Tabak, L.A. Structure-Function Analysis of the UDP-N-acetyl-D- ga1actosamine:PolypeptideN-acetylgalactosaminyltransferase. Essential residues lie in a predicted active site cleft resembling a lactose repressor fold. J. Biol. Chem. 274: 6797-6803, 1999. 26. Busch, C., Hofmann, F., Selzer, J., Munro, S., Jeckel, D., and Aktories, K. A common motif of eukaryotic glycosyltransferases is essential for the enzyme activity of large clostridial cytotoxins. J. Biol. Chem. 273: 19566-19572, 1998. 27. Hazes, B. The (QxW)3 domain: a flexible lectin scaffold. Protein Science 5 : 1490-1501, 1996. 28. Imberty A,, Piller V.. Piller F., and Breton C. Fold recognition and molecular modeling of a lectin-like domain in UDP-Ga1Nac:polypeptide N-acetylgalactosaminyltransferases. Protein Eng. 10: 1353-1356, 1997. 29. Hagen, F.K., Gregoire, C.A., and Tabak, L.A. Cloning and sequence homology of a rat UDPGa1NAc:polypeptide N-acetylgalactosaminyltransferase. Glycoconjugate J. 12: 901-909, 1995. 30. Hagen, F.K. and Nehrke, K. cDNA cloning and expression of a family of UDP-N-acetyl-Dgalactosamine: polypeptide N-acetylgalactosaminyltransferase sequence homologs from caenorhabditis elegans. J. Biol. Chem. 273: 8268-8277, 1998.

Referrnces

291

31. Albone, E.F., Hagen, F.K., VanWuyckhuyse, B.C., and Tabak, L A . Molecular cloning of a rat submandibular gland apomucin. J. Biol. Chem. 269: 16845-16852, 1994. 32. Hagen. F.K., Van Wuyckhuyse, B., and Tabak, L.A. Purification, cloning, and expression of a bovine UDP-GalNAc: polypeptide N-acetyi-galactosaminykransferase. J. Biol. Chem. 268: 18960-18965, 1993. 33. Meurer, J.A., Drong, R.F., Homa, F.L., Slightom. J.L., and Elhammer, A.P. Organization of a human UDP-GalNAc:polypeptide, N- acetylgalactosaminyltransferase gene and a related processed pseudogene. Glycobiology 6: 23 1-241, 1996. 34. Bennett, E.P., Weghuis, D.O., Merkx, G., Geurts van Kessel, A., Eiberg, H., and Clausen, H. Genomic organisation and chromosomal localisation of the three members of the UDP-Nacetylgalactosamine: polypeptide N-acetylgalactosaminyltransferase family. Glycobiology 8: 547-555, 1998. 35. Hennet, T., Hagen, F.K., Tabak, LA., and Marth, J.D. T-cell-specific deletion of a polypeptide N-acetylgalactosaminyl-transferase gene by site-directed recombination. Proc. Natl. Acad. Sci. USA 92: 12070-12074, 1995. 36. Wang, Y., Abernethy, J.L., Eckhardt, A.E., and Hill, R.L. Purification and characterization of a UDP-Ga1NAc:polypeptide N- acetylgalactosaminyltransferase specific for glycosylation of threonine residues. J. Biol. Chem. 267: 12709-12716, 1992. 37. O’Connell, B.C. and Tabak, L.A. A comparison ofserine and threonine 0-glycosylation by UDPGa1NAc:polypeptide N-acetylgalactosaminyltransferase. J. Dental Res. 72: 1554-1 558, 1993. 38. Wang, Y., Agrwal, N., Eckhardt, A.E., Stevens, R.D., and Hill, R.L. The acceptor substrate specificity of porcine submaxillary UDP- Ga1NAc:polypeptide N-acetylgalactosaminyltransferase is dependent on the amino acid sequences adjacent to serine and threonine residues. J. Biol. Chem. 268: 22979-22983, 1993. 39. Sorensen, T., White, T., Wandall, H.H., Kristensen, A.K., Roepstorff, P., and Clausen, H . UDP-N-acetyl-alpha-D-ga1actosamine:polypeptide N- acetylgalactosaminyltransferase. Identification and separation of two distinct transferase activities. J. Biol. Chem. 270: 241 66-24173, 1995. 40. Nehrke, K., Hagen. F.K., and Tabak, L.A. Charge distribution of flanking amino acids influences 0-glycan acquisition in vivo. J. Biol. Chem. 271: 7061-7065, 1996. 41. O’Connell, B.C., Hagen, F.K., and Tabak, L A . The influence of flanking sequence on the 0glycosylation of threonine in vitro. J. Biol. Chem. 267: 25010-25018, 1992. 42. Nehrke, K., Hagen, T.K.G., Hagen, F.K., and Tabak, L.A. Charge distribution of flanking amino acids inhibit 0-glycosylation of several single-site acceptors in vivo. Glycobiology 7: 1053- 1060, 1997. 43. Nehrke, K., Hagen, F.K., and Tabak, L A . Isoform-specific 0-glycosylation by murine UDPGalNAc: polypeptide N-acetylgalactosaminyltransferase-T3, in vivo. Glycobiology 8: 367-37 1, 1998. 44. de Haan, C.A., Roestenberg, P., de Wit, M., de Vries, A.A., Nilsson, T.? Vennema, H., and Rottier, P.J. Structural requirements for 0-glycosylation of the mouse hepatitis virus membrane protein. J. Biol. Chem. 273: 29905-29914, 1998. 45. Cottrell, J.M., Hall, R.L., Sturtonj R.G., and Kent, P.W. Polypeptide N-acetylgalactosaminyltransferase activity in tracheal epithelial microsomes. Biochem. J. 283: 299-305, 1992. 46. Nishimori, I.. Perini, F., Mountjoy, K.P., Sanderson, S.D.?Johnson, N., Cerny, RL. Gross, M.L.. Fontenot, J.D., and Hollingsworth, M A . N-acetylgalactosamine glycosylation of MUCI tandem repeat peptides by pancreatic tumor cell extracts. Cancer Res. 54: 3738-3744, 1994. 47. Nishimori, I., Johnson, N.R., Sanderson, S.D., Perini, F., Mountjoy, K., Cerny, RL. Gross, M.L., and Hollingsworth, M A . Influence of acceptor substrate primary amino acid sequence on the activity of human UDP-N-acetylga1actosamine:polypeptide N-acetylgalactosaminyltransferase. Studies with the MUCI tandem repeat. J. Biol. Chem, 269: 16123-16130, 1994. 48. Briand, J.P., Andrews, S.P.J., Cahill. E., Conway, N.A., and Young, J.D. Investigation of the requirements for 0-glycosylation by bovine submaxillary gland UDP-N-acetylgalactosamine: polypeptide N-acetylgalactosamine transferase using synthetic peptide substrates. J Biol Chem. 256: 12205-12207, 1981. 49. Wandall, H.H., Hassan. H., Mirgorodskaya, E.; Kristensen, A.K., Roepstorff, P., Bennett, E.P., Nielsen, P.A., Hollingsworth, M.A., Burchell, J., Taylor-Papadimitriou, J., and Clausen,

292

16 Control of Mucin-Type 0-Glycosylution

H. Substrate specificities of three members of the human UDP-N-acetyl-alpha-D-galact0samine:Polypeptide N-acetylgalactosaminyltransferase family, GalNAc-TI, -T2: and -T3. J. Biol. Chem. 272: 23503-23514, 1997. 50. Mirgorodskaya, E., Hassan, H., Wandall. H.H., Clausen, H., and Roepstorff, P. Partial VaporPhase Hydrolysis of Peptide Bonds: A Method for Mass Spectrometric Determination of 0Glycosylated Sites in Glycopeptides. Anal. Biochem. 269: 54--65,1999. 51. Matsuura, H., Greene, T.. and Hakomori, S. An alpha-N-acetylgalactosaminylation at the threonine residue of a defined peptide sequence creates the oncofetal peptide epitope in human fibronectin. J. Biol. Chem. 264: 10472--10476. 1989. 52. Bernstein, H.B., Tucker, S.P., Hunter, E., Schutzbach, J.S., and Compans, R.W. Human immunodeficiency virus type 1 envelope glycoprotein is modified by 0-linked oligosaccharides. J Virol. 68: 463-468, 1994. 53. Huang, X., Smith, M.C., Berzofsky. J.A., and Barchi. J.J.J. Structural comparison of a 15 residue peptide from the V3 loop of HIV-1IIIb and an 0-glycosylated analogue. FEBS Lett 3Y3: 280-286, 1996. 54. Bennett, E.P., Hassan, H., Hollingsworth, M.A., and Clausen, H. A novel human UDP-Nacetyl-D-Galactosamine: polypeptide N-acetylgalactosaminyltransferase, GalNAc-T7, with specificity for partial GalNAc-glycosylated acceptor substrates. FEBS lett. 460: 226- 230. 1999. 55. Liu, W., Ramachandran, V., Kang, J., Kishimoto, T.K., Cummings, R.D., and McEver, R.P. Identification of N-terminal residues on P-selectin glycoprotein ligand- 1 required for binding to P-selectin. J. Biol. Chem., 273: 7078-7087, 1998. 56. Hanisch. F.G., Her, S., Hassan, H., Clausen, H., Zachara, N., Gooley, A.A.. Paulsen, H., Alving, K., and Peter-Katalinic, J. Dynamic Epigenetic Regulation of Initial 0-Glycosylation by UDP-N-Acetylga1actosamine:Peptide N-Acetylgalactosaminyltransferases. Site- specific glycosylation of mucl repeat peptide influences the substrate qualities at adjacent or distant ser/ thr positions. J. Biol. Chem. 274: 9946 9954, 1999. 57. Iida, S., Takeuchi, H., Hassan. H., Clausen, H.. and Irimura. T. Incorporation of N-acetylgalactosamine into consecutive threonine residues in MUC2 tandem repeat by recombinant human N-acetyl-D-galactosamine transferase-TI , T2. and T3. FEBS Letters, in prc 58. Mandel, U., Hassan. H., Therkildsen, M.H., Rygaard, J., Jacobsen, M., Juhl, B.R., Dabelsteen, E., and Clausen, H. Expression of polypeptide GalNAc-transferases in stratified epithelia and squamous cell carcinomas: iminunohistological evaluation using monoclonal antibodies to three members of the GalNAc-transferase family. Glycobiology 9: 43-52. 1999. 59. Sutherlin, M.E., Nishimori, I., Caffrey, T., Bennett, E.P., Hassan, H., Mandel, U., Mack, D., lwamura, T., Clausen, H., and Hollingsworth, M. A. Expression of three UDP-N-acetyl-alphaD-ga1actosamine:polypeptide GalNAc N-acetylgalactosaminyltransferases in adenocarcinoma cell lines. Cancer Res. 57: 4744-4748, 1997. 60. Zara, J., Hagen, F.K.. Ten Hagen, K.G., Van Wuyckhuyse, B.C.. and Tabak. L.A. Cloning and expression of mouse UDP-Ga1NAc:polypeptide N- acetylgalactosaminyltransferase-T3. Biochem. Biophys. Res. Commun. 228: 38-44. 1996. 61. Lloyd, K.O., Burchell, J., Kudryashov, V., Yin, B.T.. and Taylor-Papadimitriou, J. Comparison of 0-linked carbohydrate chains in MUC-I much from normal breast epithelial cell lines and breast carcinoma cell lines. Demonstration of simpler and fewer glycan chains in tumor cells. J. Biol. Chem. 271: 33325 33334, 1996. 62. Taylor-Papadimitriou, J. and Finn, O.J. Biology, biochemistry and immunology of carcinomaassociated mucins. Immunology Today 18: 105- 107, 1997. 63. Loridon-Rosa, B., Vielh, P., Matsuura. H., Clausen. H., Cuadrado, C., and Burtin, P. Distribution of oncofetal fibronectin in human mammary tumors: immunofluorescence study on histological sections. Cancer Res. 50: 1608-1612, 1990. 64. Hassan, H., Bennett, E.P., and Clausen, H. Unpublished. 65. Bennett, E.P. and Clausen. H. Unpublished. 66. Bennett. E.P., Mandel, U., and Clausen, H. Unpublished. 67. Mandel, U., Hassan, H., and Clausen, H. Unpublished. 68. Hagen, T.K.G., Tetaert, D., Hagen, F.K., Richet, C., Beres, T.M., Gagnon, J., Balys, M.M., Van Wuyckhuyse, B., Bedi, G.S., Degand, P., Tabak, L.A. Characterization of a UDPGa1NAc:polypeptide N-acetylgalactosaminyltransferase that displays glycopeptide N-acetylgalactosaminyltransferase activity. J. Biol. Chem. 274: 27867-27874, 1999.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

17 Glycosyltransferase Inhibitors Xiungping Qian and Monica M. Palcic

17.1 Introduction Complex oligosaccharides are biosynthesized by the action of glycosyltransferase enzymes that catalyze the transfer of sugars from activated donors (usually sugar nucleotides) to glycosyl acceptors [ 1 31. Since cell-surface oligosaccharides covalently attached to glycoproteins and glycolipids act as recognition markers in diverse biological events ranging from bacterial and viral infection to cell-cell adhesion and tumor progression [4, 51, glycosyltransferases responsible for their biosynthesis have become targets for the development of inhibitors [6-91. Glycosyltransferase inhibitors are of potential therapeutic value for the treatment of diseases associated with glycosyltransferases where specific oligosaccharide sequences control a biological event such as cell differentiation or cellular adhesion [lo, 111. Additional information on the biological functions of oligosaccharides can be obtained using glycosyltransferase inhibitors by interrupting the glycosylation process thereby altering the structures of oligosaccharides on cell surfaces. Glycosyltransferase inhibitors are also useful for exploring the active sites of glycosyltransferases and elucidating molecular mechanisms of action of these enzymes. Nature itself has provided us with glycosyltransferase inhibitors such as papulacandin B [12, 131, a glucan synthase inhibitor, polyoxin D [14], a chitin synthetase inhibitor, and tunicamycin, an inhibitor of N-acetylglucosaminyl phosphotransferase (Scheme 1) [ 15-1 71. Tunicamycin is an excellent model of bisubstrate analog which inhibits protein glycosylation by blocking the biosynthesis of dolicholPP-GlcNAc from dolichol-P and UDP-GlcNAc. Many nucleotides (nucleoside mono-, di- or triphosphates). including reaction products generated in enzymatic glycosylation, are also good inhibitors of glycosyltransferases [ 181. Synthetic inhibitors of glycosyltransferases have been extensively developed to target mammalian glycosyltransferases because of their biological and medical significance. The Leloir-types of mammalian glycosyltransferases catalyze the transfer of a single sugar residue from a sugar nucleotide donor to the hydroxyl group of an ~

294

I7 Glycosyltransjkrase Inhibitors 0

OH OH OH

tunicamycin inhibitor of GlcNAc phosphotransferase IC,, = 7 nM

OH

HO

OH

papulacandin B glucan synthase inhibitor

ICso = 1.2 FM 0

OH OH

polyoxin D chitin synthase inhibitor Ki = 0.9 PM

Scheme 1. Some naturally occurring glycosyltransferase inhibitors.

acceptor with overall retention (retaining enzyme) or inversion (inverting enzyme) of configuration at the anomeric center of donor (Scheme 2). Although over 30 glycosyltransferases have been cloned [ 111, X-ray structural information on mammalian enzymes is not available [ 191. Development of glycosyltransferase inhibitors therefore relies on the preparation of large numbers of natural substrate analogs. The design of glycosyltransferase inhibitors has originally concentrated on mod-

17. I Introduction

295

Reactive OH 0

HO HO

Acceptor

OH

Donor

Oligosaccharide

+

HO

b H

Scheme 2. A generalized glycosyltransferase reaction

ifying either the sugar moiety or the pyrophosphate of sugar nucleotide donors to produce unreactive donor analogs [20, 211. Since there are more than 100 mammalian glycosyltransferases and only nine main sugar nucleotide donors, the specificity of glycosyltransferases most likely resides in the recognition of the acceptor sequence, not in the nucleotide donor [22, 231. In order to display specificity, glycosyltransferase inhibitors have been developed to incorporate structural elements of acceptors. Such inhibitors usually have modifications at the reactive OH group of the acceptor to which a glycosyltransferase transfers. The OH group is either deoxygenated, derivatized, or substituted with other functionalities. When the reactive OH group is a “key polar group” [24-261 required for recognition by the enzyme, modifications on this OH group are not tolerated, however, an immediately adjacent site can be modified to prepare potential inhibitors. Recent mechanistic studies on glycosyltransferases have led to the design of glycosyltransferase inhibitors mimicking the transition state of the reaction. Transitionstate analogs could be either mono-substrate (usually modified donor analogs) having a half-chair conformation [27],or bisubstrate having both donor and acceptor elements covalently connected [22, 281. Synthetic glycosyltransferase inhibitors have included donor analogs, acceptor analogs, or transition-state (mono- and bisubstrate) analogs. These inhibitors have been extensively studied with galactosyltransferases, fucosyltransferases, sialyltransferases, N-acetylglucosaminyltransferases, and human blood group A and B glycosyltransferases. They are discussed in turn below.

296

17 Glycosyltransjerase Inhibitors

17.2 Inhibitors of Glycosyltransferases 17.2.1 Inhibitors of Galactosyltransferases Inhibitors of pl,4-galactosyltransferase ~1,4-Galactosyltransferase(81,4-GalT, E.C. 2.4.1.22/38) is the most widely studied glycosyltransferase, since it has been commercially available for many years. It catalyzes the transfer of Gal from UDP-Gal to OH-4 of terminal p-linked GlcNAc to form N-acetyllactosamine. Some diseases are accompanied by structural changes due to a change in the level of pl,4-GalT [4]. These include the down-regulation of p1,4-GalT in multiple myeloma [29] and rheumatoid arthritis [30], while upregulation has been reported for Burkitt’s lymphoma [31] and some liver diseases [32]. A number of inhibitors of p1,4-GalT have been developed. Most of them are donor analogs. The first donor analog inhibitors, 1 and 2, were reported by Vaghefi et al. in 1987 [20, 211. They differ from the natural donor UDP-Gal in the pyrophosphate portion. The K, values for 1 and 2 were 97 and 165 pM respectively, compared to a K , value of 25 pM [33] for UDP-Gal. Malonic, tartaric acid and monosaccharide were also investigated as pyrophosphate mimics [34]. Compound 3, in which glucose was used to mimic the linkage, showed an inhibitory activity with a K, of 120 pM, while malonic and tartaric linkages were ineffective. These results support the hypothesis that during the enzyme reaction the pyrophosphate linkage binds a divalent maganese ion to form a six-membered ring complex as shown in Scheme 5 [35]. Comparison of these modifications with the lower K, value (32 pM) of UDP suggests that an intact pyrophosphate group seems to be preferred for binding. Inhibitors with modifications on the sugar portion of donor have been described [33, 36, 371. The carbocyclic analog 4 [33], where the ring oxygen atom was replaced with a methylene group, was found to be a competitive inhibitor with a K, value of 58 pM. UDP-2-deoxy-2-fluoro-galactose 5 [ 361, where 2-OH group was replaced with a strong electron-withdrawing fluorine atom, was also shown to be a competitive inhibitor. These results are consistent with the generally accepted mechanism for glycosy1 transfer that involves an oxocarbonium ion-like transition state 138, 391, which is unlikely to occur with analogs 4 and 5 . Mono-0-methyl derivatives of UDP-Gal (6-9) also exhibited inhibitory activities with estimated K, values in the range of 20270 pM [37]. It is also interesting that UDP-Fuc (10) and UDP-Man (11) are good inhibitors with K, values around 10 pM [37]. Inhibitors of pl ,4-GalT based on the putative transition state have also been reported. Compound 12 with a half-chair conformation (sp2 character at the anomeric center) and a non-cleavable C-C glycosidic bond exhibited competitive inhibition with a K, value of 62 pM [27]. More recently, Hashimoto et al. reported the synthesis of a bisubstrate analog 13, having both donor (UDP-Gal) and acceptor (GlcNAc) tethered through a methylene unit 1281. This analog is by far the most potent inhibitor reported for 81,4Gal-T with K, values of 1.35 pM and 3.3 pM with respect to GlcNAc and UDP-Gal. respectively.

Ki=97pM OH OH

OH OH

Hr+

HO

HO H % HO

OH

HO

0-UDP

3 K, = 120 pM

H

OH OH

b

0-UDP

5 K,= 149pM

4 K,=58pM

HT+

+ H Me0

HO Me0 6

0-UDP

K,=20W

HO H%

OHO-UDP

K, = 46 pM

OHO-UDP

K,=44W

K, = 270 pM

Ht%

H3r%OH

9

OHO-UDP

8

7

10

0-UDP

K, = 10 pM

11

0-UDP

K, = 8.8 pM

Scheme 3. Donor-analog inhibitors of ~1,4-galactosyltransferase

Chemical mapping studies using acceptor analogs indicate the 2-NHAc group and the 4-OH group to which the enzyme transfers are key polar groups [40-421. Modifications in acceptor are thus limited to the 3 and 6-positions. Replacement of 3-OH and 6-OH with a fluorine atom or a sulfbydryl group produced very weak competitive inhibitors 14 (421 and 15 [43]. The K, values were 2.7 mM and 1.0 mM, in the same range as the corresponding acceptor K , value (1.1 mM) [43]. Inhibitors of al,3-galactosyltransferase

al,3-Galactosyltransferase (ul,3-GalT, E.C. 2.4.1.151), a retaining glycosyltransferase, catalyzes the formation of Gala14 3 G a l epitopes [44-471, the major xenoactive antigens which cause hyperacute rejection during xenotransplantation [48, 491. A potential mechanism for retaining glycosyltransferases is a two-step double-displacement involving the formation of a glycosyl-enzyme intermediate

298

I7 Gl)~cos~ltransfL.r-ase Inhibitors

U

12

Ki=62w

I

0-UDP 13

I

K, = 1.35 pM (with respect to GlcNAc) K,= 3.3 pM (with respect to UDP-Gal)

OH OH

I Transition-State Analogs I

&:& H o

OH

NHAc

H

OMe

E

A

oNHAc H

&

14

15

K,= 2.7 mM

K,=1.0mM

OH

OMe

Scheme 4. Transition-state and acceptor analogs as inhibitors of D l .4-galactosyltransferase.

(Scheme 6) [50].The first step is the displacement of UDP from UDP-Gal by a nucleophile located in the active site to form a covalent glycosyl-enzyme intermediate. Transfer of the sugar residue to the acceptor is then completed by displacement of the glycosyl-enzyme intermediate by the OH group of the acceptor. Two inhibitors 16 [51] and 17 [52] have been developed recently based on the hypothesis that the nucleophile located in the active site is a negatively charged group (possibly a carboxylate group by analogy to glycosidases). Both inhibitors contain a nitrogen atom which can be protonated at physiological pH to form a tightly held enzyme-inhibitor complex via favorable electrostatic interaction (Scheme 7). The acceptor analog 16, where the reactive OH group normally undergoing glycoslyation is replaced with an amino group is a noncompetitive

B:

... "y

HO

HO

OR

0

HO

OH OH

Scheme 5. Proposed transition-state structure for the Dl ,4-galactosyltransferase reaction.

17.2 Inhibitors

of GIycos~~1transfL;ruses

299

Nu

t-.

Nucleotide

Scheme 6. Hypothetical two-step, double-displacement mechanism for retaining glycosyltransferases.

inhibitor with a K, value of 104 pM, about half that of the corresponding acceptor K , (190 pM). The donor analog 17 was constructed with three segments: a galactose-type 1-N-iminosugar, a vicinal diol derived from L-tartaric acid, and a 5'thiouridine, to avoid unnecessary negative charge on the molecule and to increase the stability of the inhibitor. Analog 17 has a K , of 4.4 pM, however, its mode of inhibition was noncompetitive against UDP-Gal.

17.2.2 Inhibitors of Fucosyltransferases Many antigenic oligosaccharides on the cell surface such as blood group antigens are fucosylated. Along with sialylation, fucosylation by fucosyltransferases is often the last in vivo modification of oligosaccharides. Fucose-containing glycoconjugates are considered oncodevelopmental antigens since they accumulate in numerous human cancers [53-55). lnhibitors of fucosyltransferases are potential anti-tumor agents.

Lr' Enzyme

H?

NHAc

/OH

OH

16

Scheme 7. Inhibitors of a1,3-galactosyltransferase.

Ki = 4.4 KM

OH OH

300

17 Glycosyltmnsferase Inhibitors

Inhibitors of al,2-fucosyltransferases a1,2-Fucosyltransferases (al,2-FucT, E.C. 2.4.1.69) transfer a fucosyl residue from GDP-Fuc to the 2-OH group of P-D-galactosides with inversion of configuration at the fucopyranosyl anomeric carbon. There are two distinct al,2-fucosyltransferases, ul,2-FucT I (H-enzyme) and a1,2-FucT 11 (Se-enzyme) [56-58]. The former is responsible for the production of blood group O(H) antigens on red blood cells, while the latter is thought to synthesize Fucul+2Gal structures in secretory tissues [59]. The first synthetic bisubstrate analog inhibitor 18 was successfully designed for a1,2-FucT I1 by Palcic et al., based on the proposed ion-pair mechanism shown in Scheme 8 [22].Analog 18, where the Gal unit is attached to the terminal phosfor of GD P through a flexible ethylene linkage, was found to be a competitive inhibitor with respect to both donor and acceptor substrates with K, values of 16 and 2.3 pM, respectively. Inhibition studies with the bisubstrate analog also helped establish the kinetic mechanism of the enzyme reaction. These dual competitive inhibition patterns are only consistent with a random kinetic mechanism where either substrate can bind to free enzyme. The acceptor analog 19, where the reactive 2-OH of the Gal unit was deoxygenated, was found to be a competitive inhibitor [23], suggesting that the OH undergoing fucosylation is not essential for binding to al12-FucT 11. Compound 19 has a K, value of 0.80 mM which is similar to the K,,, of the corresponding acceptor (0.20 mM). The substrate specificity of al,2-FucT I was explored using structural analogs of octyl P-D-galactoside where 3-OH, 4-OH or 6-OH were modified. The C-3 modified analogs were inhibitors with estimated K, values of 0.9-43 mM [60].

Inhibitors of al,3-fucosyltransferases There are at least five distinct human al,3-fucosyltransferases,FucT 111 to VII [3]. All of these fucosyltransferases catalyze the formation of a Fucwli3GlcNAc linkage. However, they have been differentiated from one another on the basis of their substrate specificities towards Gala1 +4GlcNAc, Galpli3GlcNAc, and NeuAcu2+3Gal~l+4GlcNAc acceptors, their tissue distributions, inhibition by ethylmaleimide, pH optima, and kinetic properties. Among them, a1,3-FucT V has gained considerable interest since it is responsible for the last step in the biosynthesis of sialyl Lewis X which is a ligand for selectins, a family of adhesion molecules [61, 621. Inhibiting al,3-FucT V may control inflammatory processes mediated by selectins. Inhibition of al,3-FucT V has been achieved by the synergistic effect of the enzymatic product, GDP and two azasugars 20 and 21 which are potent a-fucosidase inhibitors [63, 641. In the presence of GDP at its ZC5, concentration (50 pM), more than 80% of the enzyme activity was inhibited at ZC5os of the individual azasugars. This synergistic inhibition indicates a possible interaction of azasugars, GDP, and the acceptor in the active site of the enzyme mimicking the transition state as shown in Scheme 9. Based on this, the Wong group further developed an azatrisaccharide 22 where the nitrogen of P-L-homofuconojirimycin was attached to 3-OH of LacNAc through an ethylene linker [65]. In the presence of 30 pM GDP, the IC50 for 22 was reduced 77-fold to 31 pM.

17.2 Inhibitors of Glycosyltrunsferuses

30 1

K, = 2.3 pM (with respect to the acceptor) K, = 16 pM (with respect to the donor)

* ; HO &H

Scheme 8. Postulated transitionstate structure and inhibitors of al,2-fucosyltransferase 11.

NHAc O(CH&COOCHs

19

K, = 0.80 mM

More recently, a series of compounds having deoxyfuconojirimycin and Dgalactose connected though different spacers have been reported [66]. Compounds 23 and 24 displayed significantly enhanced inhibition properties towards al,3-FucT IV compared to deoxyfuconojirimycin (ZCs0= 3.5 mM) with ZC50 values of 233 and 81 yM, respectively. Two bisubstrate analogs have also been synthesized as potential inhibitors for al,3-FucTs, however, no inhibition data were reported [67, 681. 17.2.3 Inhibitors of Sialyltransferases Sialyltransferases catalyze the transfer of N-acetyl-neuraminic acid from CMPNeu5Ac to acceptors at or near the nonreducing terminus of oligosaccharide chains of glycoproteins or glycolipids [69]. Cell-surface sialic acid residues are important as antigenic determinants, as ligands for cellular adhesion and receptors for the bind-

302

17 Glycosyltransferuse Inhibitors

1

! 20

21

IC,, = 80 m M

IC,, = 34 mM

HO

&

HO

HO HO

HO&OH

H

homofuconojirimycin

OH

h>vo

HO

K, = 32.9 m M ICso = 1.54 mM (with 30 pM GDP)

OH

22

OH

K,= 2.4 m M ICso = 31 pM (with 30 pM GDP)

Scheme 9. Synergistic inhibition of ul,3-fucosyltransferase V.

ing of viruses and bacteria [70]. A correlation has been found between cell-surface sialic acid levels, sialyltransferase activity and the metastatic potential of tumor cells [71-73]. Inhibitors of sialyltransferases have been developed for a2,6-sialyltransferase (a2,6-SialT, E.C. 2.4.99.1) and a2,3-sialyltransferase (a2,3-SialT, E.C. 2.4.99.6).

Inhibitors of a2,6-sialyltransferase a2,6-Sialyltransferase catalyzes the transfer of sialic acid from CMP-sialic acid to 6OH of terminal Gal residue in GalPl-+4GlcNAcsequence. Among the inhibitors based on donor analogs, cytidine 5'-diphosphate (CDP) which shows competitive inhibition against donor with Ki values around 10 pM is the most potent inhibitor of a2,6-SialT [74, 751. CMP-quinic acid [75], phosphonate analogs 25 [76, 771 and

303

17.2 Inhibitors of’ Gl~cos!,ltl.arzsje7ruses

‘0H

OH

23 IC50= 233 pM (at 8.5

GDP-Fucose)

24 IC,, = 8 I pM (at 8.5 pM GDP-Fucose)

Scheme 10. Inhibitors of al,3-fucosyltransferase I V

26 [77]also showed competitive inhibition with somewhat higher K, values (Scheme 11). A number of analogs were recently investigated by the Schmidt group to mimic the transition state structure of CMP-NeuSAc (Scheme 12) deduced for the recently suggested Sp~1mechanism [78, 791. The transition-state analog 27 [80], with the I -yl CMP residue attached to C- 1 of flattened 2,3-dehydro-N-acetylneuraminphosphonate, exhibited more than a 100-fold higher affinity for the enzyme (K, = 0.35 pM) than the donor CMP-NeuSAc (K, = 46 pM). Analog 28 with a flat benzene ring had an even lower K, value (0.20 pM) [80]. The elimination by-product 29 in the synthesis of 27 also showed competitive inhibition with a K, value of 6 pM

HO

OH OH

OH OH

CDP

CMP-quinic acid

K,>IOlJM

K,=44pM

How utl NH2

cop-

0

P-0

AcHN

AcHN

HO 25

OH OH

K, = 250 pM

Scheme 11. Donor-analog inhibitors of a2,6-sialyltransferase

26 Kl=7X0pM

&H

bH

304

I7 Glycosyltransferase Inhibitors

K, = 0.20 pM

K,= 0.35@I

HO

HO

OH

29

Ki=6@I

OH OH

K,=40nM

OH OH

Scheme 12. Transition-state structure and inhibitors of a2,6-sialyltransferase.

[SO]. Analog 30 with combined structures of 27-29 is the most potent glycosyltransferase inhibitor reported to date with a Ki of 40 nM [Sl]. Acceptor analogs of LacNAc with OH-6' modified were also studied [82, 831. The analogs showed mixed inhibition with Ki values in the mM range. The 6'-deoxy analog had a K, value of 0.76 mM, similar to the K, value of the corresponding acceptor (0.90 mM). Inhibitors of a2,3-sialyltransferase

a2,3-Sialyltransferase transfers a sialic acid unit from CMP-sialic acid to 3-OH of terminal Gal residue in Galpli4GlcNAc or GalPl-3GlcNAc sequences. Only inhibitors based on acceptor structure have been reported. Trisaccharide acceptor analogs (31, Scheme 13) with the reactive 3-OH group of the Gal residue deoxygenated (H), substituted (F, NHz), or derivatized (OMe) showed inhibitory activities against a2,3-SialT [84]. Among them, the 3'-amino-3'-deoxy analog exhibited 95% inhibition of the activity of recombinant rat liver a2,3-SialT at a concentration

17.2 lnhihitors ojG1L’c.osL‘ltransferases

305

HO

R

HO

31

Scheme 13. Acceptor-analog inhibitors of u2.3-sialyltransferase.

R = H, F, NH2, OMe

HO O(CHP)~CH~

of 1.0 mM. This analog was also found to be an inhibitor of rat liver a2,6-SialT ~41.

17.2.4 Inhibitors of N-Acetylglucosaminyltransferases There are six N-acetylglucosaminyltransferases(GnT I to VI) involved in the synthesis of complex N-glycans [85]. N-Acetylglucosaminyltransferase V (GnT V, E.C. 2.4.1.155) has earned particular interest following the observation that GnT V activity increases when cells are transformed by tumor viruses and several oncogenes [86-881. A direct correlation between an increase in the activity of G n T V and the metastatic potential of tumor cell lines has also been reported [89, 901. Biosynthetically, GnT V catalyzes the transfer of a GlcNAc residue from UDPGlcNAc to oligosaccharide acceptors having the minimum heptasaccharide sequence, converting it to the corresponding octasaccharidc (Scheme 14) [91, 921. The trisaccharide acceptor 32 [93] was found to be an excellent acceptor for GnT V with K,,, values in the range of 23-36 pM [23. 941. The design and synthesis of acceptoranalog inhibitors was simplified by modifying trisaccharide 32, instead of the large natural heptasaccharide. The 6I-deoxy analog 33 where the reactive 6’-OH had been removed was the first reported inhibitor specific for a glycosyltransferase with K, values in the range of 30-70 pM depending on the source of GnT V [23. 951. Analog 34 with 4-OH of the mannose residue methylated sterically blocked the transfer reaction, with a K, value of 14 pM [94]. The remaining 4-OH or 6-OH of the mannose residue in 33 and 34 were substituted with an amino group, which was then further derivatized, to give two series of analogs (35 and 36, Scheme 15) [96, 971. The derivatives proved to be competitive inhibitors with K,s of 3-300 pM. Analog 37 with 4-OH and 6-OH of the mannose residue methlyated also showed 53% inhibition of GnT V from hamster kidney at a concentration of 0.8 m M [98]. Inhibitors of GnT I (E.C. 2.4.1.101) and GnT I1 (E.C. 2.4.1.143) have also been reported [99, 1001. The trisaccharide acceptor analog 38 where 6’-OH was methylated was found to be a competitive inhibitor of G n T I with a K, of 0.76 mM, similar to the K , of the corresponding acceptor (0.73 mM) [99]. Introduction of reactive groups such as iodoacetamido and diazirino groups at 6’-OH also produced inhibitors of GnT I (39, Scheme 16) with K, values of 1.1-10 mM [99]. These inhibitors can potentially be used for photolabelling residues of the enzyme active site.

306

17 Glycosyltrunsferase Inhibitors

PGlcNAc( 1+2)aMan(l-+6 PMan( 1+4)PGlcNAc( 1+4)PGlcNAc-Asn PGlcNAc(1+2)aMan(l+3/ natural acceptor

m

k

GnT V

@lcNAc( 1+2)aMan( 1-6

I

UDP-GlcNAc

PMan(l+4)PGlcNAc( 1+4)PGlcNAc-Asn pGlcNAc(I+Z)aMan( 1+3/

HO

HO

HO

NHAc

synthetic acceptor 32 PGlcNAc(l+Z)aMan( 146)PGlcNAc-OR K, = 23 - 36 pM

Scheme 14. Natural and synthetic acceptors for N-acetylglucosaminyltransferase V.

Analog 40 with the reactive OH group deoxygenated is a competitive inhibitor of GnT I1 with a K, of 0.13 niM, the same as the K, of the corresponding acceptor substrate. Competitive inhibitors of GnT I1 (41, Ki = 1.0-2.5 mM) were also obtained by introducing large pentyl or substituted pentyl groups containing reactive functionalities [ 1001. Interestingly, trisaccharide 42 which lacks the mannose residue shows relatively strong binding to the enzyme with a Ki of 0.9 mM, indicating that the mannose residue to which GnT I1 transfers GlcNAc is not required for recognition [ 1001.

17.2.5 Inhibitors of Human Blood Group A and B Glycosyltransferases Human blood group A and B glycosyltransferases (GTA and GTB, E.C. 2.4.1.40 and E.C. 2.4.1.37) are responsible for the biosynthesis of A and B blood-group antigens which are important in cell development, cell differentiation, and oncogenesis [ 101-1041. GTA catalyze the transfer of GalNAc from UDP-GalNAc to the (0)H-precursor structure (F:uca1+2Gal) to give the A determinant GalNAcw1 3[Fucal+2]Gal, while GTB uses the same acceptor but catalyze the transfer of Gal from UDP-Gal to form the B determinant Gala1 +3[Fucal42]Gal. ---f

17.2 Inhihitors oJ Gl~~osyltransferuse,s

&

HO

HO

NHAc W

HO HO &

OH

6

HO

HO

2

307

H

NHAc

HO HO &

O(CHz)7CH3

33

MOCH OH

OH

O(CHd7CH3

34

Kt=30-70pM

K, = I4 pM

HO HO

HOHO &

OH

OR2

36 (RI = NHR, Rz=Oct) K,= 21 - 297 pM 36a (RI = NHCOCHZI, R2 = a t ) : K, = 21 pM 37 (R,=OCH,,R,=PNP): 53% inhibition at 0.8 MM

Scheme 15. Inhibitors of N-acetylglucosaminyltransferase V.

Ho HO &

H

R10

GnT I

W 38 (R,=Me): K, = 0.76 mM

E

@ OH 39a-e: R1 =

y

C

H

3

0

K , = 1.1 - l0mM

Scheme 16. Inhibitors of N-acetylglucosaminyltransferase 1.

-

-----Yo

-OH

.---.N=N 9+

NHCOCH~I

reactive group

308

I7 Glycosyltransferase Inhibitors

un-

HOLoY

40(R,=R2=H):K,=0.l3mM 4 1 (R,= OH, R2 = pentyl or reactive group): K,=l.O-ZSmM

Scheme 17. Inhibitors of N-acetylglucosaminyltransferase 11.

Systematic "key polar group" mapping studies reveal that the reactive 3-OH of the Gal unit is not required for recognition by either enzyme [105]. Analogs 43a-d with 3-OH deoxygenated, substituted, or methylated are inhibitors for both enzymes as shown in Table 1. Compound 43c with 3-OH substituted with an amino group has an estimated K, of 200 nM with GTA [105]. Since both GTA and GTB are also retaining glycosyltransferases, such high inhibitory activity might result

Table 1. Inhibition constants ( K,, in pM) of analogs 43a-d with blood group A and B glycoslytransferases (GTA and GTB).

RZ Ho&O(CH&CH3

HO

&

HO

HO

OH

43a-d

Rl

R2

Inhibitor

GTA

GTB

H H H OH

H F

43a 43b 43c 43d

68 48 0.2" 22"

14 110

Estimated K,

NH2 H

5" 7.8

from favored electrostatic interaction between the positively charged amino function and a negatively charged group in the active site of the enzymes as was suggested for a1,3-GalT.

17.3 Summary The development of glycosyltransferase inhibitors is a newly emerging area, stimulated from their potential as therapeutic agents and as tools to study enzyme mechanisms. A number of substrate and transition-state analogs have been developed as glycosyltransferase inhibitors in the past decade. The inhibition constant ( K , )for most of them, however, is still in the range 10-1000 pM, similar to the K,,, values of the corresponding natural donor and acceptor substrates. The problems of specificity, stability and permeability of the inhibitors remain to be addressed. The lack of structural information on glycosyltransferases, the intrinsic low binding affinity between carbohydrates and enzymes, and laborious synthetic work makes the development of potent and specific inhibitors of glycosyltransferases an ongoing challenge. Recent advances in combinatorial carbohydrate chemistry [ 106, 1071, and high through-put screening technology [ 108-1 1 11 will promote the discovery of much more potent glycosyltransferase inhibitors which incorporate additional molecular features (hydrophobic, charged, etc.). The increasing availability of recombinant glycosyltransferases will also facilitate research in this area. Given the significant potential of such inhibitors, there can be little doubt that drugs based on glycosyltransferase inhibitors will be forthcoming.

Acknowledgments We thank Dr. Ole Hindsgaul for his helpful advice and critical reading of the manuscript. We gratefully acknowledge the Natural Sciences and Engineering Research Council of Canada for the financial support (to M.M.P) and the Alberta Research Council for a graduate scholarship in Carbohydrate Chemistry (to X.Q.).

References 1 . Palcic, M. M. Methods Enzyrnol. 1994, 230, 300. 2. Wong, C.-H.; Halcomb, R. L.; Ichikawa, Y.; Kajimoto, T. Angew. Clwnz., Int. Ed. Engl. 1995, 34, 521. 3. Brockhausen, 1.; Schachter, H. In Glycoscirnces; Gabius, H.-J., Gabius, S. Eds.; Chapman & Hall: Weinheim, 1997; pp 79- 1 13. 4. Varki, A. Glycobiology 1993, 3, 97. 5. Dwek, R. A. Clzern. Rec. 1996. 96, 683. 6. Khan, S. H.; Matta, K. L. In Glycoconjugcites: Composition, Structure, and Function, Allen, H. J.; Kieselius, E.C. Eds.; Marcel-Dekker: New York. 1992: pp 361-378.

3 10

17 Glycosyltransfrruse Inhibitors

7. Khan, S. H.; Hindsgaul, 0. In Moleculur Glycohioloqy; Fukuda, M.; Hindsgaul, 0. Eds.; IRL Press: Oxford, 1994; pp 206-229. 8. Sears, P.; Wong, C.-H. Chem. Commun. 1998, 1161. 9. Elhalabi, J. M.; Rice, K. G. Curr. Med. Chewr. 1999, 6, 93. 10. Goss, P. E.; Baker, M. A,; Carver, J. P.; Dennis, J. W. Clin. Cancer Res. 1995, I , 935. 11. Sears, P.; Wong, C.-H. Cell. Mol. L f i Sci. 1998, 54, 223. 12. Perez, P.; Varona, R.; Garcia-Acha. I.; Duran, A. FEBS Lett. 1981, 129. 249. 13. Perez, P.; Garcia-Acha, I.; Duran, A. J. G m Microhiol. 1983, 12, 245. 14. Cabib, E.; Kang, M. S.; Au-Young, J. Methods Enzyniol. 1987, 138, 643 15. Heifetz, A,; Keenan, R. W.; Elbein, A. D. Biochcwistry 1979, 18, 2186. 16. Schwarz, R. T.; Datema, R. Ado. Curhohyttr. Chem. Biochem. 1982, 40, 321. 17. Lehrman, M. A. Glycohiohqy 1991, I , 553. 18. Zollner, H. Hutidhook qf’Enzyme Inhibitors; VCH: Weinheim, 1990. 19. The crystal structure of a recombinant DNA modifying enzyme -glucosyltransferase has been reported: Vrielink, A,; Ruger, W.; Driessen, H. P. C.; Freemont. P. S. EMBO J. 1994, 13, 3413. 20. Vaghefi, M. M.; Bernacki, R. J.; Dalley, N. K.; Wilson, B. E.; Robins, R. K. J. Med. Chem. 1987, 30, 1383. 21. Vaghefi, M. M.; Bernacki, R. J.; Hennen, W. J.; Robins, R. K. J. Med. Chem. 1987, 30, 1391. 22. Palcic, M. M.; Heerze, L. D.; Srivastava, 0. P.; Hindsgaul, 0. J. Biol. Chem. 1989, 264, 17174. 23. Hindsgaul, 0.;Kaur, K. J.; Srivastava, G.;Blaszczyk-Thurin, M.; Crawley, S. C.; Heerze, L. D.; Palcic, M. M. J. Bid. Chem. 1991, 266, 17858. 24. Lemieux, R. U. In Proc. VIIIth Int1 Symp. Med. Chem.; Swedish Parmaceutical Press: Stockholm, 1985; pp. 329-351. 25. Lemieux, R. U. Chem. Soc. Rev. 1989, 18, 347. 26. Lemieux, R. U. Acc. Chem. Res. 1996, 29, 373. 27. Schmidt, R. R., Frische, K. Bioorg. Med. Chem. Lett. 1993, 3, 1747. 28. Hashimoto, H.; Endo, T.; Kajihara, Y. J. Org. Chem. 1997, 62, 1914. 29. Nishiura, T.; Fujii, S.; Kanayama, Y.; Nishikawa, A.; Tomiyama, Y.; Iida, M.; Karasuno, T.; Nakao, H.; Yonezawa, T.; Taniguchi, N.; Tarui, S. Cancer Rex 1990, 50, 5345. 30. Furukawa, K.; Matsutd, K.; Takeuchi, F.; Kosuge, E.; Miyamoto, T.; Kobata, A. Znt. Immunol. 1990,2, 105. 31. Nudelman, E.; Kannagi, R.; Hakomori, S.; Parsons, M.: Lipinski, M.; Wiels, J.; Fellous, M.; Tursz, T. Science 1983,220, 509. 32. Kim, Y. S.; Perdomo, J.; Whitehead, J. S.; Curtis, K. J. J. Clin. Invest. 1972, 51, 2033. 33. Yuasa, H.; Palcic, M. M.; Hindsgaul, 0. Can. J. Chem. 1995, 73, 2190. 34. Wang, R.; Steensma, D. H.; Takaoka Y.; Yun, J. W.; Kajimoto, T.; Wong, C.-H. Bioorg. Med. Chem. 1997, 5, 661. 35. Morrison, J. F.; Ebner, K. E. J. Biol. Chem. 1971, 246, 3977. 36. Hayashi, T.; Murray, B. W.; Wang, R.; Wong, C.-H. Bioorg. Med. Chem. 1997, 5, 497 37. Endo, T.; Kajihara, Y.; Kodama, H.; Hashimoto, H. Bioorg. Med. Chem. 1996, 4, 1939. 38. Kim, S. C.; Singh, A. N.; Raushel, F. M. Arch. Biochem. Biophys. 1988, 267, 54. 39. Sinnott, M. L. Chem. Rev. 1990, 90, 1171. 40. Sinha, S. K.; Brew, K. Curbohydr. Res. 1980, 81, 239. 41. Palcic, M. M.; Srivastava, 0. P.; Hindsgaul, 0. Curhohydr. Rex 1987, 159, 315. 42. Kajihara, Y.; Kodama, H.; Endo, T.; Hashimoto, H . Curhohydr. Rex 1998, 306, 361. 43. Kajihara, Y.; Hashimoto, H.; Kodama, H. Carhohydr. Res. 1992,229, C5. 44. Blanken, W. M.; Van den Eijnden, D. H. J. Bid. Chem. 1985,260, 12927. 45. Joziasse, D. H.; Shaper, N. L.; Salyer, L. S.; Van den Eijnden, D. H.; van der Spoel, A. C.; Shaper, J. H. Eur. J. Biochem. 1990. 191, 75. 46. Fang, J.; Li, J.; Chen, X.: Zhang, Y.; Wang, J.; Guo Z,; Zhang, W.; Yu, L.; Brew, K.; Wang, P. G. J. Am. Chem. Soc. 1998, 120, 6635. 47. Sujino, K.; Malet, M.; Hindsgaul, 0.;Palcic, M. M. Curbahydr. Rex. 1998, 305, 483. 48. Galili, U. Immunol. Today 1993, 14, 480. 49. Butler, D. Nuture 1998, 391, 320.

50. Lowary, T. L.; Hindsgaul. 0. Curhohydr. Kes. 1993, 249, 163. 51. Helland, A,-C.; Hindsgaul, 0.;Palcic, M. M.; Stults, C. L. M.; Macher. B. A. Curhohpa'r. Rex 1995. 276, 91. 52. Kim, Y. J.; Ichikawa, M.; Ichikawa, Y. J. Am. Chem. Soc. 1999, / 2 I , 5829. 53. Hakomori, S.; Nudelman, E.; Levery, S. B.; Kannagi, R. J. Biol. Clicm. 1984, 259, 4672. 54. Hakomori, S. Adu. Ccirmr Rex 1989. 52, 257 55. Feizi, T. Nature 1985, 314, 53. 56. Le Pendu J., Cartron, J. P.; Lemieux, R. U. Ani. J. Hun?. Genet. 1985, 37, 749. 57. Sarnesto, A,; Kohlin, T.; Thurin, J.; Blaszczyk-Thurin, M. J. Biol. Cllern. 1990, 265, 15067. 5 8 . Sarnesto, A,; Kohlin, T.; Hindsgaul. 0.; Thurin, J.; Blaszczyk-Thurin, M. J. Bid. Chem. 1992,267, 2737. 59. Watkins, W. M.; Greenwell, P. Yatea. A. D.; Johnson, P. H. Biochinzie 1988, 70, 1597. 60. Lowary, T. L.; Swiedler, S. J.; Hindsgaul, 0. Curho/z,vn'r. Rex 1994, 256, 257. 61. Philips, M. L.; Nudelman, E.; Gata. F. C. A,: Perez, M.; Singhal, A. K.; Hakomori, S.-I.; Paulson, J. C. Science 1990, 250, 1 130. 62. Walx, G.; Arrufo, A,; Kolanus, W.; Bevilacqua, M.; Seed, B. Science 1990, 250, 1132. 63. Wong, C.-H.: Dumas, D. P.; Ichikawa, Y.; Koseki, K.; Danishefsky, S. J.; Weston, B. W.; Lowe, J. B. J. Am. C/iern. Soc. 1992, 114. 7321. 64. Ichikawa, Y.; Lin, Y.-C.; Dumas, D. P., Shen, G. -J.; Garcia-Junceda, E.; Williams, M. A,; Bayer, R.; Ketcham, C.; Walker, L. E.; Paulsou, J. C.; Wong, C.-H. J. A m . Cliern. SOC.1992. 114, 9283. 65. Qiao, L.; Murray, B. W.; Shimazaki, M.; Schultz. J.; Wong, C.-H. J. Am. Clzern. Soc. 1996, 118, 7653. 66. Jefferies, 1.; Bowen, B. R. Bioorg. Metl. Chenz. Lett. 1997, 7 , 1171. 67. Heskamp, B. M.; Veeneman, G. H.; van der Marel, G. A,; van Boeckel, C. A. A,; van Boom, J. H. Tetrahedron 1995, 51, 8397. 68. Heskamp, B. M.; van der Marel, G. A,; van Boom, J. H. J. Curbohydr. Cliem. 1995, 14, 1265. 69. Hardium-Lepers, A,; Recchi, M.-A.; Delannoy, P. Glycohiology 1995. 5, 741. 70. Reuter, W.; Stasche, R.; Stehling, P., Baum, 0. In Glycosciences; Gabius, H.-J., Gabius, S . Eds.; Chapman & Hall: Weinheim, 1997; pp 245-259. 71. Bernacki, R. J. Science 1977, 195, 557. 72. Zhou, Q.; Hakomori. S.; Kitamura, K.; Igrashi, Y. J. Bid. Chenz. 1994, 269, 1959. 73. Gessner, P.; Riedl, S.; Quentmaier, A,; Kemmer, W. Cancer Lett. 1993, 75, 143. 74. Paulson, J . C.; Rearick, J. L.; Hill, R. L. J. Rid. Chez. 1977, 252, 2363. 75. Schaub, C.; Muller, B.; Schmidt, R. R . Glj~coconjugateJ. 1998, 15, 345. 76. Imamura, M.; Hashimoto, H. Tefruhcdron Lett. 1996, 37, 1451. 77. Miiller, B.; Martin, T. J.; Schaub, C.; Schmidt, R. R . Tetrahedron Lett. 1998, 39, 509. 78. Horenstein, B. A.; Bruner, M. J. Am. C h i w SOL..1996, 118, 10371. 79. Bruner, M.; Horenstein, B. A. Bioclzernlstrj~1998, 37, 289. 80. Amann, F.; Schaub, C.; Miiller, B.; Schmidt, R . R. Chem. Eur. J , 1998, 4. 1106. 81. Miiller, B.; Schaub, C.; Schmidt, R. R. Arzqeiv. Chem. Int. Ed. Ennyl. 1998. 37, 2893. 82. Kajihara, Y.; Hashimoto, H.; Kodama, H.; Wakabayashi, T.; Sato, K. J. Curbohydr. Cllenz. 1993, 12, 991. 83. Kajihara, Y .; Kodama, H.; Wakabayashi, T.; Sato, K.; Hashimoto, H. Curbohydr. Res. 1993, 247, 179. 84. Van Dorst, J. A. L.M.; Tikkanen, J. M.; Krezdorn. C. H.; Streiff, M. B.; Berger, E. G.; Van Kuik, J. A,; Kamerling, J. P.; Vliegnthart. J. F. G. Eur. J. Biochem. 1996, 242, 674. 85. Taniguchi, N.; Ihara, Y. Glycoconjugute J . 1995, 12, 733. 86. Yamshita, K.; Tachibana. Y.; Ohkura, T.; Kobata, A. J. Bid. Chem. 1985, 260, 3963. 87. Arango, J. Pierce, M. J. Cell. Biorhern. 1988, 37, 225. 88. Dennis, J. W.; Kosh, E.; Bryce, D.-M.; Breitman, M. L. Oncogene 1989, 4, 853. 89. Dennis, J. W.; Laferte, S.;Waghorne. C.; Breitman, M. L.; Kerbel, R. S. Science 1987, 236, 582. 90. Dennis, J. W. Cancer Surveys 1988, 7, 573. 91. Cummings, R. D.; Trowbridge, I. S. J. Bid. Chem. 1982, 257, 13421. 92. Schachter, H. Biochem. Cell Bid. 1986, 64, 163.

3 12

I 7 Glycosyltransjerase Inhibitors

93. Hindsgad, 0.;Tahir, S. H.; Srivastava, 0. P. Carbohydr. Res. 1988, 173, 263. 94. Khan, S. H.; Crawley, S. C.; Kanie, 0.;Hindsgaul, 0. J. Biol. Chem. 1993, 268, 2468. 95. Palcic, M. M.; Ripka, J.; Kaur, K. J.; Shoreibah, M.; Hindsgaul, 0.;Pierce, M. J. Bid. Chem. 1990,265, 6759. 96. Lu, P.-P.; Hindsgaul, 0.;Compston, C. A,; Palcic, M. M. Bioorg. Med. Chem. 1996, 4, 2011. 97. Lu, P.-P.; Hindsgaul, 0.;Li, H.; Palcic, M. M. Carhohydr. Rex 1997, 303, 283. 98. Brockhausen, I.; Reck, F.; Kuhns, W.; Khan, S.; Matta, K. L.; Meinjohanns, E.; Paulsen, H.; Shah, R. N.; Baker, M. A,; Schachter, H. Glycoconj. J. 1995, 12, 371. 99. Reck, F.; Springer M.; Meinjohanns, E.; Paulsen, H.; Brockhausen, I.; Schachter, H. Glycoconj. J. 1995, 12, 147. 100. Reck, F.; Meinjohanns, E.; Springer M.; Wilkens, R.; Van Drost J. A. L. M.; Paulsen, H.; Moller, G.; Brockhausen, I.; Schachter, H. Glycoconj. J. 1994, 11, 210. 101. Yamamoto, F.4.; Clausen, H.; White, T.; Marken, J.; Hakomori, S . 4 . Nutwe 1990, 345, 229. 102. Yamamoto, F.4.; Hakomori, S. J. Biol. Chem. 1990, 265, 19257. 103. Seto, N. 0. L.; Palcic, M. M.; Hindsgaul, 0.;Bundle, D. R.; Narang, S. N. Eur. J. Biochem. 1995,234, 323. 104. Seto, N. 0. L.; Palcic, M. M.; Compston, C. A.; Li, H.; Bundle, D. R.; Narang, S. A. J. Biol. Chem. 1997,272, 14133. 105. Lowary, T. L.; Hindsgaul, 0. Carbohydr. Res. 1994,251, 33. 106. Wang, Z.-G.; Hindsgaul. 0. In Glycoinzn?unoloyy2; Axford, J. S. Ed.; Plenum Press: New York, 1998; pp 219-235. 107. Ito, Y.; Manabe, S. Curr. Opin. Chem. Biol. 1998, 2, 701. 108. Liang, R.; Yan, L.; Loebach, J.; Ge. M.; Uozumi, Y.; Sekanina, K.; Horan, N.; Gildersleeve, J.; Thompson, C.; Smith, A,; Biswas, K.; Still, W. C.; Kahne, D. Science 1996, 274, 1520. 109. Takayama, S.; Martin. R.; Wu, J.; Laslo, K.; Siuzdak, G.; Wong, C.-H. J. Am. Chem. Soc. 1997,119, 8146. 110. Schriemer, D. C.; Bundle, D. R.; Li, L.; Hindsgaul, 0. Anyew. Chem. Int. Ed. Engl. 1998,37, 3383. 11 1. Schriemer, D. C.; Hindsgaul, 0. Combi. Chem. High Throughput Screening 1998, I , 155.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

18 Biosynthesis of the 0-Glycan Chains of Mucins and Mucin Type Glycoproteins Inka Brockhausen

18.1 Summary Mucins and mucin-like glycoproteins are rich in GalNAca-Ser/Thr-linked oligosaccharides (0-glycans). and are found on cell surfaces and in cellular secretions with diverse biological functions. 0-glycan structures are cell type specific and often change during growth and differentiation as well as in many disease states. The complex 0-glycans are assembled by a series of specific reactions involving glycosyltransferases and sulfotransferases that exist as families of related enzymes. The control of the expression of these enzymes as well as their distinct properties are still poorly understood. With the gene cloning and characterization of transferases, and studies of the inter-relationship of enzymes in models of disease, we are beginning to unravel the regulation of biosynthetic pathways, and ultimately the functions of complex mucin-type glycans.

18.2 Tntroduction GalNAc-Ser or -Thr 0-linked oligosaccharides (0-glycans) occur on soluble, secreted and membrane bound glycoproteins. Mucins, comprising the main class of these glycoproteins, have heavily 0-glycosylated Ser/Thr/Pro-rich tandem repeat regions and, with the exception of the membrane-bound MUC1, are secreted glycoproteins. Eight different human mucin genes (MUC1,2,3,4,5A,5B,6,7)have been sequenced and another type, MUC8, has been identified [1-3]. The expression of these glycoprotein genes is often cell-specific, and may change in disease. Mucinlike glycoproteins also have domains that are rich in 0-glycans but lack the typical mucin tandem repeat sequences and are less heavily 0-glycosylated. Examples of cell surface-bound mucin-like glycoproteins are leukosialin and PSGL- 1 , found on leukocytes, and cell adhesion molecules MadCAM-1 and GlyCAM-1, found on endothelial cells [4].

314

18 Biosynthesis of the O-Glycan Chains

A number of diseases are associated with changes in mucin expression and mucin carbohydrate structures, including inflammatory disease, ulcerative colitis, cystic fibrosis and epithelial tumors. In cancer, the expression of mucins may be up- or down-regulated, or switched to mucins that are normally expressed in other cell types [S-7]. For example, the mRNA levels for MUC2 and 3 are increased in colon cancer [S] while mRNA levels for MUC1, 2, 4 and S are increased in pancreatic cancer [9]. Carbohydrate as well as peptide antigens of cancer mucins are often changed due to altered glycosylation [3, lo]. This may lead to a drastic change in the biological and immunological properties of cancer cells. Although specific antigens of complex O-glycan chains such as ABO blood groups and sialyl-Lewis" may be increased in tumors [S, 111, O-glycans often appear to be truncated. Complex O-glycans are synthesized by series of glycosyltransferase and sulfotransferase reactions that process glycan chains to hundreds of specific O-glycan structures [5, 12, 131. In the biosynthesis of O-glycans, sugars are added individually by transfer from nucleotide sugars to the growing O-glycan chain, without the involvement of dolichol-intermediates, which distinguishes it from the biosynthesis of GlcNAc-Asn-linked oligosaccharides (N-glycans). The cloning of glycosyltransferase genes, and studies of their distinct expression as well as enzyme properties are helpful in determining the factors regulating O-glycan biosynthesis. The enzymes involved in the biosynthesis of O-glycans will be reviewed here, as well as some of the mechanisms leading to abnormal mucin expression in disease. The synthesis of other O-glycans, based on mannose-Ser/Thr, xylose-Ser/Thr, glucose-O-Ser/Thr or GlcNAc-O-Ser/Thr, will be discussed elsewhere.

18.3 Structures of O-Glycans All mucin type O-glycans have GalNAca-linked to Ser or Thr residues. The unprocessed GalNAc-Ser/Thr exists as the cancer-associated Tn antigen. GalNAcpeptide may be sialylated to form the sialyl-Tn antigen, sialyla2-6GalNAc-, or may be converted to eight different core structures of mucins (Figure 1). Core structures 1 and 2 are probably the sole structures found on non-much glycoproteins. O-glycans may be elongated by repeating Gal and GlcNAc residues (i sialic acid and I antigens), and may be terminated by Gal, GalNAc, GlcNAc, FUC, residues and sulfate esters. Many of the carbohydrate antigens found on mucins are developmentally regulated, and expressed in a tissue-specific fashion.

18.4 Functions of Mucin Type O-Glycans Due to the high degree of glycosylation of mucins, they may form gels which protect underlying epithelial cell surfaces. Mucin O-glycans may be involved in the

18.6 Synthesis of O-Glycan Core I

3 15

maintenance of protein conformations and epitopes. They form ligands for cell adhesion, microbes, sperm-egg binding, and play important roles in the immune system [ 5 , 141. O-glycans represent cancer-associated antigens and regulate the attachment and invasion of cancer cells [3, 5 , 10, 151. Altered antigenicity of mucins may be useful for cancer diagnosis; mucins also have potential in immunotherapy for cancer.

18.5 Primary O-Glycosylation The first step of O-glycosylation is catalyzed by enzymes comprising a family of polypeptide a-GalNAc-transferases. Genes encoding this ubiquitous enzyme have been cloned from several mammalian species [ 16, 171. Polypeptide a-GalNAc-transferase is located in the Golgi membrane with a membrane anchor domain; it has a short amino-terminal end directed towards the cytoplasm, and a carboxy-terminal end containing the catalytic domain within the Golgi lumen. This type I1 membrane protein domain structure is found in all cloned glycosyltransferases assembling O-glycans. Studies of Gal- and sialyltransferases [18] show that the anchor domain as well as adjacent sequences, and possibly also distant peptide regions appear to control the localization of transferases in the appropriate compartment in the Golgi. Polypeptide a-GalNAc-transferases from porcine and bovine submaxillary gland have been localized to the cis-Golgi compartment using immuno-electronmicroscopy methods [ 191. However, the subcellular distribution may depend on the cell type, and on the cellular differentiation status, and may be much broader in certain cells [20]. O-glycosylation does not appear to require a specific amino acid sequence in the acceptor substrate, in contrast to N-glycosylation which requires the Asn-X-Ser/Thr sequon. The different polypeptide a-GalNAc-transferases have mostly overlapping but distinct substrate specificities. While most enzymes prefer Thr-containing substrates in vitro, polypeptide GalNAc-transferase T4 has significant activity towards Ser as well [ 16, 17).The enzyme is discussed in more detail in Chapter 16.

18.6 Synthesis of O-Glycan Core 1 O-glycan chains based on core 1 (Figure 1) occur on most mucins and glycoproteins of many species. Core 1 may be elongated and substituted with various sugars, or can be mono- or disialylated. The unsubstituted core 1 is often found in cancer as the Thomsen-Friedenreich (T) antigen. Core 1 p3-Gal-transferase is responsible for the synthesis of core 1 and occurs in most mammalian cells 113, 211. Erythrocytes from patients with permanent mixed-field polyagglutinability. human T-lymphoblastoid cells Jurkat [22], and human colon cancer cells LSC [23] lack the enzyme and therefore highly express unsubstituted GalNAc residues (Tn antigen). In certain

3 16

18 Biosynthesis oftlie O-Gl-ycun Chuins GlcNAc

I !36 Galp3GalNAc-

2-

t. GalpGalNAc-

GlcNAc

I B6 GlcNAcp3GalNAc-

i

GlcNAcp3GalNAc-

GalNAca3GalNAc-

GlcNAcp6GalNAc-

GC2C&

GaldGalNAc-

!z?a

Ia?m

G ~ ~ N A ~ -

ThrDer-peptide

Figure 1. Biosynthesis of mucin type 0-glycans. In the biosynthesis of 0-glycan chains, GalNAc is always the first sugar added by polypeptide aGalNAc-transferase (path a) to Ser or Thr of the mucin or glycoprotein peptide substrate. Eight core structures can then be formed, depending on the expression and activities of the core synthesizing glycosyltransferases. Core 1 is formed by core 1 P3-Gal-transferase (path b), which is the precursor structure for the formation of core 2 by core 2 P6-GlcNAc-transferase (path c), In competition with core 1 synthesis, core 3 P3-GlcNAc-transferase synthesizes core 3 (path d). Core 3 may be converted to core 4 by core 4 136-GlcNAc-transferase (path e), an enzyme activity which may be due to core 2 P6-GlcNAc-transferase M. Four other core structures ( 5 to 8) have been found on mucins. However, the core 5 a3-GalNAc-transferase (path f ) , core 6 P6-GlcNAc-transferase (path g), core 7 a6-GalNAc-transferase (path h) and core 8 a3-Gal-transferase (path i) activities synthesizing cores 5 to 8, respectively, remain to be characterized.

cell types, core 1 P3-Gal-transferase activity can be induced [24] or altered [25] by the induction of differentiation. GalNAcw-aryl compounds are effective substrates for the enzyme. Since these compounds may penetrate cell membranes and interfere with core 1 synthesis of natural mucin substrates, they have been successfully used as 0-glycosylation inhibitors. Thus cells treated with GalNAc-benzyl exhibit increased exposure of GalNAc-mucin [26]. Human colon cancer cells treated with GalNAc-benzyl adhere less to E-selectin and show reduced liver metastases in nude mice [27] and in other systems. This suggests that mucin 0-glycans play an essential role in the biology of the cancer cell. Core 1 p3-Gal-transferase recognizes specific substitutents of the sugar ring of GalNAcw-benzyl substrate [28]. The activity is also greatly influenced by peptide sequences of glycopeptide substrates as well as their glycosylation patterns [29, 301. Several other glycosyltransferases of the early 0-glycosylation pathways show specificity for the substrate peptide sequence and glycosylation [31, 1251. It appears therefore that the individual glycosylation sites of a mucin substrate dictate not only the primary glycosylation but also subsequent 0-glycosylation steps.

317

18.7 Synthesis ofO-Glycan Core 2

18.7 Synthesis of 0-Glycan Core 2 0-glycans may have a number of branches, formed by the addition of GlcNAc in p1-6 linkage to Gal or GalNAc in core structures 2 and 4, and the I antigen (Figures 1 and 2) [13, 321. Linear structures of GlcNAc(31-6Gal(NAc) have also been reported. It appears that the p6-GlcNAc-transferases synthesizing these branches exist as a large family of enzymes with different specificities and tissue distributions. Core 2 P6-GlcNAc-transferase (Figure 1) appears to have a number of important functions. The enzyme is altered in leukemia, cancer and upon cell differentiation, and plays important roles in the immune system. The activity appears to be induced in the heart tissue of diabetic rats, in comparison to normal rat hearts [33].Core 2 GlcNAc

-c-

Further elonaation and terminationreactlong

I P6 Galp3GalNAc-

\

I

GlcNAc

Ga1p4GlcNAcp3Ga1p4GIcNAc

\

I 66 GlcNAcp3Galp4GlcNAcp3Galp3GalNAc-

GlcNAc e

IB6

G1cNAcp3Ga1p3Ga1NAc-

GlcNAcp3Galp4GlcNAcp3Galp3GalNAo

td tc

I Galp3GalNAcGlcNAc

IPS t

I B6 Ga1p4GlcNAcp3Ga1p4GIcNAc

GlcNAcp3Galp4GlcNAc

te

Galp4GlcNAcp3Galp3GalNAc-

I P6 Galp3GalNAc-

GlcNAc

Galp4GlcNAc

166 GlcNAcp3Galp3GalNAc-

.

I A& v-

Galp3GalNAc-

GlcNAcp3Galp3GalNAc-

tb

Galp3GalNAc-

GlcNAc I P6 ____o Galp3GalNAca

EpLez

Figure 2. Elongation and branching pathways of m u c h 0-glycan cores 1 and 2. 0-glycan core structures 1 to 4, and possibly others as well, may be elongated by repeating linear and branched Gal-GlcNAc sequences. Some of the elongation and branching reactions of core 1 and 2 structures are shown. Path a, core 2 (36-GlcNAc-transferase; path b, elongation P3-GlcNActransferase; path c, P4-Gal-transferase; path d, i (33-GlcNAc-transferase; path e, branching (36GlcNAc-transferase which may be identical to the I (36-GlcNAc-transferase; path f. internally acting I (36-GlcNAc-transferase. Other enzymes may be involved in elongation and branching. for example, b3-Gal-transferases and other (33-and (36-GlcNAc-transferase, as well as GalNAc-transferases.

3 18

18 Biosynthesis of the O-Glycun Chuins

P6-GlcNAc-transferase activity is regulated during activation of lymphocytes [ 341, and differentiation of murine and human cancer cells [25, 351. The activity is increased in leukemia cells which are aberrantly differentiated [36, 371 and is abnormal in immunodeficiency disease [38]. Repopulation of SCID mice with bone marrow cells that overexpress core 2 p6-GlcNAc-transferase results in impairment of myeloid differentiation [39]. This indicates that core 2 P6-GlcNAc-transferase is critically involved in the process of differentiation. Transgenic mice overexpressing the enzyme in T-lymphocytes show a number of malfunctions of the immune response, including reduced delayed-type hypersensitivity as well as poor proliferation and cytokine production upon stimulation [40]. This suggests that the enzyme plays an important role in the immune system, possibly related to its role in the process of cell adhesion. Lymphocytes adhere to the endothelium via selectin binding to sialyl-Lewis" and related ligands. It has been shown in several cell types that core 2 P6-GlcNAc-transferase is critical for the expression of selectin ligands which appear to be mainly linked to core 2 structures [41, 421. Thus, the branched core 2 structure regulates the expression of distant determinants; it provides additional sites for the attachment of poly-N-acetyllactosamine chains, as well as for cell adhesion ligands and antigens. In addition, the bulky core 2 structures may mask the recognition of underlying peptide by antibodies. In breast cancer cells T47D and BT20, mRNA for core 2 p6GlcNAc-transferase is not expressed [43], thus core 2 structures are not made. As a consequence, peptide epitopes are accessible by antibodies SM3. The cDNA encoding an enzyme with core 2 P6-GlcNAc-transferase activity has been cloned from HL60 cells [44] and bovine lung [45]. Three regions show significant homology to the cloned I P6-GlcNAc-transferase (Figure 2) [46]. When the soluble portion of the enzyme is expressed in insect cells, the two putative Nglycosylation sites have to be occupied for full activity and stability of the enzyme [47]. The genes of both the core 2 and the I P6-GlcNAc-transferases are located on the same chromosome (9 band q21), and may have evolved from the same ancestral gene [48]. A related pseudogene also exists [49]. The gene for core 2 P6-GlcNActransferase is divided into two exons. In contrast to P3-GlcNAc-transferases, the P6-GlcNAc-transferases do not require Mn2+ for optimal activity in vitro. The specificities of core 2 P6-GlcNActransferase activities differ slightly between cell types [50, 511. This may be due to the expression of cell type-specific P6-GlcNAc-transferases, or due to the expression of a combination of related enzymes in the same cell. The specificity of the enzyme was studied in leukocytes from patients with acute myeloid leukemia, which appear to contain only core 2 06-GlcNAc-transferase (L enzyme) but not other branching activities [36]. The enzyme recognizes most of the substituents of the sugar ring of core 1 as a substrate, and the 4- and 6-hydroxyls of both Gal and GalNAc, as well as the 2-N-acetyl group, are essential for full activity. A substrate derivative lacking the 6-hydroxyl of GalNAc was found to be a weak inhibitor [50],while GalPl3GalNAca-p-nitrophenyl substrate when activated by UV light becomes a potent irreversible inhibitor [52]. In contrast. core 2 P6-GlcNAc-transferase (M enzyme) in mucin secreting tissues is usually accompanied by other branching activities, which may originate from one or more enzyme activites [50, 511.

Recombinant human core 2 P6-GlcNAc-transferase L has been expressed in CHO cells and was immuno-localized to cis to medial Golgi compartments [53]. The enzyme has a similar distribution in the Golgi of normal and cancerous human mammary cells [ 1261.

18.8 Synthesis of O-Glycan Core 3 O-glycan chains with core 3 structures (Figure 1) occur in colonic, lung and salivary mucins [21, 541. Core 3 is synthesized by core 3 P3-GlcNAc-transferase, an enzyme that is active in colonic tissues from several species [55]. The enzyme activity is detectable but reduced in colon cancer tissue 156, 571. This may be part of the mechanism that shifts O-glycan synthesis in cancer cells towards core 1 (Figure 1) with a prevalence for the T antigen expression in cancer tissue [SS]. For reasons that remain to be determined, core 3 a3-GlcNAc-transferase activity could not be detected in cultured cells derived from colon cancer tissue [25, 51, 591. The enzyme is very specific for GalNAc-R substrates and requires all the substituents of the GalNAcring, with the exception of the 6-hydroxyl, for activity (511.

18.9 Synthesis of O-Glycan Core 4 The P6-GlcNAc-transferase activity synthesizing O-glycan core 4 (Figure 1) is always accompanied by core 2 P6-GlcNAc-transferase activity, although the ratio of these two activities differs between tissues. Cores 2 and 4 may therefore be synthesized by the same enzyme. Although core 4 P6-GlcNAc-transferase activity is relatively abundant in mucin secreting cell types, core 4 structures of mucins are found only in selected tissues, such as the intestine, lung and salivary gland. One reason for this is the relatively restricted distribution of core 3 which is the precursor structure for core 4 synthesis (Figure 1). Core 4 P6-GlcNAc-transferase activity is decreased in colon cancer tissue [56] and in tumor cells derived from of human polyposis coli cell line, relative to core 2 synthesis [59]. It remains to be established which members of the P6-GlcNAc-transferase family are affected in colon cancer. The human core 4 P6-GlcNAc-transferase has recently been cloned [59a, 59b], or [ 127, 1281 and is homologous to other P6-GlcNAc-transferases [44-471.

18.10 Synthesis of O-Glycan Cores 5-8 Mucins with core 5-8 O-glycan structures are found in various species but these core structures have not yet been described in non-much glycoproteins. U-glycans

320

18 Biosvnthesis of the 0-Glycun Chains

with core 5, containing sialic acid a2-6-linked to GalNAc-, have been described in human meconium and colonic adencarcinoma [60, 611, and in mucins from other species. A preliminary report of the synthesis of core 5 indicated that a core 5 1x3GalNAc-transferase activity is present in the mucosa of an adenocarcinoma patient and acts on GalNAc-much substrate [62]. 0-glycan core 6, GlcNAcP 1-6GalNAc-R, has been described in several human glycoproteins. This linear I -6-linked structure may be synthesized by a P6-GlcNActransferase activity that does not have the requirement for a 1-3-linked Gal or GlcNAc. An activity synthesizing core 6 has been reported in human ovarian tissue [63]. Novikoff ascites tumor cells have a P6-GlcNAc-transferase that adds GlcNAc in pl-6-linkage to Gal. These enzymes may possibly belong to a larger P6-GlcNActransferase family, The synthesis of 0-glycan core structures 7 and 8 has not yet been described. Core 7 was found on mucins from bovine submaxillary gland [64], and core 8 was found in human bronchial mucin [ 651.

18.1 1 Elongation and Branching Reactions 0-glycan core structures may be processed by glycosyl- or sulfotransferases that synthesize terminal structures, and this may terminate chain growth. Alternatively, chains are elongated by repeating GalPl-3/4GlcNAc- sequences (poly-Nacetyllactosamine chains, i antigen) before termination. These elongated structures may be branched by GlcNAcPl-6 residues (I antigen). Both the P3-Gal-[66] and P4Gal-transferases (Figure 2) [67] synthesizing poly-N-acetyllactosamine structures exist as families of related enzymes which are discussed in Chapter 10. GalNActransferases may also be involved in chain elongation. Members of the i P3-GlcNAc-transferase family, synthesizing the i antigen (Figure 2) [68], appear to be present in most cell types, and in serum, and to act on 0and N-glycans as well as glycolipids. An enzyme elongating core structures 1 and 2, by adding a GlcNAc Pl-3 residue to Gal, has been described in pig gastric mucosa [69]. This latter elongation b3-GlcNAc-transferase activity is more restricted in its occurrence and is therefore probably different from the i P3-GlcNAc-transferase. Both elongation activities are altered in leukemia cells [36]. The i P3-GlcNAc-transferase has been purified from calf serum [70]. A cDNA encoding the enzyme activity has been cloned from human lymphoma Namalwa KJM-1 cells, after transfection with cDNA libraries from human melanoma and colon cancer cells, resulting in increased i antigen reactivity [71]. The full length cDNA sequence of the enzyme predicts a typical type I1 membrane protein domain structure of 415 amino acids. However, the transmembrane domain is unusually long with 28 amino acid residues, flanked by only one basic amino acid. Most human tissues were found to express RNA that hybridized with i P3-GlcNActransferase cDNA. Another cDNA has been cloned from human and mouse that encodes enzyme with a substrate requirement for GalPl-4GlcNAc- sequences. This

I N . 12 Synthesis qf T u m i d Structures

32 1

enzyme has significant homology with members of the P3-Gal-transferase family although it transfers GlcNAc residues [72]. The GlcNAc@1-6Galbranches of elongated backbone structures (1 antigen) appear to be made by several different ()6-GlcNAc-transferases. These enzymes differ in their distribution among various tissues and in their acceptor substrate specificities. Pig gastric mucosa has an enzyme that adds the GlcNAcpl-6 branch to GalPresidues of GlcNAcp1-3 Gal structures [73]; it also has an activity that branches substituted O-glycan core 2 [74]. Other 06-GlcNAc-transferases require Gal pl -4 GlcNAcP1-3 Gal structures as a substrate, and act on centrally located Gal residues [75]. This includes a P6-GlcNAc-transferase that acts on internal Gal residues from pig gastric mucosa [76). A cDNA encoding one member of this P6-GlcNAc-transferase family, and homologous to core 2 ~6-GlcNAc-transferase,has been cloned from human teratocarcinoma PA1 cells [46]. The enzyme activity could not be detected using GlcN AcP I-3Gal-R substrate but was demonstrated by the conversion of leukosialin glycans to I antigenic chains. The recombinant enzyme was shown to act on internal Gal residues of Gal P1-4 GlcNAc PI -3 Gal-R [77].In the gene for I P6-GlcNActransferase, three exons contain the coding region, and the regions homologous to the core 2 p6-GlcNAc-transferdse gene are found on two different exons 1481.

18.12 Synthesis of Terminal Structures O-glycans of mucins are rich in blood group and tissue antigens, including the ABO blood groups and Lewis antigens. The glycosyltransferases synthesizing these terminal structures act on N-glycans, glycolipids and O-glycans. At least four different a2-Fuc-transferases exist that synthesize the blood group 0 or H determinant, Fucl-2 Gal-R [78-XI]. In addition, at least five types of a3-Fuc-transferases regulate the synthesis of Lewis antigens [S2-881. These enzymes are discussed in detail in Chapter 11. Blood groups A and B are synthesized by blood group A-dependent a3-GalNAc and blood group B-dependent a3-Gal-transferases, respectively. These two enzymes have similar acceptor substrate specificities and require the H determinant as a substrate, while another w3-Gal-transferase. that is absent from humans and old world monkeys, acts on non-fucosylated terminal @-Gal residues and makes the ‘linear B’ determinant. Galal-3Gal@-[89].The A and B transferases differ only by a few amino acids, which appear to determine the nucleotide-sugar donor specificities 190, 911. Most secreted and cell surface glycoproteins and many mucins have sialic acid termini which impart a negative charge on the sugar chains, and may protect them from being recognized by lectins or antibodies. Sialic acids themselves may also be part of determinants that constitute ligands for the binding of antibodies, carbohydrate binding molecules or microbes. For example, sialyl-Tn antigen is a cancer antigen associated with a poor clinical outcome, and sialyl-Lewisx is an important

322

18 Biosyntlzesis of the 0-Glycan Chains

SAa6 I SAa3Galp3GalNAo

GlcNAc IB6 GalpGalNAc-

d

C

GalpGalNAc-

----t

t

e

SAa3Galg3GalNAc-

2-

t b

GalNAc-

a

SAa6GalNAc-

Figure 3. Competition between chain growth and sialylation leading to truncation. 0-glycan core 1 can be acted upon by branching and elongation enzymes, as well as Fuc-transferases, blood group A and B transferases, sialyltransferases and sulfotransferases. Many of these reactions compete with each other. If GalNAc is a6-sialylated (path a), core 1 (path b) or other core structures cannot be formed and 0-glycans remain truncated. In some leukemia and breast cancer cells, and other cell types, core 1 is 1x3-sialylated (path c) in the presence of core 2 P6-GlcNAc-transferase (path d). Since core 2 06-GlcNAc-transferase and 1x3-sialyltransferase appear to be present in the same Golgi compartments, sialylation and branching reactions compete for the core 1 substrate. If chains are a3-sialylated, another sialic acid residue can be added (path e) to form the di-sialylated core 1. Chains will remain truncated since sialylated core 1 is not a substrate for elongation and branching reactions. If core 2 is formed first (path d), chain growth can occur to form complex 0-glycans. Presumably, the winner of the competition is the enzyme with the higher activity and affinity for the core 1 substrate. SA, sialyl.

ligand for cell adhesion mediated by selectins. When colon cancer cells are treated with GalNAc-benzyl which truncates mucin type 0-glycans and reduces mucin sialylation, the binding of colon cancer cells to E-selectin is reduced [92]. Sialic acid often blocks chain growth, and thus premature or increased sialylation, as seen in some cancers and leukemia cells, will lead to truncation of 0-glycans (Figure 3). There are a number of families of similar sialyltransferases, all related to each other through homology of the ‘sialylmotif’ which may be the CMP-sialic acid donor substrate binding site [93]. This class of enzymes is discussed further in Chapter 12. Two families of sialyltransferases are known to act on mucin 0-glycans, Gal a3sialyltransferases [94-981 and GalNAc 1x6-sialyltransferases [99- 1021. a3-Sialyltransferase activities (ST3Gal I, I1 and IV) act on the Gal residue of 0glycan core 1, and possibly core 2 as well. At least one of these activities is found in most cell types, and may change during differentiation [ 103, 1041, in development

18.12 Synthesis of Terminal Structures

323

[ 1051, during oncogene transfection [ 1061, and in cancer [43, 561 and leukemia [107]. The a34alyltransferase (ST3Gal 111) which acts on GalP 1-3/4GlcNAc chains may possibly be involved in the sialylation of the outer parts of O-glycan chains. Once sialyla2-3Ga1~1-3GalNAc-is synthesized from core 1, it can only be converted to the disialylated core 1 structure, which is common in serum glycoproteins, and all other pathways of O-glycan biosynthesis are blocked (Figure 3). This is due to the inhibiting effect of sialic acid on the substrate recognition by processing glycosyltransferases. If the activities of u3-sialyltransferase are high relative to other processing enzymes, O-glycan chains become highly sialylated and thus shorter. The intracellular localization of a3-sialyltransferase (ST3Gal I) has been determined by immunoelectronmicroscopy in mammary cells, after transfection with an epitope-tagged 1x3-sialyltransferase gene [ 1081. The enzyme was shown to have a broad distribution within the Golgi, with significant amounts of the enzyme in the medial and trans Golgi. Since this is also the localization site for core 2 P6-GlcNActransferase, it is conceivable that core 1 a3-sialylation competes with core 1 branching to form core 2 (Figure 3). This explains why in leukemia cells [36] and in breast cancer cells MCF7 [43], where there are high activities of both a3-sialyltransferase and core 2 ~6-GlcNAc-transferase,truncated and sialylated O-glycan chains are found on mucins and mucin-like glycoproteins. The sialyla2-6Gal linkage found in N-linked oligosaccharides does not commonly occur in mucin O-glycans but sialic acid is often 1x6-linked to GalNAc. At least three types of GalNAc u6-sialyltransferases with different specificities act on GalNAc (ST6GalNAc I, I1 and 111).ST6GalNAc I and I1 appear to have a requirement for mucin peptide in the substrate while ST6GalNAc I11 has a requirement for the sialyla2-3Galpl-3GalNAc- glycan sequence in the substrate. One of several possible modifications of sialic acids in O-glycans is O-acetylation. The O-acetyl group is apparently transferred from acetyl-CoA in the Golgi by Oacetyltransferases with various specificities for sialic acid linkages. This transfer occurs after the glycan chain has been attached to the glycoprotein [109, 1101. Oacetylation protects carbohydrate chains or the underlying peptides from degradation by intestinal bacteria, and masks underlying epitopes. In intestinal cancer, Oacetylation is reduced, which may be one reason for the increased expression of sialyl-Tn or sialyl-Lewis" antigens [ 11 1, 1121. The 5-N-acetyl group of CMP-N-acetylneuraminic acid can be modified by a hydroxylase to form CMP-N-glycolyl-neuraminic acid. The enzyme occurs in many non-human tissues and has been cloned from several species [ 1 131. N-Glycolylneuraminic acid can then be transferred by sialyltransferases to glycoproteins and mucins, resulting in a sialic acid modification which is common in mammalians but in humans is antigenic, and only found in exceptional conditions such as in cancer [ 5 ] . Gal or GlcNAc residues of mucins are often sulfated, and this imparts an additional negative charge to O-glycan chains. Sulfated O-glycans may play a role in cell adhesion through selectin binding, and in bacterial binding [ 114- 1161. Sulfate ester groups also control biosynthetic pathways of O-glycans by blocking certain glycosylation steps [ 117, 1181. Sulfated mucins are common in the intestine but their expression is reduced in colon cancer and other models of cancer [5, 15, 59, 1191. The sulfate group of O-glycans is transferred from 3'-phosophoadenosine-5'-

324

18 Biosynthesis of the 0-Glycan Chains

phophosulfate by sulfotransferases. It is not yet clear which of the sulfotransferases involved in sulfation of proteoglycan chains can also act on mucins. Several sulfotransferases have been described that can act on the Gal or GlcNAc residues of mucin substrates or partial structures found on 0-glycan chains [ 118, 120-1241. The sulfotransferase acting on core 1 is decreased in colon cancer, tumor cells derived from polyposis coli and breast cancer cells [43, 56, 591. The topic of sulfotransferases is further dealt with in Chapter 14. Acknowledgments

This work was supported by the Canadian Cystic Fibrosis Foundation and the Canadian Heart and Stroke Foundation.

References 1. S. J. Gendler, A. P. Spicer, Ann Rev Physioll995 57 607-634. 2. V. Shankar, P. Pichan, R. L. Eddy, V. Tonk, N. Nowak, S. N. J. Sait, T. B. Shows, R. E. Schults, G. Gotway, R. C. Elkins, M. S. Gilmore, G. P. Sachdev, Am J ReApir Cell Mol Biol 1997 16 232-241. 3. Y. S. Kim, J. Gum, I. Brockhausen, Glycoconj J 1996 13 693-707. 4. B. J. W. van Klinken, J. Dekker, H. A. Buller, A. W. C. Einerhand, Am J Physioll995 269 G613-G627. 5. I. Brockhausen, W. Kuhns, Glycoproteins and human disease, Medical Intelligence Unit, CRC Press and Mosby Year Book, Chapman & Hall, N. Y. 1997. 6. M. Buisine, A. Janin, V. Maunoury, J. Audie, M. Delescaut, M. Copin, J. Colombel, P. Degand, J. Aubert, N. Porchet, Gastroenterology 1996 I10 84-91. 7. P. Roussel, G. Lamblin, In: Glycoproteins, J. Montreuil, J.F.G.Vliegenthart, H. Schachter eds, Elsevier, Amsterdam 1996 pp 351-393. 8. S. B. Ho, L. L. Shekels, N. W. Toribara, Y. S. Kim, C. T. Lyftogt, D. L. Cherwitz, G. A. Niehans, Cancer Res 1995 55 2681-2690. 9. C. Balague, G. Gambus, C. Carrato, N. Porchet, J. P. Aubert, K. Y.S., F. X. Real, Gastroenterology 1994 106 1054-1061. 10. Y. Kim, A. Varki, Glycoconj J 1997 14 569-576. 11. S. Nakamori, M. Kameyama, S. Imaoka, H. Furukawa, 0. Ishikawa, Y. Sasaki, T. Kabuto, T. Iwanaga, Y. Matsushita, T. Irimura, Cancer Res 1993 53 363223637, 12. I. Brockhausen, H. Schachter, In: Glycosciences: Status and Perspectives. H.J. and S. Gabius eds, Chapman & Hall, Weinheim 1997 pp 79- I 13. 13. I. Brockhausen, In: Glycoproteins., Vol. New Comprehensive Biochemistry Volume 29A, J. Montreuil, J.F.G. Vliegenthart, H. Schachter eds, Elsevier Pub., New York, New York. 1995, pp 201-259. 14. A. Varki, Glycobiology 1993 3 97-130. 15. I. Brockhausen, I. Schutzbach, W. Kuhns, Acta Anat 1998 I61 36-78. 16. H. Clausen, E. P. Bennett, Glycobiology 1996 6 635-646. 17. E. P. Bennett, H. Hassan, U. Mandel, E. Mirgorodskaya, P. Roepstorff, J. Burchell, J. Taylor-Papadimitriou, M. A. Hollingsworth, G. Merkx, A. G. van Kessel, H. Eiberg, R. Steffensen, H. Clausen, J Biol Chem 1998 273 30472-30481. 18. K. J. Colley, Trends Glycosci Glycotech 1997 9 267-282. 19. J. Roth, Y. Wang, A. E. Eckhardt, R. L. Hill, Proc Nut1 Acad Sci USA 1994 91 89358939.

References

325

20. S. Rottger, J. White, H. H. Wandall, J. C. Olivo, A. Stark, E. P. Bennett, C. Whitehouse, E. G. Berger, H. Clausen, T. Nilsson, J Cell Sci 1998 111 45-60. 21. H. Schachter, 1. Brockhausen, In: Glycoconjugatrs. Composition, Structure and Function., H.J. Allen, E.C. Kisailus eds, Marcel Dekker, New York, NY 1992, pp 263-332. 22. J. Cartron, J. Andrev, J. Cartron, G. W. G. Bird, C. Salmon, A. Gerbal, Eur J Biochem 1978 92 111-119. 23. I. Brockhausen, J. Yang, N. Dickinson, S. Ogata, S. Itzkowitz, Glycoconj J 1998 15 595603. 24. M. Thurnher, S. Rusconi, E. G. Berger, J Clin Invest 1993 91 2103-2110. 25. 1. Brockhausen, P. Romero, A. Herscovics, Cancer Res 1991 5 3136-3142. 26. S . F. Kuan, J. C. Byrd, C. Basbaum, Y. S. Kim, J Biol Chem 1989 264 19271-19277. 27. R. S. Bresalier, Y. Niv, J. C. Byrd, Q.-Y. Duh, N. Toribara, R. W. Rockwell, R. Dahiya, Y. S. Kim, J Clin Invest 1991 87 1037-1045. 28. I. Brockhausen, G. Moller, A. Pollex-Kriiger, V. Rutz, H. Paulsen, K. L. Matta, Biochem Cell Bioll992 70 99-108. 29. I. Brockhausen, G. Moller, G. Merz, K. Adermann, H. Paulsen, Biochemistry 1990 29 1020610212. 30. M. Granovsky, T. Bielfeldt, S. Peters, H. Paulsen, M. Meldal, J. Brockhausen, I. Brockhausen, Eur J Biochem 1994 221 1039-1046. 31. I. Brockhausen, D. Toki, J. Brockhausen, S. Peters, T. Bielfeldt, A. Kleen, H. Paulsen, M. Meldal, F. Hagen, L. Tabak, Glycoconj J 1996 13 849-856. 32. M. F. A. Bierhuizen, M. Fukuda, Trends Glycosci Glycotrch 1993 6 17-28. 33. Y. Nishio, C. Warren, J. A. Buczek-Thomas, J. Rulfs, D. Koya, L. P. Aiello, E. Feener, T. B. Miller, J. W. Dennis, G. L. King, J Clin Invest 1995 96 1759-1767. 34. F. Piller, V. Piller, R. Fox, M. Fukuda, J Biol Chem 1988 263 15146-15150. 35. M. Heffernan, R. Lotan, B. Amos, M. Palcic, R. Takano, J. W. Dennis, J Biol Chem 1993 268 1242-125 1 . 36. I. Brockhausen, W. Kuhns, H. Schachter, K. L. Matta, R. Sutherland, M. A. Baker, Cancer Res 1991 51 1257-1263. 37. 0. Saitoh, F. Piller, R. I. Fox, M. Fukuda, Blood 1991 77 1491-1499. 38. F. Piller, F. Le Deist, K. I. Weinberg, R. Parkman, M. Fukuda, J Exp Med 1991 173 15011510.

39. W. J. Fellinger, P. Barran, H. Merkens, S. Y. Corbel, H. J. Ziltener, J Cellular Physiology 1998 176 350-358. 40. S. Tsuboi, M. Fukuda, EMBO J 1997 16 6364-6373. 41. S. Hemmerich, H. Leffler, S. Rosen, J Biol Chem 1995 270 12035-12047. 42. R. Kumar, R. T. Camphausen, F. X. Sullivan, D. A. Cumming, Blood 1996 88 3872-3879. 43. I. Brockhausen, J. Yang, J. Burchell, C. Whitehouse, J. Taylor-Papadimitriou, Eur J Biochem 1995 233 607-617. 44. M. F. A. Bierhuizen, M. Fukuda, Proc Nut1 Acad Sci USA 1992 89 9326-9330. 45. C. M. Li, K. B. Adler, P. W. Cheng, Am J Respir Cell Mol Bioll998 18 343-352. 46. M. Bierhuizen, M. G. Mattei, M. Fukuda, Genes Dev 1993 7 468-478. 47. D. Toki, M. Sarkar, B. Yip, F. Reck, D. Joziasse, M. Fukuda, H. Schachter, I. Brockhausen, Biochem J 1997 325 63-69. 48. M. F. A. Bierhuizen, K. Maemura, S. Kudo, M. Fukuda, Glycobiology 1995 5 417-425. 49. M. F. A. Bierhuizen, K. Maemura, M. Fukuda, GlycoconjugateJ 1995 12 857-864. 50. W. Kuhns, V. Rutz, H. Paulsen, K. L. Matta, M. A. Baker, M. Barner, M. Granovsky, I. Brockhausen, Glycoconjugate J 1993 10 38 1-394. 51. F. Vavasseur, J. Yang, K. Dole, H. Paulsen, I. Brockhausen, Glycobiology 1995 5 351-357. 52. D. Toki, M. Granovsky, F. Reck, W. Kuhns, K. L. Matta, M. A. Baker, I. Brockhausen, Biochem Biophys Res Com 1994 198 417-423. 53. D. Skrincosky, R. Kain, A. El-Battari, M. Exner, D. Kerjaschki, M. Fukuda, J Biol Chem 1997 272 22695-22702. 54. A. Klein, C. Carnoy, J. Wieruszeski, G. Strecker, A. M. Strang, H. van Halbeek, P. Roussel. G. Lamblin, Biochemistry 1992 31 6152-6165. 55. I. Brockhausen, K. L. Matta, J. Orr, H. Schachter, Biochemistry 1985 24 1866-1874.

326

18 Biosynthesis of the 0-Glycan Chains

56. J. Yang, J. Byrd, B. Siddiki, Y. Chung, M. Okuno, M. Sowa, Y. Kim, K. Matta, I. Brockhausen, Glycobiology 1994 4 873-884. 57. M. J. King, A. Chan, R. Roe, B. F. Warren, A. Dell, H. R. Morris, C. C. Bartolo, P. Durdey, A. P. Corfield, Glycobiology 1994 4 267-269. 58. G. Springer, P. Desai, W. Wise, S. Carlstedt, H. Tegtmeyer, R. Stein, E. Scanlon, Zmmunodiagnosis of Cancer 2nd ed, R. Herberman ed., Marcel Dekker, New York 1990, pp 587612. 59. F. Vavasseur, K. Dole, J. Yang, K. Matta, N. Myerscough, A. Corfield, C. Paraskeva, I. Brockhausen, Eur J Biochem 1994 222 415-424. 59a. S. Yeh, E. Ong, M. Fukuda, J Biol Chem 1999 274 3215-3221. 59b. T. Schwientek, M. Nomoto, S. B. Levery, G. Merkx, A. G. vanKesse1, E. P. Bennett, M. A. Hollingsworth, H. Clausen, J Biol Chem 1999 274 4504-4512. 60. A. Kurosaka, H. Nakajima, J. Funakoshi, M. Matsuyama, T. Nagayo, I. Yamashina, J Bid Chem 1983 258 11594-1 1598. 61. E. F. Hounsell, A. M. Lawson, J. Feeney, H. C. Gooi, N. J. Pickering, M. S. Stoll, S. C. Lui, T. Feizi, Eur J Biochem 1985 148 367-377. 62. A. Kurosaka, I. Funakoshi, M. Matsuyama, T. Nagayo, 1. Yamashina, FEBS Lett 1985 190 259-262. 63. S. Yazawa, S. A. Abbas, R. Madiyalakan, J. J. Barlow, K. L. Matta, Carboh Res 1986 149 241-252. 64. W. Chai, E. Hounsell, G. C. Cashmore, J. R. Rosankiewicz, C. J. Bauer, J. Feeney, T. Feizi, A. M. Lawson, Eur J Biochem 1992 203 257-268. 65. H. van Halbeek, A.-M. Strang, M. Lhermitte, H. Rahmoune, G. Lamblin, P. Roussel, Glycobiology 1994 2 203-219. 66. M. Amado, R. Almeida, F. Carneiro, S. B. Levery, E. H. Holmes, M. Nomoto, M. A. Hollingsworth, H. Hassan, T. Schwientek, P. A. Nielsen, E. P. Bennett, H. Clausen, J Biol Chem I998 273 12770-12778. 67. R. Almeida, M. Amado, L. David, S. B. Levery, E. H. Holmes, G. Merkx, A. G. van Kessel, E. Rygaard, H. Hassan, E. P. Bennett, H. Clausen, J Biol Chem 1997 272 31979-31992. 68. A. H. Koenderman, P. L. Koppen, D. H. van den Eijnden, Eur J Biochem 1987 166 199-208. 69. I. Brockhausen, D. Williams, K. L. Matta, Can J Biochem Cell Bioll983 61 1322-1333. 70. H. Kawashima, K. Yamamoto, T. Osawa, T. Irimura, J Biol Chem 1993 268 27118-27126. 71. K. Sasaki, K. Kurata-Miura, M. Ujita, K. Angata, S. Nakagawa, S. Sekine, T. Nishi, M. Fukuda, Proc Nut1 Acad Sci USA 1997 94 14294-14299. 72. D. Zhou, A. Dinter, R. G. Gallego, J. P. Kamerling, J. F. G. Vliegenthart, E. G. Berger, T. Hennet, Proc Nut1 Acad Sci 1999 96 406-41 1. 73. F. Piller, J.-P. Cartron, A. Maranduba, A. Veyrieres, Y. Leroy, B. Fournet, J Biol Chem 1984 259 13385-13390. 74. I. Brockhausen, K. L. Matta, J. Orr, H. Schachter, A. H. L. Koenderman, D. H. van den Eijnden, Eur J Biochem 1986 157 463-474. 75. A. Leppanen, Y. Zhu, H. Maaheimo, J. Helin, E. Lehtonen, 0. Renkonen, J Biol Chem 1998 273 17399-17405. 76. Y. Sakamoto, T. Taguchi, Y. Tano, T. Ogawa, A. Leppanen, M. Kinnunen, 0. Aitio, P. Parmanne, 0. Renkonen, N. Taniguchi, J Biol Chem 1998 273 27625-27632. 77. P. Mattila, H. Salminen. L. Hirvas, J. Niittymaki, H. Salo, R. Niemela, M. Fukuda, 0. Renkonen, R. Renkonen, J Biol Chem 1998 273 27633-27639. 78. R. J. Kelly, S. Rouquier, D. Giorgi, J Biol Chem 1995 270 4640-4649. 79. R. D. Larsen, L. K. Ernst, R. P. Nair, J. B. Lowe, Proc Nut1 Acad Sci USA 1990 87 66746678. 80. R. J. Kelly, L. K. Ernst, R. D. Larsen, J. G. Bryant, J. S. Robinson, J. B. Lowe, Proc Natl Acad Sci USA 1994 91 5843-5847. 81. S. Hitoshi, S. Kusunoki, I. Kanazdwa, S. Tsuji, J Biol Chem 1996 271 16975-16981. 82. J. F. Kukowska-Latallo, R. D. Larsen, R. P. Nair, J. B. Lowe, Genes Dew1 1990 4 12881303. 83. J. B. Lowe, L. M. Stoolman, R. P. Nair, R. D. Larsen, T. L. Behrend, R. M. Marks, Biochem Soc Trans 1991 19 649-654.

References

321

84. J. B. Lowe, J. F. Kukowska-Latallo, R. P. Nair, R. D. Larsen, R. M. Marks, B. A. Macher, R. J. Kelly, L. K. Ernst, J Biol Chem 1991 266 17467-17477. 85. S. E. Goelz, C. Hession, D. Goff, B. Griffiths, R. Tizard, B. Newman, G. Chi-Rosso, R. Lobb, Cell 1990 63 1349-1356. 86. B. W. Weston, R. P. Nair, R. D. Larsen, J. B. Lowe, J Biol Chem 1992 267 4152-4160. 87. B. W. Weston, P. L. Smith, R. J. Kelly, J Biol Chem 1992 267 24575-24584. 88. K. Sasaki, K. Kurata, K. Funavama, J Biol Chem 1994 269 14730-14737. 89. D. H. Joziasse, J. H. Shaper, D. H. Van den Eijnden, e. al., J Biol Chem 1989 264 1429014297. 90. F.-I. Yamamoto, S.-I. Hakomori, J Bid Chem 1990 265 19257-19262. 91. F.-I. Yamamoto, H. Clausen, T. White, J. Marken, S.-I. Hakomori, Nature 1990 345 229233. 92. N. Kojima, K. Handa, W. Newman, %-I. Hakomori, Biochern Biophys Res Commun 1992 182 1288- 1295. 93. A. K. Datta, J. C. Paulson, J Biol Chem 1995 270 1497-1500. 94. Y. C. Lee, N. Kurosawa, T. Hamamoto, T. Nakoaka, S. Tsuji, Eur J Biochem 1993 216 377385. 95. Y. Lee, C. Park, C. Strott, J Biol Chem 1994 269 15838-15845. 96. N. Kurosawa, T. Hamamoto, M. Inoue, S. Tsuji, Biochim Biophys Acta 1995 1244 216222. 97. Y. J. Kim, K. S. Kim, S. H. Kim, J. H. Kim, I. S. Choe, S. Tsuji, Y. C. Lee, Biochem Biophys Res Com 1996 228 324-327. 98. W. Gillespie, S. Kelm, J. C. Paulson, J Biol Chem 1992 267 21004-21010. 99. N . Kurosawa, T. Hamamoto, Y. C. Lee, T. Nakaoka, N. Kojima, S. Tsuji, J Biol Chem 1994 269 1402-1409. 100. N. Kurosawa, N. Kojima, M. Inoue, T. Hamamoto, S. Tsuji, J Biol Chem 1994 269 1904819053. 101. N. Kurosawa, M. Inoue, Y. Yoshida, S. Tsuji, J Biol Chem 1996 271 15109-15116. 102. E. R. Sjoberg, H. Kitagawa, J. Glushka, H. van Halbeek, J. C. Paulson, J Biol Chem 1996271 7450-7459. 103. A. Harduin-Lepers, M.-A. Recchi, P. Delannoy, Glycobiology 1995 5 741-758. 104. U. Gassen, S. Kelm, R. Schauer, FEBS Lett 1998 427 91-95. 105. W. Gillespie, J. Paulson, S. Kelm, M. Pang, L. Baum, J Biol Chem 1993 268 3801 -3804. 106. P. Delannoy, H. Pelczar. V. Vandamme, A. Verbert, Glycoconjugate J 1993 10 91-98. 107. M. Baker, R. Taub, A. Kanani, I. Brockhausen, A. Hindenburg, Blood 1985 66 10681071. 108. C. Whitehouse, J. Burchell, S. Gschmeissner, I. Brockhausen, K. Lloyd, J. Taylor-Papadimitriou, J Cell Bioll997 137 1229-1241. 109. A. Varki, Glycobiology 1992 2 25-40. 110. W. X. Shi, R. Chammas, A. Varki, J Biol Chem 1996 271 15130-15138. 111. B. Mann, E. Klussmann, V. Vandamme-Feldhaus, M. Iwersen, M. Hanski, E. Riecken, H. Buhr, R. Schauer, Y. Kim, C. Hanski, lnt J Cancer 1997 72 258-264. 112. J. Jass, L. Allison, S. Edgar, J Patho11995 176 143-149. 113. R. Schauer, J. P. Kamerling, In: Glycoproteins ZZ. Montreuil, J., Vliegenthart, J.F.G., Schachter, H. eds, Elsevier Science, Amsterdam 1997 pp 241-400. 114. P. Green, T. Tamatani, T. Watanabe, M. Miyasaki, A. Hasegawa, M. Kiso, C. Yuen, M. Stoll, T. Feizi, Biochem Biophys Res Comm 1992 188 244-51. 115. S. Hemmerich, S. D. Rosen, Biochemistry 1994 33 4830-4835. 116. S. Tsuboi, Y. Isogai, N. Hada, J. K. King, 0. Hindsgaul, M. Fukuda, J Biol Chem 1996 271 27213-27216. 117. I. Brockhausen, W. Kuhns, Trends G1ycosc.i Glycotecli 1997 9 379-398. 118. W. Kuhns, R. Jain, K. Matta, H. Paulsen, M. Baker, R. Geyer, I. Brockhausen, Glycobiology 1995 5 689-697. 119. T. Yamori, H. Kimura, S. K., D. Ota, K. Cleary, T. Irimura, Cancer Res 1987 47 2741-2747. 120. S. Carter, A. Slomiany, K. Gwozdzinski, Y. Liau, B. Slomiany, J Biol Chern 1988 263 1197711984.

328

18 Biosynthesis of the 0-Glycun Chains

121. J. Lo-Guidice, J. Perini, J. Lafitte, M. Ducourouble, P. Roussel, G. Lamblin, J Biol Chem 1995 270 21544-21550. 122. K. Uchimura, H. Muramatsu, K. Kadomatsu, Q. W. Fan, N. Kurosawa, C. Mitsuoka, R. Kannagi, 0. Habuchi, T. Muramatsu, J Biol Chem 1998 273 22577-22583. 123. M. Fukuta, J. Inazawa, T. Torii, K. Tsuzuki, E. Shimada, 0. Habuchi, J Biol Chem 1997 272 32321-32328. 124. S. Degroote, J. M. Lo-Guidice, G. Strecker, M. P. Ducourouble, P. Roussel, G. Lamblin, J Biol Chem 1997 272 29493-29501. 125. Brockhausen, unpublished. 126. Whitehouse et al., unpublished.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

1 9 Glycosyltransferases in Glycosphingolipid Biosynthesis Subhash Basu, Kamal Das, and Manju Basu

19.1 Introduction Glycosphingolipids (GSLs) constitute a large group of eukaryotic membrane lipids. With diversity in their oligosaccharide as well as ceramide moieties, they have been implicated in various intracellular [ 1-61 and intercellular [7-111 processes. During the last decade convincing studies have been made for sphingosine [12-151 and sphingosine- 1-phosphate [ 16, 171 as signaling molecules for various cell functions. On the other hand, the function of saccharide moieties of GSLs in normal and tumor (or cancer) cell surfaces has been studied for the last four decades [18-211. It is expected that the expression of glycosphingolipids with specific oligosaccharide structures on the different cell surfaces is tightly regulated by the transcriptional and post translational regulation of glycosyltransferase genes. Several review articles involving both GSL functions [ 18-2 11 and their regulation on cell surface expression [22-241 have been published in recent years. This review is concentrated on the characterization, cloning, regulation, and genomic organization studies of glycolipid: glycosyltransferases which have been published during the last two years. A few comprehensive review articles on this subject have been published [22, 24-30]. Functional organization of the Golgi apparatus [29] and environmental factors [31] on glycolipid biosynthesis have been the focus of recent research.

19.2 Fucosyltransferases in Glycolipid Biosynthesis In addition to a detailed review [22] on two glycolipid: fucosyltransferases (FucTs) (Table 1 [33-41]), the following reports on substrate-specificity studies with the cloned enzymes [42, 431, identification, and cloning of new FucT activities from chicken embryos [44], murine [45, 461, Caenorhabditis elegans [47, 481, and parasites

330

19 Glycosyltrunsferuses in Glycosphingolipid Biosynthesis

Table 1. Characterization of different glycolipid: fucosyltransferases (donor:GDP-fucose). Abbreviation

Other Name

Acceptor

Step in Linkage Source Figure Cata1 lyzed

References

FucT-2 FUT-1 or al,2FT or al,2 FucT

nLcOse4Cer

(12a)

[22, 33, 34, 361 [351 [371

FucT-3 FUT-3/FucT-III FUT-6/FucT-V1

SA-nLcOse4Cer (14) (LM 1)

~

a1,2

a1,3

Emb. Ch. Br. Human Serum Human Neuroblastoma Emb. Ch. Br. Human Colon Cancer Cells C. elegans

[22, 33, 38, 451 [22, 33, 38, 39, 42, 431 [471

[49, 501 have appeared in recent years. Characterization of an al,2fucosyltransferase (FucT-2; Figure 1, Step 12a), which catalyzes the formation of H- [33, 341 and B-active [36] glycolipids, was reported from bovine spleen [33, 34, 361 and human serum [ 351. At least two different glycolipid:al,3fucosyl- transferases (FucT-3/-3’; Figure 1, Steps 14a/14b) that catalyzed synthesis of LeX and SA LeX were characterized from human neuroblastoma IMR-32 [37], colon carcinoma Colo-205 [38, 391 and CHO cells [40]. The FucT-2 catalyzing H-active glycoconjugate was cloned [39] to study its stringent substrate specificity. and SA-LeX(NeuAca2,3Galp1,4(Fucal, The LeX (Gal~l-4(Fucal,3)GlcNAc-R) 3GlcNAc-R) carbohydrate epitopes are widely distributed in many tissues and are developmentally expressed in brain and lung tissues of human and rodent origin. The al,2fucosyltransferase (FucT-2), catalyzing the synthesis of H-active glycoconjugates [33-35] and B-active glycosphingolipid [36], has been cloned [41]. Identification of the a- 1,3fucosyltransferases(FucT-3) catalyzing in vitro biosynthesis of SA-LeX and LeX have been reported from human neuroblastoma IMR-32 [37], colon carcinoma Colo-205 [38, 391, and CHO mutant cells [40]. The substrate specificities of three clonal (FucT 111, Lewis type; FucT IV, myeloid type; FucT V, plasma type), and five non-clonal a1,3 FucT (Colo-205; HL-60; B-142; lymphoid; EKVX, lung carcinoma and CMLN, calf mesenteric lymph nodes) were determined with sulfated and sialylated type I/or I1 chain in the acceptor oligosaccharide moieties [42]. FucT 111, FucT IV, and FucT V formed 19%, 62%, and 47% 6-sulfo Lewis x respectively, as compared to Lewis X, whereas 6’-Sulfo LeX and 3’-sialyl6’-sulfo LeX (GLYCAM-ligand) were not synthesized from their immediate precursor by cloned FucT 111, IV, or V. All three of these fucosyltransferases were differentially active with 2’-fucosyllactose (FucT 111, 31 1%; FucT IV, 9%; FucT V 188%), but were not active with 2’-fucosyl-6’-sulfo lactose. FucT I11 and FucT V were 7- and 0.5-fold active, respectively, in forming Lewis a as compared to Lewis x, whereas FucT V was inactive. FucT I11 was 2.0-fold more active in forming 3’alpha-galactosyl Lewis a than Lewis b. FucT 111 synthesized 6-sialyl Lewis-a (40% efficiency as compared to Lewis a) from 6-sialyl type I. FucT I11 did not act on 6’-

19.2 Fucosyltransjeruses in Glycolipid Biosynthesis

33 1

sulfo or 6’-sialyl type I but was 106% and 22% active with 3’-sulfo and 6-sulfo types I. respectively. The four FucTs from CMLN, HL-60, B-142, and EKVX cells were 1.2 to 1.7 times more active with Fucu1,2Gal01,4-GlcNAc-OpNP with respect to Gal0 1,4GlcNAc~-O-A1. A domain-swapping approach demonstrated that a region of amino acids found in human ul,3/4-fucosyltransferase(FucT 111) introduced a significant increase in al,4FucT acceptor-substrate specificity in ul,3FucT-IV (FucT V). Using sitedirected mutagenesis [43], two amino acids (Asn 86 to His and Thr 87 to Ile) of the eight amino acids originally swapped from FucT I11 into FucT V were modified. The fucosyltransferase activity toward oligosaccharide lactosamine (LacAm) or LacAm-containing glycolipid (nLcOse4Cer) increased 20-fold with concomitant decrease in K , values for these substrates. The mechanisms of regulation for the expression of carbohydrate epitopes during development are unknown. In developing chicken B cells, the Lewis x is expressed in a stage-specific manner. The chicken u1,3FucT (CFTl) gene involved in Lewis x biosynthesis has been cloned [44]. It is characterized by a single long open reading frame of 356 amino acids encoding a type I1 transmembrane glycoprotein. The domain structure and predicted amino acid sequence are highly conserved between CFTl and mammalian FucT IV genes (52.8% and 46.3% identity to mouse and human. respectively). CFT 1 catalyzes in vitro LacNAc > 3’sialyl-LacNAc acceptors with almost no utilization of other neutral type I1 (lactose, 2-fucosylactosyl) or type I (lacto-N-biose I) acceptors CFT1-transfected cells make cell surface Lewis X in COS-7 cells and Lewis X + V1M-2 structure in CHO cells. CFTl gene expression can be detected in embryonic thymus and bursa and is tissue specific. Expression of the CFTl gene and cell surface Lewis x structures are proved to be closely linked during B-cell development in chicken. Cloning of a rat a1,3fucosyltransferase gene (rFucT), isolated from the rat genomic library by a PCR-cross-hybridization, has been reported [45]. This newly cloned rFucT showed the highest degree of homology to murine FucT IV (87% identity) and human FucT IV (78% identity), with lower homology (41-49% identity) to FucT 111, V, VI, or VII. COS-I cells transfected with the rFucT gene expressed a fucosyltransferase activity with type I1 (Galp 1,4GlcNAc)-containing oligosaccharide and the glycolipid acceptor nLcOse4Cer (neolactotetraosylceramide) but only low activity with sialylated substrates such as sialyllactosamine or LM 1 (sialyl-nLcOse4Cer). A novel mouse ul,3-fucosyltransferase gene (mFucT IX), from an adult mouse brain cDNA library, has recently been isolated [46] using the expression cloning method. Transfection of Namalwa cells with the mFuc-IX gene also showed a marked increase in Lewis x epitopes but not sialyl-Lewis x epitopes. The deduced amino acid sequence of mFucT IX, consisting of 359 amino acid residues, indicated a type I1 membrane protein and showed low degrees of homology to the mFuc TIV (48%), mFucT VII (39%), and the human FucT I11 (43%). Characterization and molecular cloning of a al,3fucosyltransferase (CEFT- 1) and ul,2FucT expressed by the nematode C. elegans have been reported [47]. The CEFT-1 transfected (transiently) extracts of COS7 cells catalyzed synthesis of LeX but not sialyl LeX from exogenously added respective acceptors. Although C. eleyans glycoconjugates do not express the LeX antigen, detergent extracts of adult C.

332

I 9 Glycosyltransferases in Glycosphingolipid Biosynthesis

elegans contain al,3FucT (CEFT-1), perhaps due to the presence of gene structure similar (20-30% identity) to all five cloned human al,3-FucTs. The discovery of CEFT-1 proposes the future possibility of understanding the structure-function relation of al,3FucTs and its expression during development because nematodegenetics is already well characterized [48]. The biological functions of the complex carbohydrates of parasites are not clearly understood. The molecular cloning and characterization of a fucosyltransferase of Schistosoma mansioni [49] with a DNA sequence similarity of 85%) and 64% to mouse and human fucosyltransferase type VII have been reported [49]. The parasite a1,3FucT is capable of synthesizing the SALeX determinant, but not the LeX in CHO cells. The cDNA sequence of parasite al,3fucosyltransferase contains an open reading frame encoding a protein of 351 amino acids with a predicted molecular size of 40.5 kDa. However, the transfected CHO cell extract corresponds to a fucosyltransferase at 50 kDa, indicating a glycosylated protein. Two fucosyltransferases (a2- and a3FucT) have been characterized from avian schistosome Terinchobilharzia ocillate [50].The al,3FucT resembles human FucT V and V1. It is concluded that u3-FucT and the combination of w2-FucT and u3-FucT activities are involved in the biosynthesis of the oligomeric LeX and the Fuc a1,2Fuc a1,3-GlcNAc, respectively.

19.3 Galactosyltransferases in Glycolipid Biosynthesis The glyco1ipid:galactosyltransferasesthat catalyze different positional and anomeric linkage formation in various GSLs comprise a family of more than seven individual gene products [22]. Most of these activities have been characterized and purified from normal tissues and tumor cells. The putative amino acid sequences from cDNA sequences of at least six GSL:GalTs have been reported (Table 2 [33, 51761. The recent advances will be reviewed in the following section. Two glycolipid:P-galactosyltransferases, GalT-3 and GalT-4, were characterized in embryonic chicken brain [33, 55, 57-59, 63, 70, 77-80]. Based on substrate competition studies, these two activities were believed to be due to expression of two gene products. The cDNA fragments (about 600 bp) encoding the catalytic domains of the GalT-4 (UDP-Gal: LcOse3Cer P 1,4galactosyltransferase) from ECB [78-SO] and human Colo-205 cells [79, 811 were isolated and expressed in Eschevichia coli as GST-fusion proteins. The recombinant GST-GalT-4 were folded as native enzymes because they had catalytic activity. The truncated fusion proteins (40 kDa) also had affinity for binding to GlcNAc-Sepharose, UDP-hexanolamineSepharose, and alpha-lactalbumin-Sepharose columns at 4 "C. Functionally, these were similarly to that of native GalT-4 [59]. A novel putative member of the PGalT family, designated as PGalT-4, was identified by BLAST analysis of expressed sequence tags [81, 821. Expression of the full coding sequence and a secreted form of P4Gal-T4 in insect cells showed that the gene product had UDP-Gal:LcOse3Cer 0- 1,4-galactosyltransferaseactivity. In con-

19.3 GalactosyltvansJerases in Glycolipid Biosynthesis

333

Table 2. Characterization of different glycolipid: galactosyltransferases using donor:UDPgalactose.

Abbreviation

Other name

Acceptor

Step in Figure 1

Catalyzed linkage

Source

Reference

GalT-1

CGalT

(1 5 )

81.1

GalT-2

LacCer Synthase GM1, Synthase or, 03 GdlT 04 GT p4 GalT

Ceramide-20H:fa Glc-Cer

(2)

01.4

GM2ganglioside

(SajSb)

01.3

Em. B. Ch. Br. Murine Br. Emb. Ch. Br. Hum. Kidney Em. B. Ch. Br.

LcOse3Cer

(1 1)

81 .4

[22, 33, 511 [ 52-54] [22, 33, 551 [561 [22, 33, 5759, 891 [6O, 611 [22, 33, 62, 761 [33, 55: 59, 631 [64-661 [67, 681 122, 5 5 , 69, 761 [33, 551 170,711 1731

GalT-3

GAT-4

GalT-5

a 3 GalT

nLcOse4Cer

(12b)

u1,3

GalT-6

a4 GalT

Lac-Cer

(7)

a1,4

Murine Br. Rab. Bone Marrow Emb. Ch. Br. Tumor Cells Bovine Rab. Bone Marrow Em. B. Ch. Br. Bovine tissue Murine tumor cells Tumor Cells

[74,751

trast to P4GalT-1, T-2 and T-3 [82, 831, this enzyme did not transfer galactose to GlcNAc terminal-containing fetuin, transferrin, or ovalbumin. The genomic organization of the coding region of this newly cloned P-galactosyltransferase is in six exons. BLAST analysis of expressed sequence tags (ESTs) using the coding sequence of human P3GalT-1, showed no ESTs with identical sequences but a large number with similarity. Three different sets of overlapping ESTs with sequence similarities to (33GalT-1 (UDP-Gal:P1,3-N-acetylgluco-samine pl,3GalT) were compiled, and complete coding regions of these genes were published 1841. Expression of P3GalT-1 and P3GalT-2 in the Bacculo virus system showed similar properties. However, the fourth GalT sequence, designated as a P3GalT-4, was very similar to a recently reported GDla synthesis 1611 or GM 1 synthesis by a Gal-T3 from embryonic chicken brains [59, 80, 851. The coding regions for each of the four genes are in single exons localized in lq, 31, 3a, 25, and 6, 21.3, respectively, by EST mapping. The results demonstrated the existence of a family of homologous P3-galactosyltransferase genes [84]. GalT-1 or CGalT(UDP-Ga1:Ceramide 81, lgalactosyltransferase; Table 2 [33, 5 1-54]) 2-hydroxy-fatty acid catalyzes the synthesis of galactocerebroside (galactosylceramide), which is the major constituent of myelin membranes. The localization of GalT-1 and its role in membrane sorting is recently being explored 1861.

334

19 Glycosyltrunsferuses in Glycosphingolipid Biosynthesis

Using antibodies raised against CGalT, it has been demonstrated by immunocytochemistry that the enzyme is localized in endoplasmic reticulum and the nuclear envelope but not in the Golgi on the plasma membranes 1861. All galactosylceramide synthesis catalyzed by CGalT in uiuo is concluded to occur in the lumen of the endoplasmic reticulum [86]. Recent results showed [87] that CGalT catalyzed transfer of galactose to both hydroxy and non-hydroxy fatty acid-containing ceramides and diglycerides, depending on their local availability. Perhaps the same CGalT (or GalT-1) is responsible for the synthesis of hydroxy and non-hydroxy fatty acid-containing GalCer and galactosyldiglyceride present in myelin membranes. Treatment of mouse tetratocarcinoma F9 cells for 36 hours with all-transretinoic acid (RA) caused a 9-fold increase in mRNA levels for an a1,3GalT (UDPGal: al,3-galactosyltransferase).It is expected that RA induces a1,3GalT at the transcriptional level, which results in major alterations in the surface epitope of the cells by masking the Le" antigen 1881.

19.4 N-Acetylgalactosaminyltransferasesin Glycolipid Biosynthesis At least three different gene products (GalNAcT-1, -2, and -3) are catalyzing three distinct linkage formations (Table 3) in GalNAc-containing glycosphingolipids. The Figure 1, GalNAcT- 1 (UDP-GalNAc:GM3 ~l,4N-acetylgalactosaminyltransferase; Step 4a/4b) that catalyzes formation of GM2-ganglioside (GalNAcp1-4(NeuAca2-3) Galp 1-4Glc-Cer) from GM3-ganglioside (NeuAca2-3Galp I -4Glc-Cer) has been characterized and purified from embryonic chicken brains 189-9 11, canine spleen 1921, mouse liver 1931, and guinea pig bone marrow [94], other tissues 1331, and guinea pig tumor cells [95]. Like the native enzymes, the cloned enzyme 196, 971 is also specific for GM3 or GD3 substrates and is not active with glycolipids without sialic acid (e.g., lactosylceramide; Gal(31-4Glc-cer) or a disaccharide (lactose) as substrate. Based on kinetic studies, it has been suggested that the synthase activities of GgOse3Cer (GA2), GM2 (monosialoganglioside), or GD2 (disialoganglioside), are perhaps catalyzed by the same protein 1981. Regulation of GalNAcT-1 through enzyme phosphorylation was analyzed [ I ] by determining the enzyme activity after treatment of NG108-15 cells with various protein phosphatase inhibitors (okddaic acid and orthovanadata) or protein kinase activators (phorbol ester and Forskolin). Appearance of GalNAcT- 1 during embryonic chicken brain (ECB) development has been studied exhaustively [22, 33, 89-91]. Like ECB the activity was not detectable in ECVH (embryonic chicken vitreous humor), until embryonic day 10 1991 and then increased more than six-fold until day 16, remaining at that level until birth. Similar developmental studies in developing rat brain and retina have been made with GalNAcT-1 [lOO]. The GalNAcT-1 mRNA was abundant in GMI+/ GD3+ neurons and almost absent in the flat, GMl-/GD3+ Muller glia-derived cells [ IOO]. However, the relationship between GalNAcT- 1 gene expression and the neuronal cell development is not clear. Transfection of GalNAcT- 1 (GM2 synthase) into CHO-K1 cells (lacking ability to synthesize GM2, GM1, and GDla) showed

GM2 Synthase p 1,4GalNAc-transferase (E.C. 2.4.1.92)

Gb4 Synthase, Globoside Synthase or, pl ,3Gal-NAc Transferase Gb5 Synthase Forssman GSL Synthase or, al,3-@alNAc transferase

GalNAcT-1

GalNAcT-2

GalNAcT-3

Other name

Abbreviation

(8) (9)

P1,3

a1,3

GbOse3Cer

GbOse4cer

(4a/4b)

p1,4

GM3-ganglioside

Step in Fig. 1

Catalyzed linkage

Acceptor

Em. B. Ch. Br. G. pig bone marrow Dog spl. Mouse liv. Human tissues Emb. Ch. Br. G. pig tumor cells Mouse Y l K tumor cells G. pig tissues -normal and -tumor cells.

Source

Table 3. Characterization of different glycolipid: N - acetylgalactosaminyltransferases using donor: UDP-GalNAc

[ I ' 11 [95, 1071

[22, 33, 89-91, 1081 [941 [921 [931 [96, 97, 100, 1051 [22, 33, 91, 106, 1081 195, 1091 i 1071

Reference

ul

w

w

8'

336

19 Glycosyltransferases in Glycosphingolipid Biosynthesis

the appearance of GM1 and G D l a on the surfaces [ 1011. A key regulatory role of GalNAcT- 1 (or GM2 synthase) in ganglioside biosynthesis has been suggested. CHO cells have been stably transfected with myc-epitope-tagged GalNAcT- 1 to demonstrate homodimerization in the ER [ 1021 membranes. Two protein inhibitors (27 kDa and 70 kDa) for GalNAcT-I have been purified from chicken serum [ 1031. Both polypeptides have a neuritogenic activity [103, 1041 and inhibit the incorporation of [3H]-galactose into the cell gangliosides. These peptides are 100% homologous with chick apolipoprotein A1 (apo Al). Transgenic mice with disrupted GalNAcT-1 (or GM2/GD2 synthase; EC 2.4.192) genes have been produced [105]. The mice lacked all higher ganglioside species (GM1, G D l a or GDlb). However, no biological defects in their nervous system or behavior were observed. Slight reduction in the neuronal conduction velocity from the tibia1 nerve to the somatosensory cortex, but not to the lumbar spine, was observed. It was suggested that the higher levels of GM3 and GD3 expressed in the brains of these mutant mice compensated for the lack of higher gangliosides [105].The GalNAcT-1 gene consists of at least 11 exons and spans more than eight kbp. The coding region is located in exons 2-1 1 [ 1051. The multiple transcription initiation sites and their identified promoters/enhancers are believed to be differentially involved in the cell typespecific expression of the GalNAcT-1 gene. This gene was assigned to human chromosome 12q 13.3 by means of fluorescence hybridization in situ. The GalNAcT-2 (UDP-GalNAc:GbOse3Cer 81-3N-acetylgalactos-aminyltransferase; Figure 1, Step 8), which catalyzes the synthesis of globoside from GbOse3Cer (Gala1,4Gal~l,4Glc-Cer), was initially characterized in embryonic chicken brains [ 1061, guinea pig tumor cells 104C1 [95] and Y-1-K tumor cells [ 1071. Subsequently, the activity was solubilized, purified, and separated from GalNAcT- 1 activity from ECB [91, 1081. The GalNAcT-2 activity has also been purified to homogeneity from canine spleen [ 109- 1 1 11. The GalNAcT-3 (UDP-GalNAc: GbOse4Cer al-3N-acetylgalactosaminyltransferase),which catalyzes the synthesis of Forssman GSL from globoside, has also been characterized from guinea pig tumor 104 cells [95] and canine spleen [ 110-1 111.

19.5 N-Acetylglucosaminyltransferasesin Glycolipid Biosynthesis At least three different gene products (GlcNAcT-I, T-2 and T-3), which catalyze three distinct linkage formations (Table 4) in GlcNAc-containing glycosphingolipids, have been characterized from eukaryotic cells and tissues. The GlcNAcT- 1 (UDP-GlcNAc: LcOse2Cer p 1,3N-acetylglucosaminyltransferase)that catalyzes formation of LcOse3Cer (GlcNAcb 1-3GalP 1-4Glc-Cer) from lactosylceramide (GalP1-4Glc-ceramide) has been characterized from rabbit bone marrow [22, 33, 1121, mouse lymphoma, P1798 tumors [ 1131, and human colon cancer cells [ 114, 1151. GlcNAcT-1 is proposed to regulate the expression of the sulphoglucuronylglycolipids (GlcAD1,3(S03-0-3) Gal~1,4GlcNAc~l,3-Gal~l,4Glc-Cer) in specific cell types in the cerebellum during development 11161. The biosynthesis of two pentaglycosylceramides containing a terminal GlcNAc moiety attached to nLcO-

19.6 SiulyltransJeruses in Glycolipid Biosynthesis

337

Table 4. Characterization of different glycolipid: N-acetylglucosaminyl-transferases using donor:UDP-GlcNAc. Abbreviation

Other name

Acceptor

Catalyzed Step in Fig. 1 linkage

GlcNAcT- 1 Paragloboside lactosylCore ceramide synthase (LcOse2Cer or Lac-Cer)

(31,3

(10)

GlcNAcT-2 i-core Synthase

nLcOse4Cer

(31,3

(12d)

GlcNAcT-3 i/I-core synthase

nLcOse4Cer or (31,6 iLcOse5Cer

(12e)

Source

Reference

Rabbit bone marrow P-1798 Lymphoma COIO-205 P-1798 Lymphoma COIO-205 P-1798 COIO-205

[22, 33, 76, 1121 [1131 [114, 1151 [ I 131

[ * 141 11131 w41

se4Cer by p 1,3GlcNActransferase (GlcNAcT-2) and pl ,6GlcNActransferase (GlcNAcT-3) has been reported from P-1798 mouse lymphoma [113] and human colon carcinoma Colo-205 cells [ 114, 1151. These two transferases showed different kinetic parameters including pH optima, ion effects, and also differential heatinactivation patterns at 55°C [113, 1141. The structures of the enzymatic product were established with permethylation of the radiolabeled GlcNAc-products obtained from the individual enzyme-catalyzed reaction. Purification of these enzymes (Table 4) to homogeneity is under investigation in several laboratories, and the cloning sequences are not available. Whether these glycolipid: GlcNAcTs have any sequence homology to any of the six GnTs or GlcNAcTs(1-VI) in the pathway of N-linked oligosaccharide biosynthesis [ 1 171201 is not yet known. Expression mechanisms of CD-15 antigens (sLex, SALe") have been investigated in human B lymphoid cell lines [121]. CD-15s are not expressed in mature B cells but are highly expressed in pre-B leukemia and pre-B lymphoma cell lines. The expression site has been identified on the 0-linked oligosaccharide chains of glycoproteins. However, whether the same enzyme core 2 GnT (UDP-G1cNAc:Gal p 1,3GalNAc (GlcNAc to GalNAc) p 1,6N-acetylglucosaminyItransferase)utilizes any glycolipid precursor has not been determined. The antigen from pre-B lymphoma cells carrying 0-linked oligosaccharides recognized by anti-SALe" monoclonal antibody has been characterized as a 150 kDa glycoprotein [121]. Perhaps those cells also contain a glycolipid with the similar oligosaccharide chain.

19.6 Sialyltransferases in Glycolipid Biosynthesis Sialyltransferases (SATs) are a family of glycosyltransferases that transfer sialic acid from the donor substrate CMP-NeuAc to the acceptor-oligosaccharide chains

338

19 Glycosyltransferases in Glycosphingolipid Biosynthesis

Table 5. Characterization of different glyco1ipid:sialyltransferasesusing donor:CMP-NeuAc Abbreviation

Other name

Acceptor

Catalyzed linkage

Step in Fig. 1

Source

Reference

SAT-1

(GM3 Synthase)

Lac-Cer

u2,3

(3a)

Embr. Ch. Br.

SAT-2

(GD3 Synthase)

GM3ganglioside

a2,8

(3b)

LM 1 nLcOse4Cer

a2,8 a2,3

(l3b) (13a)

Rat liver Golgi Embr. Ch. Br. Human melanoma Mouse brain Embr. Ch. Br. Embr. Ch. Br.

[22, 33, 89, I241 ~251 [I261 [127, 1561

(6a)

Colo-205 Human Placenta Human Colon Cancer Embr. Ch. Br.

SAT-3

(LDla Synthase) STZ (LM1 Synthase) ST3Gal IV

SAT-3' SAT4

GDla Synthase

LcOse4Cer

a2,3

GMIganglioside

a2,3

Rat Brain Rat Liver SAT-2'

ST8SiaV

SAT-6

ST6 Gal IT (ST6N)

GDla a2J ganglioside nLcOse4Cer a2,6

[128, 1561 ~291 [22, 33, 19, 1301 (79, 131, 132, 1331 [ 1341 [22, 33, 89, 1241 [135, 1361 1137. 1501 [138, 144, 1561 [30, 1521

bound to ceramide (GSLs) or proteins (GPs). Several comprehensive reviews have been published in recent years on glyco1ipid:SATs [22, 27-29, 331 as well as glycoprotein:SATs [30, 122, 1231. At least 13 distinct sialyltransferase [ 122, 1231 activities have been characterized and cloned. However, in this section we will review some publications on the sialyltransferases that specifically catalyze glycolipid biosynthesis and have not been reviewed before [22, 30, 33, 122-1241. A comparison of the published [22, 30, 33, 1231 nomenclatures for glyco1ipid:sialyltransferasesare given in Table 5 [124-1401. The sialo-glycolipids of ganglio-core(GalNAcp 1,4Galp 1,4Glc-Cer) and neolactocore(GlcNAc~l,3Gal~l,4Glc-Cer) families are enriched on the surfaces of neuronal and cancer cells, respectively. They are believed to be regulated during differentiation [ 139, 1401, proliferation [141], and metastasis [ 1421. A stepwise biosynthetic pathway of GDla ganglioside (Figure 1; Steps 1,2, 3a, 4a, 5a, 6a) has been reported in embryonic chicken brains [22, 27, 28, 331 and developing tissues from other species [ 143-1451 which involve at least three glycolipid:sialyltransferases, products of three different gene structures (Figure 1: SAT-I, Step 3a; SAT-2, Step 3b; SAT-4, Steps 6a and 6b). Stepwise biosynthetic pathways of Le" and SA-LeX(Figure 1:

19.6 Siulyltrunsjerases in Glycolipid Biosynthesis

339

Steps 1, 2, 10, 11, 14b or 13a and 14a) have also been reported from embryonic chicken brains 122, 27, 281, mouse lymphoma [ 1461 and human colon cancer Colo205 cells [22, 28, 381. However, little is known about the expression of Lewis-x and sialyl-Lewis-x blood-group glycolipids with multilactosamine (or poly-lactosamine) chain-bound to ceramide. A novel sialyltransferase, SAT-3 (CMP-NeuAc:nLcOse4Cer a2,3sialyltransferase: Figure 1: Step 13), has been characterized in bovine spleen [ 147, 1481, embryonic chicken brains 179, 130, 1311, human colon carcinoma 179, 131, 1321, melanoma WM266-4 cells [149], and human placenta [133]. Both sialyltransferase activities SAT-3 (Figure I , Step 13) and SAT-4 (Figure 1, Step 6a/6b) in (2010-205 cells catalyze the transfer of sialic acid to the terminal galactose of GlcNAc- and GalNAc-containing substrates, respectively. These two activities are believed to be two different gene products [22, 33, 122, 130-1331. The Colo-205 SAT-3 activity was immunoprecipitated with a polyclonal antibody produced against a truncated SAT-3 protein which was expressed in E. coli as a GST-fusion protein. The ECB cDNA [79, 1321 was homologous to the a2,3sialyltransferase (SAT-3 or STZ) cloned from human placenta 11331 and human melanoma cells [ 1491. A concentration-dependent decrease in the residual SAT-3 activity relative to SAT-4 activity was observed in the Colo-205 supernatant after precipitation of the immune complex [ 1311. Characterization of the catalytic reaction products of SAT3 and SAT-4 and the binding of specific antibodies indicated that SAT-3 and SAT-4 are two different gene products 179, 131, 1321. They catalyze the formation of an a2,3-linkage between sialic acid and the terminal galactose unit of two different GSLs with different core structures. The a2,6-sialyltransferase (SAT-6 or ST6N or ST6Gal-1 or SiaT-1) was purified and cloned from rat liver [ 151, 1521. This a2,6sialyltrans-ferase (ST6GalI) has been shown to catalyze the transfer of sialic acid from CMP-sialic acid to free disaccharide (GalP1,4GlcNAc) and onto glycoproteins and glycosphingolipids containing the same lactosamine disaccharide unit [ 1521. Kinetic analysis of mutated sialyl motifs of ST6GalI showed a change in the kinetic parameters for both the donor and the acceptor substrates [ 1521. Although many glycosyltransferases involving GSL biosynthesis have been cloned and characterized 122, 281, it is not clear exactly where in the Golgi stack individual sialylation reactions take place. The presence of several glycolipid: glycosyltransferases which are responsible for G D l a ganglioside [ 1251, or trisialoganglioside G T l a [ 143, 144, 1531, and tetrasialoganglioside, G Q l b [ 1531 biosyntheses has been shown in rat liver Golgi preparation. The same SAT-2 activity (Figure 1 , Step 3b) was shown to catalyze in vitro biosynthesis of both GD3 [127, 1291 and L D l a [129] in embryonic chicken brains. Substrate specificity studies with cloned SAT-2 (SAT11 or ST8SiaV) showed the enzyme was able to utilize GM3, GMlb, G D l a or G T l b as acceptor substrates 128, 124, 126, 154, 1551. It is believed that the same enzyme is involved in the synthesis of GD3, GDlc, G T l a and G Q l B in vivo. The SAT-2 is the active enzyme in the neolacto-glycolipid biosynthesis, such as LDla (Figure 1, Step 13b) [22, 33, 28, 1291. By transfection of the cloned human SAT-2 (a2,8sialyltransferase) cDNA,

340

I 9 Glycosyltrunsferuses in Glycosphingolipid Biosynthesis

transient and stable expression of G T l a and G Q l b was also observed in COS-7 and Swiss-3T3 cells which originally lacked SATII and SATV activities [ 1561.

19.7 Glucuronyltransferasesin Glycolipid Biosynthesis The HNK-carbohydrate (or epitope) is now considered to be the hallmark of several cell adhesion molecules. HNK-1 epitope has been implicated in the migration of neural crest cells, the adhesion of astrocytes and neurons to laminin, the outgrowth of neuritic and astrocytic processes, the preferential outgrowth of neurites from motor neurons, and the homophilic binding of neural cell to cell adhesion molecules [ 1571. In cerebellar granule neurons, it promotes differentiation and neurite growth [ 1581. HNK-epitope is highly immunogenic. Since the early 1980s, HN K- 1 antigens (glycolipids, glycoproteins, or proteoglycans) have also been found in tumor cells. The major glycolipid containing the HNK-1 epitope, SGGL-1 and its highest (3-0-sulfate GlcA~l,3Ga1(31,4GlcNAc~I,3Gal~1,4Glcl,lCer), homologue (3-0-sulfate GlcA~l,3Gal~1,4GlcNAc~l,3nlcOse4-Cer) were isolated from human peripheral nerves [ 1591. The transfer of glucuronic acid to nLcOse4Cer to form the core glycolipid of HNK-1 epitope (GlcA~1,3Gal~l,4GlcNAc~l, 3GalP1,4Glc-Cer) is catalyzed by a ~1,3glucuronyltransferase (GlcAT-1; Table 6), characterized from 19-day-old embryonic chicken brains [ 160, 1611 and developing brains of other species [162, 1631. This activity is inhibited by 10-100 pM sphingosine, a negative modulator in the signal transduction pathway [160, 1641. Subsequently, using rat brain extract as the enzyme source, it has been suggested that the glucuronyltransferase catalyzing transfer to glycolipids is different from that involved in the transfer to the glycoprotein [ 1651. The unambiguous proof will perhaps come from the sequence comparison of the two different GlcAT catalyzing glycolipid and glycoprotein HNK- 1 molecules. Partial purification and solubilization of the glycolipid glucuronyltransferase (GlcAT-1) has been reported [ 160, 1611. The glycoprotein: glucuronyl transferase has been cloned recently [ 1661. However, the report on cloning of GlcAT-1 is not yet available. Table 6. Characterization of different glycolipid: glucosyl- and Glucuronyl-transferases of animal cells. Abbreviation

Other name

Acceptor

Catalyzed linkage

Step in Source Figure 1

References

GlcT-1

Glc-cer Synthase

nFACeramide

81>1

(1)

Embr. Ch. Br.

nLcOse4Cer

fi 1,3

(12b)

Mouse Br. Embr. Ch. Br.

[22, 33, 167-169, 1711 [26, 1701 [22, 33, 160, 1611

Mouse brain

[162, 1631

GlcAT- 1 Glycolipid GlcA transferase

19.7 Glucuronyltrun,~erusesin Glycolipid Biosynthesis

IcT-1

GL~&,ICER

341

(16) S A T - 1 Gal-dE R

I a2,3

Fucal,2nLc4

SA (2) I a I T - 2 GD3

GM3

(4b) 4GaINAcT-1 GD2

(5b)

GM2 GalT-3

1

GbOse4CER

4 1b

(6b)

SAT-4

(GalNAcai GTlb Galpi .3GalNAcpi 4Galpi.4Glc-CER la2,3 SA

la2,5

Galpi.4GlcNAcl3i 3Galp1,4Gl~-CER la2,3 l a i , 3 SA Fuc ( SA-LeX)

SA

(GDla)

Figure 1.

19.8 Glycosyltransferases in Cerebroside Biosynthesis (Gluco- and Galacto-) The synthesis of cerebrosides (glucosyl- and galactosyl-) was postulated to be catalyzed by the two different gene products (glucosyl- and galactosyltransferases) expressed in developing chicken [22, 33, 51, 167-1691 and mouse brains 126, 52, 1701. However, the sequences determined from the cDNA sequences of GlcT-1 (Figure 1, Step 1) [171, 1721 and GalT-1 (Figure I , Step 15) [53, 541 show little structural homology. Glucosylceramide is proved to be the precursor of most of the longer-chain glycosphingolipids of all three families of glycolipids (ganglio-, globo-, and lacto-) (Figure 1). GlcT-1 was shown to be up-regulated at the transcriptional level during keratocyte differentiation [ 1731. During the myelination process of the neurons, GalT-1 (Figure I , Step 15) is expressed and galactosylceramide is sulfated by sulfotransferase [I741 to produce the sulfatide 11751 as an end product of the pathway. Sulfation of intermediate GSLs of the globo- and glucuronylf31,3-nLcOse4Cer is also under study [ 176. 1771. Both L-PDMP (l-phenyl-2-decanoylamino-3-

342

19 Glycosyltransferases in Glycosphingolipid Biosynthesis

morpholino- 1-propanoloHCL) and L-PPMP (1-phenyl-2-hexadecanoylamino-3morpholino- 1-propanoloHCl) inhibit GlcT-1 [26],perhaps by binding to the hydrophobic sites of the enzyme. However, mixed inhibition kinetics have been observed with the enzyme isolated from tissues [26], Substrate competition studies with the cloned and expressed protein (GlcT-l), in the presence of L-PDMP and L-PPMP, will provide better insight into the hydrophobic domain of this GlcT-1 (UDPglucose: nFA-ceramide pl ,1-glucosyltransferase).The native GlcT-1 enzyme has been reported to be heat sensitive [167]. However, no such studies have been reported with the cloned and expressed GlcT-1, until now. The gene regulation of expression of GlcT-1 by hormones would be an interesting field of research to be explored after the promoter sequence of this enzyme is known.

Acknowledgments

This article was written based on research supported by NIH grants NS-18005 (Jacob Javits Award), CA- 14764, and a grant-in-aid from the Bayer Corporation, Elkhart, IN. to SB. Our special thanks to Mrs. Dorisanne Nielsen, Mrs. Rosemary Patti, Mr. Rajit Basu and Mr. Patrick Kelly for their help in the preparation of the final draft of the manuscript.

References I . E. Bieberich, B. Freischutz, S.S. Liour and R. K. Yu. J. Neurochem. 1999, 72, 1040-1049. 2. T. Garofalo, M. Sorice, P. Misasi, B. Cinque, M. Giammatteo, G.M. Pontieri, M.G. Cifone and A. Pavan. J. Biol. Chem. 1998,273, 35153-35160. 3. S. Galadari, K. Kishikawa, C. Kamibayashi, M. C. Mumby, and Y. A. Hannun. Biochemistry 1998,37, 11232-1 1238. 4. J. Y. Lee, L. G . Leonhardt, and L. M. Obeid. Biochem. J. 1988,334, 457-461. 5. L. Riboni, P. Viani, R. Bassai, A. Prinetti, and G. Tettamanti, Prog. Lipid Res. 1997,36, 153195. 6. R. Angum, A. M. Ali, Z. Begrem, T . Vanaji, and A. Khar, FEBS Lett. 1998, 439, 81-84. 7. D. lshikawa, H. Kikkawa. K. Ogino, Y. Hirabayashi, N. Oku, and T. Taki. FEBS Lett. 1998, 441,20-24. 8. A. Gocht, G. Rutter, and B. Knie. Histochem. Cell Biol. 1998, 110, 217-220. 9. D. N. Irani, J. Neuroimmunol. 1998, 87, 1-16. 10. S. M. Crain and K. F. Shen, Ann. N.Y. Acad. Sci. 1998, 845, 106-125. 11. T. Taki, D. Ishikawa, M. Ogura, M. Nakajima and S. Handa, Cancer Res. 1997, 57, 18821888. 12. L. M. Obeid, C. Linardic, 0. L. A. Karolak, and Y. A. Hannun. Science 1993, 259, 17691771. 13. R. Kolesnick, and D. W. Golde, Cell 1994, 77, 325-328. 14. Y. Hannun, Science 1996,274, 1855-1859. 15. S. Spiegel and S. Milstein, J. Membr. Biol. 1995, 146, 225-237. 16. S. Spiegel, D. Foster and R. Kolesnick, Curr. Opin. Cell. Biol. 1996, 8, 159-167. 17. L. C. Edsall, J. R. Van Brocklyn, 0. Cuvillier, B. Kleuser, and S. Spiegel, Biochemistry 1998, 37, 12891-12898. 18. S. Hakomori, Annu. Rev. Immunol. 1984, 2, 103-126. 19. S. Hakomori a n d N. Kajima, J. Biol. Chem, 1990,265, 18713-18716.

Rejerences

343

20. Y. Nagai and S. Tsuji, Prog. Brain Res. 1993, l U 1 , 119-126. 21. T. Feizi, Curr. Opin. Struct. Biol. 1991, I , 766-770. 22. S. Basu, M. Basu, and S. Dastgheib. in Comprehensive Natural Products Chemistry. (eds: D. H. R. Barton, K., Nakanishi and 0. Meth Cohn). Vol. 3, Carbohydrate. Pergamon Press, New York, 1999, pp 107-128. 23. S. Basu, S. Ghosh, S. S. Basu, J. W. Kyle, Z . Li and M. Basu, Ind. J. Biochem. Biophys. 1993, 30, 315-323. 24. T. Ariga, and R. K. Yu, Neurochem. Res. 1998, 23, 291-303. 25. S. Ichikawa, and Y. Hirabayashi, Trends Cell Biol. 1998, 8, 198-202. 26. N. S. Radin. Biochem. Pharmacol. 1998, S7, 589-594. 27. S. Basu. Glycobiology 1991, I , 469-475. 28. S. Basu, M. Basu, and S. S. Basu, in Biology of Sialic Acid. (Ed. A. Rosenberg). 1995, pp 6994. 29. G. Echter and K. Sandhoff, J. Biol. Chem. 1993,268, 6341L6344. 30. J. C. Paulson and K. J. Colley, J. Biol. Chem. 1989, 264, 17615-17618. 31. H. Lannert, K. Gargas, I. Meissner, F. T. Wieland, and D. Jeckel, J. Biol. Chem. 1998, 273, 2939-2946. 32. P. Fredman, C. J. Wikstrand, J. E. Mansson, G. Reifenberger, S. H. Bigner, A. Rasheed, and L. Svennerholm, Glycoconj. J. 1996, 13, 391-399. 33. M. Basu, T. De, K. K. Das, J. W. Kyle, H. C. Chon, R. J. Schaeper, and S. Basu, Methods. Enzymology 1987, 138, 575-607. 34. S. Basu, M. Basu, and J. L. Chien, J. Biol. Chem. 1975, 250, 2956--2962. 35. T. Pacuzka and J. Koscielak, Eur. J. Biochem. 1976, 64, 499-506. 36. K. A. Presper, M. Basu, and S. Basu, J. Biol. Chem. 1982,257; 169-173. 37. K. A. Presper, M. Basu, and S. Basu, Proc. Natl Acad. Sci. USA 1978, 75, 289-298. 38. M. Basu, J. W. Hawes, Z. Li, S. Ghosh, A. Khan, B. J. Zhang and S. Basu, Glycobiology 1991, I , 527-535. 39. E. Holmes, G. Ostrander and S. Hakomori, J. Biol. Chem. 1985, 260, 7619-7627. 40. C. Campbell, and P. Stanley, J. Biol. Chem. 1984,259, 11208-11214. 41. B. W. Weston, R. P. Nair, R. D. Larsen, and J. B. Lowe, J. Biol. Chem. 1992, 267, 41524160. 42. E. V. Chandrasekharan; R. K. Jain, R. D. Larsen, K. Wlasichuk, R. A. DiCioccio, and K. L. Matta, Biochemistry 1996, 35, 8925-8933. 43. A. T. Nguyen, E. H. Holmes, J. M. Whitaker, S. Ho, S. Shetterly, and B. A. Macher, J. Biol. Chem. 1998,273, 25244-25249. 44. K. P. Lee, L. M. Carlson, J. B. Woodcock, N. Ramachandra, T. L. Schultz, T. A. Davis, J. B. Lowe, C. B. Thompson, and R. D. Larsen, J. Biol. Chem. 1996, 271, 32960-32967. 45. E. M. Sajdel-Sulkowska, F. I. Smith, G. Wiederschain, and R. H. McCluer, Glyconj. J. 1997, I4, 249-258. 46. T. Kudo, Y. Ikehara, A. Togayachi, M. Kaneko, T. Higara, K. Sasaki, and H. Narimatsu, J. Biol. Chem. 1998,273, 26729-26738. 47. R. A. DeBose-Boyd, A. K. Nyame, and R. D. Cummings, Glycobiology 1998,8, 905-907. 48. M. Blaxter, Science 1998, 228, 2041-2046. 49. E. T. Marques, J. B. Weiss, and M. Strand. Mol. Biochem. Parasitol. 1998, 93, 237-250. 50. C.H. Hokke, A. P. Neeleman, C. A. Koeleman, and D. H. van denEijnden, Glycobiology 1998,8, 393-406. 51. S. Basu, A. Schultz, M. Basu, and S. Roseman, J. Biol. Chem. 1971, 246, 4272-4279. 52. P. Morel1 and N. S. Radin, Biochemistry 1969, 8, 506. 53. A. Bosio, E. Binczek, M. M. Le Beau, A. A. Fernald, and W. Stoffel, Genomics 1996,34,69-75. 54. A. Bosio, E. Binczeck, and W. Stoffel, Genomics 1996, 35, 223--226. 55. S. Basu and M. Basu in “The Glycoconjugates”, ed. M. Horowitz, Academic Press, New York, 1982, Vol. 3. pp 265-285. 56. S. Chatterjee, N. Ghosh, and S. Khurana, J. Biol. Chem. 1992, 267, 7148-7153. 57. S. Basu, B. Kaufman, and S. Roseman, J. Biol. Chem. 1965,240, 4115-41 18. 58. S. Ghosh, J. W. Kyle, F. Daussin, S. Dastgheib, Z. Li, and S. Basu, Glycoconj. J. 1995, 12, 838-~ 847.

344

I 9 Glycosyltransferases in Glycosphingolipid Biosynthesis

59. S. S. Basu, S. Dastgheib, S. Ghosh, M. Basu, P. Kelly, and S. Basu, Acta Biochim. Pol. 1998, 45, 451-467. 60. F. Kaplan, and P. Hechtman, J. Biol. Chem. 1983,258, 770-776. 61. N. Miyazaki, S. Fukumoto, M. Okada, T. Hosegawa, T. Furukawa, and F. Furukawa, J. Biol. Chem. 1997,272,24794-24799. 62. M. Basu and S. Basu, J. Biol. Chem. 1972,247, 1489-1495. 63. S. Basu, S. Ghosh, S. S. Basu, J. W. Kyle, Z . Li, and S. Basu, Ind. J. Biochem. Biophys. 1993, 30, 315-323. 64. D. M. Jenis, S. Basu, and M. Pollard, Cancer Biochem. Biophys. 1982, 6, 37-45. 65. M. Basu, S.-A. Weng, H. Tang, F. Khan, B.-J. Zhang, and S. Basu, Glycoconj. J., 1996, 13, 423-432. 66. A. Passaniti and G. W. Hart, Cancer Res. 1990, 50, 7261-7267. 67. N. L. Shaper, J. H. Shaper, J. L. Merth, Proc. Natl. Acad. Sci. USA 1986,83, 1573-1577. 68. H. Narimatsu, S. Sinha, K. Brew, H. Okayama and P. Qasba, Proc. Natl Acad. Sci. USA 1986,83,4720-4724. 69. M. Basu and S. Basu, J. Biol. Chem. 1973,248, 1700-1706. 70. S. Basu, M. Basu, J. W. Kyle, T. De, K. Das and R. J. Schaeper, in Enzymes of Lipid Metabolism (eds. L. Fruyz and S. Gatt) Plenum Press, New York, 1986, pp 233-245. 71. D. H. Joziasse, J. H. Shaper, D. H. van denEijnden, A. J. van Tunen and N. L. Shaper, J. Biol. Chem. 1989,264, 14290-14297. 72. R. D. Larsen, V. P. Rajan, M. M. Ruff, J. Kukowska-Latallo, R. D. Cummings and J. B. Lowe, Proc. Natl Acad. Sci. USA 1989, 86, 8227-923 1. 73. D.H. Joziasse, N.L. Shaper, D.Y. Kim, D.H. Van den Eijnden, and J.H. Shaper, J. Biol. Chem. 1992,267, 5534-5541. 74. C. Stults, R. Larson, and B. Macher, Glycoconj. J. 1995, 12, 680-689. 75. T. Caetzee, X. Li, N. Frijita, J. Marcus, K. Suzuki, U. Francke, and B. Popko, Genomics 1966, 35, 215-221. 76. M. Basu, D.Sc. Thesis, University of Calcutta, 1974, Study of the biosynthesis of a blood group B specific glycosphingolipid. 77. J. Kyle, Ph.D. Thesis, University of Notre Dame, 1985, Characterization of glycolipid galactosyltransferases from embryonic chicken brains. 78. S. Ghosh, Ph.D. Thesis, University of Notre Dame, 1992, Studies on the regulation of two galactosyltransferases from embryonic chicken brains. 79. S. S. Basu, Ph.D. Thesis, University of Notre Dame, 1996, Cloning and expression of two glycolipid glycosyltransferases, pl,4-galactosyltransferase (GalT-4) and an alpha 2,3 sialyltransferance (SAT-3), involved in the biosynthesis of sialyl-LeXepitope in embryonic chicken brain. 80. S. Dastgheib, Ph.D. Thesis, University of Notre Dame, 1997, Characterization of p(1-3) galactosyltransferase and ceramide glycanase involved in biosynthesis and degradation of gangliosides. 81. S. Basu, M. Basu, H. Tang, S. S. Basu, J. Baker, and M. Girzadas. Proc. Glycolipid Satellite Symposium-Port Ludlow, 1995, p. 22. 82. T. Schwientek, R. Almeida, S. B. Levery, E. H. Holmes, E. Bennett, and H. Clausen, J. Biol. Chem. 1998,273, 29331-29340. 83. M. Ujita, J. McAuliffe, T. Schwientek, R. Almedia, 0. Hindsgaul, H. Clausen, and M. Fukuda. J. Biol. Chem. 1998,273, 34843-34849. 84. M. Amado, R. Almeida, F. Carneiro, S. B. Levery, E. H. Holmes, M. Nomoto, M. A. Hollingworthy, H. Hassan, and T. Schwientek. J. Biol. Chem. 1998,273, 12770-12778. 85. S. Dastgheib, S. Ghosh, M. Basu, and S. Basu, Glycoconj. J. 1997, 5 , 348. 86. H. Sprong, B. Kruithof, R. Leigendekker, J. W. Slot, G. van Meer, and P. Van der Slings. J. Biol. Chem. 1998,273, 25880-25888. 87. P. van der Bious, G. J. Strous, M. Lopes-Cardozo, J. Thomas-Oates, and G. van Meer. Biochem. J. 1996,317, 589-597. 88. S. K. Cho, J. Yeh, M. Cho, and R. D. Cummings, J. Biol. Chem. 1996,271, 3238-3246. 89. S. Basu, Ph.D. Thesis, University of Michigan, Ann Arbor, MI 1966. Studies on the biosynthesis of gangliosides.

References

345

90. J. C. Steiganvald, B. Kaufman, S. Basu and S. Roseman, J. Biol. Chem. 1975, 250, 67276734. 91. R. J. Schaeper, K. K. Das, Z . Li, and S. Basu, Carbohydrate Res. 1992,236, 227-244. 92. T. Taniguchi and A. Makita, J. Biol. Chem. 1984, 259, 5637-5642. 93. Y. Hashimoto, M. Sabine, K. Iwasaki, and A. Suzuki, J. Biol. Chem. 1993, 268, 2585725864. 94. M. Basu, S. Basu, W. G. Shanabruch, J. R. Moskal, and C. H. Evans. Biochem. Biophys. Res. Commun. 1976, 71, 385-392. 95. K. K. Das, M. Basu, C. H. Evans. and S. Basu. Carbohydrate Res. 1986, 149, 119-135. 96. K. Furukawa, H. Saejima, N. Niikawa, H. Shiku, and K. Furukawa, J. Biol. Chem. 1996, 271, 20836-20844. 97. Y. Nagata, S. Yamashiro, J. Yodoi, K. 0. Lloyd, H. Shiku, and K. Furukawa, J. Biol. Chem. 1992, 167, 12082-12089. 98. G. Pohlentz, D. Klein, G. Schwarzmann, D. Schmitz and K. Sandhoff. Proc. Natl Acad. Sci. USA 1988,85, 7044-7048. 99. M. L. Allende, R. M. Schwendiman,and W. W. Young. Curr. Eye Res. 1997,16, 1263-1269. 100. J. L. Daniotti, V. M. Rosdles Fritz, J. A. Martina, K. Furukawa and H. J. Maccioni, Neurochem. Int. 1997,31, 11-19. 101. V. M. Rosales Fritz, J. L. Doniotti and H. J. Maccioni. Biochim. Biophys. Acta. 1997, 354, 153-158. 102. G. Zhu, E. Jaskiewicz, R. Bassi, D. S. Darling and W. W. Young. Glycobiology 1997, 7,987996. 103. C. B. Conde, V. R. Grabios, S. N. Deza and B. L. Caputto, Neurochem. Res. 1997,22, 483490. 104. V. R . Grabios, C. B. Conde and R. Caputto. J. Neurosci. 1994, 3Y, 330-338. 105. K. Takamiya, A. Yamamoto, K. Furukawa, S. Yamashiro, M. Shin, M. Okada, S. Fukumoto, M. Heraguchi, N. Takeda, K. Fujimura, M. Sakae, M. Kishikawa, H. Shiku, K. Furukawa, and S. Aizawa. Proc. Natl Acad. Sci. USA 1996, 93, 10662-10667. 106. J. L. Chien, T. Williams, and S. Basu, J . Riol. Chem. 1973, 248. 1778-1785. 107. K. Yeung, J. R. Moskal, J. L. Chien, D. A. Gardner and S. Basu, Biochem. Biophys. Res. Commun. 1974,59, 252-260. 108. R. Schaeper, Ph.D. Thesis 1987, University of Notre Dame. Notre Dame, In. Studies on the acidic ganglio- and neutral globo-glycosphingolipid: N-acetylgalactosaminyltransferases in embryonic chicken brain. 109. T. Ishibashi, S. Kijimoto, and A. Makita. Biochim. Biophys. Acta. 1974, 337, 92-106. 110. N. Taniguchi, N. Yokosdwd, S. Gasa, T. Koike, and A. Makita, Proc. 6th Int. Symp. Glycoconjugate, 1981, pp 44-45. 11 1, T. Ishibashi, I. Ohkubo and A. Makita. Biochim. Biophys. Acta. 1977, 484, 24-34. 112. S. Basu, M. Basu, H. Den and S. Roseman. Fed. Proc. Fed. Am. Soc. Exp. Biol. 1970, 29, 410. 113. M. Basu and S. Basu. J. Biol. Chem. 1984, 259, 12557-12565. 114. M. Basu, F. A. Khan, K. K. Das and B. .I.Zhang. Carbohydrate Res. 1991,209, 261-277. 115. E. H. Holmes, S. Hakomori, and G. K. Ostrander. J. Biol. Chem. 1987,262, 15649-15658. 116. D. K. H. Chou and F. B. Jungalwala, J. Biol. Chem. 1993,268, 21727721733, 117. H. Schachter. Glycobiology 1991, 1, 4 5 3 ~461. 118. N . Taniguchi, M. Yoshimura, E. Miyoshi, Y. Ihara, A. Nishikawa, and S. Fujii. Glycobiology 1996, 6. 691-694. 11 9. R. Kumar, J. Yang, R. D. Larsen, and P. Stanley. Proc. Natl Acad. Sci. USA 1990, 87. 99489952. 120. M. Sarkar, E. Hall, Y. Nishikawa, R. J. Simpson, R. L. Moritz, R. Dunn. and H. Schachter. Proc. Natl Acad. Sci. USA 1991, 87, 234 238. 121. M. Nakamura, Y. Furukawa, R. Sasaki, J . Masuyarna, J. Kikuchi. S. Iwase, T. Kudo, H. Narirnatsu, S. Asakura, S. Fujiwara and J. Inokuchi. Glycobiology 1999, 9, 1-12. 122. S. Tsuji J. Biochem. 1996, 120, 1--13. 123. S. Tsuji, A. K. Dutta, and J. C. Paulson, Glycobiology 1996, 6, v-viii. 124. B. Kaufman, and S. Basu, Methods Enzymol. 1966,8, 365-368.

346

19 Glycosyltrunsferuses in Glycosphingolipid Biosynthesis

125. T. W. Keenan, D. J. Morre, and S. Basu, J. Biol. Chem. 1974,249, 310-315. 126. B. Kaufman, S. Basu and S. Roseman, J. Biol. Chem. 1968,243, 5804-5806. 127. K. Nara, Y. Watanabe, K. Maruyama, K. Kasahara, Y. Nagai, and Y. Sanai. Proc. Natl Acad. Sci. USA 1994, 91, 7952-7956. 128. Y. Yoshida, N. Kajima, N. Kurosawa, T. Hamamoto, and S. Tsuji. J. Biol. Chem. 1995,270, 14628-14633. 129. H. Higashi, M. Basu, and S. Basu, J. Biol. Chem. 1985,260, 824-828. 130. M. Basu, S. Basu, A. Stoffyn, and P. Stoffyn. J. Biol. Chem. 1982,257, 12765-12769. 131. S. S. Basu, M. Basu, Z. Li and S. Basu, Biochemistry 1996,35, 5166-5174. 132. S. S. Basu, M. Basu, S. Dastgheib, S. Ghosh, and S. Basu, Indian J. Biochem. Biophys. 1997, 34,977104. 133. H. Kitagawa and J. C. Paulson, J. Biol. Chem. 1994,269, 1394-1401. 134. V. Liepkans, A. Jalif and G. Larson, Biochemistry 1988,27, 8683-8688. 135. P. Stoffyn and A. Stoffyn. Carbohydr. Res. 1980, 78, 327-340. 136. T. J. Gu, X.-B. Gu, T. Ariga, and R . K. Yu, FEBS Lett. 1990,275, 83-86. 137. M. Pitto, P. Palestini, and M. Masserini, FEBS Lett. 1996, 383, 223-226. 138. K. Mari, Y. Yoshida, N. Kojima, S. Tsuji, J. Biol. Chem. 1996, 271, 29366-29371. 139. Y. Nagai and M. Iwamori in Biology of the Sialic Acids (ed. A. Rosenberg) Plenum Press, New York, 1995, pp 197-241. 140. A. Rosenberg in Biology of the Sialic Acids. (ed. A. Rosenberg) Plenum Press, 1995, pp 363372. 141. S. Hakomori in Biology of the Sialic Acids. (ed. A. Rosenberg) Plenum Press, 1995, pp 243260. 142. S. Hakomori, K. Handa, and K. Iwaguchi, Glycobiology 1998, 8, XI-XVIII. 143. G. VanEcten-Deckert and K. Sandhoff in Comprehensive National Products Chemistry. (eds: D. H. R. Barton, K. Nakanishi and 0. Math. Cohn) Vol. 3 (ed. B. M. Pinto) Paragamon Press, New York, pp 87-105. 144. H. C. Yohe and R. K . Yu, J. Biol. Chem. 1980,255, 608-613. 145. R. Gornati, S. Basu, G. Bernadini, A. M. Rizzo, F. Rossi, and B. Berra, Mol. and Cell. Biochem. 1997,166, 117-124. 146. M. Basu, S. Basu, and M. Potter in ACS Symp. Vol. 128, 1980, 187-212. 147. J. L. Chien, Ph.D. Thesis, University of Notre Dame, IN, 1975. Enzymatic synthesis of globoside and bovine spleen glycosphingolipids. 148. S. Basu, M. Basu, K. K. Das, F. Daussin, R. J. Schaeper, P. Banerjee, F. A. Khan and I. Suzuki, Biochimie 1988, 70, 1551-1563. 149. K. Sasaki, E. Watanabe, K. Wawashima, S. Sekine, T. Dohi, M. Oshima, N. Hannai, T. Nishi, and M. Hasegawa, J. Biol. Chem. 1993,268, 22782-22787. 150. Y.-C. Lee, N. Kojima, E. Wada, N. Kurosawa, T. Nakaokai, T. Hamamoto, and S. Tsuji, J. Biol. Chem. 1994,26Y, 10028-10033. 151. J. Weinstein, U. E. Lee, K. McEntee, L. Por-Hsiung and J. C. Paulson, J. Biol. Chem. 1987, 262, 17735-17743. 152. A. K. Dutta and J. C. Paulson, J. Biol. Chcm. 1995,270, 1497-1500. 153. H. C. Yohe, L. J. Nacala and R. K. Yu, J. Biol. Chem. 1982,257,249-252. 154. B. Freischiitz, M. Saito, H. Rahman, and R. K.Yu, J. Neurochem. 1994,62, 1965-1973. 155. Y. Yukiko, N . Kurosawa, T. Kanematsu, N. Kujima, and S. Tsuji, J. Biol. Chem. 1996, 271, 30 167-301 73. 156. N. Nara, Y. Watanabe, Y. Kawashima, T. Tai, Y. Nagai and Y. Sanai, Eur. J. Biochem. 1996,238, 647-652. 157. F. B. Jungalwala, Neurochem. Res. 1994, 19, 945-957. 158. D. K. H. Chou, S. A. Tobel, and F. B. Jungalwala, J. Biol. Chem. 1998,273, 8508-8512. 159. T. Ariga, T. Kohriyama, L. Freddo, N. Latov, M. Saito, K. Kon, S. Ando, M. Suzuki, M. E. Hemling, K. L. Rinehart, Jr., S. Kusunoki, and R. K. Yu, J. Biol. Chem. 1987, 262, 848. 160. K. K. Das, M. Basu, Z . Li, S. Basu, and F. Jungalwala, Indian J. Biochem. Biophys. 1990,27, 396-401. 161. K. K. Das: M. Basu. S. Basu, D. K. H. Chou, and F. B. Jungalwala, J. Biol. Chem. 1991,266, 5238-5243.

References

341

162. D. K. H. Chou, S. Flores, and F. B. Jungawala. J. Biol. Chem. 1991,266, 17941-17947. 163. C. Kawashima, K. Terayama, M. Li, S. Oka, andT. Kawasaki. Glycoconj. J. 1992, Y, 307314. 164. Y . A. Hanun, J. Biol. Chem. 1994,269, 3125-3238. 165. S. Oka, K. Terayama, C. Kawashima, and T. Kawasaki, J. Biol. Chem. 1992, 267, 22711 227 14. 166. K. Terayama, S. Oka, T. Seiki, Y. Miki, A. Nakamura, Y. Kozutsumi, K. Takio and T. Kawasaki, Proc. Natl Acad. Sci. USA 1997, 94, 6093-6098. 167. S. Basu, A. Schultz, and M. Basu, Proc. Fed. Am. Soc. Exp. Biol. 1969, 27, 346. 168. S. Basu, B. Kaufman and S. Roseman, J. Biol. Chem. 1968,243, 5802-5804. 169. S. Basu, B. Kaufman and S. Roseman, J. Biol. Chem. 1973,248, 1388-1394. 170. A. Shukla, and N. S. Radin. J. Lipid Res. 1991, 32, 713-722. 171. S. Ichikawa, H. Sakiyama, G. Suzuki, K. I.-P. J. Hidari, and Y. Hirabayashi, Proc. Natl Acad. Sci. USA 1996. 93, 4638-4643. 172. S. Ichikawa, K. Ozawa, and Y. Hirabayashi. Biochem. Biophys. Res. Commun. 1998, 253, 701-711. 173. R. Watanabe, K. Wu, P. Paul, D. L. Marks, T. Kobyashi, M. R. Pittelkow, and R. E. Pagano, J. Biol. Chem. 1998,273, 9651-965s. 174. K. Kamio, S. Gasa, and A. Makita, J. Lipid Res. 1992, 33, 1227-32. 175. T. Kobyashi, K. Honke, and S. Gasa, Cancer Res. 1993, 53, 24484-2489. 176. D. K. H. Chou, and F. B. Jungalwala, J. Biol. Chem. 1993,268, 330-336. 177. H. Bakker, J. Biol. Chem. 1997,272,29942-29946.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

20 Biosynthesis of Glycogen Peter J. Roach

20.1 Summary Glycogen is a branched polymer composed primarily of glucose residues connected via a-1,4-glycosidic and a- 1,6-glycosidic linkages. Its biosynthesis requires the participation of three enzymes, glycogenin, glycogen synthase and the glycogen branching enzyme. Glycogen synthase and glycogenin are glucosyl transferases that utilize UDP-glucose as the glucosyl donor. The branching enzyme catalyzes first the hydrolysis of an a-1,4-glycosidic followed by reattachment of the resulting oligosaccharide moiety via an a- 1,6-glycosidic linkage, thus creating a branchpoint. Glycogen biosynthesis can be divided into two stages, initiation and bulk polymer synthesis. Initiation is mediated by glycogenin which first catalyzes the attachment of glucose to specific tyrosine residues and thereafter adds further glucose residues, in a-1,4-glycosidic linkages, until the oligosaccharide is 8-20 in length. This “primed” glycogenin then serves as substrate for bulk synthesis by glycogen synthase and branching enzyme to generate mature glycogen. It has been proposed that there is a discrete, intermediate form of glycogen, designated proglycogen, whose concentration is independent of normal mature glycogen. However, the functional significance of proglycogen is controversial and it is not clear that a separate enzymology exists for its controlled metabolism. Glycogen is present in most cell types but, in mammals, is quantitatively highest in skeletal muscle and liver. Many microorganisms also synthesize glycogen and the related compound starch is produced by plants. The polymer is believed to function as a reserve of glucose and its deposition is linked to the nutritional status of the cell or organism.

350

20 Biosynthesis of Glycogen

U D r i b

glycogenin

UDP glycogen synthase branching enzyme

Figure 1. Pathway of glycogen synthesis. The two stages are initiation mediated by glycogenin (GN) and bulk synthesis catalyzed by glycogen synthase and the branching enzyme.

20.2 Introduction Glycogen, and its plant counterpart starch, are widely distributed in nature and are generally considered to function as reserves of glucose [ 11. The present Chapter will focus on the biosynthesis of glycogen in eukaryotes (Figure 1) and will emphasize recent information regarding the three biosynthetic enzymes, glycogenin, glycogen synthase and the branching enzyme. Glycogenin and glycogen synthase both transfer glucose from UDP-glucose to form glycosidic linkages. In mammals, glycogen and the corresponding enzymatic machinery are present in most cell types but, in absolute amount, liver and skeletal muscle are the major repositories of glycogen and hence are most important for whole body glucose metabolism. It is of interest that these two tissues use different isozymes to catalyze many of the key steps in glycogen and glucose metabolism. However, the regulatory process and the overall role of glycogen in blood glucose homeostasis goes beyond the scope of this chapter (see [2, 31 for recent reviews) which will concentrate on the enzymology and mechanism of glycogen synthesis. Glycogen is a branched polymer in which the primary linkage is via a-1,4glycosidic linkages with branchpoints created by a-l,6-glycosidic linkages. The average chain length is 13 residues. A number of considerations have led to a model with 12 tiers of outer chains, to yield a mass of lo7 Da, corresponding to -55,000 glucose residues [4-61. This would correspond to the size of P-particles that can be observed by electron microscopy [7]. Minor amounts of phosphate and glucosamine have been detected in glycogen but whether these have a physiological significance is unclear [8, 91. The reducing end of skeletal muscle glycogen is linked covalently to the initiator protein glycogenin, as is discussed in more detail below. In addition, several other proteins are associated with glycogen and the existence of “glycogen particles” was noted many years ago [lo]. Since glycogen itself is polydisperse and some of the proteins appear to interact weakly, it is unlikely that a multiprotein complex of defined stoichiometry exists. Nonetheless, it is increasingly clear that the organization of protein complexes within cells can be important and such is likely to be true for glycogen metabolism.

20.3 Glycogenin and the Initiation of Glycogen Synthesis

351

20.3 Glycogenin and the Initiation of Glycogen Synthesis 20.3.1 History Work from the laboratories of Chrisman and Whelan led to the definition of a distinct initiation phase for glycogen synthesis (reviewed in [ 1l]), mediated by a specific protein called glycogenin. The rabbit muscle protein was extensively characterized biochemically by Cohen and colleagues who also determined its amino acid sequence [ 121. Cloning of cDNAs followed [ 131 as well as the identification of two separate genes in man (GYGI and GYGZ) encoding functional glycogenins (reviewed in [ 141). The corresponding proteins are glycogenin-1, the original form found in skeletal muscle, and glycogenin-2 which is expressed preferentially in heart, liver and pancreas [ 151. Glycogenin and glycogenin-like proteins are widely distributed in nature [ 11, 14, 161. The yeast Saccharomyces cerevisiae has two genes, GLGl and GLG2, that encode functionally similar proteins. The nematode Caenorhabditis elegans genome contains four related genes, although there is no functional information about the corresponding proteins yet. At least one Drosophila melunoguster sequence is present in the databases and no fewer than eight related Arabidopsis thaliana sequences are present, one containing two copies of the catalytic domain. 20.3.2 Properties Glycogenin is a self-glucosylating protein that catalyzes the transfer of glucose from UDP-glucose, in the presence of Mn2+,to form a covalently linked oligosaccharide chain that serves as the primer for elongation by glycogen synthase. With the participation also of the branching enzyme, the bulk biosynthesis of glycogen ensues. Most biochemical studies have addressed the rabbit skeletal muscle form of glycogenin or glycogenin-l, which consists of a polypeptide with 332 residues and a subunit Mr of -37,000. The protein co-purifies with rabbit muscle glycogen synthase and is covalently attached to glycogen. The covalent linkage is through a specific tyrosine residue, Tyr194 in the rabbit muscle protein, and consists of a C-l0-tyrosyl linkage. A comparable Tyr residue can be identified in other glycogeninlike sequences even though this segment of the molecule is, interestingly, not so highly conserved as other regions. The yeast Glg2p has two Tyr residues in this region that both appear to be modified. After the initial glucosylation of Tyr, glucose residues are added in a- 1,4-glycosidic linkage from UDP-glucose until the oligosaccharide is around 10 residues long. Some glycogenins can form slightly longer chains but there appears to be a limit imposed on the length of polymer produced. At first, there was some debate as to whether a single enzyme mediated the two reactions discussed above. The best evidence to date suggests that indeed a single protein is involved. Recombinant glycogenin-1 produced in Escherischia coli is active and already glucosylated [13]. However, when expressed in an E. coli mutant that is unable to produce UDP-glucose, no glucose was detected in the glycogenin but it was still able to self-glucosylate 1171.

352

20 Biosynthesis of Glycogen

20.3.3 Reaction Mechanism

The reaction catalyzed by glycogenin is quite unusual. Glycogenin is the enzyme, the substrate and the product. Unlike some self-processing events, there is true catalysis, in that there is turnover at the catalytic site, though admittedly for only a few cycles. The mechanism of the self-glucosylation is also interesting. The glucosylated Tyr appears to serve only as a site of attachment and is not required for catalysis. Thus, mutation of Tyr194 to Phe resulted in an enzyme that was even more active in transferring glucose to small molecule acceptors 118, 191. It was known that wild type glycogenin can transfer glucose to maltose or maltose derivatives in a transglucosylation reaction. Glycogenin is a dimer as judged by gel filtration [20, 211 although analysis of its packing in crystals suggested that tetramers might also form 1181. Early in the study of glycogenin, the molecularity of the self-glucosylation reaction had been examined by measuring the concentration dependence of the reaction. It was found that the rate of self-glucosylation (as a specific activity) of rabbit muscle glycogenin-1 was constant over a significant concentration range, consistent with an intra-molecular reaction [20, 22, 231. Assuming that glycogenin is a dimer in solution, then the self-glucosylation could be inter- or intra-subunit. Lin et al. [20] addressed this issue by making two types of mutants of glycogenin-1. One was a Tyr-Phe mutation at Tyr194 to yield an active subunit that is incapable of being glucosylated. The other was a L y s i G l n mutation at Lys85, the only positively charged residue that was conserved in all glycogenin-like sequences analyzed by Roach and Skurat [14]. This mutant had very low selfglucosylating activity. Lin el al. [20] showed that, when the two mutant proteins were mixed, the Lys+Gln mutant was glucosylated after a lag. This observation supports the idea of an inter-subunit glucosylation. The lag can be explained if there is subunit exchange leading to the formation of a heterodimer which allows the transglucosylation to occur. This hypothesis is consistent with the fact that the subunit-subunit interactions seem relatively weak, with a Kd probably in the micromolar range. Alonso et a1 1191 have reported that there is concentration dependence of the self-glucosylation of the human muscle glycogenin at higher dilution, a finding which could be explained if the dimer dissociated and the intersubunit reaction consequently became bimolecular. In summary, the evidence so far suggests that self-glucosylation occurs under most conditions as an intramolecular reaction in which one subunit of glycogenin modifies its partner in a dimer. Furthermore, the relatively weak subunit-subunit interactions could have a physiological function, in allowing the individual glycogenin subunits to dissociate after the task of initiation is complete and to go their separate ways to establish different mature glycogen particles. 20.3.4 Domain Structure

Comparison of the sequences of glycogenin and related proteins enabled some provisional conclusions about domain structure to be made [ 141. Subsequent work, together with the appearance of more full-length sequences in the databases, has

20.3 Glycogenin and the Initiation of Glycogen Synthesis I

I11

I1

ir

I DXD

Y

IV

353

V

C H “DY@”

~FNXG~I / ‘‘KPW”

~n*+ UDP-Glc

t

Glycogen Synthase

Figure 2. Glycogenin structure. Several key features of glycogenin and glycogenin-like proteins are shown. The boxes correspond to regions of relatively high sequence conservation and are assigned Roman numerals (described in the text). The lines correspond to regions that are less well conserved. The NHz-terminus and the region between domains IV and V are variable in length. The arrows indicate regions of the molecule to which UDP-glucose and glycogen synthase are thought to bind.

allowed some refinement of this analysis but a three-dimensional structure has still to be developed. The catalytic portion of glycogenin consists of about 200 residues (Figure 2) and a COOH-terminally truncated form of the enzyme has been shown to be active [ 191. This region of glycogenin can be divided into five regions. Four are domains I-IV in the nomenclature of Roach and Skurat [ 141, and the other is the region between domains I11 and IV which is the site of glucosylation for those glycogenins studied functionally. As noted, the sequence around the glucosylation site(s) is not particularly well-conserved. Domain I is not conserved in all species, and is quite different in the yeast Glg proteins. Domain I1 contains some highly conserved residues, including the motif -D-X-D- which is present in all known glycogenins and almost all related sequences. This same motif has recently been reported to be common to a number of glycosyl transferase families and to be required for activity [24, 251. The residues have been implicated in binding Mn2+, consistent with the requirement of glycogenin for this divalent cation. Thus, domain I1 is likely to be involved in binding of the substrate UDP-glucose. Consistent with this idea are the results of Carrizzo and Curtino [26] who studied affinity-labeling of glycogenin with azido-UDP-glucose. Two labeled fragments were identified, one corresponding to Domains I1 plus I11 and the other starting before the glucosylation site and extending to cover domain IV. Only 16 residues NH2-terminal to the -D-XD- sequence is the highly conserved Lys residue mutated by Lin et al[20] to cause a severe loss of glycogenin activity, further supporting the hypothesis that Domain I1 is critical for activity. Thus, the presence of the -D-X-D- motif and the equivalent of rabbit muscle K85 are likely to be present in legitimate glycogenin-like proteins. Domain 111 also contains highly conserved residues, including -F-N-X-G- and -D/ N-G-X-D-Q-G-. Domain IV is short but quite well-defined by the motif -K-P-W-. There is then a region of variable length and sequence. Some small regions of limited similarity were noted previously [ 141 but the most consistent COOH-terminal feature is at the extreme COOH-terminus of mammalian and yeast glycogenins, Domain VII. For both yeast [27] and mammalian [57] glycogenin, there is now good evidence that this region is involved in protein-protein interaction with glycogen synthase. The only discernable sequence motif is -W-E-X,-D-Y-M/L and it

354

20 Biosynthesis of Glycogen

has been shown that mutation of the Tyr in yeast Glg2p affects elongation by yeast glycogen synthase [27]. 20.3.5 Function

The function of glycogenin as originally proposed by Krisman and Whelan has largely been substantiated experimentally. All the biochemical studies suggest a role as an initiator protein: the ability to self-glucosylate, to bind to glycogen synthase, to be elongated by glycogen synthase and even the ability of the subunits to dissociate. The absolute requirement of glycogenin for glycogen synthesis has been rigorously tested only in yeast, where disruption of the two yeast glycogenin genes eliminated glycogen synthesis [28]. Efforts are underway to make the same test in animals by generating glycogenin knock-out mice. Less certain is whether there might be other functions for glycogenin outside of glycogen metabolism, especially with the observation that glycogenin can transfer glucose to exogenous acceptors. It is no longer unprecedented for the same enzyme catalytic subunit to be recruited to multiple tasks by targeting and/or regulatory proteins. Furthermore, multiple glycogenin genes exist in some species. In humans, this may be explained by tissue specific function. In yeast, the reason for maintaining dual copies of some genes and not others is poorly understood. In. C. elegans and A . thaliana, there are four and perhaps eight related genes, respectively. Admittedly, there is no biochemical evidence for functionality of the corresponding proteins and not all may be legitimate glycogenins as defined here, but they could represent a family of transglucosylases of which glycogenin is the best characterized member.

20.4 Glycogen Synthase and the Bulk Synthesis of Glycogen 20.4.1 Properties Glycogen synthase was the third enzyme shown to be controlled by reversible covalent phosphorylation [29]. It catalyzes the formation of a-1,4-glycosidic linkages by transferring glucose from UDP-glucose to the non-reducing end of an existing oligosaccharide chain in a reaction chemically not dissimilar to that mediated by glycogenin (for reviews, see [30-321). Initially, the substrate is primed glycogenin and glycogen synthase-glycogenin interactions are likely to be important at this stage. Although mammalian glycogenin could complement the absence of yeast Glg proteins in whole yeast cells, there was species specificity to the glycogenin-glycogen synthase interaction. Thus, yeast glycogen synthase elongated yeast glycogenin much more effectively than mammalian glycogenin in vitro, and vice versa [28]. Later in the synthesis of the glycogen molecule, the interaction of glycogen synthase with glycogen is likely to dominate. Glycogen synthase is widely distributed in animal cells and there are distinct liver and muscle isoforms. Most tissues express the muscle form whereas liver expresses

20.4 Glycogen Synthrrse and the Bulk Synthesis of Glycogen

355

an isoform that is only 70% identical in amino acid sequence [33, 341. There is no good evidence for expression of both isoforms in the same tissue or cell. S. cerevisiae has two glycogen synthase genes, G S Y l and GSY2, that encode proteins with -50% sequence identity to the mammalian enzymes [35]. In the sequence databases, there is one sequence from D. melanogaster and one from C. elegans. Bacteria and plants have related activities but these starch/glycogen synthases differ in utilizing ADP-glucose as a substrate and have little overall sequence similarity to glycogen synthase [ 361. Glycogen synthase, from muscle or liver, exists as a homo-oligomer of a subunit of about 85,000 Da, and is probably a tetramer [30, 311. The enzyme binds glycogen and is a component of glycogen particles [lo]. Purification from rabbit muscle usually involves a step to degrade glycogen and the resulting enzyme co-purifies with glycogenin-1 [37]. Pitcher et al. [37] found a 1 : 1 ratio of glycogen synthase and glycogenin, and considered the latter to be a subunit. The enzyme activity is regulated by a combination of covalent phosphorylation and ligand binding. A number of metabolites are known to affect glycogen synthase activity. Several nucleotides, including ATP, ADP, AMP and UDP, are inhibitors but the most important metabolite is probably glucose-6-Pwhich is a powerful allosteric activator. Mg2+ions are also activators but, in contrast to glycogenin, there appears to be no absolute requirement for a metal ion. 20.4.2 Structure of Glycogen Synthase

No three-dimensional structure is known nor has there been extensive structurefunction analysis. Furthermore, there are fewer glycogen synthase sequences available than for glycogenin, in part because the plant and bacterial starch or glycogen synthases are overall unrelated in sequence. From the limited information available, the glycogen synthase molecule can be divided into five segments (Figure 3). Segment I is a short, variable region that contains two mammalian phosphorylation sites and is involved in regulation. This region is absent in the yeast proteins. Segment I1 is a region of about 40 residues that is well conserved. An early study I

I11

I1

4 K 3 8 H P-sites

I

K300

\ f \

UDP-Glc

IV

V

I""'

H

t

I '

P-sites

?Glucose-6-P

Figure 3. Glycogen synthase structure. The glycogen synthase molecule is divided into segments, designated by Roman numerals, based on knowledge of function and alignment of the sequences of the two mammalian isoforms and the two yeast proteins, which are the only full length sequences available. Unlike the corresponding analysis with glycogenin, for which related sequences from a wide range of species are available, the analysis leads generally to rather large blocks of similarity. Boxes indicate regions of relative conservation and the lines indicate less conserved regions. The filled boxes correspond to the small regions that are conserved also in bacterial and plant glycogen/ starch synthases. Note that yeast glycogen synthases lack segment I.

356

20 Biosynthesis of Glycogen

[38] had reported that pyridoxal phosphate modified lysine residue(s) in the glucose6-P binding site of liver glycogen synthase. When similar experiments were performed with the rabbit muscle enzyme, inactivation by pyridoxal phosphate was protected by UDP-glucose not glucose-6-P. Two lysines, Lys-38 and Lys-300, were identified as the residues modified [39]. In an independent study, Lys38 was implicated in UDP-glucose binding by affinity labeling [40]. A similar motif to that surrounding Lys38 (-K-V-G-G-) is present in E. coli glycogen synthase around a Lys residue involved in ADP-glucose binding (-K-T-G-G-). Segment 11 may therefore be involved in substrate binding. However, mutation of the equivalent of Lys38 in muscle (411 or yeast [56] glycogen synthase does not inactivate the enzyme. Twenty to 30 unconserved residues separate segment I1 from segment 111, a stretch of about 300 amino acids with a high degree of conservation. This region contains Lys300 which was also implicated in UDP-glucose binding. Mutation of the equivalent of this residue in yeast Gsy2p does cause inactivation of the enzyme [56]. The region also contains a residue, Pro442, which is found to be mutated in a small proportion of the human population and which has been linked to mild diabetes [41]. Expression of the mutant in COS cells indicated impaired glycogen synthase activity. The region also contains an interesting motif, @-&-H-X-W-X-X-G(where @ represents a hydrophobic residue). Although bacterial glycogen synthases and plant starch synthases in general have very little sequence homology with eukaryotic glycogen synthases, two regions of similarity can be identified, the motif mentioned above and another discussed below. The significance of this sequence is not known. Segments 111, and perhaps 11, are most likely involved in binding UDP-glucose. Segment IV begins about 60 amino acids COOH-terminal of segment 111 and extends for some 180 residues. This region contains the second motif, -P-S-X-Y/ F-E-P-W/C-G-Xg-G-Xq-G-, that is highly conserved in glycogen and starch synthases. Again, the function is unknown but the motif has absolute conservation of some residues, including a series of three Gly residues present every tenth residue. Features common to these enzymes are the reaction catalyzed and the ability to bind to polysaccharide. One speculation is that the motif might be involved in glycogen binding. Bai et al. [33] have proposed that other sequences in segment IV might be involved in glucose-6-P binding and recent experimental evidence from our laboratory supports this hypothesis [56]. Segment V is the variable, COOH-terminal portion of the molecule containing regulatory phosphorylation sites. The mammalian muscle isoform has 7 sites, the liver isoform 5 and the yeast Gsy2p 3 sites. Two of the yeast sites can readily be seen as conserved versions of mammalian sites 3a and 3b. It is possible that the yeast and mammalian enzymes share some phosphorylation controls but not others. 20.4.3 Branching Enzyme

This enzyme catalyzes the formation of a-l,6-branchpoints in glycogen by excising segments of existing oligosaccharide and reforming a-1,6-1inkages [42, 431. The mammalian enzyme is a monomer of 702 amino acids [44] with a M, of -77,000 [43]. The enzyme has 67% identity with the yeast homolog which is encoded by the

20.5 Intern2ecliute.s in the Biosynthesis of Glycogen

357

GLQ gene [45, 461. In yeast, Glc3p is required for the accumulation of glycogen. Defects in human branching enzyme are found in Type IV glycogen storage disease which is characterized by abnormal, poorly branched glycogen in the liver and muscle [47]. However, under normal conditions, branching enzyme is not thought to be limiting for glycogen accumulation despite its relatively low abundance compared with other glycogen metabolizing enzymes [43].

20.5 Intermediates in the Biosynthesis of Glycogen As noted earlier, there seem to be some upper limits to the size of the glycogen Pparticles based on physical constraints to polysaccharide packing and perhaps also the ability for metabolizing enzymes to access the outer residues. Between primed glycogenin and a mature P-particle, with -55,000 glucose residues attached, there are obviously a large number of intermediates. Indeed, using a Rat1 fibroblast model, Skurat et al. [48] reported the existence of a continuum of glycogenglycogenin species from the size of glycogenin up to species unable to enter SDS gels. Whelan and his co-workers 111, 49-51], in contrast, have proposed the existence of a discrete intermediate that they term “proglycogen” (using the term “macroglycogen” for the larger glycogen molecules). Proglycogen has been defined experimentally as a species of about 400,000 that scarcely enters SDS-PAGE and, consistent with a relatively high protein : carbohydrate ratio, as having the property of being precipitated by acids such as trichloroacetic acid. High molecular weight glycogen is soluble in trichloroacetic acid. Others have described potentially similar entities [52--551.The existence of an intermediate form of glycogen could have important implications for the regulation of glycogen metabolism and would provide another possible locus for regulation. The central issue is how glycogen intermediates accumulate (Figure 4). Several aspects of the proglycogen proposal have yet to be firmly established, in my opinion. First, there is some debate as to whether the 400,000 Da species seen in SDS-PAGE is truly a discrete molecular entity. From their analysis of glycogen in Rat1 cells, Skurat et al. [48] found that the accumulation of material just entering SDS-PAGE with Laemmli gels was due to the use of discontinuous gel electrophoresis. Upon analysis by a continuous Weber- Osborne gel system, the glycogen was smoothly distributed throughout the gel. Secondly, for the proglycogen to have a separate existence, either it must have a particular stability or a distinct enzymology. An enhanced stability would require, for example, that glycogen synthase and phosphorylase kinetic properties were such that species the size of proglycogen were not favored as substrates. Though not impossible, such behavior has not been explicitly demonstrated. Alternatively, separate proglycogen metabolizing activities would have to exist, minimally, distinct proglycogen and macroglycogen synthases (or proglycogen and macroglycogen phosphorylases). It has been reported that formation of proglycogen by “proglycogen synthase” can occur at low UDPglucose concentrations lower than needed for normal glycogen synthase activity, which has been interpreted as suggesting the presence of a different enzyme [50, 521.

358 A

20 Biosynthesis of Glycogen

TCA precipitable soluble

B

U

Figure 4. Intermediates in the biosynthesis of glycogen. The schematic is intended to illustrate the differences in two models for glycogen biosynthesis. Each panel shows the size distribution of a population of polysaccharide molecules during the biosynthesis of glycogen starting from primed glycogenin. A: The process starts with primed glycogenin (GN; n < 10) which is depicted as a triangle to symbolize a discrete size ( t = 0). Each graph shows the fraction of the population containing n glucose residues attached,f(n), as a function of n. As n, the number of attached glucose residues, increases, the ability to be precipitated by trichloroacetic acid (TCA) decreases (heavier shading depicting greater TCA-solubility-see bar at the top of the panel). B: In the discontinuous “proglycogen” model, glycogenin is converted to a discrete intermediate species, proglycogen (PG; n 2,000), before its conversion to glycogen or “macroglycogen” (MG; n 50,000). From left to right, the panels indicate different time points ( t l - 1 4 ) on the way to complete conversion of the glycogenin to glycogen in a hypothetical time course. C: A similar sequence is shown for the continuous model, in which there is no particular accumulation of any specific intermediate but the generation, with time, of a polydisperse population of ever increasing average size until the mature glycogen molecule is formed.

-

-

However, if the UDP-glucose concentration is too low, limiting substrate could prevent the formation of larger products even by conventional glycogen synthase. Molecular definition of the proglycogen synthase would be a major advance in this area but, to date, no proglycogen synthase protein has been defined by gene or protein sequence. An alternative to the proglycogen model is simply one in which there is a continuum of polysaccharide intermediates whose exact size distribution is determined by the relative activities of various metabolizing enzymes. Operational definition of proglycogen as that glycogen fraction precipitated by trichloroacetic acid will always be possible and will naturally select for molecules with a higher protein content. What is important is whether such a fraction of the glycogen has any particular metabolic or regulatory significance. More work is needed.

20.6 Conclusion Recent advances in gene cloning mean that most, if not all, of the basic enzymatic players involved in glycogen biosynthesis have been identified at the molecular level.

Rfferences

359

The exception is the putative proglycogen synthase. The future will hopefully yield more information on the three dimensional structures of these proteins and the physical basis for their interactions, information which can be correlated with improved understanding of enzymatic mechanism that should come with more analysis of defined, recombinant proteins.

Acknowledgments Work from the author’s laboratory was supported by National Institute of Diabetes Digestive and Kidney Diseases grants DK27221 and DK42576, as well as the Indiana University Diabetes Research and Training Center, DK20542. References 1. J. Preiss, D. A. Walsh, The comparative biochemistry of glycogen and starch in V. Ginsburg, P. Robbins (Eds.): Biology of Carbohydrures, Vol. I , John Wiley, New York 1981, p. 199-314. 2. P. J. Roach, A. V. Skurat, R. A. Harris, Regulation of glycogen metabolism in L. S. Jefferson, A. D. Cherrington (Eds.): The endocrine pancreas and regulation ofmetabolism, Vol. 2 1999, in

press. 3. M. Bollen, S. Keppens, W. Stalmans, Specific features of glycogen metabolism in the liver Biochem. J. 336 1998, 19-31. 4. Z. Gunja-Smith, J. J. Marshall, C. Mercier, E. E. Smith, W. J. Whelan, A revision of the Meyer-Bernfeld model of glycogen and amylopectin FEBS Lett. 12 1970, 101-4. 5. E. Goldsmith, S. Sprang, R. Fletterick. Structure of maltoheptaose by difference Fourier methods and a model for glycogen J. Mol. B i d 156 1982, 41 1-27. 6. E. Melendez-Hevia, T. G. Waddell, E. D. Shelton, Optimization of molecular design in the evolution of metabolism: the glycogen molecule Biochern. J. 295 1993, 477-83. 7. P. Drochmans, Morphologie du glycogene. Etude du miscroscope electronique de colorations negative du glycogene particulaire Journal of Ultrasctruct. Rex 6 1962, 141-63. 8 . B. R. Kirkman, W. J. Whelan, Glucosamine is a normal component of liver glycogen FEBS Lett. I94 1986, 6-1 1. 9. J. Lomako, W. M. Lomako, W. J. Whelan, R . B. Marchase, Glycogen contains phosphodiester groups that can be introduced by UDPglucose: glycogen glucose 1-phosphotransferase FEBS Lett. 329 1993, 263-7. 10. F. Meyer, L. M. Heilmeyer, Jr., R. H. Haschke, E. H. Fischer, Control of phosphorylase activity in a muscle glycogen particle. I. Isolation and characterization of the protein-glycogen complex J. Bid. Chern. 245 1970, 6642-8. 11. M. D. Alonso, J. Lomako, W. M. Lomako, W. J. Whelan, A new look at the biogenesis of glycogen FASEB J. Y 1995, 1126-37. 12. D. G. Campbell, P. Cohen, The amino acid sequence of rabbit skeletal muscle glycogenin Eur. J . Biochem. 185 1989, 119-25. 13. E. Viskupic, Y. Cao, W. Zhang, C. Cheng, A. A. DePaoli-Roach, P. J. Roach, Rabbit skeletal muscle glycogenin. Molecular cloning and production of fully functional protein in Escherichia coli J. Biol. Chern. 267 1992, 25759--63. 14. P. J. Roach, A. V. Skurat, Self-glucosylating initiator proteins and their role in glycogen biosynthesis Prog. Nuc. Acid Rex Mol. Biol. 57 1997, 289-316. 15. J. Mu, A. V. Skurat, P. J. Roach, Glycogenin-2, a novel self-glucosylating protein involved in liver glycogen biosynthesis J. Biol. Chem. 272 1997, 27589-97. 16. M. C. Gannon, F. Q. Nuttall, Glycogen in liver: characteristics and biosynthesis Trends Glycosci. Glycotechnol. 8 1996, 163-94.

360

20 Biosynthesis

of

Glycogen

17. M. D. Alonso, J. Lomako, W. M. Lomako, W. J. Whelan, J. Preiss, Properties of carbohydrate-free recombinant glycogenin expressed in an Escherichia coli mutant lacking UDP-glucose pyrophosphorylase activity FEBS Lett. 352 1994, 222-6. 18. Y. Cao, L. K. Steinrauf, P. J. Roach, Mechanism of glycogenin self-glucosylation Arch. Biochem. Biophys. 319 1995, 293-8. 19. M. D. Alonso, J. Lomako, W. M. Lomako, W. J. Whelan, Catalytic activities of glycogenin additional to autocatalytic self-glucosylation J. Biol. Chem. 270 1995, 15315-9. 20. A. Lin, J. Mu: J. Yang, P. J. Roach, Self-glucosylation of glycogenin, the initiator of glycogen biosynthesis, involves an inter-subunit reaction Arch. Biochem. Biophys. 363 1999, 163-170. 21. C. Smythe, P. Watt, P. Cohen, Further studies on the role of glycogenin in glycogen biosynthesis Eur. J. Biochem. 189 1990, 199-204. 22. Y. Cao, A. M. Mahrenholz, A. A. DePaoli-Roach, P. J. Roach, Characterization of rabbit skeletal muscle glycogenin. Tyrosine 194 is essential for function J. Biol. Chem. 268 1993, 14687-93. 23. J. Pitcher, C. Smythe, P. Cohen, Glycogenin is the priming glucosyltransferase required for the initiation of glycogen biogenesis in rabbit skeletal muscle Eur. J. Biochem. 176 1988, 391-5. 24. B. Busch, F. Hofmann, J. Selzer, S. Munro, D. Jeckel, K. Aktoties, A common motif of eukaryotic glycosyltransferases i s essential for the enzyme activity of large Clostridial cytotoxins J. Biol. Chem. 273 1998, 19566-72. 25. C. A. R. Wiggins, S. Munro, Activity of the yeast MNNI a-1,3-mannosyltransferaserequires a motif conserved in many other families of glycosyltransferases Proc. Natl Acad. Sci. USA 95 1998, 7945--50. 26. M. E. Carrizo, J. A. Curtino, Identification of two uridine binding domain peptides of the UDP-glucose-binding site of rabbit muscle glycogenin Biochem. Biophys. Rex Commun. 253 1998, 786-9. 21. J. Mu, C. Cheng, P. J. Roach, Initiation of glycogen synthesis in yeast. Requirement of multiple tyrosine residues for function of the self-glucosylating Glg proteins in vivo J. Biol. Chem. 271 1996, 26554-60. 28. C . Cheng, J. Mu, I. Farkas, D. Huang, M. G. Goebl, P. J. Roach, Requirement of the selfglucosylating initiator proteins Glglp and Glg2p for glycogen accumulation in Saccharomyces cerevisiue Mol. Cell. Biol. 15 1995, 6632-40. 29. D. L. Friedman, J. Larner, Studies on UDPG-a-glucan transglucosylase 111. Interconversion of two forms of muscle UDPG-a-glucan transglucosylase by a phosphorylation-dephosphorylation reaction sequence Biochemistry 2 1963, 669-75. 30. P. J. Roach, Liver glycogen synthase in P. D. Boyer, E. G. Krebs (Eds.): The Enzymes, Vol. 17 A , Academic Press, Orlando, FL 1986, p. 499-539. 31. P. Cohen, Muscle glycogen synthase in P. D. Boyer, E. G. Krebs (Eds.): The Enzymes, Vol. 17 A , Academic Press, Orlando, FL 1986, p. 461-497. 32. A. V. Skurat, P. J. Roach, Regulation of glycogen biosynthesis in D. LeRoith, J. E. Olefsky, S. Taylor (Eds.): Diabetes Mellitus: A fundamental and clinical text, J. B. Lippincott Company, Philadelphia 1995, p. 213-222. 33. G. Bai, Z. J. Zhang, R. Werner, F. Q. Nuttall, A. W. Tan, E. Y. Lee, The primary structure of rat liver glycogen synthase deduced by cDNA cloning. Absence of phosphorylation sites l a and l b J. Bid. Chem. 265 1990, 7843-8. 34. F. Q. Nuttall, M. C. Cannon, G. Bai, E. Y. Lee, Primary structure of human liver glycogen synthase deduced by cDNA cloning Arch. Biochem. Biophys. 31 1 1994,443-9. 35. I. Farkas, T. A. Hardy, M. G. Goebl, P. J. Roach, Two glycogen synthase isoforms in Saccharomyces cerevisiue are coded by distinct genes that are differentially controlled J. Biol. Chem. 266 1991, 15602-7. 36. J. Preiss, T. Romeo, Molecular biology and regulatory aspects of glycogen biosynthesis in bacteria Prog. Nucleic Acid Res. Mol. Biol. 47 1994, 299- 329. 37. J. Pitcher, C. Smythe, D. G. Campbell, P. Cohen, Identification of the 38-kDa subunit of rabbit skeletal muscle glycogen synthase as glycogenin Eur. J. Biochem. 169 1987, 497-502. 38. M. J. Ernest, K. H. Kim, Regulation of rat liver glycogen synthetdse D. Role of glucose 6phosphate and enzyme sulfhydryl groups in activity and glycogen binding J. Biol. Chem. 249 1974, 501 1-8.

Rrfirences

36 1

39. A. M. Mahrenholz, Y. H. Wang, P. J. Roach, Catalytic site of rabbit glycogen synthase isozymes. Identification of an active site lysine close to the amino terminus of the subunit J. Bid. Chem. 263 1988, 10561-7. 40. M. Tagaya, K. Nakano, T. Fukui, A new affinity labeling reagent for the active site of glycogen synthase. Uridine diphosphopyridoxal J. Biol. Chem. 260 1985, 6670-6. 41. M. Orho-Melander, H. Shimomura, T. Sanke: S . K . Rasmussen, K. Nanjo, 0. Pederson, L. C. Groop, Expression of naturally occurring variants in the muecle glycogen synthase gene Diabetes 48 1999, 9 18-920. 42. W. B. Gibson, B. Illingsworth, D. H. Brown, Studies of glycogen branching enzyme. Preparation and properties of a-1,4-glucan- a-1 ,4-glucan 6-glycosyltransferase and its action on the characteristic polysaccharide of the liver of children with Type IV glycogen storage disease Biochemistry 10 1971, 4253-62. 43. F. B. Caudwell, P. Cohen, Purification and subunit structure of glycogen-branching enzyme from rabbit skeletal muscle Euv. J. Biochem. 109 1980, 391-4. 44. V. J. Thon, M. Khalil, J. F. Cannon, Isolation of human glycogen branching enzyme cDNAs by screening complementation in yeast J. Biol. Chem. 268 1993, 7509-13. 45. J. F. Cannon, J. R. Pringle, A. Fiechter, M. Khalil, Characterization of glycogen-deficientglc mutants of Succharomyces cereuisiae Genetics 136 1994, 485-503. 46. D. W. Rowen, M. Meinke, D. C. LaPorte, GLC3 and GHAl of Saccharomyces cerevisiae are allelic and encode the glycogen branching enzyme Mol. Cell. Biol. 12 1992, 22-9. 47. Y. Bao, P. Kishnani, J. Y. Wu, Y. T. Chen. Hepatic and neuromuscular forms of glycogen storage disease type IV caused by mutations in the same glycogen-branching enzyme gene J. Clin. Invest. 97 1996, 941-8. 48. A. V. Skurat, S.-S. Lim, P. J. Roach, Glycogen biogenesis in Rat1 fibroblasts expressing rabbit muscle glycogenin Eu:uv. J. Biochem. (1997). 49. J. Lomako, W. M. Lomako, W. J. Whelan, Proglycogen: a low-molecular-weight form of muscle glycogen FEBS Lett. 279 1991, 223-8. 50. J. Lomako, W. M. Lomako, W. J . Whelan, R. S. Dombro, J. T. Neary, M. D. Norenberg, Glycogen synthesis in the astrocyte: from glycogenin to proglycogen to glycogen FASEB J. 7 1993, 1386-93. 51. J. Lomako, W. M. Lomako, W. J. Whelan, Glycogen metdbohsm in quail embryo muscle. The role of the glycogenin primer and the intermediate proglycogen Eur. J. Biochem. 234 1995, 343-9. 52. N. Ercan, M. C. Gannon, F. Q. Nuttall, Incorporation of glycogenin into a hepatic proteoglycogen after oral glucose administration J. Biol. Chem. 269 1994, 22328-33. 53. E. R. Lacoste, M. C. Miozzo, J. A. Curtino, A glycogen precursor associated with membrane in retina Biochem. J. 267 1990, 775-9. 54. M. C. Miozzo, E. R. Lacoste, J. A. Curtino, Characterization of the proteoglycogen fraction non-extractable from retina by trichloroacetic acid Biochrm. J. 260 1989, 287-9. 55. P. Ghosh, A. C. Heath, M. J. Donahue. R. A. Masaracchia, Glycogen synthesis in the obliquely striated muscle of Ascaris suum Ew . J. Biochem. 183 1989, 679-85. 56. Cheng, Roach, unpublished. 57. Skurat, unpublished results.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

21 Biosynthesis of Hyaluronan Paraskevi Heldin and Torvard C. Laurent

21.1 Introduction The glycosaminoglycan hyaluronan (hyaluronic acid, hyaluronate, HA) contains alternating residues of N-acetyl-D-glucosamine and D-glucuronic acid linked by p(1-4) and p(1-3) linkages [l-31. It was first described by Meyer and Palmer in 1934 and its structure was established to a large extent in the 1950s by Meyer and collaborators. When studied by various physical chemical techniques, HA has been described as a very high-molecular-weight linear polymer (mol.wt. 106-107 Da) which in infinitely dilute solution behaves as a highly expanded flexible coil. However, recent work by Scott has shown that HA chain-segments do interact with each other at higher concentrations so that a hyaluronan solution may take the form of a “cross-linked” network [2]. The rheological properties of HA solutions are rather striking-they exhibit marked viscoelasticity and shear dependence; a 10 mg/ml solution may have a relative viscosity at zero shear in the order of 105-106. Hyaluronan has been detected in varying concentrations in all vertebrate tissues analyzed [ 1-31. The highest concentrations are usually found in loose connective tissues such as synovial fluid, umbilical cord, the vitreous body and skin (0.1-10 mg/ml). HA has therefore been designated as a connective tissue polysaccharide. However, significant amounts have also been described in brain, muscles, liver etc. The lowest concentrations are found in the circulation (10-100 ng/ml) and this is due to a very effective removal of the polysaccharide from blood in the liver sinusoids. HA is usually not seen in invertebrates [%] but it was early observed that certain bacteria can synthesize HA [4]. The physiological function of HA was originally connected with its macromolecular properties. It was thought of as being an inert space-filling material between cells and a lubricant in joints. With increasing knowledge HA was assigned regulating functions in water and solute transport in the tissues [ 5 ] . The first evidence that HA was not inert came in the beginning of the 70s when the interactions between HA and cartilage proteoglycans and HA and cells were described. Cell

364

21 Biosynthesis of Hyaluronan

surface receptors for HA were isolated ten years later. We now know a number of proteins, hyaladherins, which specifically recognize HA and which endow the polysaccharide with important structural functions as well as cell-regulating activities [61.

21.2 Site of Biosynthesis HA distinguishes itself from other glycosaminoglycans in connective tissues in that it is not formed in the Golgi. Markovitz and Dorfman [7] showed that it was synthesized by the protoplast membrane of group A streptococci and subsequently Philipson and Schwartz [S] and Prehm [9] described the synthesis in the cell membrane of eucaryotic cells. According to Prehm, HA is synthesized on the interior side of the membrane and translocated to the outside. In bacteria HA is deposited in the surrounding capsule after synthesis. In eucaryotic cells it is located in the pericellular space, where it is difficult to detect by ordinary morphological techniques, but where it can be visualized by an ingenious method described by Clarris and Fraser [ lo]. These investigators showed that when a suspension of particles are added to a fibroblast culture they become excluded from an extensive hyaluronidasesensitive structure around each cell (Figure 1).

21.3 Biosynthetic Precursors HA as other polysaccharides is synthesized from nucleotide sugars; in this case UDP-glucuronic acid and UDP-N-acetylglucosamine [ 1 11. The initiation mecha-

Figure 1. Visualization of the HA containing pericellular coat around a human mesothelial cell from pleura, by the exclusion of formalinfixed erythrocytes.

21.4 Hyuluronan Synthuses

365

nism of HA synthesis is unknown. Other glycosaminoglycans exist as proteoglycans and a core protein, to which the first sugar in the chain is bound by glycosidic linkage, is required for the initiation. Blockage of protein synthesis therefore blocks the formation of polysaccharide. However, it has repeatedly been shown that inhibition of protein synthesis does not have this effect on HA synthesis [ 111 which argues for a different starting mechanism than for proteoglycans. Although diverging reports have occurred in the literature regarding protein contents in HA preparations it has not been shown that the reducing end of hyaluronan is bound covalently to protein. Some authors have looked for the possibility that HA-synthesis utilizes lipid intermediates but have been unsuccessful [ 1 11.

21.4 Hyaluronan Synthases Although a considerable amount of work on the purification of the HA synthesizing enzymes (HAS) have been carried out through the years, it took 40 years from the pionering work on HA biosynthesis by Glaser and Brown and Dorfman and his colleagues [ 111, before the enzymes were identified and cloned. In the following few years a wealth of information has accumulated.

21.4.1 Microbial Enzymes The HA synthase in streptococci was identified when acapsular mutants lacking HA synthesizing capacity, with about 100-fold less virulence than the encapsulated srreptococci, were studied [ 121. These studies led to definition of a gene locus, designated has (hyaluronic acid synthase), responsible for production of the HA capsule in group A streptococci [ 13, 141. The has operon was shown to be composed of three genes: has A, has B and has C. Further studies revealed that the has A gene encodes a 42 kDa protein which was shown to be a HAS possessing both UDPGlcNAc and UDP-GlcA glycosyltransferase activities [ 15, 161. The enzyme was designated spHAS (Streptococcus pyogmes HAS) 1171. The other two genes in the has operon, encodes UDP-glucose dehydrogenase (converts UDP-glucose to UDP-GlcA) and UDP-glucosepyrophosphorylase (converts glucose 1-phosphate and UTP to UDP-glucose). spHAS shows about 20% similarity to the NodC protein from Rhizohium, which synthesizes N-acetylglucosamine oligomers, and the DG42 protein which is a protein Differentially expressed at Gastrulation in embryos of the frog (Xenopzo laeuis) as well as about 10% similarity to chitin synthases from Sacchuromyces [18]. DeAngelis and Weigel [ 191 confirmed that spHAS is a functionally active HAS by expressing recombinant spHAS in Escherichia coli and by showing that the recombinant enzyme alone polymerizes the UDP-sugar moieties to high molecular mass HA. Recently, a HAS from group C Streptococcus equisimilis was also identified and designated seHAS [20]. Similar to spHAS, the enzyme migrates in SDS-PAGE as a

366

21 Biosynthesis of’Hyaluronan

component of molecular mass about 42 kDa; moreover, the two enzymes exhibit about 72% amino acid sequence similarity and possess four transmembrane segments and probably one membrane associated domain. Interestingly, seHAS exhibits a two-fold faster rate of chain elongation than spHAS. Furthermore, DeAngelis and his colleagues [21] reported a unique HAS from Pasteurella multocida (designated pmHAS) which is twice the size of and structurally different from the other known HAS. More recently, DeAngelis and collaborators while sequencing the genome of Paramecium bursaria chlorella virus (PBCV-1) found an open reading frame, A98R, that exhibited similarity to known HAS. Expression of the A98R gene in E. coli revealed that the encoded recombinant protein is an authentic HAS [22]. This was the first report of a virus capable of synthesizing a polysaccharide, and of a HAS outside the animal Kingdom. 21.4.2 Vertebrate Synthases In 1996 the first mammalian HAS genes were cloned using functional cloning approaches. Itano and Kimata [23], and Briskin and his colleagues [24] independently identified the first putative mammalian HAS isoenzyme from mouse (mHAS1) and human (hHASl), respectively. The encoded mouse and human proteins consist of 578 and 543 amino acid residues, respectively, and have no apparent NH2-terminal signal sequences. Simultaneously, Spicer and his colleagues [25], taking advantage of the strong similarity of regions among spHAS, DG42 and NodC, and Watanabe and Yamagushi [26], using a degenerate RT-PCR approach, cloned the second HAS isoenzyme from mouse (mHAS2) and human (hHAS2), respectively, both encoding 552 amino acid residues. One year later, Spicer et al. [27] cloned a third mouse HAS isoform (mHAS3). The homologous isoforms of mouse and human HAS exhibit about 99% sequence identity, whereas the various isoforms within a species show 55%0 (HASl/HAS2), 57% (HASl/HAS3) and 71% (HAS2/HAS3) amino acid sequence similarities. Hydrophilicity analysis of the three mammalian HAS isoforms revealed that they share structural features with bacterial HAS. All mammalian HAS proteins possess two membrane spanning regions located towards the N-terminus and a cluster of C-terminal transmembrane regions separated by a large central domain which resides inside the cell (Figure 2). This intracellular loop contains amino acid residues conserved between bacterial and vertebrate HAS proteins suggesting that the residues may be required for the catalytic activity of the HAS. Furthermore, the HAS proteins possess HA binding motifs that likely anchor the growing HA chain to HAS as well as putative phosphorylation motifs for protein kinases. Consensus phosphorylation sites for protein kinase C and CAMPdependent protein kinase A are present in the intracellular domains of HAS1 , HAS2 and HAS3 and likely provide a mechanism through which the activities of the three HAS isoforms are regulated (see below). Expression of any one of the three HAS genes led to HA synthesis suggesting that a single HAS protein possesses the ability to both synthesize HA chains and extrude them through the plasma membrane. By expressing the recombinant DG42 protein from Xenopus laeuis in yeast,

21.5 Mechanism of Synthesis

367

Figure 2. Schematic depiction of mammalian HAS protein membrane topology. All three HAS proteins exhibit similar structure. Regions in bold have high amino acid similarity between the HAS proteins. A HA binding motif is depicted by n.Open and filled circles represent consensus phosphorylation sites for protein kinase C and CAMP-dependent protein kinase, respectively. A conserved Cys-residue between bacterial and mammalian HAS proteins is marked by

DeAngelis and Achyuthan [28] confirmed that the DG42 protein synthesizes hyaluronan and transfers both GlcUA and GlcNAc to the growing polymer. This was the first characterization of a vertebrate HA synthase. The HAS genes have evolved from a common ancestral gene by sequential gene duplication and divergence. Since the three HAS genes are located on separate chromosomes in human as well as in mouse, it is likely that the HAS gene family arose early in vertebrate evolution [29]. Further characterization of the vertebrate HAS gene family was achieved by degenerate PCR cloning from genomic DNAs of different animals [30]. These studies revealed a fourth HAS gene which is expressed in X. lueuis, (xHAS-us), and that HAS1 and DG42 are orthologs (DG42 is therefore designated xHAS 1).

21.5 Mechanism of Synthesis At present, due to lack of structural data on HAS, we have only indirect information on the biosynthetic mechanism of HA. As mentioned above it is presumably very different from the synthesis of polysaccharide chains in proteoglycans: i) the polymers are synthesized at different sites. E.g. monensin which blocks transport of secretory proteins does inhibit synthesis of proteoglycans but not HA [31]; ii) they utilize different precursor pools [32, 331; iii) hyaluronan biosynthesis is, in contrast to proteoglycan biosynthesis, less sensitive to inhibition of protein synthesis [32, 34, 351; iv) oligosaccharides of sulphated glycosaminoglycans can in vitro accept transfer of sugar residues from corresponding UDP-sugars but it has not been possible to use oligosaccharides of HA as substrates for HAS [ 1 I].

The fact that each mammalian HAS protein can function independently as syn-

368

21 Biosynthesis of Hyaluronan

thase suggests that HAS is a single protein with many features; it interacts with the plasma membrane, it is able to bind two different substrates, it has two different glycosyl transferase activities, it has regulative phosphorylation sites and a capacity to translocate HA over the membrane. 21.5.1 Chain Elongation

In 1983 Prehm [36] proposed a mechanism by which HA is synthesized on the interior side of the plasma membrane by addition of intact UDP-sugars to the reducing end of the chain and concommittant release of the UDP attached to the chain in the preceeding sugar transfer. The non-reducing end of the chain was translocated through the membrane into the pericellular space. The evidence for this mechanism was the following: i) 3H-Uridine from labeled UDP-sugars could be incorporated into the HA-chain during synthesis in a membrane preparation of the enzyme; ii) When cells were grown in the presence of 32P,this isotope was found in isolated and purified high-molecular-weight HA. Hyaluronidase but not nucleases or pronase turned the 32P-labeledmaterial into a low molecular weight product; iii) When the 32P-labeled HA was incubated with membrane-bound enzyme and unlabeled UDP-sugars, 32P-UDPseemed to be released from the chain; iv) Acid and alkaline hydrolysis of 32P-labeledHA indicated that UDP was bound in the reducing end by an a-glycosidic linkage, which also is the case for the nucleotide substrates; v) Evidence for an elongation by addition of sugars to the reducing end was found in pulse-chase experiments [ 371. Radioactive sugars, which were incorporated in HA before unlabeled sugars were added to the incubation mixtures, could preferentially be removed by p-glucuronidase and P-hexosaminidase, which digest the chain from the non-reducing end. A similar result has later been described by us [38]. Evidence for translocation into the pericellular space was similarly shown by double pulse-labeling [9]. Intact cells were grown in 3Hglucosamine which was incorporated into high-molecular-weight HA. The cells were then disrupted and incubated with unlabeled UDP-glucosamine and 14Clabeled UDP-glucuronic acid which yielded a 3H- and I4C-labeled hybrid of high-molecular-weight HA. If intact cells before disruption had been treated with hyaluronidase the product was essentially 14C-labeled and of lower molecular weight indicating that the original 'H-labeled polysaccharide had been translocated extracellularly and that the chain still was growing at the membrane: vi) UDP-GlcNAc and UDP-GlcUA periodate oxidized in the ribose-residues acted as irreversible inhibitors of HA-synthesis, which could be expected if they were incorporated in the chain terminus and stopped further transfers [39]. So far the presence of UDP-chain intermediates in the HA synthesis have only been shown by isotope labeling. It would be desirable if a UDP-HA oligosaccharide

21.5 Mechanism ojsynthesis

369

from a synthesizing system could be isolated and chemically identified. Furthermore, such a fragment should be tested for acceptor activity in the presence of HAS. Recent work on the biosynthesis of HA in Pusteurellu multocidu has shown that in this organism HA is elongated by addition of sugar residues to the non-reducing end [40]. The Pusteurellu synthase has a distinctly different amino acid sequence than other HA synthases and it may be that there exists two different pathways to make HA from the same UDP-sugars. However, the observation by DeAngelis [40] may necessitate a reanalysis of the proposed mechanism in streptococci and eukaryotic cells.

21.5.2 Translocation Does a specific channel exist for the translocation of HA over the membrane? HAS itself is of relatively low molecular weight and the membrane organization of bacterial and mammalian HAS proteins does not enable the formation of a pore through which HA could be translocated across the membrane; usually membrane channels have more than six transmembrane domains [41]. One would expect that a membrane channel constitutes a larger complex. It is therefore of interest that we found HAS as part of a larger aggregate when solubilized by detergent [38]. Very recently, Tlapak-Simmons et al. [42] in a series of experiments on recombinant spHAS and seHAS showed that the active HAS enzym is a monomer which forms a complex with about 16 cardiolipin molecules thereby forming a pore through which the growing HA chain is extruded. HA carries hydrophobic patches [43] which could interact with membrane lipids during the passage [44].

21.5.3 Shedding A final mechanism of interest is the termination of polymerization and subsequent shedding of the polymer from the cell surface. It is known from numerous investigations that the molecular weight of HA in vivo and in cell cultures varies and that often inflammatory conditions result in lower degree of polymerization. Prehm [45] found that in several cell lines the cell-associated hyaluronan and that dissociated from the cell had similar and high molecular weights indicating that the shedding was due to a process at the cell membrane. However, from one cell line fragments of lower molecular weights were released indicating intrachain scissions. This degradation could be prevented by free radical scavengers and was presumably due to release of free oxygen radicals from the cells. The exact mechanism leading to the shedding from the cell membrane is unknown, but we have in time-lapse movies observed that cells often shed the whole pericellular cloud at the same time (unpublished) and it is therefore possible that the process could be actively regulated by the cells. Prehm and collaborators have recently proposed that a 56 kDa protein isolated from group C streptococci, which originally was thought to be the synthase, is actually a kinase taking part in the shedding [46].

370

21 Biosynthesis of Hyaluronan

21.6 Regulation of HA Synthesis Under physiological conditions HA synthesis is affected by the growth phase of the cells. HA biosynthesis is high in rapidly growing cells in culture such as foreskin fibroblasts and synovial cells and low in slowly growing cells. In addition, there is an intrinsic coupling between the HA production and the immediate environment of the cells. A large number of growth factors, cytokines and hormones which are produced in a tissue- and cell type-specific manner during inflammation and neoplasia have been shown to regulate HA synthesis (see e.g. [ 3 ] ) The stimulatory effects on HA biosynthesis of platelet-derived growth factor-BB (PDGF-BB), basic fibroblast growth factor (bFGF) and transforming growth factor-pl (TGF-PI) were not correlated to their mitogenicity; PDGF-BB was a more potent stimulator of HA synthesis than bFGF although they were equipotent mitogens, and TGF-P1 inhibits cell growth under the culture conditions used [47]. Furthermore, epidermal growth factor (EGF), follicle stimulating hormone and the oocyte factor are potent inducers of HA synthesis by cumulus cells, a prerequisite for a successful ovulation and fertilization in mammals [48]. Moreover, it has been demonstrated that insulin growth factor-I and EGF as well as IL-1p increase HA production in mesothelial cells from pericardium and from peritoneum of patients receiving continuous ambulatory peritoneal dialysis, respectively [49, 501. The potency of a cytokine on the stimulation of HA production differ between fibroblasts derived from different tissues. For example, leukoregulin, a lymphocyte released lymphokine, exhibits a much higher potency to induce HA synthesis in orbital fibroblasts than in dermal fibroblasts [51]. It is likely that leukoregulin is the molecular trigger of the abundant accumulation of HA in Graves’ ophthalmopathy. Interestingly, whereas the combination of IFN-)I and TNF-a led to production of high molecular mass HA, combination of IL-1 and TNF-a increased the amounts of low molecular mass HA in fibroblast cultures [52]. These observations suggest that IL-1 and TNF-a may be responsible for the production of polydisperse HA chains by rheumatoid synovial cells. Most of the growth factors and cytokines tested mediate their stimulatory activities on hyaluronan synthesis at least in part through activation of protein kinase C, but also other kinases seem to be involved in a direct or indirect activation of preexisting HAS molecules [35, 48, 531. The cytoplasmatic region of each HAS protein possesses consensus protein kinase C and CAMP-dependent protein kinase phosphorylation sequences, suggesting that the activities of HAS isoforms may be regulated partly through phosphorylation by kinases (Figure 3). The molecular cloning of HAS genes made it possible to investigate the HAS isoform(s) responsible for HA accumulation in cultured cells in response to different stimuli ([53]and our own unpublished observations). The expression of mRNA for the three HAS isoforms was found to be regulated independently in a cell type- and cell proliferative state-specific manner. Interestingly, TGF-P induced expression of HAS2 mRNA more rapidly than HAS1 mRNA in human skin fibroblasts suggesting different functions of the synthesized HA [53]. The three HAS proteins exhibit intrinsic differences in vitro in their ability to synthesize HA; plasma membranes

21.7 Concluding Remarks

371

Figure 3. Schematic representation of the regulation of HAS protein activity.

isolated from CHO cells transfected with cDNA for HAS2 synthesized HA chains of high molecular mass whereas plasma membranes from HAS1 and HAS3 transfected cells synthesized polydisperse HA chains [54]. It is possible that the HAS isoforms interact with different components which may have accessory or regulatory roles in HA biosynthesis. Most interestingly, the migration ability of untransfected and CHO transfected cells with the three HAS isoforms was HA dependent; clones of CHO transfected cells with high HA synthesizing capacity migrated much slower than clones synthesizing low amounts of HA [54]. As an attempt to elucidate possible differences in the transcription factors that can regulate the promotor region of HAS genes, Yamada et al. [55] analyzed the promotor region of mHAS1 and identified binding sites for certain transcription factors including AP-2, p53, sox-5, SPY, and MyoD. Further analysis of the promotors of the three HAS genes as well as elucidation of the signal transduction pathways that regulate their activities are necessary for understanding their different biological roles.

21.7 Concluding Remarks The increase in our knowledge of HA synthesis has accelerated especially with the isolation and characterization of several synthases which are differentially expressed both in the developing embryo and adult tissues. This is welcome information in many fields: where HA has aroused an increasing interest, notably among developmental biologists, tumor biologists, pathophysiologists and others. The accumulation of knowledge affords new possibilities to understand not only the synthetic mechanism but also the biological function of the polysaccharide. What are the roles of the three HAS isoforms in vivo? We will probably soon know. Mice deficient in HAS1, HAS2 and HAS3 genes are under investigation. HAS2 homozygous mutant mice die during embryonic development from a number

372

21 Biosynthesis qf Hyaluronan

of defects, including failure to form the endocardia1 cushions of the heart and yolk sac defects, whereas HAS1 and HAS3 knock-outs are viable [59]. The new information opens up the possibilities to intervene both in normal and pathological processes. For example, it is known that pathological HA accumulation may impair organ functions e.g. in rheumatoid joints, lung disease and transplanted organs [ 1-31. Increased knowledge about the regulation of HA biosynthesis should make it possible to design specific inhibitors for each one of the HAS isoforms which could be used in these condition. Polydispersity of HA chains is an important parameter in several biological processes such as angiogenesis. Low molecular weight HA induces differentiation of capillary endothelial cells whereas high molecular mass polymer prevents it (for references see [56]).As discussed above, different HAS may produce HA of different molecular weights which could be involved in regulation of angiogenesis. Another practical development is within reach: the large scale in vitro synthesis of HA. Since HA is a medically important product there is a need for production of the polysaccharide without protein and nucleic acid contaminations, which are difficult to remove in preparations from tissues. Attempts to produce in vitro have already been made [57].

Acknowledgments Supported by grants from The Swedish Cancer Society (3446-896-03XBB), Medical Research Council (K97-03X and 03X-4), Goran Gustafsson Foundation, The Swedish Gustav Vis 80-5rs Fond, BioPhausia and Q-Med.

References 1. Laurent, T.C.; Fraser, J.R.E. Hyaluronan. FASEB J. 1992, 6, 2397-2404. 2. Laurent, T.C. (ed.) The Chemistry, Biology and Medical Applicatioiis of Hyaluronan and Its Derivatives. Portland Press, London, 1998. 3. Fraser, J.R.E.; Laurcnt. T.C. Hyaluronan. In: Extracellular Matrix. Volume 2. Molecular Coinponents and Interactions. (Comper, W.D. ed.) Harwood Academic Publ. 1996, 141-199. 4. Kendall, F.E.; Heidelberger, M.; Dawson, M.F. A serologically inactive polysaccharide elaborated by mucous strains of Group A hemolytic streptococcus. J. Biol. Chem. 1937, 118, 61-69. 5. Comper, W.D.; Laurent, T.C. Physiological function of connective tissue polysaccharides. Physiol. Reu. 1978, 58, 255-315. 6. Toole, B.P. Hyaluronan and its binding proteins, the hyaladherins. Curr. Opin. Cell Biol. 1990. 2, 839-864. 7. Markowitz, A,; Dorfman, A. Synthesis of capsular polysaccharide (hyaluronic acid) by protoplast membrane preparations of group A streptoccocus J. Bid. Chem. 1962, 237, 273-279. 8. Philipson, L.H.; Schwartz, N.H. Subcellular localization of hyaluronate synthetase in oligodendroglioma cells. J. Biol. Chenz. 1984, 259, 5017-5023. 9. Prehm, P. Hyaluronate is synthesized at plasma membranes. Biochem. J. 1984, 220, 597-600. 10. Clarris, B.J.; Fraser, J.R.E. Barrier around synovial cells in vitro. Nature 1967, 214, 1159. 1 1. Roden, L.; Structure and metabolism of connective tissue proteoglycans. In: The Bioclzemistry qf Glycoproteins and Proteoglycuns (Lennarz. W.J. ed). Plenum Publ. New York 1980, 267371.

References

313

12. Wessels; M. R.; Moses, A. E.; Goldberg, J. B.; DiCesare, T. J. Hyaluronic acid capsule is a virulence factor for mucoid group A streptococci. Proc. Nut1 Acud. Sci. USA 1991, 88, 83178321. 13. Dougherty, B.A.; van de Rijn, 1. Molecular characterization of a locus required for hyaluronic acid capsule production in group A streptococci. J. Exp. Med. 1992, 175, 1291-1299. 14. DeAngelis, P. L.; Papaconstantinou, J .; Weigel, P.H. Isolation of a Streptococcus pyogenes gene locus that directs hyaluronan biosynthesis in acapsular mutants and in heterologous bacteria. J. Biol. Chem. 1993, 268, 14568-14571. 15. DeAngelis, P. L.; Papaconstantinou, J.; Weigel, P.H. Molecular cloning, identification, and sequence of the hyaluronan synthase gene from group A Streptococcus pyogenes. J. Biol. Chem. 1993,268, 19181-19184. 16. Dougherty, B. A,; van de Rijn, I. Molecular characterization of hus A from an operon required for hyaluronic acid synthesis in group A streptococci. J. Bid. Chem. 1994, 269, 169-175. 17. Weigel, P. H.; Hascall, V. C.; Tammi, M. Hyaluronan synthases. J. Bid. Chern. 1997, 272, 13997-14000. 18. DeAngelis, P. L.; Yang, N.; Weigel, P. H. The streptococcus pyogenes hyaluronan synthase: Sequence comparison and conservation among various group A strains. Biochem. Biophys. Rrs. Commun. 1994,195, 1-10 19. DeAngelis, P. L.; Weigel, P. H. Immunochemical confirmation of the primary structure of streptococcal hyaluronan synthase and synthesis of high molecular weight product by the recombinant enzyme. Biochemistry 1994, 33, 9033-9039. 20. Kumari, K.; Weigel, P. H. Molecular cloning, expression, and characterization of the authentic hyaluronan synthase from group C streptococcus equisimilis. J. Bid. Chern. 1997, 272, 3253932546. 21. DeAngelis, P. L.; Jing, W.; Drake, R. R.; Achyuthan, A. M . Identification and molecular cloning of a unique hyaluronan synthase from Pasteurella multocida. J. Biol. Chem. 1998, 273, 8454-8458. 22. DeAngelis, P. L.; Jing. W.; Graves, M. V.; Burbank, D. E.; Van Etten, J. L. Hyaluronan synthase of chlorella virus PBCV-1. Science 1997, 278, 1800-1803. 23. Itano, N.; Kimata, K. Expression cloning and molecular characterization of HAS protein, a eukaryotic hyaluronan synthase. J. Biol. Chem. 1996,271, 9875-9878. 24. Shyjan. A. M.; Heldin, P.; Butcher, E. C.; Yoshino, T.; Briskin, M. J. Functional cloning of the cDNA for a human hyaluronan synthase. J. Biol. Chern. 1996,271, 23395-23399. 25. Spicer, A. P.; Augustine, M. L.; McDonald, J . A. Molecular cloning and characterization of a putative mouse hyaluronan synthase. J. Biol. C/irw. 1996,271, 23400-23406. 26. Watanabe, K.; Yamagushi, Y. Molecular identification of a putative human hyaluronan synthase. J. Biol. Chem. 1996, 271, 22945-22948. 27. Spicer, A. P.; Olson, J.S.; McDonald, J. A. Molecular cloning and characterization of a cDNA encoding the third putative mammalian hyaluronan synthase. J. Biol. Chern. 1997, 272, 89578961. 28. DeAngelis, P. L.; Achyuthan, A. M. Yeast-derived recombinant DG42 protein of Xenopus can synthesize hyaluronan in vitro. J. Biol. Chem. 1996. 271, 23657-23660. 29. Spicer, A. P.; Seldin, M. F.; Olsen, A. S.; Brown, N.; Wells, D. E.; Doggett, N. A,; Itano, N.; Kimata, K.; Inazawa, J.; McDonald, J. A. Chromosomal localization of the human and mouse hyaluronan synthase genes. Genornics 1997, 41>493- 497. 30. Spicer, A. P.; McDonald, J. A. Characterization and molecular evolution of a vertebrate hyaluronan synthase gene family. J. Bid. Chcm. 1998, 273, 1923-1932. 31. Mitchell, D.; Hardingham, T. Monensin inhibits synthesis of proteoglycan but not hyaluronate in chondrocytes. Biochem. J. 1982, 202. 249-254. 32. Kleine, T.O. Hyaluronate proteoglycan complex: evidence for separate biosynthesis mechanisms of the macromolecules. Connectiw Tisx Res. 1978, 5, 195- 199. 33. von Figura, K.; Kiowski, W.; Buddecke, E. Differently labelled glucosamine- precursor pools for the biosynthesis of hyaluronate and heparan sulphate. Eur. J. Biochem. 1973, 40, 89-94. 34. Mapleson, J.L.; Buchwald, M. Effect of cycloheximide and dexamethasone phosphate on hyaluronic acid synthesis and secretion in cultured human skin fibroblasts. J . Cell Physiol. 1981, 10Y3215-222.

374

21 Biosynlhesis of Hyaluronan

35. Suzuki, M.; Asplund, T.; Yamashita, H.; Heldin, C.-H.; Heldin, P. Stimulation of hyaluronan biosynthesis by platelet-derived growth factor-BB and transforming growth factor-PI involves activation of protein kinase C. Biochem. J . 1995, 307, 817-821. 36. Prehm, P. Synthesis of hyaluronate in differentiated teratocarcinoma cells. Mechanism of chain growth. Biochem. J. 1983,211, 191-198. 37. Prehm, P. Synthesis of hyaluronate in differentiated teratocarcinoma cells. Characterization of synthase. Biochem. J. 1983, 211, 181-189. 38. Asplund, T.; Brinck, J.; Suzuki, M.; Briskin, M.J.; Heldin, P. Characterization of hyahronan synthase from a human glioma cell line. Biochim. Biophys. Actu 1998, 1380, 377-388. 39. Prehm, P. Inhibition of hyaluronate synthesis. Biochem. J. 1985,225, 699-705. 40. DeAngelis, P. Molecular Directionality of Polysaccharide Polymerization by the Pasteurella multiocida Hyaluronan Synthase. J. Biol. Chem. 1999, 274, 26557-26562. 41. Deves, R.; Boyd, C. A. Transporters for cationic amino acids in animal cells: discovery, structure, and function. Physiol. Rev. 1998, 78, 487-545. 42. Tlapak-Simmons, V.L.; Kempner, E.S.; Baggenstoss, B.A.; Weigel, P.H. The active streptococcal hyaluronan synthases (HASs) contain a single HAS monomer and multiple cardiolipin molecules. J. Biol. Chem. 1998, 273, 26100-26109. 43. Heatley. F.; Scott, J.E. A water molecule participates in the secondary structure of hyaluronan. Biochem. J. 1988,254,489-493. 44. Gosh, P.; Hutadilok, N.; Adam, N.; Lentini, A. Interactions of hyaluronan (hyaluronic acid) with phospholipids as determined by gel permeation chromatography, multi-angle laser-lightscattering photometry and ‘H-NMR spectroscopy. Intl J. Biol. Mucromol. 1994, 16, 237-244. 45. Prehm, P. Release of hyaluronate from eukaryotic cells. Biochem. J. 1990, 267, 185-189. 46. Nickel, V.; Prehm, S.; Lansing, M.; Mausolf, A.; Podbielski, A,; Deutscher, J.; Prehm, P. An ectoprotein kinase of group C streptococci binds hyaluronan and regulates capsule formation. J. Biol. Chem. 1998,273, 23668-23673. 47. Heldin, P.; Laurent, T.C.; Heldin, C-H. Effect o f growth factors on hyaluronan synthesis in cultured human fibroblasts. Biochem. J. 1989, 258, 919-922. 48. Tirone, E.; D’Alessandris, C.; Hascall, V. C.; Siracusa, G.; Salustri, A. Hyaluronan synthesis by mouse cumulus cells is regulated by interactions between follicle-stimulating hormone (or epidermal growth factor) and a soluble oocyte factor (or transforming growth factor-81). J. Biol. Chem. 1997, 272, 4787-4794. 49. Honda, A,; Noguchi, N.; Takehara, H.; Ohashi, Y.; Asuwa, N.; Mori, Y. Cooperative enhancement of hyaluronic acid synthesis by combined use of IGF-I and EGF; and inhibition by tyrosine kinase inhibitor genistein, in cultured mesothelial cells from rabbit pericardial cavity. J. Cell Sci. 1991, 98, 91-98. 50. Yung, S.; Coles, G. A,; Davies, M. IL-18, a major stimulator of hyaluronan synthesis in vitro of human peritoneal mesothelial cells: Relevance to peritonitis in CAPD. Kidney Intern. 1996, 50, 1337-1343. 51. Smith, T. J.; Wang, H.-S.; Evans, C. H. Leukoregulin is a potent inducer of hyaluronan synthesis in cultured human orbital fibroblasts. Am. J. Physiol. 1995, 268, C382ZC388. 52. Sampson; P. M.; Rochester, C. L.; Freundlich, B.; Elias, J. A. Cytokine regulation of human lung fibroblast hyaluronan (hyaluronic acid) production. J. Clin. Invest. 1992, 90, 1492-1 503 53. Sugiyama, Y.; Shimada, A.; Sayo, T.; Sakai, S.; Inoue, S. Putative hyaluronan synthase mRNA are expressed in mouse skin and TGF-P upregulates their expression in cultured human skin cells. J. Invest. Dermutol. 1998, 110, 116-121. 54. Brinck, J.; Heldin, P. Expression of Recombinant Hyaluronan Synthase (HAS) Iosforms in CHO Cells Reduces Cell Migration and Cell Surface CD44. Exp. Cell Res. 1999,252, 342-351. 55. Yamada,Y.; Itano, N.; Zako, M.; Yoshida, M.; Lenas, P.; Niimi, A,; Ueda, M.; Kimata, K. The gene structure and promoter sequence of mouse hyaluronan synthase 1 (mHAS1). Biochem. J. 1998, 330, 1223-1237. 56. Rooney, P.; Kumar, S.; Ponting, J.; Wang, M. The role of hyaluronan in tumour neovascularization. Intl J. Cancer 1995, 60, 632-636. 57. O’Reagan, M.; Martini, I.; Crescenzi. F.; De Luca, C.; Lansing, M. Molecular mechanisms and genetics of hyaluronan biosynthesis. Intl J. Biol. Mucromol. 1994, 16, 283-286. 58. Laurent, T.; Lilja, K. Unpublished observation. 59. Spicer, A. Personal communication.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

22 Biosynthesis of Chondroitin Sulfate and Dermatan Sulfate Proteoglycans Geetlza Sugumaran and Barbara M. Vertel

22.1 Introduction Proteoglycans are macromolecules that contain a protein core with one or more covalently linked glycosaminoglycan (GAG) chains. The GAG chains are linear polymers of repeating disaccharides attached to the core protein through a specific oligosaccharide linkage. They are variably substituted with sulfate, leading to the generation of a high degree of negative charge. This chapter focuses on the biosynthesis of chondroitin sulfate (CS) and dermatan sulfate (DS) proteoglycans, which are characterized by GAG chains consisting of repeating hexosamine ( N acetylgalactosamine (GalNAc)) and uronic acid disaccharides (Figure 1, Section 22.2). Proteoglycan structure, function, distribution and metabolism are not addressed here in detail (for review, the reader is referred to [ 1-81 and other Chapters in this volume). As a group, the CS and DS proteoglycans are widely distributed and structurally and functionally diverse, the common property being simply the presence of at least one CS or DS chain. In fact, the core proteins differ greatly, and the GAG chains vary widely in number, length and structural complexity. Many of the CS and DS proteoglycans are prominent in extracellular matrices (ECM), while some are a part of the plasma membrane and others are stored in intracellular granules. A range of structural and metabolic functions have been established for proteoglycans in cartilage, bone, ligaments, tendons, skin and blood vessels, and new roles continue to be discovered. For instance, aggrecan is uniquely designed to concentrate negative charges in large and abundant aggregates, and is thereby able to maintain high levels of hydration, effectively occupy a large tissue volume in the cartilage ECM, and absorb compressive forces. Decorin, a small interstitial ECM proteoglycan that bears a single DS or CS chain, is involved in the regulation of collagen fibril formation and also implicated in growth factor regulation and cell adhesion. As a third example, serglycin, which is a CS proteoglycan in mucosal mast cells but may alternatively be a heparan sulfate (HS)/heparin or mixed proteoglycan in other myeloid cells, is found in the secretory granules of immmuno-

316

22 Biosynthesis of Chondroitin Suljute and Dermatan Sulfate Proteoglycans

Chondroitin 4-sulfate /

I

OS03H

\

Chondroitin 6-sulfate

Dermatan sulfate

1,QIdoA-a -1S-GalNAc

Figure 1. Repeating disaccharide residues of chondroitin 4-sulfate, chondroitin 6-sulfate and dermatan sulfate GAG chains.

logically active cells where it functions in granule condensation and protease binding. Cell surface proteoglycans, such as the syndecans or thrombomodulin mediate cellular activities through interactions with growth factors and other bioactive molecules. Thrombomodulin is also one of the group of “part-time” proteoglycans, because it occurs both with and without its CS chain. Representative proteoglycans are described and shown in Table 1 and Figure 2. Despite the diversity in structure and function, CS and DS proteoglycans share features of their biosynthesis. The synthetic process begins with the translation of the protein core and its translocation into the lumen of the rough endoplasmic reticulum (ER). Subsequent to the activation of sugars and formation and translocation of the precursor sugar nucleotides, xylose (Xyl) is added to serine residues of the core protein, and a tetrasaccharide linkage region is completed by the sequential addition of GlcA and Gal residues to produce GlcA-Gal-Gal-Xyl-Ser (Figure 3 ) . The addition of GalNAc as the fifth sugar establishes the beginning of a bona fide CS or DS galactosaminoglycan chain. As the nascent proteoglycan is transported through the Golgi, the repeating disaccharides of the GAG chain are added and sulfated, and in the case of DS, GlcA residues are epimerized to form IdoA. Throughout the process, each monosaccharide is added individually to the growing chain. These events are shown in Figure 4. Concurrently, glycoprotein-like oligosaccharides and other types of GAGS are often attached to the same core

22. I Introduction

377

Table 1. Some CS and DS Proteoglycans.

Extracellular Matrix Hyalectans Aggrecan Versican Neurocan Brevican

SLRPs Decorin Big1ycan Epiphycan Type IX collagen Bamacan

Core protein (kDa)

GAG type ( # chains)

Location

220 265-370 136 I00

CS (100) CS (10-30) cs (3-7) CS (1-3)

Cartilage Most soft tissues Brain Brain

DS (1) DS ( 2 ) DS (2-3)

All connective tissues All connective tissues Epiphyseal cartilage

cs (1)

Cartilage, vitreous humor Basement membranes

40 40 35 68 138

CS (1)

Cell Surface Syndecan-1

31

CS/HS (1-4)

Syndecan-4 NG2

20 25 1

CS/HS (1 -4) CS (2-3)

Betaglycan (P/T) Thrombomodulin (PIT) CD44 (P/T)

110 58-60 32-38/49

CS/HS (1-2) CS (1) CS (0 4)

Epithelial cells, developing mesenchyme Ubiquitous Brain, developing mesenchyme Fibroblasts Endothelial cells Epithelial cells, lymphocytes

CS/HS (10-15)

Myeloid cells

lntracellular Serglycin

10-15

proteins. The enzymes responsible for each of these sugar transfers are organized within membranes of the secretory pathway and function in different compartments of the ER and Golgi apparatus. Thus, numerous opportunities exist for regulating this complex, multistep biosynthetic process. Since the specific structures of both the protein core and the attached GAG chains are important for function of the mature proteoglycan, the outcome of this regulated biosynthesis has significant consequences. (It should be noted that critical control mechanisms operating at the levels of transcription and translation are not considered in this chapter; the reader is referred to [7-91). Both CS and DS are synthesized as proteoglycans and generally appear in tissues in this form, but free GAG chains are sometimes found as products of proteoglycan processing or degradation. Our knowledge of proteoglycan biosynthesis has grown through studies of intact cells, cell-free soluble systems, microsomal/Golgi preparations, and more recently, transfected cells. The structural analysis of GAG chains has been facilitated by the family of GAG lyases that have specificities for hexosamine, sulfate and uronic acid, and by GAG mapping methods that are proving to be quite useful in elucidating fine structural details [lo]. Molecular cloning and genetic analysis of the biosynthetic enzymes and the proteoglycans themselves are powerful approaches

318

22 Biosynthesis of Chondroitin Suljate and Dermatan SuFfate Proteoglycans

N

4

\

decorin

serglycin

I ' syndecan

Figure 2. A few common CS and DS proteoglycans. Decorin, aggrecan, syndecan-1, and serglycin are represented diagrammatically. Thicker solid lines represent core proteins, with a box toward the C-terminus of syndecan to indicate its transmembrane domain. Thinner solid lines indicate CS or DS chains, and dashed lines are HS chains. For serglycin, the GAG chains may be CS or HS. It is not shown, but aggrecan may also have KS chains, and most proteoglycans will have N-linked oligosaccharides and may have 0-linked oligosaccharides.

that have been instrumental in the current advancement of our knowledge base, and will allow interesting manipulations in the future. It is becoming apparent that several of the enzymes involved in the formation of GAG chains are members of families of related genes. The existence of multiple isozymes for individual transferase reactions provides a biological basis for developmental and tissue-specific regulation, and a potential mechanism for generating specific GAG microstructures. Overall, proteoglycan biosynthesis is an area of active investigation with a promising future. Although many features of CS and DS biosynthesis are understood, basic questions remain unanswered. For example, why are certain GAG chain initiation sites

Figure 3. Structure of the common oligosaccharide linkage region.

22.2 Proteoylycan Structure

rER

319

Golgi

>

Figure 4. Proteoglycan biosynthesis and the secretory pathway. The rough ER (rER) and Golgi compartments of the exocytic pathway and the extracellular space are indicated. Aggrecan synthesis and processing is represented below to illustrate modifications characteristic of each processing stage. Folding begins and N-linked oligosaccharides (Y) are added in the rER, xylose addition occurs in the late ER-early Golgi, followed by completion of the linkage region in the cis-medial Golgi, and repeating disaccharides are added and sulfated in the late Golgi. The mature aggrecan is secreted into the extracellular space where it associates to form aggregates with hyaluronan and link protein. Proteoglycans such as decorin are also secreted and function extracellularly, while others such as syndecan or thrombomodulin are inserted into the cell membrane during translation and become cell surface-associated upon completion of synthesis and processing. Serglycin remains in intracellular storage granules after processing in the Golgi is completed, and is released after cell activation.

and not others selected for modification? How is the choice made between adding a CS or DS or an HS chain? How is GAG chain length regulated? and How are unique GAG microstructures synthesized? Among the CS and DS proteoglycans, the biosynthesis of aggrecan, serglycin and decorin has been studied extensively, and will be emphasized in this Chapter.

22.2 Proteoglycan Structure 22.2.1 Proteoglycans and Their Core Proteins When they were first discovered, proteoglycans were named mucopolysaccharides because the most prominent characteristic of the substance was its slimy viscosity,

380

22 Biosynthesis of Chondroitin Sulfate and Dermatan Sulfate Proteoglycans

and it was not understood that, with the exception of hyaluronan, the GAGS are covalently attached to proteins (discussed in [ 1 11). Subsequently, proteoglycan classification was based on the type of attached GAG chain, i.e., as CS, DS, HS, or keratan sulfate proteoglycans. Since molecular cloning strategies have been applied, however, commonalties in the organization of proteoglycan domains and genes have emerged that define proteoglycan families. Despite the relationships among the various proteoglycans, the subject of proteoglycan biosynthesis in the context of this volume lends itself to presentations organized according to GAG chain type, and this chapter considers specifically the CS and DS proteoglycans. Thus, the CS or DS chain modification is the common feature among these proteoglycans, and otherwise they bear little similarity to each other. In general, proteoglycans are extensively modified proteins, with the core protein representing in some cases less than 10% of the molecular mass. For this reason, the chemical purification of intact core proteins free of GAG chains proved to be difficult, and the initial characterization of core proteins was accomplished through the cell-free translation of mRNAs from proteoglycan-synthesizing tissues such as cartilage. Since that time, molecular cloning has served to more systematically define core protein structures and gene families. The core proteins range in size from ten to several hundred kDa, while the corresponding proteoglycans are 80 kDa to over 3,000 kDa. A brief introduction to some CS and DS proteoglycans and their protein cores follows, but for a more complete description of proteoglycans, core proteins and gene structure and organization, the reader is again directed to several excellent reviews [ 1-81. The two families that predominate among the CS and DS proteoglycans are found in the ECM. The modular arrangements, common structural features and genomic organization within the families suggest relationships that reflect exon shuffling and duplications during evolution. Aggrecan, the large aggregating CS proteoglycan of cartilage, is the prototypic member of one major family whose members are distinguished by the ability to form aggregates with hyaluronan, through the highly conserved N-terminal hyaluronan-binding domain ([7, 121 and references therein). Each member is also characterized by a highly conserved Cterminal selectin-like domain and a central GAG attachment domain. Other CS proteoglycans in this family include: versican, which is widely distributed and whose mRNA is subject to considerable alternative splicing; neurocan, which is developmentally regulated in neurons; and brevican, which is the smallest of the group, brain-specific, and sometimes present in a non-glycosylated form as a part-time proteoglycan. In the small leucine-rich repeat proteoglycan (SLRP) family, decorin, biglycan, and epiphycan are the three DS/CS proteoglycans ([7] and references therein). These non-aggregating proteoglycans contain one to three GAG chains attached to a core protein characterized by a central region of multiple leucine-rich repeats flanked by small cysteine clusters. Similar motifs are also characteristic of unrelated proteins (e.g., ribonuclease inhibitor). It has been suggested that the domains of the core protein mediate protein-protein and protein-membrane interactions; indeed, the ability of decorin to bind collagen fibrils appears to be a property of the core protein. A variety of other CS and DS proteoglycans have been described. The ones listed here and in Table 1 are presented to indicate the range of macromolecular types

22.2 Proteoylycan Structure

381

and are not a complete representation. Additional CS proteoglycans of the ECM include bamacan, a multidomain basement membrane proteoglycan [ 131 and type IX collagen, which is found in cartilage associated with type I1 collagen, and also found in the vitreous, where it has a CS chain roughly ten times larger than its CS chain in cartilage [14].Some CS and DS proteoglycans reside at the cell surface, and among these are syndecans -1 and -4, which are also HS proteoglycans (see Chapter 42, this volume), betaglycan, which is a proteoglycan form of the type I11 receptor for transforming growth factor-B [15] and NG2, which is expressed in neural cells and is the homologue of the human melanoma proteoglycan [ 161. Generally, these proteoglycans have a large N-terminal, GAG-substituted ectodomain, a transmembrane domain, and short C-terminal cytoplasmic domain. They are known to interact with growth factors and other biologically active molecules, and sometimes with ECM constituents, and may function in cell interactions, adhesion, migration and proliferation. Within this group, CD44 and thrombomodulin exist as part-time proteoglycans, sometimes with attached GAG chains, and sometimes not [ 17- 191. Serglycin is an important and well-characterized representative of the intracellular storage proteoglycans. first appreciated because of its multiple Ser-Gly repeats [20]. Taken together, the CS and DS proteoglycans are a heterogeneous and distinctive group of complex macromolecules.

22.2.2 What Initiates GAG Chain Addition? At present it is not entirely clear what commits a protein to become a proteoglycan. Studies with synthetic peptides have been unable to define a single consensus sequence for chain initiation, but generally, a substituted Ser is followed by Gly and the Ser-Gly pair is preceded by acidic amino acids (reviewed in [ 12, 211). Recently, the alignment of amino acid sequences of over 50 CS attachment sites from 19 different core proteins generated the consensus sequence a-a-a-a-Gly-Ser-Gly-a-b-a (with a=Glu or Asp and b=Gly,Glu or Asp), which was further tested by in vitro xylosylation reactions [22]. Despite these correlations, potential Ser-Gly sites for xylose addition commonly escape substitution. As a further complication, results from site-directed mutagenesis studies indicate that either of the two Gly residues in the Ser-Gly-Ile-Gly sequence of the decorin core protein could be replaced by Ala residues, without any appreciable decrease in G A G substitution [23]. Properties such as proximity to other substituted sites, downstream sequence, or secondary structure may also be important for xylose addition to singly and multiply substituted core proteins, and in the latter case, may facilitate a processive mechanism for G A G initiation [ 12, 211. Other sequence and structural features contribute to the specification of HS or CS chains on the common linkage region (Section 22.2.3).

22.2.3 The Linkage Region

A common tetrasaccharide sequence consisting of GlcA-Gal-Gal-Xyl links both galactosaminoglycan (CS and DS) and glucosaminoglycan (HS and heparin) chains to the Ser residues of the core proteins (Figure 3). Phosphate groups have been

382

22 Biosynthesis of Chondroitin Suyute und Dermatan Suljute Proteoglycans

identified on Xyl residues of aggrecan and decorin [24-261, and sulfation of some Gal (predominantly the second Gal) has also been reported, notably for aggrecan [27]. Interestingly, all of the Gal residues are sulfated in the carbohydrate-protein linkage region of the chondroitin 4-sulfate chain attached to urinary trypsin inhibitor and inter-a-trypsin inhibitor [28, 291. The addition of GlcA as the last residue signifies the completion of the common tetrasaccharide linkage region for both galactosaminoglycans and glucosaminoglycans. Because mutant CHO cells lacking Xyl transferase [30], Gal transferase I [31], or GlcA transferase 1 [32] fail to synthesize both types of GAG chains, it has been concluded that the same enzymes catalyze the addition of the linkage sugars that initiate the formation of both galactosaminoglycan and glucosaminoglycan chains (Section 22.3). The subsequent addition of the first hexosamine residue commits the growing GAG chain to become a galactosaminoglycan or glucosaminoglycan chain, and for this reason represents a key regulatory step in determining GAG chain type. Under some conditions, core proteins have been reported to be modified only by linkage sugars, without the continued polymerization of GAG chains (Section 22.3.3). What determines whether a galactosaminoglycan or glucosaminoglycan chain will be added to a given GAG substitution site is an unresolved question of great interest. As a rule, specific proteoglycans can be classified as CS and DS or HS and heparin proteoglycans, but variability in the GAG substitution is also documented. In the case of serglycin, for example, the same core protein is substituted with CS chains when expressed in mucosal mast cells, but is substituted with heparin in connective tissue mast cells, and may be substituted with varying proportions of heparin and CS chains in different cells and under different growth conditions [20, 331. For proteoglycans such as syndecan-1 and betaglycan, the same core protein can be substituted with both CS and HS chains. Some sequence and structural features of the core protein, such as clusters of flanking acidic amino acids, an adjacent Trp, (Ser-Gly)z units, or a hydrophobic pocket appear to preferentially direct the addition of HS rather than CS chains, but the relative proportion of HS versus CS chains is also dependent upon the specific expressing cell and growth conditions [21, 341. Although certain amino acid sequences, other distal signals and non-sequence based features favor HS addition, other factors such as interactions with the glycosyl transferases or targeting of core proteins to subcellular sites containing HS- or CS-synthesizing enzymes may influence the outcome as well.

22.2.4 CS and DS Chains

CS contains GlcA and N-acetylgalactosamine (GalNAc) residues as the repeating disaccharides, linked pl-3 and p1-4 (Figure 1). The GalNAc residues are predominantly sulfated in the 4- or 6-position, with a few non-sulfated residues. CS is heterogeneous, containing variable proportions of non-sulfated chondroitin, chondroitin 4-sulfate, and/or chondroitin 6-sulfate residues. The extent of sulfation and amounts of 4-sulfate and 6-sulfate vary significantly in different species and tissues. In general, the sulfation is almost exclusively 6-sulfate or 4-sulfate, so that CS is not ordinarily found with equal amounts of both types of sulfate. Usually a single GalNAc residue will have only one sulfate, either 4 or 6, but disulfated 4, 6 GalNAc

22.3 Biosynthesis of CS and DS Proteoglycans

383

residues are occasionally found. Additionally, GlcA may be sulfated at the 2- and 3-positions. DS is even more heterogeneous, since all DS contains the IdoA that defines this substance as DS (Figure 1) and some GlcA more common to CS. When IdoA and GlcA are present they are interspersed, and as a result, some regions are DS-like and others are CS-like within a single chain. Thus, DS is always a CS/DS hybrid. This is an interesting organization because the biosynthesis of IdoA, an epimerization process catalyzed by uronosyl epimerase, requires GlcA residues as precursors. The GalNAc of the dermatan disaccharide (IdoA-GalNAc) appears to be always 4-sulfated and rarely non-sulfated or 6-sulfated, despite 6-sulfated chondroitin disaccharide residues in the same GAG chain. In addition, DS may have IdoA 2-sulfate in fairly high amounts. At least nine different hexuronic acid-GalNAc disaccharide units have been identified, with sulfate residues located at the different positions mentioned and containing either GlcA or IdoA residues. The sulfated GAG is frequently the “business end” of the molecule and its heterogeneity and microstructure can be of paramount importance in directing function. The best characterized and most striking examples of unique GAG microstructure functions have been described for HS/heparin; the first of these selective structures to be described was for the antithrombin 111 binding sequence in heparin (see Chapter 42, Volume IV). For CS/DS chains, specific interaction with heparin cofactor I1 requires the 2-sulfated IdoA-containing disaccharide units [35]. The importance of CS/DS microstructure is also suggested for other biological interactions involving hepatocyte growth factor/scatter factor, promotion of fibroblast growth factor-2 during wound repair, and neurite outgrowth (see [36] for references). Although the basic steps involved in the biosynthesis of CS/DS have been established in some detail, there is little information concerning the control and consequences of the high degree of variability seen in the number and length of galactosaminoglycan chains, the extent and type of sulfate substitution, and the degree of epimerization of the uronic acid residues. It is likely that these regulatory features will reflect such factors as the core protein sequence, structure and abundance, and may be influenced by post-translational modifications that affect the rate of core protein transit through the Golgi complex and accessibility of the GAG-synthesizing transferases and sugar/sulfate donor substrates. A role for accessory proteins or perhaps molecular chaperones is an intriguing but essentially unexplored possibility.

22.3 Biosynthesis of CS and DS Proteoglycans 22.3.1 Biosynthesis of the Core Protein Once individual proteoglycan core proteins were identified, the investigation of early stages in proteoglycan synthesis and processing progressed. As is the case for most membrane and secreted glycoproteins, nascent core protein biosynthesis is initiated by translation in the cytosol and translocation of the nascent polypeptide into the lumen of the ER through the translocon, via a signal sequence-mediated

384

22 Biosynthesis o j Chondroitin Sulfute and Dermutan Sulfate Proteoglycans

process. After the nascent core protein is translocated to the ER lumen, N-linked oligosaccharides are added co-translationally to Asn residues from dolichol phosphate intermediates. Chaperone-mediated folding, formation of disulfide bonds and the initial trimming of the N-linked oligosaccharides presumably occur before the modified core protein exits the ER. Some of these events have been detailed specifically for the aggrecan and decorin core proteins (137, 381 and references therein), and they are probably operative for other core proteins as well. The necessity of exit from the ER for the continued progress of proteoglycan intermediates through the secretory pathway is underscored by the recessive genetic defect, nanomeliu, a lethal dwarfism with skeletal abnormalities caused by the absence of aggrecan in the cartilage ECM ([371 and references therein). As a result of a premature stop codon within the CS2 domain of aggrecan, nanomelic chondrocytes synthesize a truncated core protein that fails to become a proteoglycan because it does not arrive at the Golgi complex. Although the abnormal precursor undergoes ER-mediated processing events such as the addition of N-linked oligosaccharides and even xylosylation, pulse/chase labeling studies have shown that it exhibits a time-dependent loss which most likely involves disposal by ER-associated degradation. What minimal folding or conformational requirements of the aggrecan core protein are necessary to meet quality control criteria for exit from ER are not yet clear, but the three globular domains of normal aggrecan certainly undergo extensive folding. The problem of aggrecan folding and trafficking through the secretory pathway has been addressed in expression studies using cells transfected with selected aggrecan domains, focused on the role of globular domains and interactions with molecular chaperones. Results suggest that proteins containing G1 as the only globular domain move slowly out of the ER, while intracellular trafficking and GAG addition is expedited by the C-terminal G3 domain (1391 and references therein). This experimental approach should be helpful for the further characterization of proteoglycan synthesis, processing and trafficking. Decorin has been emphasized in biosynthetic studies of DS proteoglycans, particularly with respect to co- and early post-translational processing. One interesting feature of the decorin core protein is the presence of a 14 amino acid propeptide. While evidence has been presented for cleavage of the propeptide as an early posttranslational event in bovine articular chondrocytes [40], other studies demonstrate that decorin is secreted from fibroblasts with the propeptide intact [41]. The expression of decorin constructs with deletions in the N-terminal propeptide resulted in the synthesis and secretion of decorin with shorter than normal GAG chains [42], suggesting that the propeptide may influence interactions with the modifying glycosyltransferases or affect intracellular transport. The propeptide of biglycan, which is not removed, has also been implicated in facilitating GAG chain addition 1431. 22.3.2 Origin of Sugar and Sulfate Precursors Synthesis of the CS/DS chains proceeds by the transfer of individual sugars from uridine sugar nucleotide (UDP) precursors. The major precursor activation path-

22.3 Biosynthesis of CS and D S Pvoteoglycuns

385

way is from glucose to sugar phosphates, followed by formation and modification of UDP-linked sugars (for further information, refer to Chapters 1 and 2, this volume). The intermediate required for sulfation of chondroitin/dermatan chains is 3’phosphoadenosine 5’-phosphosulfate (PAPS), formed from sulfate and ATP via two sequential reactions. These two reactions are catalyzed by ATP sulfurylase and APS kinase, activities that reside on a single bifunctional protein in mammals [44]. The reactions involved in the activation of sugars and sulfate take place in the cytosol, catalyzed by soluble enzymes. Subsequently, the sugar nucleotides and PAPS are transported from the cytosol across the membrane into the lumen of the Golgi by means of an antiport mechanism [45]. Although transport of UDP-Xyl has been reported [46], its formation has been shown to take place in particulate fractions [47] and within the lumen of the ER or Golgi [48], where the membrane-bound UDP-GlcA decarboxylase is located. An adequate supply of specific nucleotide sugars and PAPS within the ER or Golgi lumen at the site of enzymatic addition is critical for the biosynthesis of proteoglycans. Consequently, the availability (or lack) of sufficient amounts of these molecules is likely to bear upon the regulation of GAG chain initiation, final chain length, and the type and extent of sulfation.

22.3.3 Addition of the Linkage Oligosaccharides Xylosylation The biosynthesis of the polysaccharide portion of the GAG chain begins with the addition of Xyl from its donor substrate, UDP-Xyl, to specific serine moieties in a core protein. Xyl transferase reactions have been investigated using microsomal preparations from several tissues, with cartilage serving as a prototype. Initially, immunocytochemical and subcellular fractionation studies placed the site of xylosylation between the ER to Golgi (reviewed in [49]). Based on the examination of radiolabeled semi-intact avian chondrocytes in combination with electron microscopic autoradiography and subcellular fractionation, it was determined more directly that xylosylation begins in the ER and continues in the early Golgi [48, 501. The ER-to-Golgi transport step is necessary for a core protein to progress through the secretory pathway, and it has been suggested that the core protein-Xyl transferase interaction and subsequent interaction of Xyl transferase with Gal transferase I (see below) could serve as a docking mechanism to deliver the core protein to the Gal transferase 1 in the early Golgi compartment [49]. Although the easily solubilized membrane-bound Xyl transferase has been purified from cartilage and rat chondrosarcoma, and some of its molecular and catalytic properties have been examined (reviewed in [49]),further progress has been impeded by a lack of molecular cloning and analysis. Since xylosylation of serine residues is the first step in the sequence of reactions leading to the biosynthesis of GAG chains, Xyl transfer is an important step in the transition from a protein to a proteoglycan. However, because the same Xyl transferase initiates chain synthesis for both HS/heparin and CS/DS, this modification is not the determinative event in specifying the synthesis of a particular type of G AG chain.

386

22 Biosynthesis of Chondroitin Suljate and Derrnatan Sulfate Proteoglycans

The phosphorylation of Xyl has been reported for both aggrecan and decorin. Phosphorylated Xyl is evident on aggrecan molecules even in the ECM, suggesting that the modification is stable in chondrocytes [24]. This addition has been followed as an early post-translational event [51]. In the case of decorin, phosphorylation is transient. Under normal circumstances, the phosphate group is removed upon addition of the first GlcA residue. However, treatment with brefeldin A, an inhibitor that interferes with anterograde transport from the ER to the Golgi, and competition experiments using p-D-xylosides to prime galactosaminoglycans demonstrate that the dephosphorylation step can be blocked ([38] and references therein). It will be interesting to learn more about the enzymatic machinery involved in Xyl phosphorylation and dephosphorylation, and the specific details and biological relevance of these modifications for aggrecan and decorin. It has been suggested that phosphorylated Xyl may serve as a signal during intracellular trafficking or in regulating subsequent GAG modification.

Galactosylation Extension of the GAG linkage region on the xylosylated core protein continues by the sequential addition of two Gal residues from UDP-Gal. Xylosylated core proteins provide the natural substrate, but p-D-xylosides have also proven to be useful for priming galactosaminoglycan synthesis in cultured cells (as in the example discussed above). Cell-free microsomal systems derived from cartilage were first used to demonstrate the transfer of Gal to free Xyl and P-D-xylosides. Although several properties of the two Gal transfer reactions were similar, competition experiments with mixed substrates indicated that the reactions were catalyzed by either two separate enzymes or by two independent catalytic centers on the same enzyme. Solubilization of the Gal transferases has been more difficult to achieve than for Xyl transferase, suggesting that these enzyme(s) are more firmly attached to the Golgi membranes. Interestingly, Gal transferase I but not Gal transferase I1 was coprecipitated with immune serum against Xyl transferase. Subcellular fractionation studies of chick cartilage support the existence of two Gal transferase enzymes residing in different locations of the cis or medial Golgi. Studies in rat liver using “freeze-frame” Golgi incubations and a diffusible xyloside acceptor also support the notion of two separate enzymes (reviewed in [52]). Finally, mature DS proteoglycan failed to be produced by fibroblasts from a progeroid syndrome patient with reduced Gal transferase I activity, suggesting that the second Gal transferase does not compensate for the enzymatic defect [53]. Just recently, a cDNA clone encoding human Gal transferase I that restored GAG synthesis in Gal transferase I mutants was isolated using a clone from the EST data base [54]. Evidence clearly supports two independent Gal transferases for the addition of the two Gal to the linkage region on normal core proteins. However, the observation that GAG chains are assembled on P-D-xylosides in Gal transferase I-defective CHO cells suggests that alternative pathways for these glycosyl transferase reactions may also exist [31]. (Apparently, an alternative pathway is not available in fibroblasts, since GAG chain formation failed to occur on P-D-xylosides added to the progeroid patient cells with reduced Gal transferase I activity described above

22.3 Biosyntkesis of CS and DS Proteoglycans

387

[53].)It will be of interest to learn how the organization of the enzymatic machinery within Golgi subcompartments impacts biosynthesis. In this regard, the identification of a pool of decorin intermediates with attached Gal-Gal-Xyl (similarly noted for the proteoglycan precursor in brefeldin A-treated melanoma cells) and the observation that the highest levels of phosphorylated Xyl are found on the linkage trisaccharide of decorin suggest that biosynthesis progresses in discrete stages. The inability of p-D-xylosides to prime galactosaminoglycan synthesis or inhibit decorin glycosylation and phosphorylation in brefeldin A-treated fibroblasts indicates that some biosynthetic steps may occur independently of one another and utilize separately organized sites of multi-enzyme complexes [ 381. Finally, the sulfation of linkage Gal has been reported for CS and DS proteoglycans. It appears that this modification is catalyzed by the sulfotransferases that add sulfate to the GalNAc of CS chains since the 6-sulfotransferase that sulfates GalNAc in CS also catalyzes the 6-sulfation of Gal in keratan sulfate (Section 22.3.4) and is capable of sulfating the linkage region Gal residues [55]. If this is the case, the modification would occur relatively late in the biosynthetic process. Addition of GlcA and completion of the common tetrasaccharide linkage region The first GlcA is transferred from UDP-GlcA by a GlcA transferase that appears to be distinct from the GlcA transferase involved in the formation of the repeating disaccharide units [ 521. Sucrose density gradient centrifugation studies of Golgi membrane fractions indicate that this first GlcA transfer takes place in both medial and trans Golgi regions separate from the region where the two Gal residues are added [52]. Recently, a human GlcA transferase I was cloned and expressed [56].A hamster cDNA 95Y0 identical to the human cDNA clone was also reported [57],and shown to correct mutants lacking GlcA transferase I activity that were isolated using a new selection strategy [32]. Results presented in the latter study suggest that the GlcA transferase I catalyzes additions to a common linkage region for both CS and HS biosynthesis in CHO cells. Interestingly, the hamster cDNA was also 65% identical to the GlcA transferase (GlcAT-P) involved in the GlcA modification known to be involved in generating the carbohydrate epitope HNK-1 (human Datural killer cell carbohydrate antigen1), an epitope highly abundant in glycoproteins, proteoglycans, and glycolipids of the nervous system. An evolutionary relationship between the enzymes was strongly suggested, and also supported by the restoration of GAG assembly in mutant CHO cells deficient in GlcA transferase I after transfection with either the hamster GlcA transferase I or GlcAT-P cDNAs [57]. Substrate specificities for the two recombinant enzymes exhibited significant overlap, although the GlcA transferase I was highly selective for substrates resembling the linkage region tetrasaccharide and the GlcAT-P was considerably less selective. Transfection experiments further established that both enzymes were able to catalyze GAG biosynthesis and produce HNK- 1 carbohydrate epitopes. These results are particularly intriguing in light of the expression of the GlcA transferase I in all tissues tested, and the expression of GlcAT-P only in brain and neurons.

388

22 Biosynthesis of Chondroitin Sulfate and Dermatan Sulfate Proteoglycans

In the case of the part-time proteoglycan, thrombomodulin, the a-thrombomodulin form was found to be a core protein modified by only the linkage tetrasaccharide [58]. The block to further GAG assembly at this point demonstrates that addition of the fifth hexosamine can be the critical regulatory step for continued GAG polymerization. It remains to be established if this property is unique to parttime proteoglycans. Although thrombomodulin with attached tetrasaccharide linkage region was able to promote galactosaminoglycan chain elongation, tetrasaccharide linkage region-Ser and tetrasaccharide linkage region-Ser-Gly-Gly were not, suggesting that the core protein itself may be important for continued GAG assembly [591.

Initiation of CS/DS chains by addition of the first GalNAc When GAG polymerization continues, the respective transfer of GalNAc or GlcNAc to the linkage oligosaccharide is the first step that provides specificity for CS/DS or HS/heparin GAG formation. Evidence has been presented to support the notion that the first GalNAc and the first GlcNAc are added by enzymes different from the GalNAc and GlcNAc transferases involved in polymerization [60621. However, in the thrombomodulin studies discussed above, the success of a chondroitin-synthesizing enzyme preparation in promoting GAG assembly on a thrombomodulin-linkage tetrasaccharide substrate raises the possibility that the same enzyme can catalyze the transfer of GalNAc for both the linkage region and repeating disaccharide units. These issues should be resolved as the cloning and molecular characterization of the enzymes progress. An alternative a-GalNAc transferase activity has been described that could function in a capping mechanism to block the further addition of GAG chains or in the initiation of HSlheparin rather than CS/DS chain formation. This unique aGalNAc transferase was recently purified from a human sarcoma cell line. Peptide analyis revealed 100% identity to the multiple exostoses-like gene EXTL2/EXTR2 family of tumor suppressors, and expression of a soluble recombinant form of the protein was shown to catalyze the transfer of GalNAc or GlcNAc to the serinelinkage tetrasaccharide from activated UDP precursors [63]. Although small amounts of a-GalNAc-terminated Gl~A-Gal-[~H]Gal-xyloside were identified in several different cell lines grown with p-D-xylosides and ['H]galactose 1641, the presence of a-GalNAc in CS or DS chains has never been described. It remains to be determined whether or not the a-GalNAc serves as a stop signal to preclude further chain elongation or as a shunting mechanism for HS biosynthesis, or whether it serves yet another function.

22.3.4 Formation of the CS/DS Chains Addition of the repeating disaccharides Upon completion of the linkage region and the addition of the first acetyl hexosamine, the growth of CS chains continues by the addition of alternating residues of GlcA and GalNAc from activated precursors. Formation of the repeating dis-

22.3 Biosynthesis of CS und D S Proteoylycans

389

accharide units occurs in the Golgi by alternating the transfer of sugars to the nonreducing end of the growing nascent proteoglycan primer in a highly organized fashion, with both N-acetylhexosaminyl transferase and GlcA transferase acting in concert to elongate individual chains. Polymerization on endogenous primers can result in CS chains as large as 70 kDa or more. However, exogenously added oligosaccharides serve as substrates for the addition of only one or a few sugars, and do not serve as primers for substantial chain polymerization. This significant discrepancy suggests that positioning the nascent proteoglycan in juxtaposition to the enzymes is essential for extensive polymerization. Density gradient subfractionation studies, immunocytochemical and cytochemical studies have suggested that the polymer-forming glycosyl transferases in chondrocytes are contained in both medial and truns Golgi fractions ([52] and references therein). What interactions and mechanisms hold the nascent proteoglycan in place in relation to the synthesizing glycosyl transferases and serve to regulate chain lengths are not understood. In this regard, the observation that brefeldin A-treated cells add shorter and undersulfated chains to CS/DS proteoglycans such as decorin and serglycin, and perhaps no GAG chains to other CS proteoglycans, suggests that organization within the Golgi membranes is important and may differ in specific Golgi subcompartments and in the various specialized cells [37, 65-69]. It will be of interest to determine what relationships exist between Golgi organization and chain elongation for different CS proteoglycans and in different cells. The polymer-forming enzymes involved in CS biosynthesis have been partially purified. A photoaffinity technique for labeling with [ P-”P]5N3UDP-GlcA was used to demonstrate that the GlcA transferase I1 is an 80 kDa protein [70]. Separate proteins appear to be involved in the transfer of GlcA and GalNAc residues in CS/ DS sulfate synthesis (Sugumaran unpublished), in contrast to the finding that a single protein catalyzes the transfer of both sugar residues in HS/heparin synthesis [71, 721. It remains to be established whether the same or separate polymer-forming glycosyl transferases exist for CS and DS synthesis and how the enzymes are regulated.

Epimerization of GlcA to IdoA to form DS As mentioned in Section 22.2.4, DS is defined as a CS-like GAG with IdoA as well as GlcA. Usually the IdoA content is greater than 50% of the total uronic acid, but smaller percentages are also seen. The remainder consists of GlcA-containing CS regions with small but variable amounts of nonsulfated chondroitin disaccharide residues. The IdoA is formed by the epimerization of GlcA after it has been incorporated into the GAG chain, and not from an IdoA nucleotide (reviewed in [55]).It has been shown using microsomal systems that there is little epimerization before the addition of PAPS to the incubation mixtures and experiments performed later with cultured human skin fibroblasts grown under conditions of sulfate deprivation confirmed and extended these findings. All the DS-like regions with IdoAcontaining disaccharide units were found to have 4-sulfated GalNAc adjacent to the epimerized GlcA, while all the CS-like regions with GlcA-containing disaccharides remained non-sulfated. Thus, the epimerization of GlcA to IdoA in the formation

390

22 Biosynthesis of Chondroitin Suljate and Dermatan Sulfate Proteoglycans

of DS is intimately linked to GalNAc sulfation, although the order of the relationship has not been established definitively (reviewed in [ 5 5 ] ) . The precise subcellular location for epimerization is not known, but studies with monensin, which blocks the secretory pathway in the medial Golgi region, have indicated that 6-sulfation of decorin is less affected than the epimerization and 4sulfation of decorin [73]. The experiments suggest that the 4-sulfotransferase and epimerase must be in close proximity to each other and that the 6-sulfation of decorin occurs in an earlier Golgi compartment. The epimerase has yet to be characterized and cloned. Sulfation of GalNAc Sulfation of CS and DS occurs with the direct transfer of sulfate groups from PAPS to appropriate sites on the GAGS during or after polymerization ([55] and references therein). The same microsomal preparations that are involved in GAG polymerization contain the enzymes for the addition of sulfate, and separation of the sulfotransferases from the site of polymerization results in a lower efficiency of sulfation. Partial GAG and oligosaccharide chains that are not in juxtaposition to the sulfotransferases are much less efficient as sulfate acceptors. Sulfation in vitro tends to occur in an “all or nothing” fashion, resulting in chains that rapidly become highly sulfated or remain unsulfated. Specific hexosaminyl 4- and/or 6-0sulfotransferases are involved in the transfer to each type of sulfation site. Whether the specific sulfation is a random or programmed process is not known. Sulfation appears to occur while the nascent GAG chains are actively growing rather than after the chains are completed. Like polymerization, 6-sulfation of chondroitin/ dermatan takes place in the medial and early trans Golgi regions while 4-sulfation appears to occur in a later trans Golgi region [85]. Sulfation has been suggested as one of the factors regulating GAG chain length (reviewed in [55]).Cell-free studies have shown that the presence of a 4-sulfated GalNAc at the end of a CS chain, or a GalNAc that is preterminal to a GlcA at the end of a CS chain, blocks the further addition of GalNAc or GlcA residues. Naturally occurring CS chains as well as endogenous microsomal primers terminate in GalNAc 6-sulfate, GalNAc 4-sulfate, or GlcA, but in vitro 6-sulfation of GalNAc at the non-reducing end has not been demonstrated. It is probable that in vivo the terminal GalNAc 6-sulfate modifications arise through the action of a distinct enzyme found in a variety of tissues from different vertebrate species that is capable of adding a 6-sulfate to a terminal GalNAc 4-sulfate. The resulting GalNAc 4,6disulfate may next be 4-desulfated by another enzyme, producing a terminal GalNAc 6-sulfate that can then allow further polymerization to proceed. This sequence of events may provide a salvage mechanism for prematurely terminated chains. Accordingly, aggrecan from chicken cartilage and rat chondrosarcoma has been shown to contain significant proportions of terminal GalNAc 4-sulfate and GalNAc 4,6disulfate. Whether or not such sulfated terminal GalNAc residues regulate chain lengths in vivo needs to be examined rigorously. A chondroitin 6-sulfotransferase has been purified to homogeneity from chicken embryos [74, 751. Substrate competition experiments, photoaffinity labeling [75] and

22.4 Concluding Remarks/Perspectiws

39 1

subsequent cloning and expression of this enzyme [76] have confirmed that a single enzyme catalyzes the sulfation of both the GalNAc residues of chondroitin and Gal residues of keratan. Mammalian chondroitin 6-sulfotransferases with varying degrees of homology to the avian chondroitin 6-sulfotransferase have also been cloned and expressed [77-791. Recently, a novel 6-sulfotransferase has been identified in fetal bovine serum that specifically transfers sulfate to the C6 position of GalNAc residues adjacent to IdoA residues of DS [SO]. It remains to be established if these individual 6-sulfotransferases have different substrate specificities like the HS 3-0sulfotransferase isoforms [811. With respect to 4-sulfation, a 4-sulfotransferase has been purified to homogeneity from rat chondrosarcoma [82]. This enzyme is capable of sulfating only GalNAc residues that are adjacent to GlcA residues in CS and not those next to IdoA residues in DS. This observation suggests that separate 4-sulfotransferases may be involved for CS and DS or, alternatively, 4-sulfation may ordinarily occur before epimerization to form DS. Cloning and characterization of the 4-sulfotransferase will help to resolve this issue.

Sulfation of uronic acid A portion of the IdoA in DS is usually 2-sulfated, while in contrast, the 2-sulfation of GlcA in CS is only occasional. Recently, a uronic acid 2-sulfotransferase with considerable homology to the heparin/HS epimerase was cloned and expressed [ 361. The expressed enzyme was capable of sulfating IdoA residues in DS with high efficiency and GlcA residues in chondroitin 6-sulfate with lower efficiency. Neither chondroitin nor chondroitin 4-sulfate functioned as substrates for 2-sulfation. The reaction appears to take place after the GalNAc residues are sulfated and is likely to be the last step in the biosynthesis of DS and CS. Thus, it presumably occurs in a relatively late Golgi compartment. Although 3-sulfation has not been demonstrated in mammalian CS or DS chains, novel sulfated chondroitin oligosaccharide structures containing 3-O-sulfated GlcA that are recognized by the HNK-1 monoclonal antibody have been isolated from king crab cartilage [83]. The significance of this modification is not yet established.

22.4 Concluding Remarks/Perspectives The concerted effort now underway to systemmatically clone and characterize the biosynthetic enzymes and individual proteoglycans will continue to drive the progress in understanding proteoglycan biosynthesis. The fine structure mapping of oligosaccharides and glycosaminoglycans will enhance these efforts. Cell biological studies using cells from different tissues and species should reveal similarities and differences in the organization and activities of the enzymes in the ER and Golgi subcompartments and the trafficking of the different proteoglycans. Since several of the enzymes involved in the formation of GA G chains are members of Families of related genes, and others have yet to be cloned and characterized, it will be of

392

22 Biosynthesis of Chondroitin Surfate und Dermutun Sulfute Proteoglycans

interest to determine the regulatory mechanisms operating in specific tissues and during growth and development. Gene knockouts and mutational analysis will add new dimensions to our understanding of these processes. Methodologies such as the Cre-loxP recombination system offer the possibility of targeted mutations that should provide a more selective approach to the disruption of activities [84]. These combined investigations should also lead to an understanding of how specific GAG microstructures are generated. The current interest and activity in proteoglycan biosynthesis offer promise for the future.

Acknowledgments The authors thank John Keller, Jerry Silbert and Marvin Tanzer for their helpful comments and suggestions on the manuscript, and Tung-Ling Chen for her help with the figures. Our apologies to those investigators whose work was referenced only indirectly due to space limitations. Support for the original research was provided by National Institutes of Health grants DK28433 (BMV), AR45909 (BMV), and AR41649 (GS) and VA Medical Research Services (GS). References I . Hascall VC, Heinegird DK, Wight T N (1991). In: Cell Biology of Extracellular Matrix, E.D. Hay, ed., Plenum, New York, 149-175. 2. Wight TN, Heinegird DK, Hascall VC (1991). In: Cell Biology of Extracellular Matrix, E.D. Hay, ed., Plenum, New York, 45-78. 3. KjellCn L, Lindahl U (1991). Annu Rev Biochem 60, 443-475. 4. Esko JD (1991). Curr Upin Cell Biol 3, 805-816. 5. Hardingham TE. Fosang AJ (1992). FASEB J. 6, 861-870. 6. Silbert JE, Bernfield M, Kokenyesi R (1997). In: Glycoproteins 11, J. Monreuil, J.F.G. Vliegenthart, and H. Schachter, eds., Elsevier, New York, 1-31. 7. Iozzo RV (1998). Annu Rev Biochem 67, 609-652. 8. Iozzo, ed. (2000). Proteoglycans: Structure, Biology und Moleculur Interactions. Marcel Dekker, Inc, New York. 9. Wight T N (1999). In: Comprehensive Natural Products Chemistry, B.M. Pinto, ed., Elsevier, New York, 161L177. 10. Turnbull JE, Hopwood JJ, Gallagher JT (1999). Proc Nut1 Acud Sci USA 96, 2698-2703. 11. Yanagishita M (1993). Experientiu 49, 366-368. 12. Schwartz NB, Pirok EWI, Mensch JRJ, Domowicz MS (1999). Progr Nucl Acid Rrs Mol Bid 62, 177-225. 13. Wu RR, Couchman JR (1997). J Cell Biol 136, 433-444. 14. Yada T, Suzuki S, Kobayashi K, Kobayashi M, Hoshino T, Horie K, Kimata K (1990). J Biol Chem 265, 6992-6999. 15. Lopez-Casillas F, Cheifetz S, Doody J, Andres JL, Lane WS, MassaguC J (1991). Cell 67, 785795. 16. Nishiyama A, Dahlin KJ, Prince JT, Johnstone SR, Stallcup WB (1991). J Cell Biol 114, 359371. 17. Fransson L-A (1987). Trends Biochem Sci 12, 406-411. 18. Brown TA, Bouchard T, St. John T, Wayner E, Carter WG (1991). J Cell Biol 113, 207-221. 19. Bourin M-C, Akerlund EL, Lindahl U (1990). J Biol Chem 265, 15424- 1543I . 20. Humphries DE, Stevens RL (1992). Adv Expt Med Biol313, 59-67.

References

393

Esko JD, Zhang L (1996). Curr Opin Struet Biol 6, 663-670. Brinkmann T, Weilke C, Kleesiek K (1997). J Biol Chetn 272, 11171-5. Mann DS, Yamaguchi Y, Bourdon MA, Ruoslahti E (1990). J Biol Chem 265, 5317-5323. Oegema TRJ, Kraft EL, Jourdian GW, VanValen TR (1984). J. Bid. Chetn. 259, 1720-1726. Gloss1 J, Hoppe W, Kresse H (1986). J B i d Chrm 261, 1920-1923. Shibata S. Midura RJ, Hascall VC (1992). J. Biol. Chem. 267, 6548--6555. Suguhard K, Yamashina I, De Waard P, Van Ilalbeek H, Vliegenthart JFG (1988). J. Bid. Chem. 263, 10168-10174. 28. Yamada S, Oyama M, Yuki Y, Sugahara K (1995). Eur J Biochem 233, 687-693. 29. Yamada S, Oyama M, Kinugasa H, Nakagawa T, Kawasaki T; Nagasdwa S, Khoo KH, Morris HR, Dell A, Sugahara K (1995). Glycobioloyy 5, 335-341. 30. Esko JD, Stewart TE, Taylor WH (1985). Proc Nut/ Acad Sci USA 82, 3197-3201. 31. Esko JD, Weinke JL, Taylor WH, Ekborg G, Roden L, Anantharamaiah G, Gawish A (1987). J Biol Chem 262, 12189-12915. 32. Bai X, Wei G, Sinha A, Esko J D (1999). J Biol Chem 274, 13017-13024. 33. Stevens RL, Austin K F (1989). Immunol Toduj~10, 381-386. 34. Habuchi H, Habuchi 0, Kimata K (1998). Trendy Glycosci Glycotech 10, 65-80. 35. Maimone MM, Tollefsen DM (1990). J Biol Chem 265, 18263-18271. 36. Kobayashi M, Sugumaran G, Liu J, Shworak NW, Silbert JE, Rosenberg R (1999). J Biol Chem 274, 10474-10480. 37. Vertel BM (1995). Trend.s Cell Biobogy 5 , 458-464. 38. Moses J, Oldberg A, Fransson L-A (1999). Eur J Biochem 260, 874-884. 39. Luo W, Gno C, Zheng J, Chen T-L, Wang PY, Vertel BM, Tanzer ML (2000). J Bone Mineral Metab 18, in press. 40. Sawhney RS, HeriQg TM, Sandell LJ (1991); J Biol Chem 9231-9240. 41. Moses J, Oldberg A, Cheng F, Fransson L-A (1997). Eur J Biochem 248, 521-526. 42. Oldberg A, Antonsson P, Moses J, Fransson L-A (1996). FEBS Lett 386, 29-32. 43. Hocking AM, McQuillan DJ (1996). Glyc~ohioloyy6, 71 7. 44. Lyle S, Stanczak J, Ng K, Westley J, Schwartz NB (1994). Biochemistry 33, 5920-5925. 45. Abeijon C, Mandon EC, Hirschberg CB ( I 997). Trends Biochem Sci 22, 203 -207. 46. Nuwayhid N , Glaser JH, Johnson JC. Conrad HE, Hauser SC, Hirschberg CB (1986). J Biol Chem 261, 12936-12941. 47. Silbert JE, DeLuca S (1967). Biochem Biophys Actu 141, 193-196. 48. Kearns AE, Vertel BM, Schwartz NB (1993). J Biol Chem 268, 11097-1 1104. 49. Schwartz N B (1995). Trends Glycosci Glycotechnol7, 429-445. 50. Vertel BM, Walters LM, Flay N, Kearns AE, Schwartz NB (1993). J Biol Chem 268, 1110511112. 51. Lohmander LS, Hascall VC. Yanagishita M; Kuettner KE, Kimura JH (1986). Arch Biochem Biophys 250, 21 1-227. 52. Silbert JE, Sugumaran G (1995). Biochnn Biophys Acta 1241, 371-384. 53. Quentin E, Gladen A, Roden L, Kresse H (1990). Proc Nut/ Acad Sci ( U S A ) 87, 1342--1346. 54. Okajima T, Yoshida K, Kondo T, Furukawa K (1999). J Biol Chem 274, 22915-22918. 55. Silbert JE (1996). Glycoconj J 13, 907T912. 56. Kitagawa H, Tone Y, Tamura J, Neumann KW, Ogawa T, Oka S, Kawasaki T, Sugahara K (1998). J Biol Chem 273, 6615--6618. 57. Wei G, Bai X, Sarkar A, Esko JD (1999). J Biol Chem 274, 7857-7864. 58. Nadanaka S, Kitagawa H, Sugahara K (1998). J Biol Chern 273, 33728-33734. 59. Naddnaka S, Kitagawa H, Furnitaka F, Tamura J, Neumann KW, Ogawa T, Sugahara K (1999). Biochrm J340, 353-357. 60 Rohrmann K, Niemann R , Buddecke E (1985). Eur J Biochem 148,463-469. 61 Fritz TA. Gabb MM. Wei G. Esko JD (1994). J Biol Chem 269. 28808-28814. 62. Zhang L,’ David G, Esko JD’(1995).J Biol C‘iiem 27127-27135.’ 63. Kitagawa H, Shimakawa H , Sugahara K (1999). J Biol Chem 274, 13933-13937. 64. Etchinson JR, Freeze HH (1996). Glycohioloyy 6, 177Ll89. 65. Sugumaran G, Katsman M, Silbert JE (1992). J Bid Chem 267, 8802-8806. 66. Calabro A, Hascall VC (1994). J Biol Chem 269, 22771-22778.

21. 22. 23. 24. 25. 26. 27.

394

22 Biosynthesis of Chondroitin Sulfate and Derrnatan Sulfate Proteoglycuns

67. Wong-Palms S, Plaas AH (1995). Arch Biochem Biophys 319, 383-392. 68. Uhlin-Hansen L, Kusche-Gullberg M, Berg E, Eriksson I, Kjellen L (1997). J Biol C%em272, 3200-3206. 69. Moses J, Oldberg Eklund E, Fransson L-A (1997). Eur J Biochrm 248, 767-774. 70. Sugumaran G, Katsman M, Sunthankar P, Drake RR (1997). J Biol Chem 272, 14399-14403. 71. Lind T, Lindahl U, Lidholt K (1993). J Biol Chem 268, 20705-20708. 72. Lind T, Tufaro F, McCormick C, Lindahl U, Lidholt K (1998). J Biol Chem 273, 2626526268. 73. Hoppe U, Gloss1 J, Kresse H (1985). Eur J Biochem 152, 91-97. 74. Habuchi 0, Matsui Y, Kotoya Y, Aoyama Y, Yasuda Y, Noda M (1993). J Biol Chem 268, 21968-21974. 75. Sugumaran G , Katsman M, Drake RR (1995). J Biol Chem 270, 22483-22487. 76. Fukuta M, Uchimura K, Nakashima K, Kato M, Kimata K, Shinomura T, Habuchi 0 (1995). J Biol Chem 270, 18575-18580. 77. Fukuta M, Kobayashi Y, Uchimura K, Kimata K, Habuchi 0 (1998). Biochirn Biophys Actu 1399, 57-61. 78. Uchimura K, Kadomatsu K, Fan QW, Muramatsu H, Kurosawa N, Kaname T, Yamamura K, Fukuta M, Habuchi 0, Muramatsu T (1998). Glycobiology 8, 489-96. 79. Mazany KD, Peng T, Watson CE, Tabas I, Williams KJ (1998). Biochim Biophys Actu 1407, 92-7. 80. Nadanaka S, Fujita M, Sugahara K (1999). FEBS Lett 452, 185-189. 81. Liu JA, Shworak NW, Sinay P, Schwartz JJ, Zhang LJ, Fritze MS, Rosenberg R D (1999). J Biol Chem 274, 5185-5192. 82. Yamauchi S, Hirahara Y, Usui H, Takeda Y, Hoshino M, Fukuta M, Kimura JH, Habuchi 0 (1999). J Biol Chem 274, 2456-2463. 83. Kitagawa H, Tanaka Y, Yamada S, Seno N , Haslam SM, Morris HR, Dell A, Sugahara K (1 997). Biochemistry 36, 3998-4008. 84. Marth JD (1996). J Clin Invest 97, 1999-2002.

A,

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

23 Biosynthesis of Heparin and Heparan Sulfate Proteoglycans Lena Kjellin and Ulf Lindahl

23.1 Introduction Heparin and the related polymer, heparan sulfate, are sulfate-substituted, strongly negatively charged polysaccharide chains that are synthesized in mammalian (and other) cells, covalently linked to proteins. The aim of this chapter is to outline current views on the biosynthesis of such chains. with special regard to the generation of saccharide sequences specifically designed for interactions with proteins of biological interest. The only well-defined protein-binding sequence described so far is the antithrombin-binding pentasaccharide structure that is responsible for the blood anticoagulant action of heparin and heparan sulfate. However, recent results point to the occurrence of other regions capable of more-or-less specific interactions with a variety of proteins, including enzymes, enzyme inhibitors, extracellular matrix molecules, growth factors and other cytokines. The biosynthesis of heparin and heparan sulfate (HS) is initiated by the formation of a (GlcA~l,4-GlcNAcal,4),,polymer, that is subsequently modified through partial N-deacetylation/N-sulfationof GlcNAc units, C5-epimerization of GlcA to L-iduronic acid (IdoA) residues, and 0-sulfation primarily at C2 of IdoA and C6 of GlcN units. 0-Sulfation may also occur at C2 of GlcA and C3 of GlcN units. Heparin undergoes more extensive modification than HS, and thus contains more IdoA and sulfate residues. Generally, polymer modification is incomplete, in the sense that each reaction will involve only a fraction of the potentially available substrate residues. This selectivity results in structural variability, that is typically expressed in a domain-type display of more or less modified sequences. This variability appears to be strictly regulated, and of critical importance to the generation of specific protein-binding regions. Recent studies on heparin/HS biosynthesis have been focused on the isolation and molecular cloning of the various enzymes that catalyze this process. Such enzymes include the glycosyltransferase(s) responsible for generating the initial polysaccharide chain, as well as a variety of enzymes involved in subsequent mod-

396

23 Biosynthesis of Heparin and Heparan Sulfate Proteoglycans

ifications, such as the GlcA C5-epimerase and different sulfotransferases. Recent findings indicate that several of these enzymes occur in genetically distinct isoforms, at least some of which differ with regard to substrate specificity. For reviews see [79, 801.

23.2 The Proteoglycans: Structure, Location and Functions Contrary to heparin which is synthesized exclusively by connective tissue type mast cells, HS is produced by most mammalian cell types [l].A varitety of core proteins carry the polysaccharide chains; the heparin proteoglycan of mast cells is named serglycin due to the occurrence in its core protein of a central region of repeating Ser-Gly residues [2]. The function of this proteoglycan may be to package cationic proteases and other positively charged mast cell mediators [3]. Indeed, our recent results imply a role for heparin in mast cell homeostasis, since the connective tissue type mast cells of mice devoid of heparin (due to a targeted mutation in NDST-2, see below) are reduced in number and have an abnormal morphology with a dramatically decreased number of dense granulae, normally present in these cells [81, 821. The HS proteoglycans perlecan and agrin are found in basement membranes [4, 51 while proteoglycans belonging to the syndecan and glypican families are cell surface bound [6]. Proteoglycans carrying HS chains have been implicated in diverse physiological processes including organogenesis in embryonic development, angiogenesis, regulation of blood coagulation and growth factor/cytokine action, cell adhesion and lipid metabolism [7-121.

23.3 Biosynthesis of the Polysaccharide Backbone The enzymes responsible for HS/heparin biosynthesis are located largely in the Golgi apparatus, to which the core proteins are transported following translation in the RER. Before chain elongation commences, the protein-glycosaminoglycan linkage region is synthesized. The first step involves xylosylation of selected serine residues of the core protein. Although a consensus sequence for GAG addition has been difficult to identify, substituted serines are found in Ser-Gly (and more rarely Ser-Ala) dipeptides that have one or more acidic amino acids in close proximity (see [ 131). At least in CHO-cells, the same xylosyltransferase appears to initiate both HS and chondroitin sulfate biosynthesis [ 141. A xylosyltransferase involved in chondroitin sulfate initiation was purified in the 1970s [ 151, but the enzyme has not been cloned. After xylosylation, two galactose units are added by two different galactosyltransferases [ 161. A mutant CHO-cell defective in the galactosyltransferase which transfers the first galactose unit, showed decreased chondroitin sulfate as well

23.4 Outline of Polymer-Modijkation Reactions

397

as HS production, indicating that also this enzyme participates in the biosynthesis of both GAGS [17]. After transfer of the two galactose units, glucuronyl transferase I adds a glucuronic acid (GlcA) residue from UDP-GlcA to the Xyl-Gal-Gal trisaccharide. This enzyme has recently been cloned [ 18, 191. Glucuronyl transferase I may also participate in the biosynthesis of the carbohydrate epitope HNK-1, whereas, conversely, a previously characterized glucuronyl transferase involved in HNK- 1 biosynthesis has the ability to add GlcA to the glycosaminoglycan-protein linkage region [19]. Based on the results obtained with CHO-cell mutants (see above) and the fact that the linkage regions in HS/heparin and chondroitin sulfate/ dermatan sulfate are similar, it has been assumed that the two biosynthetic pathways utilize the same enzymes to assemble these first four sugar residues. The next step, transfer of an N-acetylglucosamine (GlcNAc) residue from UDP-GlcNAc to the linkage region appears to be the committing step towards HS/heparin rather than chondroitin sulfate/dermatan sulfate biosynthesis. Obviously, this enzyme must be able to distinguish a HS core protein from a core protein designed for chondroitin sulfate attachment, conceivably through recognition of the peptide structural features previously shown to favour HS addition (i.e. a cluster of acidic residues, hydrophobic amino acids, and a close spacing of glycosylation sites 1131). The GlcNAc transferase has been partially purified [20], and was recently cloned [21]. Interestingly, this enzyme is a HexNAc transferase with the ability to transfer also an N-acetylgalactosamine (GalNAc) unit to the same acceptor. Remarkably, the GalNAc is incorporated in a-anomeric configuration, contrary to the PGalNAc residue that normally occupies this position in chondroitin or dermatan sulfate. The aGalNAc residue is tentatively regarded as a regulatory stop signal that will prevent further GAG formation. Phosphorylation or sulfation of the linkage region components are other proposed means of regulating the biosynthetic machineries. While the xylose residue of both HS and chondroitin sulfate chains may be phosphorylated 122-241, sulfation of galactose has so far been reported only for the chondroitin sulfate linkage region [24, 251. In the case of the chondroitin sulfate proteoglycan decorin, the 2-phosphorylation of xylose has been shown to be transient and suggested to be involved in intracellular transport and/or in the control of modifications of the glycan chain [26]. It remains to be seen whether the enzymes involved in formation of the linkage region, similar to most of the polymer-modifying enzymes (see below), occur in isoforms. Polymerization of the actual HS precursor chain is catalyzed by a GlcA/GlcNAc copolymerase, that appears to exist in at least two isoforms. Intriguing recent data suggest that these enzymes are identical to the putative tumor suppressors EXTl and EXT2 [27, 281.

23.4 Outline of Polymer-Modification Reactions Modification of the (GlcA-G1cNAc)n polymer appears to be initiated while the chain is still growing 1291. The initial modification step, deacetylation of GlcNAc

398

23 Biosynthesis of Heparin and Heparan Suljate Proteoglycans

residues followed by N-sulfation, is a prerequisite for the following reactions. The bifunctional GlcNAc N-deacetylaselN-sulfotransferase(NDST) enzyme catalyzes both the N-deacetylation and the subsequent N-sulfation steps (see [30-321 and references therein). The N-deacetylationlN-sulfation reaction is followed by C-5 epimerization of GlcA to IdoA, 2-0-sulfation of the hexuronic acid residues, and 6-0- and 3-0-sulfation of GlcN residues (Figure 1).

c70000010 NDST

nooomomo 1 OSTs

cooHNAc

OSO;

HNSOj

Figure 1. Scheme of polymer-modification reactions in the biosynthesis of heparin/HS. In general, about half of the GlcN residues in HS, but 90Yn or more of those in heparin, will be N-sulfated during biosynthesis. How the N-sulfation pattern of a certain polysaccharide is regulated is not known but may in part depend on the type of NDST isoform(s) involved. The GlcA CS-epimerase (EPI) has the capacity to convert a GlcA to an IdoA residue, provided that the GlcN neighbour at its nonreducing end (to the left) is N-sulfated. Due to the still poorly understood selectivity of polymer modification only some of the GlcA residues that satisfy the substrate recognition requirements of the epimerase will actually be epimerized; note that only one of the two potentially susceptible GlcA units in the scheme has actually been epimerized. Following the epimerization step, three different 0-sulfation reactions take place, starting with 2 - 0 sulfation of IdoA residues and followed by 6 - 0 sulfation and 3-0 sulfation of GlcN units. While several isoforms with different substrate specificities have been found both for the 6-0- and the 3-0-sulfotransferases (see the main text), only one mammalian 2-0-sulfotransferase has so far been identified (45). This enzyme species appears to catalyze also 2-0-sulfation of GlcA units, a less common reaction. The precise relation of this latter step to the other polymer-modification reactions is unclear.

23.4 Outline of' Polymer-Modijication Reactions

399

23.4.1 The N-DeacetylaselN-Sulfotransferases Two NDSTs, NDST-1 and NDST-2, have so far been described in some detail [333.51, and the cloning of a third isoform was recently reported 1361. It was previously speculated that NDST-2, first identified in mastocytoma tissue [37], might be responsible for heparin biosynthesis while NDST-1, purified from rat liver [38] would be committed to the biosynthesis of HS. However, the occurrence of both transcripts in a variety of tissues and cell-types seemed to indicate that both isoforms may be involved in HS production [30, 3 I , 361. The heparin-producing connectivetissue type mast cells, on the other hand, express little or no NDST-1 but abundant NDST-2 transcript [30]. Overexpression of the two NDST isoforms in 293 cells demonstrated isoform-specific HS N-sulfation patterns, such that the polysaccharide modified by NDST-2 contained longer regions of consecutive N-sulfated disaccharide units [85]. Recent studies of mice deficient in the NDST-2 isoform, generated through targeted gene disruption, indicate that this isoform is primarily committed to the biosynthesis of heparin [81, 821, despite the widespread distribution of its transcript [30]. The NDSTs are large type I1 membrane proteins with apparent molecular weights around 100 kDa [36-381. Recent results demonstrate that the Nterminal half of the protein contains the N-deacetylase active site [39], whereas the N-sulfotransferase site is located in the carboxy-terminal half of the protein [40]. 23.4.2 The CS-Epimerase The HS CS-epimerase that catalyzes the conversion of GlcA to IdoA has been cloned from bovine lung 1411. This enzyme is distinct from the epimerase involved in dermatan sulfate biosynthesis [42]. IdoA has a more flexible conformation than GlcA [43], and the formation of IdoA in GAGS is therefore believed to generally promote binding of the polysaccharides to proteins. The GlcA C5-epimerization is the only modification reaction in heparin/HS biosynthesis that cannot be reproduced without an enzyme catalyst [44]. 23.4.3 The 2-O-Sulfotransferase So far only one isoform of heparan sulfate 2-O-sulfotransferase has been cloned from mammalian cells [45]. This enzyme, as purifed from CHO cells [46], transfers sulfate to position 2 of IdoA in an -1doAal-4GlcNS03- disaccharide sequence, provided that the GlcN unit is not 6-O-sulfated. A highly similar mouse homolog of this enzyme was found to transfer sulfate also to C2 of GlcA residues [83]. Studies of a cell mutant defective in HS 2-O-sulfation afforded similar conclusion [47]. Is the substrate specificity of this enzyme sufficiently broad to account for all 2-0sulfation in heparin and HS biosynthesis? The severe phenotype of mice homozygous for a gene trap mutation in the gene encoding the 2-O-sulfotransferase, the mice dying during the neonatal period with renal agenesis and defects of the eye and

400

23 Biosynthesis of Heparin and Heparan Sulfate Proteoglycans

skeleton [ l l ] , may support this notion. On the other hand, recent results suggest that two distinct 2-O-sulfotransferases may be present in Drosophila [48]. 23.4.4 The 6-O-Sulfotransferases The sequence of a human 6-O-sulfotransferase and the partial sequence of its hamster homologue have been published [49]. The substrate specificity of this enzyme suggested the occurrence of additional 6-O-sulfotransferase isoform(s) [ 501, as required to account for the different 6-O-sulfated structures present in heparin/HS. 23.4.5 The 3 - 0 Sulfotransferases Recent extensive work by Rosenberg and coworkers has demonstrated the presence of multiple isoforms of human 3-O-sulfotransferase (3-OST) genes [ 511. Northern blot analysis of five of these variants showed some to be widely distributed, whereas others had a restricted distribution [51]. While 3-OST-1 is the key enzyme in generating anticoagulant-active HS [ 521, 3-OST-2 and 3-OST-3A/3-OST3B recognize other saccharide structures around the target GlcN residue [53]. The 3-OSTs are much smaller than the NDSTs and contain typically between 300 and 400 amino acids [51, 521. The sulfotransferase domains are homologous to those of the NDSTs [51] Interestingly, 3-OST-1, contrary to the other OSTs, appears to be a freely soluble protein lacking membrane anchoring [ 5 1, 521.

23.5 The Products, Heparin and Heparan Sulfate The basic structural features of the products of the biosynthetic process are best understood in terms of the substrate specificities of the implicated enzymes. Since the GlcA C5-epimerization and O-sulfation reactions all depend on the presence of preformed N-sulfate groups, the extensively N-sulfated heparin (with at most a few remaining N-acetyl groups) is also abundant in IdoA and O-sulfate residues ([32] and references therein). Likewise, these components are found in the N-sulfated domains of HS chains (Figure 1). However, analysis of the domain organization of HS molecules revealed further complexity. Typically, three types of domains occur in about equal proportions: N-acetylated (NA) domains (or “blocks”) composed of 2 2 consecutive N-acetylated disaccharide units, N-sulfated (NS) domains composed of 2 2 N-sulfated disaccharide units, and mixed domains of alternating N-acetylated and N-sulfated (NA/NS) disaccharide units [54]. The NA blocks lack IdoA units and O-sulfate groups, except at the immediate transition zones to adjacent, Nsulfated, structures. The NS domains contain IdoA and both 2-O- and 6-O-sulfate groups, whereas the alternating NA/NS sequences show IdoA and 6-O-sulfate residues but few, if any, 2-O-sulfate groups [54, 551. The trisulfated disaccharide unit,

23.6 Inteructions with Proteins

40 1

-I~OA(~-OS~~)-G~~NSO~(~-OSO~)-, abundant in heparin, thus is essentially restricted to the NS domains in HS. Indeed, heparin may be conceived as a special type of HS, consisting, by and large, of an extended NS structure. Detailed compositional analysis of HS domains revealed that the distribution of IdoA and 2-Osulfate residues between NS-domains of different length is remarkably similar in different HS preparations, whereas 6-O-sulfation may be highly variable [54, 841. Clearly, the overall variability in amount and distribution of the major structural components of HS may give rise to a large number of sequence permutations. Additional complexity is introduced through less common (“unique”) building blocks, such as N-unsubstituted GlcN units 1561, 2-O-sulfated GlcA (rather than IdoA), or 3-O-sulfated GlcN residues (see reviews in [8, lo]). Many biological functions of HS are believed to critically depend on the fine structure of the polysaccharide chains [8, 571. Such selective dependence is presumably reflected by the difference in composition of HS species isolated from different mammalian organs [58, 541, as well as by the remarkably distinct display of epitopes recognized by different anti-HS monoclonal antibodies in tissue sections [56, 59, 601. The most striking association of a defined cell with a specific polysaccharide structure is seen for the connective-tissue type mast cell, that is strongly committed to the generation of heparin [61].Conversely, changes in composition of HSs produced by a given tissue or cell have been documented in relation to ageing [57, 621, “differentiation” or “transformation” of cells in culture (see e.g. [63, 641 and references therein), fetal development [ 651, and certain disease conditions such as systemic amyloidosis [66]. The functional incitement to, as well as the consequences of these changes in HS structure are, with few exceptions, unknown.

23.6 Interactions with Proteins While all current information strongly suggests that the biological functions of HSs are primarily mediated through binding of polysaccharide chains to proteins, the question as to the specificity of such interaction remains unclear. Since this problem relates to the issue of HS biosynthesis and its regulation it will be considered in general terms. A large number of potential protein ligands have been identified through their ability to bind heparin. Because of this common property there has been a tendency to dismiss the interactions as “nonspecific”. On the other hand, binding selectivity expressed by a particular HS species in relation to a given protein ligand may be obscured in a heavily substituted heparin NS domain, where some sulfate residues contribute to binding whereas others are redundant [8]. The problem is underscored by the realization that physiologically important interactions may indeed be truly nonspecific. Examples of such interaction include the binding of thrombin to heparin. an essential step toward its inactivation by antithrombin (671, and the “facilitated diffusion” of growth factors across cell surfaces through interactions with HSPGs [68]. Conversely, it seems reasonable to assume that the “unique” substituents found in selected H S species represent highly specific binding

402

23 Biosynthesis of Heparin and Hepurun Sulfate Proteoglycans

sites for selected proteins. The best known example of such a site is the antithrombin-binding pentasaccharide sequence with its internal 3-0-sulfated GlcN unit. This residue, in conjunction with adjacent N-, 2-, and 6-0-sulfated monosaccharide units, provides exquisite specificity to the interactions of heparin and HS with antithrombin [67, 691. The genetically distinct 3-OSTs that catalyze 3-0-sulfation of GlcN residues in different structural surroundings [ 5 11, are presumably designed to create binding sites in HS for specific interactions with different proteins. Not only the rare substituents, but also the common IdoA 2-0- and GlcN 6-0sulfate groups appear to be incorporated in a strictly regulated fashion during HS biosynthesis, as demonstrated by the studies of HS composition and immunochemical properties cited above. These findings suggest that (more or less) specific protein-binding sites may also be expressed through defined combinations of such relatively abundant constituents. This concept is indeed supported by characterization of interactions between HSs and a variety of proteins, including enzymes [70] growth factors (and their receptors?) [71-73] and cytokines [74-761.

23.7 Regulation of HS Biosynthesis While much useful information has been obtained for many of the enzymes involved in heparin and HS biosynthesis, regarding primary structure as well as substrate specificity, additional isoforms are likely to be discovered. To further understand the process of biosynthesis, the organization of all enzyme species within the Golgi compartment, and their mode of interaction, with each other and with the polysaccharide substrates, need to be elucidated. It should be emphasized that substrate recognition by most of the enzymes generally depends on structural modifications introduced in previous reaction steps, thus providing a clue as to the organization of the enzymes in relation to each other. Altered expression of any given enzyme is therefore likely to affect subsequent modification reactions. Of particular interest, selected enzyme isoforms may preferrably associate with certain isoforms of other enzymes, to generate isoform-specific enzyme complexes. The availability of the sulfate donor PAPS [77] is yet another factor that could influence the biosynthesis, due to differences in K, of the different sulfotransferases. Finally, regulatory interactions with so far unidentified Golgi components may have either inhibitory or activating effects on a target enzyme. For example, NDST-2 depends on a cationic cofactor for expression of N-deacetylase activity [ 351. Little is known of the actual mode of regulation of HS biosynthesis in vivo. However, studies of interactions of HS with antithrombin and with certain growth factors point to finely tuned regulatory mechanisms in control of specific sulfation reactions. For instance, 3-0-sulfation of a defined GlcN residue concludes the formation of the antithrombin-binding pentasaccharide sequence, in a reaction that strictly depends on the acceptor recognition properties of the corresponding 3-OST [78]. Indeed, the proportion of antithrombin-binding, and thus blood anticoagulant, HS chains appears to be controlled by the activity of the appropriate 3-OST iso-

References

403

form (3-OST-1) [lo]. Moreover, analysis of HS from human aorta revealed a subtle but distinct increase in 6-0-sulfation with increasing subject age, and a concomitant increase in the proportions of HS chains capable of binding platelet-derived growth factor as well as acidic fibroblast growth factor (FGFl) [62, 72). By contrast, interaction with basic fibroblast growth factor (FGF2), that does not require 6-0-sulfate groups, remained essentially unchanged with age. These observations underline the need of defining the mechanisms in control of the expression levels and catalytic action of the corresponding sulfotransferases.

References 1. L. Kjellen & U. Lindahl Annu. Reo. Bioclzem. 1991 60, 443-475. 2. E. Ruoslahti Annu. Rev. Cell Biol. 1988 4, 229-255. 3 . R. Matsumoto, A. Sali, N. Ghildyal, M. Karplus & R.L. Stevens J. Biol. Chem. 1995 270, 19524-1953 1. 4. D.J. Glass & G.D.Yancopoulos Curr. Opin. Neurohiol. 1997 7, 379-384. 5. R.V. lozzo Matrix Bid. 1994 14, 203-208. 6. G. David & M. Bernfield Matrix Biol. 1998. 17, 461-463. 7. U. Lindahl, K. Lidholt, D. Spillman & L. Kjellen Thromb. Rex 1994 75, 1 --32. 8. M. Salmivirta, K. Lidholt & U. Lindahl FASEB J. 1996 10, 1270-1279. 9. K.M. Cadigan & R. Nusse Genes Dev. 1997 11, 3286&3305. 10. R.D. Rosenberg, N.W. Shworak, J. Liu, J.J. Schwartz & L. Zhang J. Clin. Incest. 1997 99, 2062-2070. 11. S.L. Bullock, J.M. Fletcher, R.S.P. Beddington & V.A. Wilson Genes & Den 1998 12. 18941905. 12. A. Woods & J.R. Couchman Trends Cell Biol. 1998 8, 189-192. 13. J.D. Esko & L. Zhang Curr. Opin. Struct. Bid. 1996 6. 663-670. 14. J.D. Esko, T.E. Stewart & W.H. Taylor Proc. Nut1 Acad. Sci. U.S.A. 1985 X2, 3197-3201. 15. N.B. Schwartz & L. Roden Carhohydr. Res. 1974 37, 167-180. 16. T. Helting J. Biol. Chem. 1971 244, 815-822. 17. J.D. Esko, J.L. Weinke, W.H. Taylor, G. Ekborg, L. Roden, G. Anantharamaiah & A. Gawish J. Bid. Chem. 1987 262, 12189-12195. 18. H. Kitagawa, Y. Tone, J-i. Tamura, K.W. Neumann, T. Ogawa, S. Oka, T. Kawasaki & K. Sugahara, K. J. Biol. Chem. 1998 273, 6615--6618. 19. G. Wei, X. Bai, A.K. Sarkar & J.D. Esko J. Biol. Chem. 1999 274. 7857. 20. T.A. Fritz, P.K. Agrawal, J.D. Esko, & N.R. Krishna Glycohiolo~jy1997 7, 587-595. 21. H. Kitagawa, H. Shimakawa & K. Sugahara J. Biol. Chem. 1999 274, 13933-13937. 22. L.-A. Fransson, I. Silverberg & I. Carlstedt J. Biol. Chem. 1985 240, 4722-4726. 23. T.R. Oegema Jr, E.L. Kraft, G.W. Jourdian & T.R. Van Valen J. Bid. Chem. 1984 259, 17201726. 24. K. Sugahara, Y. Ohi, T. Harada, P. de Wadrd & J.F. Vliegenthart J. Biol. Chem. 1992 247, 6027-6035. 25. F. Cheng, D. Heinegbrd, L-A. Fransson. M. Bayliss, J. Bielicki, J. Hopwood. & K. Yoshida J. Biol. Chenz. 1996 271, 2857-2858. 26. J. Moses, A. Oldberg, F. Cheng & L-A. Fransson Eur. J. Biochem. 1997 248, 521-526. 27. T. Lind: F. Tufaro, C. McCormick, U. Lindahl & K. Lidholt J. Biol. Chem. 1998 273, 2626526268. 28. C. McCormick, Y. Leduc, D. Martindale, K. Mattison, L.E. Esford, A.P. Dyer & F. Tufaru Nature Genet. 1998 19, 158-161. 29. K. Lidholt & U. Lindahl Biochem. J. 1992 287, 21-29. 30. M. Kusche-Gullberg, 1. Eriksson, D. Sandback Pikas & L. Kjellen J. Bid. Chem. 1998 273, 11902-11907.

404

23 Biosynthesis of Hepurin and Hepuran SuEfate Proteoglycans

L. Toma, P. Berninsone & C.B. Hirschberg J. Bid. Chem. 1998 273, 22458-22465. U. Lindahl, M. Kusche-Gullberg & L. Kjellen J. Biol. Chem. 1998 273, 24979-24982. Y. Hashimoto, A. Orellana, G. Gil: & C.B. Hirschberg J. Biol. Chem. 1992 267, 15744-15750. A. Orellana, C.B. Hirschberg, Z. Wei, S.J. Swiedler, M. Ishihara J. Biol. Chem. 1994 269, 2270-2216. 35. I. Eriksson, D. Sandback, B. Ek, U. Lindahl & L. Kjellen J. Bid. Chem. 1994 269, 1043810443. 36. J-I. Aikawa & J.D. Esko J. Bid. Chem. 1999 274,2690-2695. 37. 1. Pettersson, M. Kusche, E. Unger, H. Wlad, L. Nylund, U. Lindahl & L. Kjellen J. Bid. Chem. 1991 266, 8044-8049. 38. E. Brandan & C.B. Hirschberg J. Biol. C h e m 1988 263, 2417--2422. 39. Z. Wei & S.J. Swiedler J. Biol. Chem. 1999 274, 1966-1970. 40. P. Berninsone & C.B. Hirschberg J. Biol. Chelern. 1998 273, 25556-25559. 41. J-p. Li, A. Hagner-McWhirter, L. Kjellen, J. Palgi, M. Jalkanen & U. Lindahl J. Biol. Chem. 1997 272, 28158-28163. 42. A. Malmstrom & L. Aberg Biochem. J. 1982 201, 489-493. 43. B. Casu, M. Petitou, M. Provasoli & P. Sinay Trends Biochem. Sci. 1988 13; 221-225. 44. N. Razi, E. Feyzi, I. Bjork, A. Naggi, B. Casu & U. Lindahl Biochem. J. 1995 309,465-472. 45. M. Kobayashi, H. Habuchi, M. Yoneda, 0. Habuchi & K. Kimata J. Biol. Chem. 1997 272. 13980-1 3985. 46. M. Kobayashi, H. Habuchi, 0. Habuchi, M. Saito & K. Kimata J. Biol. Chem. 1996 271, 7645-7653. 47. X. Bai. & J.D. Esko J. Bid. Chem. 1996 271, 17711-17717. 48. J. Sen, J.S. Goltz, L. Stevens & D. Stein Cell 1998 95, 471L481. 49. H. Habuchi, M. Kobayashi & K. Kimata J. Bid. Chem 1998 273, 9208-9213. 50. H. Habuchi. 0. Habuchi & K. Kimata TrCG 1998 10, 65-80. 51. N.W. Shworak, J. Liu, L.M. Petros, L. Zhang, M. Kobayashi, N.G. Copeland, N.A. Jenkins & R.D. Rosenberg J. Biol. Chem. 1999 274, 5170-5184. 52. N.W. Shworak, J. Liu, L.M.S. Fritze, J.J. Schwartz, L. Zhang, D. Logeart &. R.D. Rosenberg J. Bid. Chem. 1997 227, 28008-28019. 53. J. Liu, N.W. Shworak, P. Sinay, J.J. Schwartz, L. Zhang, L.M. Fritze & R.D. Rosenberg J. Bid. Chem. 1999274, 5185-5192. 54. M. Maccarana, Y. Sakura, A. Tawada, K. Yoshida & U. Lindahl J. Biol. Chem. 1996 271, 17804- 17810. 55. J.E. Turnbull & J.T. Gallagher Biochem. J. 1991 273, 553-559. 56. J. van den Born, K. Gunnarsson, M.A.H. Bakker, L. Kjellen, M. Kusche Gullberg, M. Maccarana, J.H.M. Berden & U. Linddhl J. Bid. Chem. 1995 270, 31303-31309. 57. J.T. Gallagher Biochem. Soc. Truns. 1997 25, 1206-1209. 58. B. Lindahl, L. Eriksson & U. Lindahl Biochem. J. 1995 306, 177-184. 59. J. van den Born, K. J a m , K.J. Assmann, U. Lindahl & J.H. Berden J. Biol. Chem. 1996 271, 22802-22809. 60. T.H. van Kuppevelt, M.A. Dennissen, W.J. van Venrooij, R.M. Hoet & J.H. Veerkamp J. Biol. Chem. 1998 273, 12960-12966. 61. R.L. Stevens Proy. Clin. Bid. Res. 1989 297, 131-143. 62. E. Feyzi, T. Saldeen, E. Larsson, U. Lindahl, M. Salmivirta J. Bid. Chem. 1998 273, 1339513398. 63. G.C. Jayson, M. Lyon, C. Paraskeva, J.E. Turnbull, J.A. Deakin, J.T. Gallagher J. Biol. Chenz 1998 273, 51-57. 64. F. Safaiyan, U. Lindahl & M. Salmivirta, M. Eur. J. Biochem. 1998 252, 576-582. 65. Y.G. Brickman, M.D. Ford, J.T. Gallagher, V. Nurcombe, P.F. Bartlett & J.E. Turnbull J. Bid. Chem. 1998 273,4350-4359. 66. B. Lindahl & U. Lindahl J. Biol. Chem. 1997 272, 26091-26094. 67. M.C. Bourin & U. Lindahl Biochem. J. 1993 289, 313-330. 68. A.D. Lander Matrix- Biol. 1998 17, 465-472. 69. P.D. Grootenhuis, P. Westerduin, D. Meuleman, M. Petitou & C.A. van Boeckel Nnt. Stnlct. Biol. 1995 9, 736-739.

31. 32. 33. 34.

70. N. Parthasarathy: I.J. Goldberg, P. Sivaram, B. Mulloy, D.M. Flory & W.D. Wagner J. Bid. Chenl. 1994 269, 22391-22396. 71. E. Feyzi, F. Lustig, G. Fager, D. Spillmann, U . Lindahl & M. Salmivirta J. Biol. Ciicn?. 1997 272, 5518-5524. 72. J . Kreuger, K. Prydz, R.F. Pettersson, U . Lindahl & M. Salmivirta G l j ~ n h i o l o y1999 ~ ~ in press. 73. D.A. Pye, R.R. Vives, J.E. Turnbull, P. Hyde. J.T. Gallagher J. Biol. Chern. 1998 273 2293622942. 74. S.E. Stringer & J.T. Gallagher J. Biol. Chcni. 1997 272, 20508-~20514. 75. D. Spillmann, D. Witt & U. Lindahl J. Biol. C/zenz. 1998 273, 15487-15493. 76. H. Lortat-Jacob, J.E. Turnbull & J.A. Grimaud Biockern. J. 1995 310, 497-505. 77. C. Abeijon, E.C. Mandon & C.B. Hirschberg 7rends Biochem. Sci. 1997 22, 203-207. 78. M. Kusche, G. Biickstrom, J. Riesenfeld, M. Petitou, J. Choay & U. Lindahl J. Bid. Clien7. 1988 263, 15474- 15484. 79. M. Salmivirta, K. Lidholt & U. Lindahl F A S E B J. 1996 10, 1270-1279. 80. R.D. Rosenberg, N.W. Shworak, J. Liu. J.J. Schwartz & L. Zhang J. Clin. Inuest. 1997 99, 2062-2070. 81. E. Forsberg, G. Pejler, M. Ringvall, C. Lunderius. B. Tomasini-Johansson, M. Kusche-Gullberg, 1. Eriksson, J. Ledin, L. Hellman & L. Kjellen Nature 1999 400, 773-776. 82. D.E. Humphries, G.W. Wong, D.S. Friend, M.F. Gurish, W.T. QIu, C. Huang, A.H. Sharpe & R.L. Stevens Nature 1999 400, 769-772. 83. J. Rong, H. Habuchi, K. Kimata, U . Lindahl & M . Kusche-Gullberg Biochem. J. in press. 84. Safaiyan et al. Unpublished. 85. Sandback-Pikas et al. Unpublished.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

24 Biosynthesis of Proteoglycans with Keratan Sulfates Jumes L. Funderhuryh

24.1 Introduction: Keratan Sulfate Renaissance In 1939 a Gal-containing polysaccharide was identified in digests of cornea by Suzuki [ I ] . This material was independently isolated and characterized by Karl Meyer in 1953, who named it keratosulfate [2] and showed it to be a linear polymer of Gal(pl-4)-GlcNAc( pl-3) (polylactosamine) with sulfate esterified at C-6 in each monosaccharide. Due to its macromolecular size and acidic character keratan sulfate (KS) was classified as a glycosaminoglycan. Although originally identified in and named for the cornea, KS was soon found in skeletal tissues resulting in designations of KSI (corneal) and KSII (skeletal) KS. The development of monoclonal antibodies to KS in 1986 led to the observation that KS-containing molecules are widely distributed in animal tissues [3]. The sensitivity and selectivity of immunochemical identification of KS has helped contribute to a rapid increase in publications over the last decade documenting new classes of KS-containing molecules with novel tissue distribution and biological roles. The characterization of these new KS compounds has complemented ongoing structural studies of corneal and cartilage KS in expanding our understanding of these complex glycoforms. Additionally, recent work includes a revival of interest in characterization of the glycosyltransferases involved in KS biosynthesis. Several reviews have examined structure and biosynthesis of KS through 1994 [4-7]. This Chapter, therefore, will focus on the most recent developments in our knowledge of KS structure, the proteins that carry KS, and the enzymatic mechanisms for KS synthesis.

24.2 Keratan Sulfate Structure and Distribution KS structure is characterized by wide variety in chain length and degree of sulfation and also by the capping structures on the non-reducing terminals of the chains. The

408

24 Biosynthesis of Proteoylycans with Keratan Sulfates

KSI Cornea NeuAc2 Gala GalNAcR(S0 GlcNAcR(S0 NeuGc2

‘Man, 2

6

3Man-’GlcNAcR- :GlcNAcR-N-Asn

,’

NeuAc2dmLMan

’

FUC

KSll Articular Cartilage NeuAc Fuc’

‘$2alNAc-O-SeiiThi NeuA~~-~Gal3’

Figure 1. Structure of keratan sulfate. A summary of protein linkage, sulfate distribution, and capping groups is shown for (A) KSI from cornea and (B) KSII from articular cartilage KSII. Circles represent Gal(PI-4) and squares are GlcNAc(P1-3) moieties. Shading represents 6-O-sulfation and half shaded figures, partial sulfation.

original designations of KSI and KSII were based on structural differences between KS from cornea and that from cartilage. Corneal KS is N-linked to the protein core and cartilage KS is O-linked. The N- and O-linked KS types, however, are not tissue-specific in their localization. In this review KSI will refer to all N-linked KS and KSII will refer to KS linked to protein through GalNAc-O-SerlThr. A third type of KS linkage (Man-O-Ser) has been described and might, therefore, be termed KSIII.

24.2.1 Corneal K S Corneal KS is the prototype for KSI and has been most extensively characterized. KS in cornea is about 10-fold more abundant than in cartilage and 2-4 orders of magnitude more abundant than in other tissues [ 3 ] . Features of corneal KS structure are summarized in Figure 1A. Sulfate is distributed along the polymer with a characteristic pattern. Lactosamine disaccharides nearest the reducing end are unsulfated followed by a region of disaccharides sulfated only on the GlcNAc moiety. The non-reducing end of corneal KS consists of a variable domain of from 8-34 disulfated disaccharides [ 81. This domain is responsible for the heterogeneity in the charge and size characteristic of corneal KS. There is also evidence of N-sulfation in the highly sulfated domain of corneal KS 191. The non-reducing terminus of each chain is “capped”, about 70% of corneal chains terminating with neuraminic acid, the remainder with PGalNAc or aGal [lo]. KSI linkage to protein involves a complex-type biantennary oligosaccharide Nlinked to asparagine in the core protein. Sensitivity of corneal KS synthesis to the antibiotics tunicamycin and deoxynojirimycin demonstrated dolichol-linked highmannose oligosaccharides to be precursors to the linkage oligosaccharide [ 1 I]. Structural analyses of purified subfractions of porcine corneal KS generated the

24.2 Kerutun Sulfute Structure and Distribution

409

structure in Figure lA, showing KS extending only the C-6 branch of the linkage oligosaccharide with the C-3 branch terminating with a single lactosamine capped by sialic acid [8]. Subsequent studies support the idea that C-3 branches may also be extended with KS chains. Comparisons of the molecular weight of the KS released by N-glycanase to KS chain length support the existence of biantennary extension of the linkage [lo]. Even more compelling is the observation that synthesis of neither corneal KS nor KS on fibromodulin (a type I KS proteoglycan from cartilage) was eliminated by swainsonine, an inhibitor that blocks processing of the six-linked arm of the precursor high-mannose oligosaccharide [ 1 1, 121. We radiolabeled corneal KS proteoglycans either on the GlcNAc “stub” from which the KS chain had been cleaved or on sialic acids terminating the non-extended arm 1131, then separated tryptic peptides containing the labeled oligosaccharides. Some oligosaccharides were labeled for both KS and sialic attachment, indicating a single-arm extension of KS as shown in Figure 1A, but some peptides were labeled only for KS attachment, indicating likelihood of biantennary KS extension at these sites. Thus both mono- and biantennary extensions of the linkage can occur in corneal KS proteoglycans, and the location within the core protein may influence the type of extension. Heterogeneity of KS extension of the linkage region is also observed in the zona pellucida PZP3 protein. In this protein KSI was demonstrated extending only C-3, only C-6, or both arms of the linkage oligosaccharides [ 141. 24.2.2 Non-corneal KSI Fibromodulin in cartilage and osteoadherin in bone are also modified with N-linked KS chains, as is the PZP3 protein from zona pellucida [14-161. The cartilage proteoglycan aggrecan was recently shown to contain 2-3 N-linked KS chains in addition to the 20 or more O-linked K S attachment sites [ 171. KSI was also found in dermis of the Pacific mackerel [ 181. In addition to these clearly defined examples of KSI, N-linked polylactosamine is a component of numerous cell surface and extracellular glycoproteins, many of these sulfated. It seems likely that future studies will add to the list of KSI-bearing compounds. KSI chains in fibromodulin and osteoadherin appear to be relatively short (8-9 disaccharides) and more highly sulfated than corneal KS [ I 91. Fibromodulin KS lacks the clear domain structure of corneal KSI but is reduced in Gal sulfation near the reducing terminus. Capping groups on fibromodulin are more typical of those on cartilage KS (Figure lB) than of those on corneal KS [19]. These findings suggest that KS structure is dictated primarily by tissue-specific factors rather than by the linkage or the core protein. 24.2.3 KSII The structure of KSII from articular cartilage is illustrated in Figure 1B. The KS chains are typically shorter than corneal KS (5-11 disaccharides) and lack its domain structure. KSII is highly sulfated, consisting almost completely of di-sulfated

410

24 Biosynthesis of Proteoglycans with Keratan SuEfates

monomers interrupted occasionally by single lactosamine monomers, mono-sulfated on GlcNAc [5]. Linkage to the protein is via a “core 2” mucin linkage to serine or threonine. Sialylation of Gal on the linkage oligosaccharide is only partial. KSII chains are capped at the non-reducing ends by sialic acids linked to the C-3 or C-6 of the terminal GlcNAc. Alpha fucose is linked to the C-3 of sulfated GlcNAc throughout the chain excluding the terminus [20]. Interestingly, KSII from tracheal cartilage does not exhibit fucosylation, and carries C3-linked but no C6-linked sialic acids at the chain terminus [21, 221 indicating, as with KSI, a tissue-specific, not core protein-specific, determination of chain structure and modification. 24.2.4 KSIII A third chemical linkage of KS to protein has been identified in proteoglycan from brain [23]. In these molecules, the KS chains are linked directly to mannose; i.e., KS-Man-0-Ser. Several brain KS proteoglycans have been identified in the past few years. Apparently this Man linkage occurs in the KS chains modifying phosphocan. Full characterization of other neural KS structures is pending.

24.3 Keratan Sulfate Proteoglycans The primary control point in KS biosynthesis is the availability of an appropriate core protein. It has become apparent that the enzymatic machinery for KS synthesis is present in a number of cells and tissues, but KS modifies only a limited number of proteins. Understanding the genes and gene expression patterns for KS proteoglycan core protein, therefore, is essential to complete understanding of KS biosynthesis. Additionally, understanding of the primary and tertiary architecture of the core proteins can provide insight into initiation signals for KS biosynthesis. The list of KS proteoglycans has increased rapidly in the last few years. 24.3.1 SLRPs In cornea KS modifies five proteins of the Small Leucine-Rich Proteoglycan (SLRP) gene family. These are lumican, keratocan, and mimecan [24, 251. Fibromodulin in cartilage and osteoadherin in bone are also products of this gene family [15, 161. These proteins share several conserved features, primarily a series of 24-amino acid leucine rich-repeats (LRRs) that make up the central portion of each protein molecule. X-ray crystallographic and computer modeling studies suggest that the LRRs fold the protein into a series of beta-sheets forming a bow or archshaped three-dimensional structure [26]. The N-linked KS linkage sites are located in the LRR domains of these proteins and positioned on the external (convex) side of the arch. Amino acid sequencing of KS-containing tryptic peptides from fibro-

24.3 Keratan Sulfate Protcoglycans

41 1

modulin showed all four consensus N-glycosylation sites in this protein to be modified with KS chains. However, size and mass considerations suggest that not more than one site is substituted on each protein molecule [27]. A similar approach identified three consensus N-glycosylation sequences in chicken corneal lumican and keratocan as bearing KS [28]. Each of these sites is positioned immediately Nterminal to the first leucine of an LRR, suggesting that this position represents the “outside” of the folded protein. Our studies of tryptic fragments of bovine corneal lumican, keratocan, and mimecan, labeled at sites of KS attachment, suggested a quantitative difference within each protein in the frequency of utilization of individual sites for KS attachment [ 131. Additionally, this study suggested that some linkage sites support biantennary extension of the linkage region, whereas at other sites a single arm extension may occur. The KS-containing members of the SLRP family of proteins all exhibit a consensus for tyrosine sulfation (tyrosine adjacent to acidic amino acids) in the Nterminal domain of the protein. In fibromodulin these sequences do in fact contain tyrosine sulfate [ 151. Tyrosine sulfation consensus is lacking in the products of the SLRP family that bear chondroitin/dermatan sulfate chains (decorin, biglycan, and epiphycan). This feature of the protein could serve as a signal for addition of KS to these molecules. 24.3.2 Aggrecan The major proteoglycan of cartilage is aggrecan, a very large protein with a welldefined domain structure. KS is attached to aggrecan near the N-terminal region of the GAG-linkage domain, in a region of a repeated six amino acid motif. The sequence of this motif is highly conserved in different vertebrate species but the numbers of repeated units varies [29]. This variation appears to account for some of the differences in aggrecan KS content in different species. KS is also found linked both 0- and N-linked in a separate region of the aggrecan protein, the HA binding domain [17]. The HA-binding KS chains have different length and sulfation than the KS chains from the GAG-binding region of the molecule. This suggests protein conformation in this huge molecule may control access to the glycosyl- or sulfotransferase enzymes during passage through the Golgi. 24.3.3 Cell-Associated K S A number of recent reports have identified KS associated with epithelial surfaces. In adult tissues, keratinocytes, uterine endometrial cells, corneal endothelium, sebaceous gland, salivary gland, and sweat gland epithelia exhibit KS immunoreactivity [30-321. Additionally KS is expressed by epithelial-derived carcinoma cells [33, 341. Recently the endometrial protein MUCl was shown to bear KS chains [35]. This glycoprotein is a common component of the mucin layer associated with apical surfaces of secretory epithelia and could be responsible for the KS observed in many of these glandular surfaces. Another cell-surface molecule CD44 has been

412

24 Biosynthesis of Proteoylycuns with Keratun Sulfutes

shown to contain KS [36]. This protein is found in a number of alternately spliced forms and can also bear heparan sulfate. It represents the first example of an integral membrane protein to be modified with KS. A third type of cell-associated KS was identified associated with intracellular keratin molecules of keratinocytes [ 371. These examples of cell-associated KS clearly demonstrate the previously unrecognized variety in the distribution of KS glycoforms and of the cell types involved in its biosynthesis. Like chondroitin and heparan sulfates, these molecules are clearly not restricted to interstitial connective tissues but serve as post-translational modifications of a variety of proteins in many tissues.

24.3.4 Brain One of the most active areas of recent KS research has focused on proteoglycans of the central nervous system. After cornea and skeletal tissues, brain appears to exhibit the most abundant KS and is one of the tissues most rich in enzymes of KS biosynthesis. The major cartilage proteoglycan aggrecan is present in neural tissues, but KS may not modify CNS aggrecan [38]. Several proteoglycans that appear to be unique to nervous tissue have been described, including ABAKAN [39], SV2 [40], calustrin [41], and phosphocan-KS [42]. Each appears to be a unique KSlinked protein with highly specific localization produced by a limited number of cells. Several other KS-linked proteins are present in neural tissue that have yet to be fully characterized [43].

24.4 Enzymatic Reactions of K S Biosynthesis Extension of the KS chain occurs via the action of glycosyltransferases that alternately add Gal and GlcNAc to the growing polymer. Original characterization of Gal-transferase activity in corneal cells showed the enzyme to resemble the pgalactosyltransferase (PGalT) abundant in serum and milk [44]. In many tissues PGalT is considered a “housekeeping” gene and is expressed at levels independent of cell activity. In cornea, however, expression of PGalT is upregulated during development and is maintained at unusually high levels in adult corneal cells [45]. Interestingly, PGalT activity continues at high levels in corneal cells in culture that have lost KS synthesis [44]. Recently a family of five related PGalT genes has been identified [46]. Although milk PGalT seems to be the enzyme involved in corneal KS synthesis, there is no data linking this enzyme to KS biosynthesis in other tissues. KS N-acetylglucosaminyltransferase (GnT) activity has not been extensively characterized. Two recent reports have presented candidates for KS GnT. A widely distributed enzyme, rich in neural tissue, called iGnT was shown to be important in the synthesis of linear polylactosamine [47]. Another potentially relevant GnT enzyme that can generate linear polylactosamine has been identified and cloned, but

24.5 Metabolic Control

ef K S Synthrsis

413

at the current time, there is no experimental evidence linking either of these enzymes to KS biosynthesis [48]. Sulfation of KS in cornea is accomplished by at least two separate enzymes [49]. Recent studies have identified and cloned two sulfotransferase enzymes that modify K S [50, 511. One of these proteins adds sulfate to GalNAc moieties of chondroitin sulfate and to Gal moieties in KS also. The second enzyme also catalyses sulfation of Gal in KS but does not use chondroitin sulfate as an acceptor. Messenger RNA for the KS-specific sulfotransferase is particularly abundant in brain and cornea. It would therefore appear likely that this sulfotransferase represents an enzyme involved in KS biosynthesis. As with Gal sulfation, a KS-specific enzyme is probably responsible for transfer of sulfate to the GlcNAc moieties of KS. Two recent reports have described GlcNAc 6-0-sulfotrdnsferase enzymes specific for non-reducing terminal GlcNAc in lactosamine-containing oligosaccharides [52, 531. One of these enzymes is enriched in eye and brain tissues. This raises the possibility that KS GlcNAc sulfation may occur simultaneously with elongation and only on the terminus of the growing chain. The idea of coordinated elongation and sulfation of KS is supported by biosynthetic studies with cell-free corneal extracts that showed a coordinate change in the V,,, of both elongation and sulfation activities with respect to KS chain length [54]. KS biosynthesis appears to be a late event in protein processing, occurring in the trans Golgi or trans Golgi network (TGN). In chondrocytes, aggrecan is detected throughout the ER/Golgi apparatus, but KS is detected by immunohistology only in the trans Golgi and TGN [55], suggesting these compartments as site of its biosynthesis. Brefeldin A, an inhibitor of transport through the Golgi, eliminates KS substitution on aggrecan and fibromodulin but does not block secretion of decorin modified with chondroitin chains [56]. KS addition may therefore be one of the final posttranslational modifications proteoglycans undergo before secretion.

24.5 Metabolic Control of K S Synthesis Recent KS literature contains numerous reports of alterations in KS resulting from metabolic, pathologic, or developmental changes in tissues. KS expression is clearly under close control, but neither the mechanism nor the function of the alterations in KS expression is well understood. Two generalized patterns of KS alteration appear repeatedly in these reports. The first is a consistent developmental alteration in KS. Typically, KS concentrations in organisms appear to increase with developmental age. This is best documented in the cornea. In chicken, sulfated epitopes of KS are not detected until approximately 12 hours after invasion of this cell population into the primary stroma [57]. After that time, KS accumulates rapidly in the stroma, progressing in a posterior to anterior gradient during embryogenesis and continuing after hatching. In the mouse, corneal lumican is found primarily as a non-sulfated glycoprotein at birth, and only begins to be modified with KS about the time of eye

4 14

24 Biosynthesis of Proteoglycuns with Keratan Sulfates

opening at 10-14 days, increasing in charge and size for at least one year [58]. Analysis of human corneas suggests a similar increase in sulfated KS in the stroma throughout life [59]. In cartilage, the KS content of aggrecan also undergoes an agerelated increase in abundance, associated with increased KS chain length and sulfation [60]. Rat brain, as well, shows little embryonic KS, developing the majority of KS activity after birth [61]. This pattern of age-dependent accumulation seems to be primarily a feature of KS associated with extracellular connective tissues. Cellassociated KS and the cell type specific KS of neural tissues display individual patterns of KS expression unrelated to this trend. Another widely observed property of KS biosynthesis is its loss in vitro. Cell types that secrete the most abundant KS in vivo (chondrocytes and keratocytes), when placed in tissue culture typically revert to fibroblastic cells that secrete little KS. Human corneal fibroblasts in culture appear to lose expression of the corneal proteins modified by KS [62]. However, cultured bovine cultured keratocytes do show expression of all three KSI-linked proteins modified with truncated KS chains of low sulfation [63]. These experiments suggest that downregulation of KS biosynthesis in vitro relates to activities of KS-specific glycosyl- and sulfotransferases. This idea is supported by a recent study showing chicken keratocytes to be markedly reduced in their ability to transfer sulfate to GlcNAc residues in KS after short-term culture [64]. Specific enzymes required for polymerization and sulfation of KS may, therefore, be key regulators of KS biosynthesis, both in the in vitro context and possibly in the highly specific tissue localization exhibited by this molecule in vivo. Development of chondrocyte and recently of keratocyte cultures that secrete full-length, sulfated KS for extended periods in vitro provides important tools for future studies to identify the KS-specific biosynthetic enzymes [65, 661. Identification of these enzymes represents the next essential stage in our understanding the factors regulating KS biosynthesis in the diverse situations in which this interesting molecule is found.

Acknowledgments

The author thanks Martha Funderburgh for critical reading of the manuscript. This work was supported by PHS Grants EY09368 and EY00952 (to Gary W. Conrad). References 1. 2. 3. 4. 5.

M. Suzuki, J Biochem, 1939,30, 185-191. K. Meyer, A. Linker, E.A. Davidson and B. Weissman, J Biol Chem, 1953,205, 61 1-616. J.L. Funderburgh, B. Caterson and G.W. Conrad, J Biol Chem, 1987,262, 11634-40. V.C. Hascall, Prog Clin Biol Res, 1982, 110 Pt B, 3-15. I.A. Nieduszynski, T.N. Huckerby, J.M. Dickenson, G.M. Brown, G.H. Tai and M.T. Bayliss, Biochem SOCTrans, 1990, 18, 792-3. 6. J.L. Funderburgh, M.L. Funderburgh, M.M. Mann and G.W. Conrad, Biochem Soc Trans, 1991, 19, 871-876. 7. H. Greiling, Exs, 1994, 70, 101-22.

Rejerences

4 15

8. M. Oeben, R. Keller, H.W. Stuhlsatz and H. Greiling, Biochem J , 1987, 248, 85-93. 9. P.W. Tang, P. Scudder, H. Mehmet, E.F. Hounsell and T. Feizi. Eur J Biochem, 1986, 160, 537-45. 10. G.H. Tai, T.N. Huckerby and I.A. Nieduszynski, J B i d Chem, 1996, 271, 23535-46. 1 1. C. Ziegler and G. Mersmann, Biochim Biophys Acfu, 1984; 799, 203-8. 12. A.H. Plaas and S. Wong-Palms, J Biol Clzenz, 1993, 268, 26634-44. 13. J.L. Funderburgh, M.L. Funderburgh, M.M. Mann and G.W. Conrad, J Biol Chem, 1991, 266, 14226-31. 14. S. Noguchi and M. Nakano, Eur J Biochern, 1992,209, 883-94. 15. P. Antonsson, D. Heinegard and A. Oldberg, J Biol C h m , 1991, 266, 16859-61. 16. Y. Sommarin, M. Wendel, Z. Shen, U. Hellman and D. Heinegard, J Biol Chem, 1998, 273, 16723-9. 17. F.P. Barry, L.C. Rosenberg, J.U. Caw. T.J. Koob and P.J. Neame, J Biol Chem, 1995, 270, 20516-24. 18. M. Ito, M. Kitamikado and T. Yamagata, Biochini Biophys Actu, 1984, 797, 221-30. 19. R.M. Lauder. T.N. Huckerby and LA. Nieduszynski, Glycoconj J , 1997, 14, 651-60. 20. G.M. Brown, T.N. Huckerby, B.L. Abram and LA. Nieduszynski, Biochem J , 1996,319, 13741. 21. LA. Nieduszynski, T.N. Huckerby, J.M. Dickenson, G.M. Brown, G.H. Tai, H.G. Morris and S. Eady, Biochem J , 1990, 271, 243-5. 22. J.M. Dickenson, T.N. Huckerby and LA. Nieduszynski, Bioclzem J , 1991,278, 779-85. 23. T. Krusius, J. Finne, R.K. Margolis and R.U. Margolis. J Biol Chetn, 1986, 261, 823742. 24. L.M. Corpuz, J.L. Funderburgh, M.L. Funderburgh, G.W. Bottomley, S. Prakash and G.W. Conrad, J Biol Chem, 1996,271, 9759-9753. 25. J.L. Funderburgh, L.M. Corpuz, M.R. Roth, M.L. Funderburgh, E.S. Tasheva and G.W. Conrad, J Biol Chrm, 1997, 272, 2809% 95. 26. R.V. Iozzo, Crit Rev Biochem Mol B i d , 1997, 32, 141-74. 27. A.H. Plaas, P.J. Neanie, C.M. Nivens and L. Reiss, J Biol Chem, 1990,265, 20634-40. 28. J.R. Dunlevy, P.J. Neame, J.P. Vergnes and J.R. Hassell, J Biol Chem, 1998, 273, 961 5 -2 1. 29. C.R. Flannery, C.B. Little and B. Caterson, Matrix B i d , 1998, 16, 507-1 I . 30. T. Shiozawa, Y. Tsukahara, J. Nakayama, K . Ishii and T. Katsuyama, Gynecol Ohstet Inuest, 1991, 32, 239-42. 31. J.M. Sorrell and B. Caterson, Histochemistry, 1990, 94, 269-75. 32. N.J. Fullwood, Y. Davies, LA. Nieduszynski, B. Marcyniuk, A.E. Ridgway and A.J. Quantock, Inoest Ophthulnd Vis Sci, 1996, 37, 1256-70. 33. N . Ito, M. Yokota, C. Nagaike, Y. Morimura, K. Hatake, 0. Tanakd and T. Matsunaga, Histochem J , 1996, 28, 613-23. 34. S. Cooper, M.F. Pera, W. Bennett and J.T. Finch, Biochem J , 1992, 286, 959-66. 35. J.D. Aplin, N.A. Hey and R.A. Graham, Glycohioloqy, 1998, 8, 269-76. 36. K. Takahashi, I. Stamenkovic, M. Cutler, A. Dasgupta and K . K . Tanabe, J B i d Cheni, 1996. 271, 9490-6. 37. LA. Schafer and J.M. Sorrell, E.xp Cell Res, 1993, 207, 213-9. 38. M. Domowicz, H. Li, A. Hennig, J. Henry, B.M. Vertel and N.B. Schwartz, Deu Biol, 1995, 171, 655-64. 39. H. Seo and E.E. Geisert, Jr., Neurosci Lett, 1995, 197, 13-6. 40. T.W. Scranton, M. Iwata and S.S. Carlson, J Neurochem, 1993; 61; 29-44. 41. M.A. Burg and G.J. Cole, J Neurohiol, 1994, 25, 1-22. 42. R.K. Margolis, U. Rauch, P. Maurel and R.U. Margolis, Perspect Deu Neurohiol, 1996, 3, 273-90. 43. B. Miller, A.M. Sheppard and A.L. Pearlman, J Comp Neurol, 1997, 380, 533-52. 44. J.E. Christner, J.J. Distler and G.W. Jourdian. Arcl7 Biochem Biophys, 1979, 192, 548-58. 45. C.X. Cai, E. Gibney, M.K. Gordon, J.K. Marchant, D.E. Birk and T.F. Linsenmayer, E - Y ~ Eye Res, 1996, 63, 193-200. 46. R. Almeidd, M. Amado, L. David, S.B. Levery, E.H. Holmes, G . Merkx, A.G. van Kessel; E. Rygaard, H. Hassan, E. Bennett and H. Clausen, J Biol Chem, 1997,272, 31979-91.

41 6

24 Biosynthesis of Proteoglycans with Keratan Sulfates

47. K. Sasaki, K. Kurata-Miura, M. Ujita, K. Angata, S. Nakagawa, S. Sekine, T. Nishi and M. Fukuda, Proc Nut1 Acud Sci U S A , 1997, 94, 14294-9. 48. D. Zhou, A. Dinter, R.G. Gallego, J.P. Kamerling, J.F.G. Vliegenthart, E.G. Berger and T. Hennet, Proc Nut1 Acud Sci U S A , 1999, 96, 406-1 1. 49. E.R. Ruter and H. Kresse, J Biol Chern, 1984, 259, 11771-6. 50. M. Fukuta, J. Inazawa, T. Torii, K. Tsuzuki, E. Shimada and 0. Habuchi, J Bid Chem, 1997, 272, 32321-8. 51. 0. Habuchi, Y. Hirahara, K. Uchimura and M. Fukutd, Glycobiology, 1996, 6, 51-7. 52. K. Uchimura, H. Muramatsu, K. Kadomatsu, Q.W. Fan, N. Kurosawa, C. Mitsuoka, R. Kannagi, 0. Habuchi and T. Muramatsu, J Biol Chem, 1998,273,22577-83. 53. S. Degroote, J.M. Lo-Guidice, G. Strecker, M.P. Ducourouble, P. Roussel and G. Lamblin, J Biol Chem, 1997,272, 29493-501. 54. R. Keller, R. Driesch, T. Stein, M. Momburg, H.W. Stuhlsatz, H. Greiling and H. Franke, Hoppe Seylers Z Physiol Chem, 1983, 364, 239-52. 55. B.E. Vertel, Trends Cell Biol, 1995, 5, 458-464. 56. S. Wong-Palms and A.H. Plaas, Arch Biochem Biophys, 1995, 319. 383-92. 57. J.L. Funderburgh, B. Caterson and G.W. Conrad, Dev Bid, 1986, 116, 267-17. 58. S. Ying, A. Shiraishi, C.W. Kao, R.L. Converse, J.L. Funderburgh, J. Swiergiel, M.R. Roth, G.W. Conrad and W.W. Kao, J Biol Chem, 1997, 272, 30306- 13. 59. R. Praus and I. Brettschneider, Ophthal Res, 1975, 7, 452-458. 60. G.M. Brown, T.N. Huckerby, M.T. Bayliss and I.A. Nieduszynski, J Biol Chem, 1998, 273, 26408- 14. 61. B. Meyer-Puttlitz, P. Milev, E. Junker, I. Zimmer, R.U. Margolis and R.K. Margolis, J Neurochern, 1995,65, 2327-37. 62. J.R. Hassell, P.K. Schrecengost, J.A. Rada, N. SundarRaj, G. Sossi and R.A. Thoft, Invest Ophthulmol Vis Sci, 1992, 33, 547-57. 63. J.L. Funderburgh, M.L. Funderburgh, M.M. Mann, S. Prakash and G.W. Conrad, J Bid Chem, 1996,271, 31431-6. 64. K. Nakazawa, I. Takahashi and Y. Yamamoto. Arch Biochem Biophys, 1998,359, 269-82. 65. A.H. Plaas, A.L. Ison and J. Ackland, J Biol Chern, 1989,264, 14447-54. 66. M.P. Beales, J.L. Funderburgh, J.V. Jester and J.R. Hassell, Invest Ophthulrnol Vis Sci,1999, 40, (in press).

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

25 The Biosynthesis of GPI Anchors Y a m S. Morita, Alvaro Acosta-Serrano, and Paul T. Enylund

25.1 Introduction Glycosyl phosphatidylinositols, or GPIs, are a family of molecules that contain the -phospholipid, and they are structural motif Manal-4GlcNal-6myo-inositol-1 found in all eukaryotes examined to date. The first GPIs discovered were phosphatidylinositol-based. Thus they were named glycosyl phosphatidylinositols, with the abbreviation GPI. This nomenclature is no longer strictly correct due to the discovery of inositol-phospho-ceramide-based GPIs. GPIs are usually covalently linked to proteins, functioning to attach plasma membrane proteins to the cell surface. However, not all GPIs are protein anchors. Protozoan parasites use GPIs to anchor carbohydrate chains on the cell surface, and they also express a variety of abundant free GPIs (up to lo7 copies/cell) that do not anchor proteins or polysaccharides [ l , 21. Free GPIs, of unknown function, are also found in animal cells [3-51. In this review we will focus on protein-anchoring GPIs, first discussing their structural variations and then their biosynthesis. Recent reviews on GPIs are [6-111.

25.2 Structure of GPI Anchors The core structure of protein-anchoring GPIs is EtN-P-6Manal-2Manal-6Manal4GlcNal-6myo-inositol-l-phospholipid, with the ethanolamine amino group linked to the protein’s C-terminal a-carboxyl group [12]. However, anchors have a wide variety of modifications that we will discuss in this section. 25.2.1 Glycan Core Modifications The simplest known GPI anchor, in the gp63 protein of the protozoan parasite Leishmania major (Figure 1A) [ 131: has no modification to the linear core structure.

L. major

E;N

4‘H,

I

EtN Ethanolamine

Sialic Acid

GalNAc

PGal

@ aGal

GlcNAc

GlcN

aMan

T. brucei

C18:O-AryI Cltl:O-Alkyl

C16:O

Figure 1. Examples of some GPI anchor structures from protozoan parasites (A-C), yeast (D) and mammals (E). The number of Gal residues in the T. hrucei VSG anchor varies depending on the VSG variant [ 171. The number of N-acetyllactosamines and sialic acids in the anchor of PARP (C) is variable, but the average composition of the side chain is: Sialic Acids-(Gal-GlcNAc)9 [16]. Branch points in the PARP GPI side chain are hypothetical. In the yeast anchor (D), only one of the two terminal Man residues is present. Many S. cereuisiae proteins contain a ceramide-based GPI (D) with the exception of some proteins such as Gaslp which contain a lyso- or diacylglycerol [15].

A)

25.3 CPI Precursor Synthesis

4 19

Other anchors, in contrast, contain cores that are modified with various substituents such as extra EtN-Ps or sugars, either mono- or oligosaccharides (Figure 1B-E). The additional EtN-Ps, found in mammalian and yeast anchors, do not appear to be used for linkage to proteins. One of the most common sugar side chains, especially in mammals, is a GalNAc linked (31-4 to the first Man (Figure 1E) [14]. Another frequent modification is an additional Man linked al-2 to the third Man (Figures 1D and 1E) [14, 151. The oligosaccharide side chain can sometimes dwarf in size the anchor core, and in contrast to the defined structure of the core, the side chain can be highly heterogeneous. One side chain of this type is found in PARP, the major surface glycoprotein of the insect stage of the protozoan parasite Trypanosoma brucei (Figure 1C) [16].A somewhat less complicated side chain is found in the VSG anchor, from the mammalian bloodstream life cycle stage of the same parasite. The extent of galactosylation of the trimannosyl core varies depending on the type of VSG (Figure 1B) [17]. Although the functions of GPI modifications are in general unknown, the Gal residues in the T. brucei VSG anchor have been proposed to fill the space between the densely packed VSG molecules [ 181.

25.2.2 Variations in Anchor Lipid Structure Most GPI anchors are glycerolipids, containing either a diacylglycerol or a l-alkyl2-acyl glycerol. The acyl and alkyl groups vary in chain length, degree of saturation, and level of heterogeneity, depending on the protein or cell type. However, some GPI anchors contain homogeneous fatty acids. The GPIs on human CD52-I and porcine membrane dipeptidase contain two stearoyl residues linked to glycerol [ 19, 201. The T. hrucei VSG anchor contains dimyristoyl glycerol (Figure 1B) [21]. In contrast, the T. brucei PARP GPI contains a lyso-glycerolipid (Figure 1C) [ 161. A less common variation, found in some proteins in Dictyostelium discoideum, yeast, and T. cruzi, is an anchor that contains ceramide instead of a glycerolipid (Figure 1D) [15, 22, 231. Some anchors contain an additional fatty acid that is ester linked to the 2hydroxyl group of the myo-inositol. Examples are human erythrocyte acetylcholinesterase [24], T. brucei PARP [16, 251 (Figure lC), and a population of the human CD52 antigen [ 191. In some inositol-acylated proteins this acyl chain is a palmitate whereas in some others (e.g. PARP) mass spectrometry analysis has showed that it is quite heterogeneous, containing a mixture of fatty acids (Figure 1C) [16].

25.3 GPI Precursor Synthesis GPI biosynthesis occurs in the ER. It is a stepwise process in which sugars and an EtN-P are sequentially added to PI, resulting in a complete GPI precursor which can be linked to a protein. In this section we will review pathways of GPI biosynthesis mainly in the three major experimental systems, T. brucei, human, and yeast

420

25 The Biosynthesis of GPI Anchors 0 Man

1

1 Man

ii------ir-r

2Man

3 Man

3Men

H7

+ Ethanolamine

H8

HUMAN

GlcN G A C GlcNAc

lnosiiol AcyUAlk

Figure 2. Biosynthetic pathways for GPI precursors in T. brucei, human, and yeast. Note that some intermediates in each species have an acylated inositol. In T brucei, early mannosylated intermediates as well as glycolipid A are in equilibrium with inositol-acylated forms (not shown) [53]. The insect form of T. brucei undergoes neither inositol deacylation nor the last three reactions of the fatty acid remodeling, resulting in a lyso GPI precursor containing an acylated inositol. This lyso GPI is also found in PARP (Figure 1C).

(see Figure 2). These pathways have been elucidated by two major experimental approaches. Cell free systems using T brucei crude cell membranes and substrates such as sugar nucleotides first revealed the intermediates and the steps involved in the pathway [26, 271. This approach has subsequently been used successfully with many other cell types. Another powerful approach has been to characterize mutants in mammalian and yeast cells that are defective at different steps in the GPI biosynthetic pathway. The availability of mutants has then allowed cloning of genes that correct the defects. A striking success of this approach was the cloning of human PIG-A gene [28], the product of which is involved in the first step in the pathway. Identification of PIG-A was important because a mutation in this gene appears to be the primary cause of an acquired clonal blood disease, paroxysmal nocturnal hemoglobinuria (291.

25.3.1 GlcNAc-PI Synthesis The committed step of GPI biosynthesis is the transfer of GlcNAc from UDPGlcNAc to PI (Figure 2) [30]. In mammalian cells the synthesis of GlcNAc-PI is mediated by an enzyme complex containing at least four polypeptides, encoded by the genes PIG-A, PIG-H, PIG-C, and GPIl (311. PIG-A is an ER transmembrane protein with a large cytoplasmic N-terminal domain (321. It is probably the catalytic subunit because of its homology to a bacterial GlcNAc transferase [33, 341 and because its yeast homolog can be specifically photolabeled with a uridine nucleotide derivative designed to mimic UDP-GlcNAc ( 1171. First cloned in yeast, PIG-C [35]

25.3 GPI Precursor Synthesis

421

and GPI1 [36] are both ER membrane proteins with several putative transmembrane segments. PIG-H associates with the cytoplasmic leaflet of the ER membrane [32], without having any obvious transmembrane domains [37]. The functions of PIG-C, PIG-H, and GPI 1 are unknown. Interestingly, there appears to be no PZGH gene homolog in the yeast genome database. It is unclear why at least four subunits are needed for the GlcNAc transferase step. One possibility is that the multiprotein structure is required for regulation of the pathway, as control of this initial step could alter flux through the pathway. Although regulation of GPI biosynthesis has not been studied in detail, the rate of mammalian GPI biosynthesis does appear to be enhanced by treatment of cells with phytohemagglutinin [38]. In this regard, it will be interesting to evaluate the structure of the GlcNAc transferase in T. brucei. This protozoan parasite makes massive amounts of GPIs, apparently even more than it needs to anchor its lo7 VSG molecules, raising the possibility that GPI biosynthesis is constitutively active and possibly unregulated [39, 401. Another possibility is that the multiprotein structure of GlcNAc transferase enables selection of PI substrates with specific fatty acids. Indeed, in vitro studies on the mammalian enzyme complex showed that it can distinguish PIS of different fatty acid composition [ 3I]. 25.3.2 GlcNAc-PI Deacetylation

The second step of GPI biosynthesis is de-N-acetylation of GlcNAc-PI (Figure 2), a reaction that occurs before either mannosylation [41] or inositol acylation. Information about the de-N-acetylase has been obtained from biochemical (the enzyme has been partially purified from T. brucei [42]) and from genetic studies. The active site of de-N-acetylase is in PIG-L, an ER membrane protein with most of its polypeptide facing the cytosol [43, 441. GTP stimulates GlcNAc-PI de-N-acetylation in crude mammalian microsomes [45], but it has no apparent effect on isolated PIG-L [44],suggesting that other factors may be involved in this reaction. A comparison of the substrate specificity of the de-N-acetylase in T. brucei, L. mujor, and human cellfree systems using synthetic N-acyl GlcN-PIS with various chain length N-acyl groups suggested that the substrate specificity of this enzyme is conserved throughout evolution [41, 461. PIG-L does not associate with the GlcNAc transferase complex under conditions in which the four subunits of GlcNAc transferase can be co-immunoprecipitated [31], suggesting that there may not be an interaction in vivo between the GlcNAc transferase and the de-N-acetylase. Furthermore, the two enzymes apparently do not even co-localize in vivo. In mammalian cells the de-N-acetylase activity, unlike that of the GlcNAc transferase, appears to be concentrated in a sub-compartment of the ER that can be co-isolated with mitochondria [118]. 25.3.3 Inositol Acylation

Inositol acylation plays an essential role in the biosynthesis of GPIs in mammals, yeast, and T. hrucei (Figure 2). In yeast and mammals (but not T. brucei), inositol

422

25 The Biosynthesis of GPI Anchors

acylation is a prerequisite for GPI mannosylation (Figure 2) [47-501. In yeast, inositol acylation is acyl-CoA-dependent [47]. For mammalian cells, however, the literature is contradictory. In one report involving murine lymphoma cells, inositol acylation is thought to be a CoA-dependent transacylation reaction that is enhanced by GTP [51]. In another paper regarding Chinese hamster ovary cells, inositol acylation of a synthetic dioctanoyl GlcN-PI analog depends on acyl-CoA with no GTP requirement [49]. Whether these inconsistent results are due to different experimental conditions remains to be determined. Many GPI anchors on proteins in mammals and yeast do not contain acylated inositol. Since biosynthesis of all GPIs appears to require inositol acylation and since the complete GPI precursor appears to be inositol acylated, there must be a mechanism to selectively deacylate the inositol. Recent studies on mammalian cells indicate that deacylation occurs in the ER, immediately after linkage of the acylated GPI precursor to the protein. Deacylation appears to be cell type- and proteinspecific [ 521. Unlike in mammals and yeast, inositol acylation in T. brucei is not necessary for mannosylation, but instead is required for efficient transfer of EtN-P to the third mannose of the glycan core (Figure 2) [50, 531. In this parasite the acyl donor does not appear to be acyl-CoA [54]. In the bloodstream form of T. brucei the complete GPI precursor (glycolipid A) is in a dynamic equilibrium with its inositol-acylated form (glycolipid C), with the equilibrium apparently maintained by an inositol acyltransferase and an inositol deacylase [53]. The function of glycolipid C is not yet known. Since the pool size of glycolipid C is several times larger than that of glycolipid A [55, 561, one possibility is that glycolipid C may be a reservoir for the complete precursor. Alternatively, since a putative GPI catabolic intermediate in T. brucei also has an acylated inositol, glycolipid C could serve as a catabolic intermediate [57]. In the insect form of T. brucei, inositol deacylation does not occur, and therefore the complete GPI precursor is inositol-acylated [58, 591, as is the anchor of PARP (Figure 1C). 25.3.4 GPI Mannosylation

The next step in GPI biosynthesis is the addition of three mannose residues (Figure 2), and Dol-P-Man is the donor of all three [60, 611. In a T. brucei cell free system, the first mannosyltransferase requires GlcN-inositol as the minimum substrate structure but the presence of the lipid moiety strongly enhances substrate recognition [62]. As mentioned above, in T. brucei and in the closely related organism L. major, mannosylation occurs immediately following the formation of GlcN-PI, without a requirement for inositol acylation. Interestingly, these organisms seem to have developed a system to channel the GlcN-PI substrate between the GlcNAc de-N-acetylase and the first mannosyltransferase [46, 62, 631 so that GlcNAc-PI cannot be mannosylated without de-N-acetylation. In contrast, in the mammalian cell free system, where inositol acylation is required for mannosylation, there is no evidence for substrate channeling [41]. PIG-B, a gene required for the third mannosylation reaction, has been cloned in

25.3 GPI Precursor Synthesis

423

mammals and yeast [64, 651. The product of PIG-B is a 554 amino acid polypeptide with an apparently large C-terminal domain (470 amino acids) oriented towards the lumen of the ER [65]. PIG-B is homologous to a protein involved in mannose transfer in yeast N-glycan synthesis [9, 661, but in neither case has enzymatic activity been detected.

25.3.5 Transfer of EtN-P In T. brucei and yeast, the donor of EtN-P is phosphatidylethanolamine [67, 681. In T. brucei, EtN-P is transferred only to the third mannose (Figure 2). In contrast, in mammals and yeast the first mannose can also be modified by EtN-P. Furthermore, in mammals, the second mannose can also receive this modification (Figure 2) [64, 69-71]. In all cases, however. it appears that only the EtN-P on the third mannose can link to a protein. Little is known about the enzymes catalyzing the transfer of EtN-P. However, mammalian PIG-F, a very hydrophobic protein which can correct a mutant that accumulates Manz-(EtN-P-)Man-GlcN-(acyl)PI,is thought to be involved in the transfer of EtN-P specifically to the third mannose, though its exact function is not known [72]. A variation in the core structure of GPI protein anchors, found in mucin-like proteins in T. cruzi, is the presence of aminoethylphosphonate instead of EtN-P. In this parasite, protein can be anchored to a GPI containing either EtN-P or aminoethylphosphonate [23, 731. The corresponding GPI precursors appear to be synthesized in a cell-free system without any obvious precursor/product relationship [74].

25.3.6 Lipid Remodeling Lipid remodeling, in which the lipid moiety on a GPI is exchanged for a different lipid species, is a widespread phenomenon. In the bloodstream form of T. brucei, fatty acid remodeling of the GPI precursor involves sequential removal of the longer fatty acids at the sn-2 and sn-1 positions of glycerol and replacement by myristate (Figure 2) [75].Myristoyl CoA is the fatty acid donor. Why a dimyristoyl GPI is essential for this parasite is not known, but there is an additional myristoylation mechanism, termed myristate exchange, that appears to act on protein-attached GPIs in a post-ER compartment [76]. Myristate exchange is possibly a proofreading mechanism, ensuring that GPIs are fully myristoylated. Consistent with the importance of GPI myristoylation, analogs of myristate are highly toxic to the parasite V71. There are other examples of remodeling acyl residues on glycerol. In the insect form of T. buucei, a partial remodeling occurs, whereby the sn-2 fatty acid is removed to form a lyso GPI that is then attached to proteins [59]. In L. mexicana free GPIs, fatty acid remodeling involves myristate replacing a longer sn-2 fatty acid [78]. A similar event occurs in yeast, but in this case a shorter fatty acid is replaced by a longer one (C26 : 0) [79]. The yeast remodeling takes place in the ER after attachment of the GPI to a protein. The presence of distearoyl glycerol, a rare species

424

25 The Biosynthesis of GPI Anchors

in total cellular PI, in anchors for human CD52-1 and porcine membrane dipeptidase implies that there could be fatty acid remodeling of those anchors as well [ 19, 201. In another type of lipid remodeling, found in yeast and other species, the entire diacylglycerol moiety is replaced by ceramide after the addition of the complete GPI precursor to a protein [SO]. In T. cruzi, the major cell surface mucin-like proteins are anchored by an sn-1 -alkyl-2-acyl GPI in a non-infective stage of the parasite’s life cycle, but this same protein becomes mostly ceramide-based in a different stage, which is infective [23]. In the latter stage, there appears to be a selective remodeling in which ceramide is introduced specifically into GPIs of mucin-like proteins that were initially synthesized with a glycerol-based anchor. In mammalian cells, a third type of lipid remodeling might occur to introduce the sn-l-alkyl group that is often found on mature GPI anchors. One study suggests that most GPI precursors contain diacyl glycerol, whereas some mature proteins contain alkylated anchors [S 11, implying lipid remodeling after the attachment of the GPI precursor to the protein. In contrast, another study suggests the presence of a substantial amount of alkyl group-containing precursors [71]. The reason for this difference is unclear. If this type of lipid remodeling exists in mammalian cells, it will be interesting to find out how some proteins evade alkylation of their GPI anchors.

25.3.7 Addition of Carbohydrate Side Chains GPIs receive various types of carbohydrate modifications to the trimannosyl core (Figures 1B-E). The modification of T. brucei VSG GPI by Gal residues (Figure 1B) takes place after the protein attachment and occurs possibly in both the ER [82] and the Golgi [83]. In yeast, the fourth Man is added in the ER before the addition of the GPI to the protein (Figure 2) whereas the fifth is added to the protein anchor in the Golgi (Figure 1D) [84]. In mammalian cells, the GalNAcD1-4 side chain is presumably linked to the first Man after the attachment of the precursor to the protein since the candidate precursors do not contain the GalNAc residue (Figure 2). On the other hand, in a protozoan parasite Toxoplusmu gondii, the GalNAc addition to the first Man apparently takes place before the EtN-P is transferred to the third mannose [85].

25.3.8 Topology of Biosynthetic Pathways In mammalian cells, studies with membrane-impermeable probes have revealed that GlcNAc-PI and GlcN-PI are found on the cytoplasmic leaflet of the ER membrane [86]. Consistent with that result, the active sites of the putative GlcNAc transferase (PIG-A) and de-N-acetylase (PIG-L) apparently face the cytoplasm. At some later point in the biosynthetic pathway, the GPI must flip to the lumen of ER where the complete precursor is attached to a protein. However, it is not known which GPI intermediate(s) undergo flipping.

25.4 Attaclzment of the GPI Precursor to u Protein

425

In Trypanosoma brucei, studies with membrane-impermeable probes revealed that mannosylated GPI intermediates are also enriched in the cytoplasmic leaflet, suggesting that this is the site for mannosylation [87]. A similar conclusion was reached in L. major [63]. Because of the proposed topology of mannosylated intermediates in these parasitic protozoa, the lumenal orientation of mammalian PIG-B, a protein involved in the transfer of the third mannose, was unexpected [65]. In mammalian cells, however, the topology of PIG-B is consistent with the fact that the donor of all three core Man is Dol-P-Man, which is also the donor of the last four Man residues in N-glycan biosynthesis in the lumenal leaflet. Moreover, a putative Dol-P-Man flippase mutation inhibits the incorporation of the first Man into GlcN-(acyl)PI [SS]. Thus, all evidence so far suggests that in mammalian cells Man is added to the GPI in the lumen of ER. The topology of the biosynthetic pathway could be different between protozoa and mammals. More likely, these mannosylated intermediates may flip freely in T. hrucei, and the observed topology of these substrates may not represent the topology of enzymatic reactions.

25.4 Attachment of the GPI Precursor to a Protein 25.4.1 Basic Features The primary translation product of proteins destined to be anchored by GPI contains both an N-terminal secretory signal sequence and a C-terminal signal sequence that specifies GPI addition. The latter is cleaved and replaced by the GPI precursor in the ER (Figure 3A). This transfer reaction is mediated by a transamidase and involves a nucleophilic attack by the EtN amino group of the GPI precursor on a carbonyl of a specific amino acid residue near the C-terminus. This residue, residing at the ‘‘asite”, will become the new C-terminus of the GPI anchored protein (Figure 3B). In a cell free system that reconstitutes anchor attachment, an alternative nucleophile such as hydrazine or hydroxylamine can substitute for a GPI in this reaction [89]. In a T. brucei cell free system, GPI addition to VSG does not require ATP or other cofactors [90]. In a mammalian system. the transfer of the GPI precursor to a protein seems to be enhanced by an ATP-dependent chaperone [91, 921.

B Figure 3. Anchoring of GPI to a protein (A) Transamidase reaction (B) Features in the primary structure of GPI-anchored proteins See text for details

I

,Secretory Signal

tl)s’te

7 Hydrophiltc Spacer

] Hydrophobic Region

C

426

25 The Biosynthesis of GPI Anchors

25.4.2 Protein Machinery for GPI Addition Genetic studies, on yeast mutants that synthesize the complete GPI precursor but do not attach it to a protein, have lead to the discovery of two genes involved in anchor addition. The GAAI gene encodes a 614 amino acid ER integral membrane protein with a large C-terminal lumenal domain. Its enzymatic function is unknown [93]. Another yeast gene, GPI8, encodes a 41 1 amino acid ER integral membrane protein containing a large N-terminal lumenal domain [94]. GPISp is likely to have transamidase activity because it is homologous to Jack Bean asparaginyl endopeptidase which expresses transamidase activity in uitro [95]. Homologs of yeast GAAI and GPI8 genes have also been identified in humans [96, 971.

25.4.3 Signal Sequence for GPI Addition A comparison of GPI signal sequences from different GPI-anchored proteins, as well as mutagenesis studies of model proteins, have revealed the major sequence requirements for GPI anchor addition (Figure 3B) [98-1061. These sequences are up to 35 residues long, and all information specifying anchor addition seems to reside between the w site and the C-terminus. GPI anchoring requires a hydrophobic sequence of 10-20 residues at the extreme C-terminus and it is the hydrophobicity of this site, rather than its specific sequence, that is important for GPI transfer. Another crucial feature is that the amino acid residues at the o,o 1, and w 2 positions must be small, usually Gly, Asp, Asn, Ala, Ser, or Cys. Finally, there is a spacer between the w 2 position and the hydrophobic region, usually containing 5-12 hydrophilic amino acids. Although GPI anchor addition signals appear to be similar in all species, there may be subtle differences as signals from protozoa sometimes do not function efficiently when tested in mammalian cells [107]. See [ 1081 for a more detailed description of the mechanism of GPI addition.

+

+

+

25.5 Evolution of GPI Biosynthesis How such complex pathways of GPI biosynthesis evolved is an intriguing question [ 1091. The simplest structure related to GPI, GlcNal-6myo-inositol-1-phosphodialkylglycerol, has been identified in Methanosarcina barkeri [ 1lo], an organism that belongs to the Archaea [ 1 1 11. The biosynthetic pathway may have evolved step by step from such a simple structure. In this regard, it is interesting that the 6 position of GlcN can be substituted by an aminoethylphosphonate in T. cruzi mucinlike proteins [23, 731. This ethyl amino group might have been used to anchor proteins in ancient organisms before the GPI was fully elaborated with its trimannosyl core. Perhaps such primitive protein anchors still exist in Archea! We then sip hot sake and wonder if the trimannosyl core evolved later to provide additional advantages, for example an elastic cushion for proteins so as to accommodate mechanical stress, as was recently suggested to be a property of polysaccharides [ 1121.

Rejkrences

427

25.6 Future Studies The regulation of GPI biosynthesis should be one important aspect of the future research. Is the GlcNAc transferase reaction regulated? How are the syntheses of proteins and GPI precursors coordinated? How are excess GPIs degraded? What is the quality control mechanism for GPI anchored proteins? How are the syntheses of free, protein-anchoring, and polysaccharide-anchoring GPIs coordinated, for example in Leishmania species? Some clarification regarding these questions is only beginning to emerge. Drug development is another important area for future efforts. Parasitic infections are becoming more serious in developing countries, with the most alarming being Plasmodium julciparum, the causative agent of malaria. GPIs are the major type of surface carbohydrates in this species [ 113, 1 141. Since there are striking differences in the GPI biosynthetic pathways of mammals as compared to those of pathogenic protozoa and fungi, we should be able to target drugs to the pathogen’s GPI biosynthetic pathways. There are already promising leads for selective inhibitors [50, 77, 115, 1161. Future identification of genes and characterization of enzymes and regulatory proteins will allow the design of drugs based on the mechanistic and/or structural differences between these proteins in pathogens and their hosts.

Acknowledgments

Work in the author’s laboratory was supported by NIH grant AI21334. We thank Drs. Jay Bangs, Tamara Doering, Mike Ferguson, Lucia Guther, David Jiang, Taroh Kinoshita, Anant Menon, Kojo Mensa-Wilmot, Ken Milne, Jim Morris, and Kim Paul for comments on the manuscript. Drs. Menon and Kinoshita shared unpublished results. References B. J. Mengeling & S. J. Turco. Microbial glycoconjugates. Curr Upin Struct Biol, 1998, 8, 512-517. M. J. McConville. The surface glycoconjugates of parasitic protozoa: potential targets for new drugs. Aust N Z J Med, 1995, 25, 168-776. T. L. Rosenberry, D. Sevlever & M. E. Medof. Metabolism of GPls in mammalian cells. BraJ Med B i d Res, 1994,27, 151-159. W. van’t Hof, E. Rodriguez-Boulan & A. K. Menon. Nonpolarized distribution of glycosylphosphatidylinositols in the plasma membrane of polarized Madin-Darby canine kidney cells. J Bid Chem, 1995,270, 24150-24155. N. Singh, L. N. Liang, M. L. Tykocinski & A. M. Tartakoff. A novel class of cell surface glycolipids of mammalian cells. Free glycosyl phosphatidylinositols. J Biol Chem, 1996, 271, 12879--12884. V. L. Stevens. Biosynthesis of glycosylphosphatidylinositol membrane anchors. Brochem J , 1995,310, 361-310.

428

25 The Biosynthesis of GPI Anchors

7. P. T. Englund. The structure and biosynthesis of glycosyl phosphatidylinositol protein anchors. Annu Rev Biochem, 1993, 62, 121-138. 8. M. A. Ferguson. The surface glycoconjugates of trypanosomatid parasites. Philos Trans R Soc Lond B Biol Sci,1997, 352, 1295-1302. 9. T. Kinoshita, K. Ohishi & J. Takeda. GPI-anchor synthesis in mammalian cells: genes, their products, and a deficiency. J Biochem (Tokyo), 1997, 122, 251-257. 10. A. K. Menon, N. A. Baumann, W. van’t Hof & J. Vidugiriene. Glycosylphosphatidylinositols: biosynthesis and intracellular transport. Biochem Soc Trans, 1997,25, 861-865. 11. M. A. Ferguson, J. S. Brimacombe, S. Cottaz, R. A. Field, L. S. Guther, S. W. Homans, M. J. McConville, A. Mehlert, K. G. Milne, J. E. Ralton & et al. Glycosyl-phosphatidylinositol molecules of the parasite and the host. Parasitology, 1994, 108, S45-54. 12. M. A. Ferguson, S. W. Homans, R. A. Dwek & T. W. Rademacher. Glycosylphosphatidylinositol moiety that anchors Trypanosoma hrucei variant surface glycoprotein to the membrane. Science, 1988,239, 753-759. 13. P. Schneider, M. A. Ferguson, M. J. McConville, A. Mehlert, S. W. Homans & C. Bordier. Structure of the glycosyl-phosphatidylinositol membrane anchor of the Leishmania major promastigote surface protease. J Biol Chem, 1990,265, 16955-16964. 14. S. W. Homans, M. A. Ferguson, R. A. Dwek, T. W. Rademacher, R. Anand & A. F. Williams. Complete structure of the glycosyl phosphatidylinositol membrane anchor of rat brain Thy-1 glycoprotein. Nature, 1988, 333, 269-272. 15. C. Fankhauser, S. W. Homans, J. E. Thomas-Oates, M. J. McConville, C. Desponds, A. Conzelmann & M. A. Ferguson. Structures of glycosylphosphatidylinositol membrane anchors from Sacchuromyces cerevisiae. J Biol Chem, 1993,268, 26365-26374. 16. A. Treumann, N. Zitzmann, A. Hulsmeier, A. R. Prescott, A. Almond, J. Sheehan & M. A. Ferguson. Structural characterisation of two forms of procyclic acidic repetitive protein expressed by procyclic forms of Trypanosoma brucei. J Mol B i d , 1997, 269, 529-547. 17. A. Mehlert, J. M. Richardson & M. A. Ferguson. Structure of the glycosylphosphatidylinositol membrane anchor glycan of a class-2 variant surface glycoprotein from Trypanosoma hrucei. J Mol B i d , 1998,277, 379-392. 18. S. W. Homans, C. J. Edge, M. A. Ferguson, R. A. Dwek & T. W. Rademacher. Solution structure of the glycosylphosphatidylinositol membrane anchor glycan of Trypanosoma hrucei variant surface glycoprotein. Biochemistry, 1989, 28, 2881-2887. 19. A. Treumann, M. R. Lifely, P. Schneider & M. A. Ferguson. Primary structure of CD52. J Biol Chem, 1995,270, 6088-6099. 20. 1. A. Brewis, M. A. Ferguson, A. Mehlert, A. J. Turner & N. M. Hooper. Structures of the glycosyl-phosphatidylinositol anchors of porcine and human renal membrane dipeptidase. Comprehensive structural studies on the porcine anchor and interspecies comparison of the glycan core structures. J Biol Chem, 1995, 270, 22946-22956. 21. M. A. Ferguson, M. G. Low & G. A. Cross. Glycosyl-sn-l,2-dimyristylphosphatidylinositol is covalently linked to Trypanosoma hrucei variant surface glycoprotein. J Bid Chem, 1985, 260, 14547- 14555. 22. J. Stadler. T. W. Keenan, G. Bauer & G. Gerisch. The contact site A glycoprotein of Dictyostelium discoideum carries a phospholipid anchor of a novel type. EMBO J , 1989, 8, 371377. 23. A. Acosta Serrano, S. Schenkman, N. Yoshida, A. Mehlert, J. M. Richardson & M. A. Ferguson. The lipid structure of the glycosylphosphatidylinositol-anchored mucin-like sialic acid acceptors of Trypanosoma cruzi changes during parasite differentiation from epimastigotes to infective metacyclic trypomastigote forms. J Biol Chem, 1995, 270, 27244-27253. 24. W. L. Roberts, S. Santikarn, V. N. Reinhold & T. L. Rosenberry. Structural characterization of the glycoinositol phospholipid membrane anchor of human erythrocyte acetylcholinesterase by fast atom bombardment mass spectrometry. J Biol Chem, 1988, 263, 18776-18784. 25. M. A. Ferguson. Site of palmitoylation of a phospholipase C-resistant glycosylphosphatidylinositol membrane anchor. Biochem J , 1992,284, 297-300. 26. W. J. Masterson, T. L. Doering, G. W. Hart & P. T. Englund. A novel pathway for glycan assembly: biosynthesis of the glycosylphosphatidylinositol anchor of the trypanosome variant surface glycoprotein. Cell, 1989, 56, 793-800.

References

429

27. A. K. Menon, R. T. Schwarz, S. Mayor & G . A. Cross. Cell-free synthesis of glycosylphosphatidylinositol precursors for the glycolipid membrane anchor of Trypanosornu hrucei variant surface glycoproteins. Structural characterization of putative biosynthetic intermediates. J Biol Chem, 1990, 265, 9033-9042. 28. T. Miyata, J. Takeda, Y. Iida, N. Yamada, N. Inoue, M. Takahashi, K. Maeda, T. Kitani & T. Kinoshita. The cloning of PIG-A, a component in the early step of GPI-anchor biosynthesis. Science, 1993, 259, 13 18- 1320. 29. J. Takeda, T. Miyata, K. Kawagoe, Y. Iida, Y. Endo, T. Fujita, M. Takahashi, T. Kitani & T. Kinoshita. Deficiency of the GPI anchor caused by a somatic mutation of the PIG-A gene in paroxysmal nocturnal hemoglobinuria. Cell, 1993, 73, 703-71 1. 30. T. L. Doering, W. J. Masterson, P. T. Englund & G. W. Hart. Biosynthesis of the glycosyl phosphatidylinositol membrane anchor of the trypanosome variant surface glycoprotein. Origin of the non-acetylated glucosamine. J Biol Chem, 1989, 264, 11 168-1 11 73. 31. R. Watanabe, N. Inoue, B. Westfall, C. H. Taron. P. Orlean, J. Takeda & T. Kinoshita. The first step of glycosylphosphatidylinositol biosynthesis i s mediated by a complex of PIG-A, PIG-H, PIG-C and GPII. EMBO J , 1998,17, 877-885. 32. R. Watanabe, T. Kinoshita, R. Masaki, A. Yamamoto, J. Takeda & N . Inoue. PIG-A and PIG-H, which participate in glycosylphosphatidylinositol anchor biosynthesis, form a protein complex in the endoplasmic reticulum. J Biol Chem, 1996, 271, 26868-26875. 33. M. Bessler, P. Hillmen, L. Longo, L. Luzzatto & P. J. Mason. Genomic organization of the X-linked gene (PIG-A) that i s mutated in paroxysmal nocturnal haemoglobinuria and of a related autosomal pseudogene mapped to 12q21. Hunz Mol Genet, 1994, 3, 751-757. 34. K. Kawagoe, J. Takeda. Y. Endo & T. Kinoshita. Molecular cloning of murine pig-a, a gene for GPI-anchor biosynthesis, and demonstration of interspecies conservation of its structure, function, and genetic locus. Genomics, 1994, 23, 566-574. 35. S. D. Leidich, Z. Kostova, R. R. Latek, L. C. Costello, D. A. Drapp, W. Gray, J. S. Fassler & P. Orlean. Temperature-sensitive yeast GPI anchoring mutants gpi2 and gpi3 are defective in the synthesis of N-acetylglucosaminyl phosphatidylinositol. Cloning of the GP12 gene. J Biol Chef%, 1995, 270, 13029-13035. 36. S. D. Leidich & P. Orlean. Gpil , a Saccl~arotnycescereuisiue protein that participates in the first step in glycosylphosphatidylinositol anchor synthesis. J Biol Cliem, 1996. 271, 27829-27837. 37. T. Kamitani, H. M. Chang, C. Rollins, G. L. Waneck & E. T. Yeh. Correction of the class H defect in glycosylphosphatidylinositol anchor biosynthesis in Ltk-cells by a human cDNA clone. J Biol Chem, 1993,268,20733 -20736. 38. J. Schubert, R. E. Schmidt & M. E. Medof. Regulation of glycoinositol phospholipid anchor assembly in human lymphocytes. Absent mannolipid synthesis in affected T and natural killer cell lines from paroxysmal nocturnal hemoglobinuria patients. J Biol Chern, 1993,268, 628 1-6287. 39. W. J. Masterson & M. A. Ferguson. Phenylmethanesulphonyl fluoride inhibits GPI anchor biosynthesis in the African trypanosome. EMBO J , 1991, 10; 2041 -2045. 40. J. E. Ralton, K. G. Milne, M. L. Guther, R. A. Field & M. A. Ferguson. The mechanism of inhibition of glycosylphosphatidylinositol anchor biosynthesis in Trj,p~no~oma hvucei by mannosamine. J Biol Chem, 1993,268,241 83- 24189. 41. D. K. Sharma, T. K. Smith, A. Crossman, J. S . Brimacombe & M. A. Ferguson. Substrate specificity of the N-acetylglucosaminyl-phosphatidylinositolde-N-acetylase of glycosylphosphatidylinositol membrane anchor biosynthesis in African trypanosomes and human cells. Biochem J , 1997, 328, 171-177. 42. K. G. Milne, R. A. Field, W. J. Masterson, S. Cottaz, J. S. Brimacombe & M. A. Ferguson. Partial purification and characterization of the N-acetylglucosaminyl-phosphatidylinositol de-N-acetylase of glycosylphosphatidylinositol anchor biosynthesis in African trypanosomes. J Biol Chenz, 1994,269, 16403-16408. 43. N. Nakamura, N. Inoue, R. Watanabe, M. Takahashi, J. Takeda, V. L. Stevens & T. Kinoshita. Expression cloning of PIG-L, a candidate N-acetylglucosaminylphosphatidylinositol deacetylase. J Biol Chem, 1997, 272, 15834-15840. 44. R. Watanabe, K. Ohishi, Y. Maeda, N . Nakamura & T. Kinoshita. Mammalian PIG-L and its yeast homologue Gpi 12p are N-acetylglncosaminyl-phosphatidylinositolde-N-acetylases essential in glycosylphosphatidylinositol biosynthesis. Biorlzem J , 1999, 33Y, 185- 192.

430

25 The Biosynthesis of GPI Anchors

45. V. L. Stevens. Regulation of glycosylphosphatidylinositol biosynthesis by GTP. Stimulation of N-acetylglucosamine-phosphatidylinositoldeacetylation. J Biol Chem, 1993, 268, 9718-9724. 46. T. K. Smith, F. C. Milne, D. K. Sharma, A. Crossman, J. S. Brimacombe & M. A. Ferguson. Early steps in glycosylphosphatidylinositol biosynthesis in Leishmania mujor. Biochem J , 1997, 326, 393-400. 47. L. C. Costello & P. Orlean. Inositol acylation of a potential glycosyl phosphoinositol anchor precursor from yeast requires acyl coenzyme A. J Biol Chem, 1992,267, 8599-8603. 48. M. Urakaze, T. Kamitani, R. DeGasperi, E. Sugiyama, H. M. Chang, C. D. Warren & E. T. Yeh. Identification of a missing link in glycosylphosphatidylinositol anchor biosynthesis in mammalian cells. J Biol Chem, 1992, 267, 6459-6462. 49. W. T. Doerrler, J. Ye, J. R. Falck & M. A. Lehrman. Acylation of glucosaminyl phosphatidylinositol revisited. Palmitoyl-CoA dependent palmitoylation of the inositol residue of a synthetic dioctanoyl glucosaminyl phosphatidylinositol by hamster membranes permits efficient mannosylation of the glucosamine residue. J Biol Chem, 1996,271, 27031-27038. 50. T. K. Smith. D. K. Sharma, A. Crossman, A. Dix, J. S. Brimacombe & M. A. Ferguson. Parasite and mammalian GPI biosynthetic pathways can be distinguished using synthetic substrate analogues. EMBO J , 1997, 16, 6667-6675. 51. V. L. Stevens & H. Zhang. Coenzyme A dependence of glycosylphosphatidylinositol biosynthesis in a mammalian cell-free system. J Biol Chem, 1994, 269, 31397-31403. 52. R. Chen, E. I. Walter, G. Parker, J. P. Lapurga, J. L. Millan, Y. Ikehara, S. Udenfriend & M. E. Medof. Mammalian glycophosphatidylinositol anchor transfer to proteins and posttransfer deacylation. Proc Nut1 Acad Sci U S A , 1998, 95, 9512-9517. 53. M. L. Guther & M. A. Ferguson. The role of inositol acylation and inositol deacylation in GPI biosynthesis in Trypanosoma hrucei. EMBO J , 1995, 14, 3080-3093. 54. M. C. Field, A. K. Menon & G. A. Cross. A glycosylphosphatidylinositol protein anchor from procyclic stage Trypanosoma hrucei: lipid structure and biosynthesis. EMBO J , 1991, 10, 273 1-2739. 55. T. L. Doering, M. S. Pessin, G. W. Hart, D. M. Raben & P. T. Englund. The fatty acids in unremodelled trypanosome glycosylphosphatidylinositols. Biochem J , 1994, 299, 741-746. 56. M. L. Guther, A. Treumann & M. A. Ferguson. Molecular species analysis and quantification of the glycosylphosphatidylinositol intermediate glycolipid C from Trypanuwma brucei. Mol Biochem Parasitol, 1996, 77, 137-145. 57. K. G. Milne, M. A. J. Ferguson & P. T. Englund. A Novel Glycosylphosphatidylinositol in African Trypanosomes. A possible catabolic intermediate. J Biol Chem, 1999,274, 1465-1471. 58. M. C. Field, A. K. Menon & G. A. Cross. Developmental variation of glycosylphosphatidylinositol membrane anchors in Trypanosoma brucei. Identification of a candidate biosynthetic precursor of the glycosylphosphatidylinositol anchor of the major procyclic stage surface glycoprotein. J Biol Chem, 1991,266, 8392-8400. 59. M. C. Field, A. K. Menon & G. A. Cross. Developmental variation of glycosylphosphatidylinositol membrane anchors in Trypanosomu hrucei. In vitro biosynthesis of intermediates in the construction of the GPI anchor of the major procyclic surface glycoprotein. J Biol Chem, 1992,267, 5324-5329. 60. R. DeGasperi, L. J. Thomas, E. Sugiyama, H. M. Chang, P. J. Beck, P. Orlean, C. Albright, G. Waneck, J. F. Sambrook, C. D. Warren & et al. Correction of a defect in mammalian GPI anchor biosynthesis by a transfected yeast gene. Science, 1990, 250, 988-991. 61. A. K. Menon, S. Mayor & R. T. Schwarz. Biosynthesis of glycosyl-phosphatidylinositol lipids in Trypanosoma brucei: involvement of mannosyl-phosphoryldolichol as the mannose donor. EMBO J , 1990, 9,4249-4258. 62. T. K. Smith, S. Cottaz, J. S. Brimacombe & M. A. Ferguson. Substrate specificity of the dolichol phosphate mannose: glucosaminyl phosphatidylinositol alphal-4-mannosyltransferase of the glycosylphosphatidylinositol biosynthetic pathway of African trypanosomes. J Biol Chem, 1996,271, 6476-6482. 63. K. Mensa-Wilmot, J. H. LeBowitz, K. P. Chang, A. al-Qahtani, B. S. McGwire, S. Tucker & J. C . Morris. A glycosylphosphatidylinositol (GPI)-negative phenotype produced in Leishmunia major by GPI phospholipase C from Trypanosoma brucei: topography of two GPI pathways. J Cell Biol, 1994, 124, 935-947.

Rejerences

43 1

64. E. Canivenc-Gansel, I. Imhof, F. Reggiori, P. Burda, A. Conzelmann & A. Benachour. GPI anchor biosynthesis in yeast: phosphoethanolamine is attached to the alphal,4linked mannose of the complete precursor glycophospholipid. Glycobioloyy, 1998. 8, 761 770. 65. M. Takahashi, N . Inoue, K. Ohishi, Y. Maeda, N. Nakamura, Y. Endo, T. Fujita, J. Takeda & T. Kinoshita. PIG-B, a membrane protein of the endoplasmic reticulum with a large lumenal domain, is involved in transferring the third mannose of the GPI anchor. EMBO J , 1996, 15, 4254-4261. 66. P. Burda, S. te Heesen, A. Brachat, A. Wach, A. Dusterhoft & M. Aebi. Stepwise assembly of the lipid-linked oligosaccharide in the endoplasmic reticulum of Succizuronzyces cereoisiae: identification ofthe ALG9 gene encoding a putative mannosyl transferase. Proc Natl Acud Sci U S A , 1996, 93,7160-7165. 67. A. K. Menon & V. L. Stevens. Phosphatidylethanolamine is the donor of the ethanolamine residue linking a glycosylphosphatidylinositol anchor to protein. J Biol Chrm, 1992, 267, 15277-1 5280. 68. A. K. Menon, M. Eppinger, S. Mayor & R . T. Schwarz. Phosphatidylethanolamine is the donor of the terminal phosphoethanolamine group in trypanosome glycosylphosphatidylinositols. EMBO J , 1993; 12, 1907-1914. 69. S. Hirose, G. M. Prince, D. Sevlever, L. Ravi, T. L. Rosenberry, E. Ueda & M. E. Medof. Characterization of putative glycoinositol phospholipid anchor precursors in mammalian cells. Localization of phosphoethanolaniine. J Biol Chem, 1992, 267, 16968-16974. 70. T. Kamitani, A. K. Menon, Y. Hallaq, C. D. Warren & E. T. Yeh. Complexity of ethanolamine phosphate addition in the biosynthesis of glycosylphosphatidylinositol anchors in mammalian cells. J Biol Chem, 1992,267, 2461 1-24619. 71. A. Puoti & A. Conzelmann. Structural characterization of free glycolipids which are potential precursors for glycophosphatidylinositol anchors in mouse thymoma cell lines. J Biol Chern, 1992,267, 22673-22680, 72. N. Inoue, T. Kinoshita, T. Orii & J. Takeda. Cloning of a human gene, PIG-F, a component of glycosylphosphatidylinositol anchor biosynthesis, by a novel expression cloning strategy. J Biol Chern, 1993,268, 6882-6885. 73. J . 0. Previato, C. Jones, M. T. Xavier, R. Wait, L. R. Travassos, A. J. Pdrodi & L. Mendonca-Previato. Structural characterization of the major glycosylphosphatidylinositol membrane-anchored glycoprotein from epimastigote forms of Trypano.sonm cruzi Y-strain. J Biol Chem, 1995,270, 7241-7250. 74. N. Heise, J. Raper, L. U. Buxbaum, T. M. Peranovich & M. L. de Almeida. Identification of complete precursors for the glycosylphosphatidylinositol protein anchors of Tr.vpunosornu cruzi. J Biol Clleni, 1996, 271, 16877-16887. 75. W. J. Masterson, J. Raper, T. L. Doering, G. W. Hart & P. T. Englund. Fatty acid remodeling: a novel reaction sequence in the biosynthesis of trypanosome glycosyl phosphatidylinosito1 membrane anchors. Cell, 1990, 62, 73-80. 76. L. U. Buxbaum, K. G. Milne, K . A. Werbovetz & P. T. Englund. Myristate exchange on the Trypanosoma brucei variant surface glycoprotein. Proc Natl Acad Sci U S A , 1996, 93, 11781183. 77. T. L. Doering, J. Raper, L. U. Buxbaum, S. P. Adams, J. I. Gordon, G . W. Hart & P. T. Englund. An analog of myristic acid with selective toxicity for African trypanosomes. Science, 1991, 252, 1851-1854. 78. J . E. Ralton & M. J. McConville. Delineation of three pathways of glycosylphosphatidylinositol biosynthesis in Leishmania mr.xicana. Precursors from different pathways are assembled on distinct pools of phosphatidylinositol and undergo fatty acid remodeling. J Biol Chem, 1998, 273, 4245-4257. 79. G. Sipos, F. Reggiori, C. Vionnet & A. Conzelmann. Alternative lipid remodelling pathways for glycosylphosphatidylinositol membrane anchors in Saccizuromyces cerevisiae. EMBO J , 1997, 16, 3494-3505. 80. F. Reggiori, E. Canivenc-Gansel & A. Conzelmann. Lipid remodeling leads to the introduction and exchange of defined ceramides on GPI proteins in the ER and Golgi of Saccharomyces cerevisiae. EMBO J , 1997, 16, 3506-3518. -

432

25 The Biosynthesis of GPI Anchors

81. N. Singh, R. A. Zoeller, M. L. Tykocinski, P. B. Lazarow & A. M. Tartakoff. Addition of lipid substituents of mammalian protein glycosylphosphoiuositol anchors. Mol Cell Bid, 1994, 14, 21-31. 82. S. Mayor, A. K. Menon & G. A. Cross. Galactose-containing glycosylphosphatidylinositols in Trypanosoma brucei. J Biol Chem, 1992,267, 754-761. 83. J. D. Bangs, T. L. Doering, P. T. Englund & G. W. Hart. Biosynthesis of a variant surface glycoprotein of Trypanosoma brucei. Processing of the glycolipid membrane anchor and Nlinked oligosaccharides. J Biol Chem, 1988,263, 17697-17705. 84. G. Sipos, A. Puoti & A. Conzelmann. Biosynthesis of the side chain of yeast glycosylphosphatidylinositol anchors is operated by novel mannosyltransferases located in the endoplasmic reticulum and the Golgi apparatus. J Biol Chem. 1995,270, 19709-19715. 85. S. Tomavo, J. F. Dubremetz & R. T. Schwarz. Biosynthesis of glycolipid precursors for glycosylphosphatidylinositol membrane anchors in a Toxoplasma gondii cell-free system. J Biol Chem, 1992,267, 21446-21458. 86. J. Vidugiriene & A. K. Menon. Early lipid intermediates in glycosyl-phosphatidylinositol anchor assembly are synthesized in the ER and located in the cytoplasmic leaflet of the ER membrane bilayer. J Cell Biol, 1993, 121, 987-996. 87. J. Vidugiriene & A. K. Menon. The GPI anchor of cell-surface proteins is synthesized on the cytoplasmic face of the endoplasmic reticulum. J Cell Biol, 1994, 127, 333-341. 88. L. A. Camp, P. Chauhan, J. D. Farrar & M. A. Lehrman. Defective mannosylation of glycosylphosphatidylinositol in Lec35 Chinese hamster ovary cells. J Biol Chem, 1993, 268, 6721-6728. 89. S. E. Maxwell, S. Ramalingam, L. D. Gerber, L. Brink & S. Udenfriend. An active carbonyl formed during glycosylphosphatidylinositol addition to a protein is evidence of catalysis by a transamidase. J Biol Chrm, 1995,270, 19576-19582. 90. S. Mayor, A. K. Menon & G. A. Cross. Transfer of glycosyl-phosphatidylinositol membrane anchors to polypeptide acceptors in a cell-free system. J Cell Biol, 1991, 114, 61-71. 91. R. Amthauer, K. Kodukula, L. Gerber & S. Udenfriend. Evidence that the putative COOHterminal signal transamidase involved in glycosylphosphatidylinositol protein synthesis is present in the endoplasmic reticulum. Proc Nut1 Acad Sci U S A , 1993, 90, 3973-3977. 92. J. Vidugiriene & A. K. Menon. Soluble constituents of the ER lumen are required for GPI anchoring of a model protein. EMBO J , 1995, 14, 4686-4694. 93. D. Hamburger, M. Egerton & H. Riezman. Yeast Gaalp is required for attachment of a completed GPI anchor onto proteins. J Cell Biol, 1995, 129, 629-639. 94. M. Benghezal, A. Benachour, S. Rusconi, M. Aebi & A. Conzelmann. Yeast Gpi8p is essential for GPI anchor attachment onto proteins. EMBO J , 1996, 15, 6575-6583. 95. Y. Abe, K. Shirane, H. Yokosawa, H. Matsushita, M. Mitta, 1. Kato & S. Ishii. Asparaginyl endopeptidase of jack bean seeds. Purification, characterization, and high utility in protein sequence analysis. J Biol Chem, 1993$268,3525-3529. 96. J. Yu, S. Nagarajan, J. J. Knez, S. Udenfriend, R. Chen & M. E. Medof. The affected gene underlying the class K glycosylphosphatidylinositol (GPI) surface protein defect codes for the GPI transamidase. Proc Nut1 Acad Sci U S A , 1997, 94, 12580 -12585. 97. Y. Hiroi, I. Komuro, R. Chen, T. Hosoda, T. Mizuno, S. Kudoh, S. P. Georgescu, M. E. Medof & Y. Ydzaki. Molecular cloning of human homolog of yeast GAAl which is required for attachment of glycosylphosphatidylinositols to proteins. FEBS Lett, 1998, 421, 252-258. 98. I. W. Caras, G. N. Weddell & S. R. Williams. Analysis of the signal for attachment of a glycophospholipid membrane anchor. J Cell Biol, 1989, 108, 1387-1396. 99. I. W. Caras & G. N. Weddell. Signal peptide for protein secretion directing glycophospholipid membrane anchor attachment. Science, 1989,243, 1196-1 198. 100. R. Micanovic, L. D. Gerber, J. Berger, K. Kodukula & S. Udenfriend. Selectivity of the cleavage/attachment site of phosphatidylinositol-glycan-anchoredmembrane proteins determined by site-specific mutagenesis at Asp-484 of placental alkaline phosphatase. Proc Nafl Acad Sci U S A , 1990, 87, 157-161. 101. P. Moran & I. W. Caras. Fusion of sequence elements from non-anchored proteins to generate a fully functional signal for glycophosphatidylinositol membrane anchor attachment. J CeN Bid, 1991, 115, 1595-1600.

References

433

102. P. Moran & I. W. Caras. A nonfunctional sequence converted to a signal for glycophosphatidylinositol membrane anchor attachment. J Cell Biul, 1991, 115, 329-336. 103. P. Moran, H. Raab, W. J. Kohr & I. W. Caras. Glycophospholipid membrane anchor attachment. Molecular analysis of the cleavage/attachment site. J Biol Chem, 1991, 266, 1250-1257. 104. K. Kodukula, D. Cines, R. Amthauer, L. Gerber & S. Udenfriend. Biosynthcsis of phosphatidylinositol-glycan (PI-G)-anchored membrane proteins in cell-free systems: cleavage of the nascent protein and addition of the PI-G moiety depend on the size of the COOHterminal signal peptide. Proc Nut/ Acud Sci U S A , 1992, 89; 1350-1 353. 105. K. Kodukula, L. D. Gerber, R. Amthauer, L. Brink & S. Udenfriend. Biosynthesis of glycosylphosphatidylinositol (GP1)-anchored membrane proteins in intact cells: specific amino acid requirements adjacent to the site of cleavage and GPI attachment. J Cell Bid, 1993, 120, 657664. 106. K. E. Coyne, A. Crisci & D. M. Lublin. Construction of synthetic signals for glycosylphosphatidylinositol anchor attachment. Analysis of amino acid sequence requirements for anchoring. J Biol Chem, 1993,268, 6689-6693. 107. P. Moran & I. W. Caras. Requirements for glycosylphosphatidylinositol attachment are similar but not identical in mammalian cells and parasitic protozoa. J Cell Biol, 1994, 125, 333-343. 108. S. Udenfriend & K. Kodukula. How glycosylphosphatidylinositol-anchored membrane proteins are made. Annu Rev Biochern, 1995, 64, 563- 591. 109. M. A. Ferguson, W. J. Masterson, S. W. Homans & M. J. McConville. Evolutionary aspects of GPI metabolism in kinetoplastid parasites. Cell Biul Int Rep, 1991, 1.5,991-1005. 110. M . Nishihara, M. Utagawa, H. Akutsu & Y. Koga. Archaea contain a novel diether phosphoglycolipid with a polar head group identical to the conserved core of eucaryal glycosyl phosphatidylinositol. J Biol Chem, 1992, 267, 12432-12435. 11 1. C . R . Woese, 0. Kandler & M. L. Wheelis. Towards a natural system of organisms: proposal for the domains Archaea, Bacteria, and Eucarya. Pvoc Null Acud Sci U S A , 1990, 87, 45764519. 112. P. E. Marszalek, A. F. Oberhauser, Y . P. Pang & J. M. Fernandez. Polysaccharide elasticity governed by chair-boat transitions of the glucopyranose ring. Nczture, 1998, 396, 661--664. 1 13. P. Gerold, L. Schofield, M. J. Blackman. A. A. Holder & R. T. Schwarz. Structural analysis of the glycosyl-phosphatidylinositol membrane anchor of the merozoite surface proteins-1 and -2 of Plasmodiumfakiparum. M o l Biochenz Parasiiol, 1996, 75, 131 143. 114. D. C . Gowda, P. Gupta & E. A. Davidson. Glycosylphosphatidylinositol anchors represent the major carbohydrate modification in proteins of intraerythrocytic stage Plusmodiunt falcipurttm. J Bid Chem, 1997, 272, 6428~-6439. I 15. C. Sutterlin, A. Horvath, P. Gerold, R. T. Schwarz, Y . Wang, M. Dreyfuss & H. Riezman. Identification of a species-specific inhibitor of glycosylphosphatidylinositol synthesis. EMBO J. 1997, 16, 6374-6383. 116. T. L. Doering, T. Lu, K. A. Werbovetz. G. W. Gokel, G. W. Hart, J. I. Gordon & P. T. Englund. Toxicity of myristic acid analogs toward African trypanosomes. Pvoc Nut1 Acad Sci U S A , 1994; 91, 9735 9739. 1 17. Z. Kostova et al. Personal communication. 118. J. Vidugiriene, D. K. Sharma, T. K. Smith, N. A. Baumann & A. K. Menon. Segregation of glycosylphosphatidylinositol biosynthetic reactions in a subcompartment of the endoplasmic reticulum. J Biol Cheni, 1999, 274, 15203 12. -

~

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

26 Escherichia coli Lipid A: A Potent Activator of Innate Immunity Teresa A . Garrett and Christian R. H. Raetz

26.1 Introduction Gram-negative bacteria are covered with a protective envelope. The inner membrane is a bilayer composed of phospholipids, primarily phosphatidylethanolamine, phosphatidylglycerol and cardiolipin (Figure 1A) [ I]. The rigid peptidoglycan, which determines the shape of the bacterial cell, encloses the inner membrane (Figure 1A) [l]. The outer membrane is an asymmetrical lipid bilayer. The inner leaflet is similar to the inner membrane, composed primarily of phospholipids. The outer leaflet, on the other hand, consists almost entirely of a complex glycolipid called lipopolysaccharide (LPS) [ 11. LPS is important for several reasons. It helps form an essential permeability barrier inhibiting the diffusion of certain large and hydrophobic molecules into the cell [2]. LPS is also required for the assembly of outer membrane proteins such as porins [ 3 ] .A minimal LPS, consisting of lipid A and Kdo, is required for the viability of most Gram-negative bacteria [ l]. LPS also plays an important role in the pathogenesis of animal infections [ 11 and in the development of the symbiotic relationship between plants and nitrogen fixing bacteria, such as those of the bacterial family Rhizobiaceae [4]. In addition, the hydrophobic portion of LPS (lipid A), is a potent immunostimulant leading to production ofzytokines such as TNF-a, IL- 1p, and IL-6 [ I , 5 , 61. Left unchecked, the overproduction of these cytokines during severe Gram-negative infections can lead to shock [ I , 5 , 61. For this reason, lipid A is also refered to as endotoxin.

26.2 Structure of Lipopolysaccharide The lipopolysaccharide molecule can be divided into three distinct domains: the lipid A region, the core sugars, and the 0-antigen repeat. Figure 1B shows a schematic representation of the LPS molecule.

Figure 1. Molecular organization of the Gram-negative envelope and schematic structure of E. coli K- 12 lipopolysaccharide. A: The Gram-negative envelope of E. coli consists of two lipid bilayers, the inner and outer membranes. The inner membrane is composed of glycerophospholipids, primarily phosphatidylethanolamine and phosphatidylglycerol. The periplasmic space contains the peptidoglycan. The outer membrane is an assymetric bilayer. The inner leaflet is composed of phospholipids, and the outer leaflet is composed of lipopolysaccharide. B: This figure shows the sugars that make up the core and 0-antigen of E. coli LPS. Abbreviations are as follows: Kdo, 3-deoxy-~-manno-octu~osonic acid; Hep, L-gly'cero-u-mannoheptose; Glc, D-glucose; Gal, u-galactose; GlcNAc, N-acetyl-~-gIucosamine;Rha, L-rhamnose; GaK u-galactofuranose. The Rs indicate proposed substoichiometric modifications to the molecule as follows: R1, phosphate; R2, Kdo, rhamnose, or phosphoethanolamine; R3,phosphate or ethanolamine pyrophosphate; R4, heptose; R5,heptose or GlcNAc (adapted from [ 11).

Lipid A

Core Sugars

" = 1-10

0-Antigen Kcpeat

0

5. Q

33

b

e I .

0

26.3 Lipid A Biosyntlzesis in E. coli

437

Lipid A is the hydrophobic portion of LPS that anchors it to the outer membrane. Escherichia coli lipid A is a disaccharide of glucosamine that is phosphorylated at the 1 and 4’ positions and is acylated at the 2, 3, 2‘, and 3’ positions with R-3-hydroxymyristate (Figures 1B and 2). Two additional (secondary) fatty acyl chains are esterified to the 2’ and 3’ hydroxymyristoyl moieties to form the acyloxyacyl units that are characteristic of lipid A [I]. The core region of LPS consists of a non-repeating, branched oligosaccharide. Figure 1B shows the sugars that make up the core region in E. coli. A dissacharide of 3-deoxy-~-n?unno-octulosonicacid (Kdo) is linked to the 6’ position of lipid A. The rest of the core is attached to the inner Kdo of Kdoz-lipid A (Figure 1B). In E. coli, L-glycero-D-manno-heptose is linked to the inner Kdo but in Rhizobium leguminosarum and Legionellu pneumophilu. mannose, which is very similar in structure to L-glycero-D-manno-heptose, is attached to Kdo. The core plays an important role in maintaining an effective permeability barrier. E. coli mutants lacking heptose and distal core sugars do not correctly insert all of the outer membrane porins and are more susceptible to antibiotics [7].In other Gram-negative bacteria, such as Huemophilus injuenzue, the LPS core is hyper variable and important in the organism’s ability to evade immune detection [8, 91. In Rhizohium, disruption of core biosynthesis inhibits bacteriod formation in plant root nodules [lo]. The O-antigen repeat is the most highly variable portion of the LPS molecule among different E. coli strains [ l , 1 1 , 121. Over 160 different forms of O-antigen have been identified for E. coli [ 111. In E. c d i K12, the O-antigen consists of the five sugar unit shown in Figure 1 [7, 13, 141. This sugar motif is found in 1-40 repeats among LPS molecules of a single cell [ 151. The presence of the O-antigen is important in evasion of host defense mechanisms [ 151. E. coli lacking the O-antigen are more easily killed by complement [ 151. As noted above the entire LPS molecule is not required for growth of E. coli. The minimal portion that promotes growth under laboratory conditions is lipid A with at least one of its acyloxyacyl moieties (i.e. a penta-acylated species), and two Kdo sugars (see Figures 1 and 2). However, the core sugars and the O-antigen plays a role in virulence and other specialized functions, such as symbiosis in Rhizohia [4, 8, lo].

26.3 Lipid A Biosynthesis in E. coli

The biosynthetic pathway of lipid A in E. coli is well understood [ l ] . The nine enzymes required to synthesize Kdoz-lipid A have been identified, and the genes encoding them are known [ 1, 16, 171. Many of the enzymes have been purified to homogeneity allowing detailed biochemical analysis. Because lipid A is required for Gram-negative viability, many of the enzymes involved in its biosynthesis are potential targets for the development of novel anti-microbial agents [25, 301.

438

26 Escherichia coli Lipid A : A Potent Activator oflnnate Immunity

26.3 Lipid A Biosjnthrsis in E. coli

439

26.3.1 Acylation of UDP-GlcNAc LPS biosynthesis begins with the acylation of the sugar nucleotide UDP-Nacetylglucosamine (UDP-GlcNAc) with R-3-hydroxymyristate [ 1, 181. The UDPGlcNAc 3-O-acyltransferase, encoded by the lpxA gene, catalyzes this reaction. Hydroxymyristate activated as a thioester on acyl carrier protein (ACP) is transferred to the 3-OH of the glucosamine of UDP-GlcNAc (Figure 2). The reaction is specific for acyl-ACP. Acyl-CoA does not serve as an efficient acyl donor. The reaction is also highly specific for acyl chain length and the presence of the R-3hydroxyl group [ 1, 181. lnterestingly, LpxAs from different Gram-negative bacteria have different chain length specificities. For instance, LpxA from Pseudomonas azireginosa is selective for R-3-hydroxydecanoate, and LpxA from Neisseria meningitidis is selective for R-3-hydroxylaurate 119, 20, 211. E. coli LpxA has been purified to homogeneity, crystallized and its structure solved to 2.6 A resolution [22]. It has a unique protein fold known as a left handed parallel beta helix. LpxA is required for growth 1231. Mutants in 1pxA are temperature sensitive for growth and hypersensitive to certain antibiotics, like erythromycin, that are not usually very effective against E. coli 120, 231. This suggests that inhibitors of LpxA may make current antibiotic therapies more effective. With the high resolution structure of this enzyme, LpxA becomes a promising target for rational design of inhibitors. Following the first acylation of UDP-GlcNAc, the 2-acetyl group is removed by UDP-3-O-(R-3-hydroxymyristoyl)-GlcNAc deacetylase (Figure 2). The deacetylase is encoded by the essential lpxC gene [ 1, 241. Promising antibacterial agents have been identified which target the deacetylase [25]. These inhibitors are chiral hydroxamic acids with various hydrophobic or aromatic substituents. The mechanism by which they inhibit the deacetylase may be through the binding of an active site zinc [26]. These compounds are bactericidal against E. coli, Enterobacter, Klebsiella, and Proteus. However. Pseudomonas and Serratia were not affected [25]. This may due to the inability of the drug to enter these cells, or its rapid secretion or metabolism once inside the cell. Nonetheless, these inhibitors were able to protect mice from lethal challenges of Escheuichia coli and serve as excellent lead compounds for the development of more potent deacetylase inhibitors 1251. Treatment of E. coli cells with the deacetylase inhibitor leads to a ten-fold increase in LpxC protein in the cell [27]. Inhibition of lpxA also leads to an increase in LpxC specific activity 1281. This suggests that the deacetylase is a point of regulation in lipid A biosynthesis. The deacetylation of the UDP-3-O-(R-3-hydroxymyristoyl)GlcNAc is the first thermodynamically favorable reaction in lipid A biosynthesis [28]. The O-acylation reaction catalyzed by LpxA has an unfavorable equilibrium constant of -0.01 [28] whereas the deacetylation is strongly favored. This helps to commit the UDP-GlcNAc (also a substrate for peptidoglycan and O-antigen biosynthesis) and the R-3-hydroxymyristoyl ACP (also a substrate for fatty acid elongation) for use in the biosynthesis of lipid A 111. Regulation at this point in the pathway may maintain a balance between phospholipid, LPS and peptidoglycan biosynthesis.

440

26 Escherichia coli L@id A: A Potent Activator of Innate Immunity

After deacetylation at position 2, the acyl sugar nucleotide is N-acylated by the enzyme, UDP-3-O-(R-3-hydroxymyristoyl)-a-~-glucosamine N-acyltransferase, the product of the lpxD gene (Figure 2) [ l , 241. This gene has sequence homology to lpxA and is expected to have the same overall protein fold [22, 241. Like LpxA, this enzyme is required for growth, and is specific for acyl-ACP and R-3hydroxymyristate [ 11. 26.3.2 Disaccharide Formation UDP-diacylglucosamine (UDP-DAG), the product of the LpxD catalyzed reaction (Figure 2), is the substrate for a specific hydrolase that removes the UMP group to form 2,3-diacylglucosamine 1-phosphate (also known as lipid X) (Figure 2) [ 11. The gene encoding this enzyme has recently been identified as a open reading frame of previously unknown function and is now called lpxH [ 161. Preliminary work shows that the enzyme is specific for UDP-DAG and does not efficiently hydrolyze CDP[ 161. This estabdiacylglycerol [29] or UDP-3-O-(R-3-hydroxymyristoyl)-GlcNAc lishes that hydrolysis occurs after the second acylation reaction catalyzed by LpxD. The LpxH protein is required for growth, and like LpxA, LpxC, and LpxD is a potential target for novel antibacterial agents [ l , 16, 23, 25, 301. The next step in lipid A biosynthesis is disaccharide formation. lpxB encodes the disaccharide synthase [1, 311. This enzyme catalyzes the condensation of one molecule of UDP-DAG and one molecule of lipid X in a pl’-6 linkage to form tetraacyldisaccharide 1-phosphate (DS-1-P) (Figure 2) [31].The UDP-DAG becomes the distal (non-reducing) sugar and the lipid X becomes the proximal (reducing) sugar of lipid A (Figure 2). The disaccharide synthase will not condense two UDP-DAG or two lipid X molecules directly, consistent with the requirement for LpxH activity prior to the disaccharide synthase [32].A point mutation in lpxB (recently shown to be a Glu326 to Lys change) leads to the accumulation of high steady state levels of lipid X in cells of strain MN7, providing a useful source of both radioactive and non-radioactive lipid X [ 1, 32, 331. 26.3.3 Phosphorylation by the Lipid A 4’ Kinase After disaccharide formation, the lipid A 4’ kinase, catalyzes the transfer of the y-phosphate from ATP to the 4’ position of DS-1-P to form tetraacyldisaccharide 1,4’-his-phosphate (also known as lipid I V , ) (Figure 2) [34]. Incorporation of the 4’ phosphate is required for subsequent reactions in lipid A biosynthesis [35]. Also, lipid A-like molecules are maximally active as activators of signal transduction in mammalian cells when both the 1 and 4’ phosphates are present [13, 36, 371. The 4’ kinase activity was first identified in 1987 [34]. It has since proven useful in the preparation of (4’-32P)-lipidIVA, and (4’-32P)-Kdo2-lipidIVA for the study of late reactions in lipid A biosynthesis (see below) [34, 381. Such probes are also useful for the detection of lipid A binding proteins in mammalian cells [37, 391.

26.3 Lipid A Biosynthesis in E. coli

441

The gene encoding the lipid A 4’ kinase is lpxK [ 171. When overexpressed, LpxK is capable of phosphorylating dissacharides with 2, 3,4, or 6 acyl chains [ 171. LpxK has not been purified to homogeneity due to the membrane protein’s instability in detergents required to solubilize it, but its overexpression greatly increases the yield of (4’-”P)-lipid IVA precursors. ZpxK is an essential gene [40] that forms an operon with an essential upstream gene, msbA. MsbA may be involved in lipid A transport from the inner membrane to the outer membrane (see below) [41]. The co-expression of msbA and lpxK raises the interesting possibility that LpxK is also involved in thc function of MsbA in lipid A transport, perhaps as a separate subunit of the transport apparatus.

26.3.4 Kdo Addition and the Late Acyltransferases

Following incorporation of the 4’ phosphate, the Kdo transferase, encoded by the essential kdtA gene [42], adds two 3-deoxy-~-nzanno-octulosonic acid (Kdo) sugars from the activated sugar CMP-Kdo to the 6’ position of lipid IVA (Figure 2) [35]. The single polypeptide chain encoded by kdtA is responsible for the sequential addition of both sugars [35].First, an a2-6‘ linkage is made between the inner Kdo and lipid IV,. Next, an a2-4 linkage is formed between the two Kdo sugars. Kdo transferase homologs from different Gram-negative bacteria have slightly different specificities. The H. injuenzae Kdo transferase adds only one Kdo residue to lipid IVA [43] while the Chlamydia trachomatis Kdo transferase, encoded by yseA, adds 3 or 4 Kdo sugars to lipid IVA [44]. The Kdo transferase from E. coli has been cloned, overexpressed and purified [35]. The purified protein displays at least 1000-fold preference for substrates which contain a 4’ phosphate group, indicating the importance of the 4’ kinase in lipid A biosynthesis (351. htrB and msbB encode the late acyltransferases involved in the addition of laurate and myristate, respectively, in acyloxyacyl linkages to Kdo2-lipid IVA (Figure 2) [45-471. These acyl chains are esterified to the hydroxyl group of the 2’ and 3’ hydroxymyristates of Kdo2-lipid I V , . Like the early acyltransferases involved in the acylation of UDP-GlcNAc, HtrB and MsbB are specific for acyl-ACP [45]. HtrB is selective for lauroyl-ACP over decanoyl- or myristoyl-ACP [46]. Hydroxymyristoyl ACP is not an effective acyl donor 1461. HtrB requires both Kdo residues on the acceptor substrate, as neither Kdol-lipid IVA nor lipid IVA are very effective substrates for the reaction 1461. htvB is required for growth above 33 “ C [48]. Multicopy suppression analysis of the htrB- temperature sensitive phenotype led to the identification of msbB [49]. insbB is not required for growth and was identified as a potential late acyltransferase by its sequence homology to htrB 146, 47, 491. Biochemical characterization of msbB mutant and overexpressing strains established that the gene encoded the myristoyltransferase 1471. It is believed that laurate addition normally precedes myristate addition in vivo. MsbB efficiently acylates Kdo2-(lauroyl) lipid IV,, but Kdoz-lipid I V , can be acylated -100-1000 times more slowly (Figure 2) 1471.

442

26 Escherichia coli Lipid A: A Potent Activator of Innate Immunity

26.3.5 Other Acyltransferases In 1987 Brozek et al., characterized an enzyme found in membranes of E, coli that catalyzes the transfer of a palmitate moiety from the sn-1 position of glycerophospholipids, such as phosphatidylethanolamine or phosphatidylglycerol, to the hydroxyl of the N-linked acyl chain of lipid X [50]. This enzyme is specific for glycerophospholipids as acyl donors. Acyl-ACP and acyl-CoA are not substrates [50].The reaction is very specific for acyl chain length. No other fatty acyl chain can substitute for palmitate [50]. The palmitoyl transferase will utilize lipid X, lipid IVA, Kdoz-lipid IVA (Figure 3), or lipid A as acceptor substrates [50, 511. Palmitate addition to lipid A is a substoichiometric modification found in E. coli cells treated with ammonium metavanadate [52]. It can occur in addition to the acylations performed by HtrB and MsbB to produce a hepta-acylated lipid A. Other bacteria, such as Salmonella, have very active palmitoyl transferases even in the absence of ammonium metavanadate treatment and a higher proportion of their lipid A contains palmitate [ l , 531. The role of palmitate or heptaacylation in the function of LPS is unknown. The gene encoding the palmitoyl transferase has recently been identified in Salmonella as the PhoPQ-activated gene pagP [51, 54, 551. Overexpression of the E. coli homolog, crcA (now refered to as IpxY), leads to massive overproduction of palmitoyl transferase activity [5 11. Analysis of inner and outer membrane fractions for palmitoyl transferase activity shows that LpxY (Figure 3) is located in the outer membrane [51]. This is in contrast to the inner membrane localization of all other enzymes involved in lipid A biosynthesis. An additional late acyltransferase has also recently been identified. A gene originally designated ddg (now known as b x P ) is -60% identical to htrB suggesting it encodes a late acyltransferase [56].A role for this protein is suggested by the change in the fatty acid composition of lipid A in E. coli or Salmonella grown at cold temperatures [57, 581. When E. coli cells are grown at 12"C, the fraction of laurate in the lipid A drops from 16% to 5% while the fraction of palmitoleate increases from a trace amount to 10% [57]. This would suggest that at low temperatures an additional acyltransferase is induced that is specific for palmitoleate. Palmitoleoyl transferase activity is seen in extracts of cells grown at 12°C but not in extracts of cells grown at 30 "C (Figure 3) [59]. This activity is highly specific for palmitoleoylACP over palmitoyl-, myristoyl- or lauroyl-ACP [59]. Overexpression of ZpxP on a multicopy plasmid leads to overproduction of this cold-induced activity, indicating that lpxP encodes a cold-induced lipid A palmitoleoyl transferase [59]. The incorporation of an unsaturated acyl chain into the lipid A may help maintain outer membrane fluidity at cold temperatures. This finding opens an interesting avenue of study into the temperature regulation of lipopolysaccharide biosynthesis. 26.3.6 Transport of Lipid A and the Role of MsbA The enzymes required for the biosynthesis of Kdoz-lipid A and the addition of the core sugars are localized to the cytoplasm or the inner membrane of the cell [l].All of the substrates (e.g. acyl-ACP, sugar nucleotides) are also found in the cytoplasm

26.3 Lipid A Biosynthrsis in E. coli

443

-Lipid IVA IVA Kdo2-Lipid

phosphatidylethanolamine

2-monoacyl-glycerophospho. ethanolamine

Figure 3. Other regulated acylation reactions involved in E. coli lipid A biosynthesis under special conditions. The reactions catalyzed by LpxP (formerly known as Ddg) [59] and LpxY (also known as PagP or CrcA) [51] are shown. LpxP is a cold induced protein which incorporates an unsaturated 16 carbon fatty acyl chain in place of the laurate normally incorporated by HtrB. LpxY, the expression of which is activated by the PhoP/PhoQ system [51], is an outer membrane protein that incorporates palmitate to make hepta-acylated lipid A. LpxY is capable of acylating lipid X, lipid IVA, and lipid A, as well as Kdoz-lipid IVA (as shown). The physiological substrate is probably lipid A , given that LpxY is an outer membrane protein.

444

26 Escherichia coli Lipid A: A Potent Actiuutor of Innute Immunity

of the cell [ 11. The 0-antigen polymer is thought to be transferred to core-lipid A on the periplasmic face of the inner membrane [ 1, 601. There must be a mechanism for the transport of core-lipid A from the inner leaflet of the inner membrane to the periplasmic face where 0-antigen addition can occur [I]. Finally, the complete LPS molecule must move to outer leaflet of the outer membrane (Figure 4A). Little is known about this process in part because biochemical assays for transport are difficult and effective genetic screens for mutants in LPS transport have not been developed. However, a potential protein in this process has been identified through a screen for multicopy suppressors of the htrB- temperature sensitive phenotype. A gene called mshA was identified that has homology to mammalian Mdr proteins [61]. Proteins of this class are involved in the expulsion of hydrophobic drugs from cells and some members have been directly implicated in the secretion of phospholipids into the bile [62]. One could hypothesize that MsbA is involved in the transport of lipid A from the inner membrane to the outer membrane. Evidence to support the hypothesis that MsbA may be involved in lipid A transport is growing. htrB mutants accumulate tetraacylated lipid A in the inner membrane at the non-permissive temperature of 42 "C [41, 631. However, when msbA is provided in trans on a low copy plasmid, the tetraacylated lipid A is transported to the outer membrane [41]. In cells lacking msbA, hexaacylated lipid A accumulates in the inner membrane indicating that the transport of lipid A has been affected by the lack of mshA [41]. Taken together this strongly suggests that msbA plays a role in the movement of lipid A from the inner to the outer membrane, but direct biochemical assays remain to be developed.

26.4 Lipid A Activation of Signal Transduction in Animal Cells The lipid A moiety of LPS elicits rapid cellular responses from a number of immune responsive cells such as macrophages, lymphocytes, monocytes and endothelial v

Figure 4. Models of LPS transport from the inner to the outer membrane, and of lipid A activated signal transduction. A: The transport of LPS is divided into four steps [l]. In step 1, an 0-antigen unit is flipped across the inner membrane. Once exposed to the periplasmic space 0-antigen polymerization is thought to occur [ 1, 60, 811. RfbX is the proposed flippase required for this event [82]. Concurrently, core-lipid A transport occurs, perhaps facilitated by MsbA (step 2) [41, 631. The 0antigen repeat is transferred by RfaL to the core-lipid A molecule on the periplasmic surface of the inner membrane to make a complete LPS molecule. In step 3, LPS is proposed to movc to the inner surface of the outer membrane and then to be flipped across the outer membrane (step 4) [I]. B LPS may dissociate from the bacterium and either remain free in solution or be bound by an acute phase serum protein, called LPS binding protein (LPB). LBP can deliver LPS to lipid A responsive cells and mediate transfer into the plasma membrane or to CD14, a surface expressed glycoprotein [65, 83, 841. LPS is then proposed to interact with the putative lipid A signaling receptor (possibly TLR2/4) [72-741. In LBP-independent signaling LPS is proposed to bind directly to the lipid A receptor. Upon binding intracellular signaling is initiated. NF-KB is activated and moves to the nucelus to promote transcription of cytokines and interleukins. When CD14 is prescnt, an additional signal transduction cascade is initiated. MKK3 is thought to activate p38 by phosphorylation. p38, a MAP kinase homolog, can then activate the transcription factor MEF2C [ 8 5 ] .

445

26.4 Lipid A Activation .f Signal Transduction in Animal Cells

A Exterior Lipid A rrl.. ................

.................

[n\n -In-

Periolasm

Polymerase

04

1

Y?

\/

h

................

........ .............................

0L ohgosaccharide antigen 0

Mil

y;i

Inner membrane

--

MsbA7

.............

-.

Kdo

Outer core

........................

Cytoplasm

_..

446

26 Escherichia coli Lipid A: A Potent Activator of’lnnate Iinmunity

cells. Cellular responses include phagocytosis of bacteria, endocytosis of LPS, production of cytokines such as tumor necrosis factor-a, interleukin 1 (IL-1), and IL-6, production of leukotrienes, expression of surface IgM and arachidonic acid release [ l , 6, 13, 641. The overproduction of these inflammatory mediators can lead to septic shock, an acute physiological condition which can result in organ failure, catastrophic drop in blood pressure, and death [ 1, 5 , 131. Effective treatments for lipid A induced shock are not available, in part because the cellular signaling events induced by lipid A/LPS in mammalian cells are not fully understood. However, recent work has revealed some of the important signaling events that occur upon LPS activation. LPS can dissociate from the bacterial surface and may bind to a serum protein. termed LBP [65] (Figure 4B). LPS is then presented to animal cells as an LPS-LBP complex or, in the absence of LBP, as LPS monomers or aggregates. LPS-LBP complexes interact with a glycosyl-phosphatidyl-inositol-anchored glyco-protein called CD-14 [66, 671. CD-14 binds LPS and is then thought to interact with a lipid A receptor protein which initiates signal transduction inside the cell. LBP is also able to catalyze the transfer of LPS into membrane bilayers [68701. Signal transduction may be initiated through recognition of LPS patches in the plasma membrane. Recent data also shows that fluorescent LPS can move through the membrane into the cell, suggesting that signal transduction might be initiated via interaction with an intracellular lipid A receptor [71]. Even in the absence of LBP, LPS micelles or monomers are thought to initiate signaling by binding directly to the lipid A receptor. Recent evidence suggests that the Toll-like receptor proteins TLR2 and/or TLR4 might function as lipid A receptor proteins [72-741 and that LBP and CD-14 function as a lipid A delivery system. One of the earliest responses to LPS is the activation of a MAP kinase isoform, p38, by phosphorylation of threonine 180 and tyrosine 182 (within 5 minutes of LPS exposure) [75, 761. Interestingly, in the case of 70Z/3 mouse cells, p38 activation only occurs in cells which express CD14 and utilize LPS-LBP complexes. Cells which respond to LPS via direct binding of LPS micelles or monomers to the lipid A receptor do not activate p38 [77]. The best characterized response to LPS is the activation of the inducible transcription factor nuclear factor-Kl3 (NF-KB)NF-KB promotes the transcription of many LPS response genes such as the kappa genes in 70Z/3 cells and the interleukins or turmor necrosis factor in macrophages [78]. The overproduction of inflammatory mediators caused by LPS activation of mammalian cells is responsible for the complications of septic shock [ l , 6, 13, 301. One strategy to treat this condition focuses on the use of specific antagonists that block LPS signaling [30, 77, 79, 801. The properties of lipid A biosynthetic precursors and lipid A molecules from non-toxic Gram-negative bacteria have been of great interest. Synthetic analogs of lipid A based on the structures of naturally occurring non-toxic lipopolysaccharides were made and found to act as antagonists of LPS cell activation [79]. This strategy lead to the development of a Rhodobacter cupsulutus lipid A analog called E5531 [79]. This compound has shown great promise in clinical trials. It prevents the systemic response induced by a low dose endotoxin challenge in healthy volunteers [81, 821.

Rgferences

447

While our knowledge of lipid A activated cell signaling has increased dramatically over the last 15 years, we are still far from fully understanding the complex series of events initiated upon recognition of LPS by the cell. The verification of TLR2 and/or TLR4 as the lipid A signaling receptor(s) will serve as the starting point for the dissection of this complex signal transduction pathway. Studies showing the structural specificity of the LPS response for lipid A agonists versus antagonists [36] in TLR2/4 transfected cells would help define the protein as the true lipid A receptor.

26.5 Summary LPS is an important biomolecule. It is absolutely necessary, in some form, for the viability of E. coli and most other Gram-negative bacteria. Ongoing genomic sequencing projects demonstrate that the enzymes of lipid A biosynthesis (Figure 2) are encoded by relatively conversed single copy genes present in virutally all Gramnegative pathogens, further supporting the lipid A pathway as a target for new antibiotic discovery [30]. The lipid A biosynthetic pathway in E. coli is now fully elucidated. The nine enzymatic activities needed to synthesize Kdo2-lipid A from acyl-ACPs and activated sugar nucleotides in E. coli have been identified. The regulation of lipid A biosynthesis and why such a complex molecule is necessary for Gram-negative viability remain to be worked out. The lipid A portion of LPS is recognized by the innate immune system of animal systems as evidence of Gram-negative infection. Lipid A stimulates the production of diverse inflammatory mediators, especially by macrophages. Overproduction of these inflammatory mediators can lend to endotoxic shock during severe Gramnegative sepsis. Antibiotics that target lipid A biosynthesis and antagonist of lipid A triggered signal transduction [ 301 might provide useful new therapies for Gramnegative sepsis.

Acknowledgments This research was supported by NIH grants GM-51310 and GM-51796 to C.R.H.R. T.A.G. is supported by a special grant from the Mitzutani Foundation for the Glycosciences.

References 1. Raetz, C. R. H. (1996). Bacterial Lipopolysaccharides: a Remarkable Family of Bioactive Macroamphiphiles. In Esclierichia coli and Salmonella: Cellular and Molecular Biology, F. C. Neidhardt, ed. (Washington, D. C.: American Society for Microbiology), pp. 1035-1063. 2. Vaara, M. [ 1993). Antibiotic-supersusceptible mutants of Eschrricliia coli and Sulmonellu typliimurium. Antimicrob. Agents Chemother. 37, 2255-2260.

448

26 Escherichia coli Lipid A: A Potent Activator of Innate Immunity

3. Sen, K., and Nikaido, H. (1991). Trimerization of an in iiitro synthesized OmpF porin of Escherichiu coli outer membrane. J. Biol. Chem. 266, 11295-1 1300. 4. Stacey, G., Burris, R. H., and Evans, H. J. (1992). Biological Nitrogen Fixation (New York: Routledge, Chapman and Hall, Inc.), pp. 943. 5. Barron, R. L. (1993). Pathophysiology of septic shock and implications for therapy. Clin. Pharm. 12, 829-845. 6. Glauser, M. P., Heumann, D., Baugartner, J. D., and Cohen, J. (1994). Pathogenesis and potential strategies for prevention and treatment of septic shock: an update. Clin. Infect. Dis. 18, S203-216. 7. Schnaitman, C. A,, and Klena, J. D. (1993). Genetics of lipopolysaccharide biosynthesis in enteric bacteria. Microbiol. Rev. 57, 655-682. 8. Hood. D. W.) Deadman, M. E., Allen, T., Masoud, H.: Martin, A,, Brisson. J. R., Fleischmann, R., Venter, J. C., Richards, J. C., and Moxon, E. R. (1996). Use of the complete genome sequence information of Huemophilus influenzae strain Rd to investigate lipopolysaccharide biosynthesis. Mol. Microbiol. 22, 951-965. 9. Weiser, J. N., Love, J. M., and Moxon, E. R. (1989). The molecular mechanism of phase variation of H. injluenzae lipopolysaccharide. Cell 59, 657-665. 10. Kadrmas, J. L., Allaway, D., Studholme, R. E., Sullivan, J. T., Ronson, C. W., Poole, P. S., and Raetz, C. R. H. (1998). Cloning and overexpression of glycosyltransferases that generate the lipopoysaccharide core of Rhizobium leguminosarum. J. Biol. Chem. 273, 26432-26440. 11. Reeves, P. (1993). Evolution of Salmonella 0 antigen variation by interspecific gene transfer on a large scale. Trends Genet. 9, 17-22. 12. Whitfield, C., Amor, P. A,, and Koplin, R. (1997). Modulation of the surface architecture of Gram-negative bacteria by the action of surface po1ymer:lipid A-core ligase and by detrminants of polymer chain length. Mol. Microbiol. 23, 629-638. 13. Rietschel, E. T., Kirikae, T., Schade, F. U., Mamat, U., Schmidt, G., Loppnow, H., Ulmer, A. J., Zahringer, U., Seydel, U., Di Padova, F., Schreier, M., and Brade, H. (1994). Bacterial endotoxin: molecular relationships of structure to activity and function. FASEB J. 8, 217225. 14. Stevenson, G., Neal, B., Liu, D., Hobbs, M., Packer, N. H., Batley, M., Redmond, J. W., Lindquist, L., and Reeves, P. (1994). Structure of the 0-antigen of Eschericlziu coli K-12 and the sequence of its rfh gene cluster. J. Bacteriol. 176, 4144-4156. 15. Goldman, R. C., and Hunt, F. (1990). Mechanism of 0-antigen distribution in lipopolysaccharide. J. Bacteriol. 172, 5352 -5359. 16. Babinski, K. J., and Raetz, C. R. H. (1998). Identification of a gene encoding a novel Eschevichiu coli UDP-2,3-diacylglucosamine hydrolase. FASEB J. 12, ,41288. 17. Garrett, T. A,, Kadrmas, J. L., and Raetz, C. R. H. (1997). Identification of the Gene Encoding the Escherichiu coli Lipid A 4’-Kinase. Facile Phosphorylation of Endotoxin Analogs with Recombinant LpxK. J. Biol. Chem. 272, 21855-21864. 18. Anderson, M. S., and Raetz, C. R. H. (1987). Biosynthesis of lipid A precursors in Esckerichiu coli: a cytoplasmic acyltransferase that converts UDP-N-acetylglucosamine to UDP-3-0-(R-3hydroxymyristoy1)-N-acetylglucosamine.J. Biol. Chem. 262, 51 59- 5 169. 19. Dotson, G. D., Kaltashov, I. A , , Cotter, R. J., and Raetz, C. R. (1998). Expression cloning of a Pseudomonus gene encoding a hydroxydecanoyl-acyl carrier protein-dependent UDP-GlcNAc acyltransferase. J. Bacteriol. 180, 330-337. 20. Odegaard, T. J., Kaltashov, I. A . . Cotter, R. J., Steeghs, L., van der Ley, P., Khan, S., Maskell, D. J., and Raetz, C. R. H. (1997). Shortened hydroxyacyl chains on lipid A of Escherichiu coli cells expressing a foreign UDP-N-acetylglucosamine 0-acyltransferase. J. Biol. Chem. 272, 19688-19696. 21. Williamson, J. M., Anderson, M. S., and Raetz, C. R. H. (1991). Acyl-acyl carrier protein specificity of UDP-GlcNAc acyltransferases from Gram-negative bacteria: relationship to lipid A structure. J. Bacteriol. 173, 3591-3596. 22. Raetz, C. R. H., and Roderick, S. L. (1995). A left-handed parallel p helix in the structure of UDP-N-acetylglucosamine acyltransferase. Science 270, 997-1000. 23. Galloway, S. M., and Raetz, C. R. H. (1990). A mutant of Escherichiu coli defective in the first step of endotoxin biosynthesis. J. Biol. Chem. 265, 6394-6402.

Rgfevences

449

24. Kelly, T. M., Stachula, S. A., Raetz, C. R. H., and Anderson, M. S. (1993). The,firA gene of Escherichia coli encodes LJDP-3-O-(R-3-hydroxyniyristoyl)-a-D-glucosamine N-acyltransferase: the third step of endotoxin biosynthesis. J. Biol. Chem. 268, 19866- 19874. 25. Onishi, H. R., Pelak, B. A., Gerckens, L. S., Silver, L. L., Kahan, F. M., Chen, M. H.. Patchett, A. A., Galloway, S. M., Hyland, S. A., Anderson, M. S., and Raetz, C. R. H. (1996). Antibacterial agents that inhibit lipid A biosynthesis. Science 274, 980-982. 26. Jackman, J. E., Raetz, C. R. H., and Fierke, C. A. (1999). UDP-3-O-(R-3-Hydroxymyristoyl)N-acetylglucosamine deacetylase of Escherichia coli is a zinc metalloenzyme. Biochem., in press. 27. Sorensen, P. G., Lutkenhaus, J., Young, K., Eveland, S. S., Anderson, M . S., and Raetz, C. R. €1. (1996). Regulation of UDP-3-O-[R-3-hydroxymyristoyl]-N-acetyl-glucosamine deacetylase in Escherichiu coli. The second enzymatic step of lipid A biosynthesis. J. Biol. Chem. 271, 25898-25905. 28. Anderson, M. S., Bull, H. S., Galloway, S. M., Kelly. T. M., Mohan, S., Radika, K., and Raetz, C. R. H. (1993). UDP-N-acetylglucosamine acyltransferase of Escherichiu coli: the first step of endotoxin biosynthesis is thermodynamically unfavorable. J. Biol. Chem. 268, 19858-~19865. 29. Ray, B., and Raetz, C. R. H. Unpublished. 30. Wyckoff, T. J. O., Raetz, C. R . H., and Jackman, J. E. (1998). Anti-bacterial and antiinflammatory agents that target endotoxin. Trends Microbiol. 6 , 154- 159. 31. Ray, B. L., Painter, G., and Raetz, C. R. H. (1984). The biosynthesis of gramnegative endotoxin: formation of lipid A disaccharides from monosaccharide precursors in extracts of Escherichiu coli. S. Biol. Chem. 259, 4852--4859. 32. Radika, K., and Raetz, C. R . H. (1988). Purification and properties of lipid A disaccharide synthase of Escherichiu coli. J. Biol. Chem. 263, 14859-14867. 33. Raetz, C. R. H. (1992). Lipid A disaccharide synthase from Excherichia coli. Meth. Enzymol. 209, 455- 466. 34. Ray, B. L.. and Raetz, C. R. H. (1987). The biosynthesis of Gram-negative endotoxin: a novel kinase in Escherichiu coli membranes that incorporates the 4‘ phosphate of lipid A. J. Biol. Chem. 262, 1122-1 128. 35. Belunis, C. J., and Raetz, C. R. H. (1992). Biosynthesis of endotoxins: purification and catalytic properties of 3-deoxy-D-manno-octulosonic acid transferase from Escherichiu coli. J . Biol. Chem. 267, 9988-9997. 36. Golenbock, D. T., Hampton, R. Y., Qureshi, N., Takayama, K., and Raetz, C. R. 11. (1991). Lipid A-like molecules that antagonize the effects of endotoxins on human monocytes. J. Biol. Chem. 266, 19490-19498. 37. Hampton, R. Y . , and Raetz, C. R. H. (1991). Macrophage catabolism of lipid A is regulated by endotoxin stimulation. J. Biol. Chem. 266, 19499-1 9509. 38. Hampton, R. Y., and Raetz, C. R. H. (1992). Lipid A 4’-kinase of Escherichiu coli: enzyme assay and preparation of 4’-3?P-labeled probes of high specific radioactivity. Meth. Enzymol. 209, 466- 475. 39. Hampton, R . Y., Golenbock, D. T., and Raetz, C. R. H. (1988). Lipid A binding sites in membranes of macrophage tumor cells. J. Biol. Chem. 263, 14802-14807. 40. Garrett, T. A., Que, N. L. S., and Raetz, C. R. H. (1998). Accumulation of a lipid A precursor lacking the 4’-phosphate following inactivation of the Escherichiu cdi /pxK gene. J . Biol. <:hem. 273, 12457-12465. 41. Zhou, Z., White, K. A,, Polissi. A., Georgopoulos, C., and Raetz, C. R. H. (1998). Function of the E,schrrichiu coli msbA gene, an essential ABC family transporter, in lipid A and phospholipid biosynthesis. J. Biol. Chem 273, 12466 -12475. 42. Belunis, C. J., Clementz, T., Carty, S. M., and Raetz, C. R. H. (1995). Inhibition of lipopolysaccharide biosynthesis and cell growth following inactivation of the kdtA gene in Esc~herichiu coli. S . Biol. Chem. 270, 27646-27652. 43. White, K. A., Kaltashov, I. A., Cotter, R. J., and Raetz, C. R. H . (1997). A mono-functional 3deoxy-D-manno-octulosonic acid (Kdo) transferase and a Kdo kinase in extracts of Huemophilus irifluenzue. J. Biol. Chem. 272. 16555-63. 44. Belunis, C. S., Mdluli, K. E., Raetz, C. R. H., and Nano, F. E. (1992). A novel 3-deoxy-Dmanno-octulosonic acid transferase from Chlurnydiu trachoniutis required for expression of the genus-specific epitope. J. Biol. Chem. 267, 18702- 18707.

450

26 Escherichia coli Lipid A: A Potent Activator o j Innate Immunity

45. Brozek, K. A,, and Raetz, C. R. H. (1990). Biosynthesis of lipid A in Escherichia coli: acyl carrier protein-dependent incorporation of laurate and myristate. J. Biol. Chem. 265, 1541015417. 46. Clementz, T., Bednarski, J. J., and Raetz, C. R. H. (1996). Function of the htrB high temperature requirement gene of Escherichia coli in the acylation of lipid A. HtrB catalyzed incorporation of laurate. J. Biol. Chem. 271, 12095-12102. 47. Clementz, T., Zhou, Z., and Raetz, C. R. H. (1997). Function of the Escherichia coli msbB Gene, a Multicopy Suppressor of htrB Knockouts, in the Acylation of Lipid A. Acylation by MsbB follows laurate incorporation by HtrB. J. Biol. Chem. 272, 10353-10360. 48. Karow, M., Fayet, O., Cegielska, A,, Ziegelhoffer, T., and Georgopoulos, C. (1991). Isolation and characterization of the Escherichiu coli htrB gene, whose product is essential for bacterial viability above 33 "C in rich media. J. Bacteriol. 173, 741 -750. 49. Karow, M., and Georgopoulos, C. (1992). Isolation and characterization of the Escherichiu coli msbB gene, a multicopy suppressor of null mutations in the high-temperature requirement gene htrB. J. Bacteriol. 174, 702-710. 50. Brozek, K. A,, Bulawa, C. E., and Raetz, C. R. H. (1987). Biosynthesis of lipid A precursors in Escherichiu coli: A membrane bound enzyme that transfers a palmitoyl residue from a glycerophospholipid to lipid X. J. Biol. Chem. 262, 5170-5179. 51. Bishop, R. E., Gibbons, H. S., Miller, S. I., and Raetz, C. R. H. (1999). Transacylation between glycerophospholipids and lipid A in the outer membrane of Escherichiu coli. FASEB J, in press. 52. Zhou, Z., and Raetz, C. R. H. Unpublished. 53. Raetz, C. R. H., Purcell, S., Meyer, M. V., Qureshi, N., and Takayama, K. (1985). Isolation and characterization of eight lipid A precursors from a 3-deoxy-D-munno-octulosonic aciddeficient mutant of Salmonella typhimurium. J. Biol. Chem. 260, 16080-1 6088. 54. Guo, L., Lim, K. B., Gunn, J. S., Bainbridge, B., Darveau, R. P., Hackett, M., and Miller, S. I. (1997). Regulation of lipid A modifications by Sulmonellu typhimurium virulence genes phoPphoQ. Science 276, 250-253. 55. Guo, L., Lim, K. B., Poduje, C. M., Daniel, M., Gunn, J. S., Hackett, M., and Miller, S. I. (1998). Lipid A acylation and bacterial resistnce aganst vertebrate antimicrobial peptides. Cell 95, 189-198. 56. Carty, S. M., Sreekumar, K., and Raetz, C. R. H. (1997). A cold-induced, palmitoleoyl transferase involved in lipid A biosynthesis in Escherichia coli. FASEB J. 11. 57. Van Alphen, L., Lugtenberg, B., Rietschel, E. T., and Mombers, C. (1979). Architecture of the outer membrane of Escherichiu coli: phase transitions of the bacteriophage K3 receptor complex. Eur. J. Biochem. 101, 571-579. 58. Wollenweber, H.-W., Schlecht, S., Liideritz, O., and Rietschel, E. T. (1983). Fatty acid in lipopolysaccharides ofSalmonella species grown at low temperatures. Eur. J. Biochem. 130,167-171. 59. Carty, S. M., Sreekumar, K. R., and Raetz, C. R. H. (1999). Effect of cold shock on lipid A biosynthesis in Eschericiu coli. Induction at 12 "C of an acyltransferase specific for palmitoleoylACP. J. Biol. Chem., in press. 60. Mulford, C. A,, and Osborn, M. J. (1983). An intermediate step in translocation of lipopolysaccharide to the outer membrane of Sulmonellu typhimurium. Proc. Natl Acad. Sci. USA 80, 1159-1163. 61. Karow, M., and Georgopoulos, C. (1993). The essential Escherichia coli msbA gene, a multicopy suppressor of null mutations in the htrB gene, is related to the universally conserved family of ATP-dependent translocators. Mol. Microbiol. 7, 69-79. 62. Oude Elferink, R. P. J., Ottenhoff, R., van Wijkdnd, M., Smit, J. J. M., Schinkel, A. H., and Groen, A. K. (1995). Regulation of biliary lipid secretion by the mdr2 P-glycoprotein in the mouse. J. Clin. Invest. 95, 31-38. 63. Polissi, A., and Georgopoulos, C. (1996). Mutational analysis and properties of the mshA gene of Escherichia coli, coding for an essential ABC family transporter. Mol. Microbiol. 20, 1221-1233. 64. Tracey, K. J.: and Lowry, S. F. (1990). The role of cytokine mediators in septic shock. Adv. Sur. 23,21-56. 65. Grube, B. J., Cochane, C. G., Ye, R. D., Green, C. E., McPhail, M. E., Ulevitch, R. J., and Tobias, P. S. (1994). Lipopolysaccharide binding protein expression in primary human hepatocytes and HepG2 hepatoma cells. J. Biol. Chem. 269, 8477-8482.

66. Haziot, A., Chen, S., Ferrero, E., Low, M. G., Silber, R., and Goyert, S. M. (1988). The monocyte differentiation antigen, CD14, is anchored to the cell membrane by a phosphatidylinositol linkage. J. Immunol. 14, 547-552. 67. Haziot, A., Tsuberi, B., and Goyert, S. M. (1993). Neutrophil CD14: Biochemical properties and role in secretion of TNF-a in response to LPS. J. Immunol. 150, 5556. 68. Wurfel, M. M., and Wright, S. D. (1997). Lipopolysaccharide-binding protein and soluble CD14 transfer lipopolysaccharide to phospholipid bilayers. J . Immunol. 158, 3925-3934. 69. Yu, B., and Wright, S. D. (1996). Catalytic properties of lipopolysaccharide (LPS) binding protein. Transfer of LPS to soluble CD14. J. Biol. Chem. 271, 4100--4105. 70. Yu; B., Hailman, E., and Wright, S. D. (1997). Lipopolysaccharide binding protein and soluble CD14 catalyze exchange of phospholipids. J. Clin. Invest. 99, 315-324. 71. Detmers, P. A., Thieblemont, N., Vasselon, T., Pironkova, R., Miller, D. S., and Wright, S. D. (1996). Potential role of membrane internalization and vesicle fusion in adhesion of neutrophils in response to lipopolysaccharide and TNF. J. Immunol. 157, 5589-96. 72. Kirschning, C. J., Wesche, H., Merrill-Ayers, T., and Rothe, M. (1998). Human Toll-like receptor 2 confers responsiveness to bacterial lipopolysaccharide. J. Exp. Med. 188, 2091-2097. 73. Poltorak, A,, He, X., Smirnova, I., Liu, M. Y., Huffel, C. V., Du, X., Birdwell, D., Alejos, E., Silva, M., Galanos, C., Freudenberg, M., Ricciardi-Castagnoli, P., Layton, B., and Beutler, B. (1998). Defective LPS signaling in C3H/HeJ and C57BL/IOScCr mice: mutations in Tlr4 gene. Science 282, 2085-2088. 74. Yang, R. B., Mark, M. R., Gray, A,, Huang, A., Xie, M. H., Zhang, M.. Goddard, A., Wood, W. I., and Gurney, A. L. ( I 998). Toll-like receptor-2 mediates lipopolysaccharide-induced cellular signalling. Nature 395, 284-288. 75. Han, J., Lee, J. D., Bibbs, L., and Ulevitch, R. J. (1994). A MAP kinase targeted by endotoxin and hyperosmolarity in mammalian cells. Science 265, 808-81 1. 76. Raingeaud, J., Guptd, S., Rogers, J. S., Dicken, M., Han, J., Ulevitch, R . J., and Davis, R. J. (1995). Pro-inflammatory cytokines and environmental stress cause p38 mitogen-activated protein kinase activation by dual phosphorylation on tyrosine and threonine. J. Biol. Chem. 270, 7420-7426. 77. Garrett, T. A,, Rosser, M. F. N., and Raetz, C. R. H. (1999). Signal transduction triggered by lipid A-like molecules in 702/3 pre-B lymphocyte tumor cells. Biochem. Biophys. ACTA, in press. 78. Baldwin, A. S. (1996). The NF-KB and IKB proteins: New discoveries and insights. Ann. Rev. Immunol. 14, 649-68 1. 79. Christ, W. J., Asano, O., Robidoux, A. L., Perez, M., Wang, Y., Dubuc, G. R., Gavin, W. E., Hawkins, L. D., McGuinness. P. D., Mullarkey, M. A., Lewis, M. D., Kishi, Y., Kawata, T., Bristol. J. R.. Rose, J. R., Rossignol, D. P., Kobayashi, S., Hishinuma, I., Kimura, A,, Asakawa, N., Katayama, K., and Yamatsu, I. (1995). E5531, a pure endotoxin antagonist of extraordinary potency: chemistry and biology. Science 265, 80-83. 80. Takayama, K., Qureshi, N., Beutler, B., and Kirkland, T. N. (1989). Diphosphoryl lipid A from Rhodopseudomonas sphaeroides ATCC 17023 blocks induction of cachectin in macrophages by lipopolysaccharide. Infect. Immun. 57, 1336-1338. 81. Bunnell, E., Lynn, M., Parillo, J. E., Habet, K., Friedhoff, L. T., and Rogers, S. L. (1995). Effect of E5531 on systemic responses to endotoxin in healthy volunteers. Critical Care Med. (Suppl.) 23, A147. 82. Osborn, M. J. (1979). Biosynthesis and assembly of the lipopolysaccharide of the outer membrane. In Bacterial Outer Membranes, M. Inouye, ed. (New York: Wiley), pp. 15--34. 83. Bunnell, E., Neumann, A,, Lynn, M., Friedhoff, L. T., Rogers, S. L., Habet, K., and Parillo, J . E. (1995). E5531, an endotoxin antagonist, blocks the hyperdynamic and depressant cardiovascular effects of endotoxin in healthy subjects. Critical Care Med. (Suppl.) 23, A151. 84. Liu, D., Cole, R. A., and Reeves, P. R. (1996). J. Bacteriol. 178, 2102-2107. 85. Ulevitch, R. J., and Tobias, P. S. (1995). Receptor-dependent mechanisms of cell stiumulation by bacterial endotoxin. Annu. Rev. Immunol. 13, 437-457. 86. Wright, S. D., Ramos, R. A,, Tobias, P. S., Ulevitch, R. J., and Mathison, J. C. (1990). CD14: a receptor for complexes of lipopolysaccharide (LPS) and LPS binding protein. Science 249, 1431-1433. 87. Han, J., Jiang, Y., Li, Z., Kravchenko, V. V., and Ulevitch, R. J. (1997). Activation of the transcription factor MEF2C by the MAP kinase p38 in inflammation. Nature 386, 296-299.

Part I1

Volume 3

I1 G1y cosidases

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

27 Lysosomal Degradation of Glycolipids Thomas Kolter and Konrad Sandho#

27.1 Summary Glycosphingolipids are amphiphilic components of cellular membranes. In mammalian cells, their constitutive degradation occurs in the acidic compartments, the endosomes and the lysosomes. Here, glycosidases sequentially cleave off the terminal carbohydrate residues. For glycolipid substrates with short oligosaccharide chains, the additional presence of sphingolipid activator proteins is required. A considerable part of our current knowledge about glycolipid degradation is derived from a class of human diseases, the sphingolipidoses, which are caused by inherited defects within this pathway. Current research focusses on topology and mechanism of lysosomal glycolipid digestion; but also approaches to a better understanding and therapy of sphingolipidoses are undertaken.

27.2 Introduction Glycolipids are components of biological membranes and are composed of a carbohydrate moiety linked to a hydrophobic aglycon. Together with glycoproteins and glycosaminoglycans they contribute to the glycocalix on eukaryotic cell surfaces. According to the structure of the membrane anchor they can be divided into in glycosphingolipids, the major glycolipids in animals, and glycoglycerolipids that are abundant in bacteria and plants. Most animal glycolipids contain a hydrophobic ceramide building block and a hydrophilic, extracellular oligosaccharide chain [I]. Ceramide, a N-acylated D-erythro-sphingosine, is also component of sphingomyelin, a plasma membrane phospholipid. Variations in the type, number and linkage of sugar residues within the oligosaccharide chain lead to the wide range of naturally occurring glycosphingolipids (GSLs). They can be classified into series which are

456

27 Lysosomal Degradation of Glycolipids

characteristic for a group of evolutionary related organisms. E.g. sialic acid containing GSLs of the ganglio-series, the gangliosides, are abundant on neuronal cells of higher animals where they contribute to the function of the nervous system (compare Chapter 9). Beside the species dependence, GSLs form cell-type specific patterns which change with cell growth, differentiation, viral transformation, ontogenesis and oncogenesis (compare Chapter 47). GSLs form a protective layer on biological membranes protecting them from inappropriate degradation (see below). Their biosynthesis (Chapter 19) starts at intracellular membranes with the formation of ceramide from the amino acid L-serine and palmitoyl coenzyme A. In the Golgi-apparatus, carbohydrate residues are added to the growing glycolipid chain by the action of membrane-bound glycosyltransferases (compare Chapter 19). Within the course of exocytosis, GSLs are transported to the plasma membrane where they can interact with toxins, viruses, and bacteria [2] as well as with membrane bound receptors (lectins) and enzymes (compare Chapter 47). After a certain time, GSLs enter the cell and are transported to the acidic cellular compartments [ 31. Within this process, the anticytosolic orientation of the carbohydrate head groups is preserved. In addition to their function on the cell surface, animal glycolipids may serve as metabolic precursors for sphingolipids that contribute to the water permeability barrier of the skin [4]. Sphingolipids are also involved in the transduction of extracellular signals into the interior of cells [I]. In humans, inherited defects of glycosphingolipid and sphingolipid catabolism give rise to lysosoma1 storage diseases, the sphingolipidoses [ 51 (see also Chapter 57).

27.3 Mechanisms of Lysosomal Glycolipid Degradation The constitutive degradation of most cellular macromolecules takes place in the lysosomes. Cellular components can reach these organelles by endocytosis, autophagy, or, eventually, by direct transport from the cytosol and are cleaved by hydrolytic enzymes into their building blocks. Different hydrolases with acidic pH optima are involved in this process, e.g. proteases, glycosidases, lipases, phospholipases, nucleases, phosphatases, and sulfatases. The liberated building blocks are able to leave the lysosomes via diffusion or with the aid of specialized transport systems.to be utilized within salvage processes or to be further degraded. Within the course of membrane recycling, also glycolipids are endocytosed and traffic through the endosomal compartment to reach the lysosome for degradation. 27.3.1 Glycosidases Mammalian glycolipids are degraded within the acidic compartments, the endosomes and the lysosomes. Here, glycosidases sequentially cleave off the terminal carbohydrate residues from the nonreducing ends of glycolipids and also other glycoconjugates. Via lower glycosylated GSLs, monosaccharides, sialic acids, fatty acids, and sphingosine are produced. The hydrolytic enzymes involved in glycolipid

27.3 Mechanisms of Lysosomal Glycolipid Degrudution

457

degradation are proteins without transmembrane regions or GPI-anchors and, therefore, occur as soluble enzymes in the lumen of the lysosomes. Usually they consist of a single polypeptide chain, with the exception of p-hexosaminidases which are dimeric enzymes of two identical or different subunits. Also acid ceramidase, the enzyme that cleaves the membrane anchor, consists of two subunits albeit derived from a single precursor by proteolytic cleavage. Lysosomal enzymes are synthesized as prepropeptides which are post-translationally modified to the mature proteins, e.g. by proteolysis, glycosylation and glycan trimming. In contrast to many proteases, the propeptides of glycosidases can already show enzymatic activity. Lysosomal enzymes are usually targeted to the lysosomes by the mannose-6phosphate receptor although exceptions exist, e.g. in the case of glucocerebrosidase. The SAP-precursor protein is secreted within the secretory pathway and can be taken up by cells by at least three independent receptor mechanisms, the low density lipoprotein receptor-related protein (LRP), the mannose-6-phosphate receptor, and the mannose receptor [6]. Lysosomal glycosidases operate by general acid catalysis [ 71 and information on the active sites has been obtained by photoaffinity labeling, e.g. in the case of p-hexosaminidase B [S] or affinity labeling [9] in the case of glucocerebrosidase. In the mechanism of hexosaminidases, neighbor group participation of the acetamido group has to be considered [lo].

27.3.2 Topology of Endocytosis and Lysosomal Glycolipid Degradation Glycosphingolipids, together with other components of the plasma membrane, are transported to the lysosomal compartment by endocytosis. It is not clear whether components of the plasma membrane reach the lysosomes as part of the lysosomal membrane, which would be a direct consequence of endocytotic membrane-flow (Figure 1A). Lipids of the limiting lysosomal membrane are not or at least much slower degraded than plasma membrane-derived lipids entering the lysosomal compartment. The inner leaflet of the limiting membrane is covered with a glycocalix that is largely composed of lactosamine units [ l l ] which are thought to protect the perimeter membrane from the attack of the lysosomal hydrolases. A model of the topology of lysosomal lipid digestion (Figure 1B) suggests that parts of the endosomal membranes-obviously those enriched in components derived from the plasma membrane-invaginate and bud off into the endosomal lumen and thus form intra-endosomal vesicles. These vesicles, in turn, become intralysosomal vesicles or other intralysosomal membrane structures by successive processes of membrane fission and fusion along the endocytic pathway [12]. Accordingly, glycolipids originating from the outer leaflet of the plasma membrane face the lysosol on the outer leaflet of intralysosomal vesicles. In this model there is no glycocalix hindering the access of lysosomal proteins. It explains at least in part why membrane-derived lipids are degraded while the integrity of the limiting membrane is preserved. This model is supported by a series of observations [ 3 , 131 and should be taken into account when details of the lysosomal degradation machinery are investigated in vitro.

458

27 Lysosomal Degradation of Glycolipids

0

A

cz coated pit

Golgi apparatus

L

J

<-

f

G o b apparatus

Late endosome Primary

Transient fusion

q

GIycocaIix

Lysosomal membrane

Secondary lysosome

/ Lysosomal

Transient fusion and transfer

Glycocalix

membrane

Figure 1. Two models for the topology of endocytosis and lysosomal digestion of glycosphingolipids (GSLs) derived from the plasma membrane [3]. Conventional model (A): degradation of GSLs derived from the plasma membrane occurs selectively within the limiting lysosomal membrane. Alternative model for the topology of endocytosis and digestion of GSLs (B):during endocytosis glycolipids of the plasma membrane become incorporated into the membranes of intraendosomal vesicles (multivesicular bodies). The vesicles are transferred into the lysosomal compartment when the late endosome fuses with primary lysosomes. PM, plasma membrane; 0 , glycosphingolipid.

27.3.3 Sphingolipid Activator Proteins In contrast to the lysosomal degradation of soluble biomolecules like many proteins and oligosaccharides, the digestion of glycolipids with short carbohydrate head groups requires a three-component system composed of the glycolipid to be degraded, a hydrolytic enzyme-in most cases a glycosidase-and a sphingolipid activator protein. Glycolipids with glycan chains of sufficient length, e.g. the sialogangliotetraose moiety in ganglioside GM 1, can be degraded without the requirement for an activator-protein, at least in vivo. Presumably, the carbohydrate part of ganglioside GM1 extents far enough into the cytosol and is accessible by the watersoluble P-galactosidase. As illustrated by the degradation of ganglioside GM2, the product of the P-galactosidase-reaction, the branched tetrasaccharide moiety of this glycolipid requires the additional presence of an activator protein for degradation. The sphingolipid activator proteins known to date are encoded by just two genes: one carries the information for the GM2-activator [ 141 and the other for the socalled SAP-precursor or prosaposin, which is proteolytically processed to four ho-

27.3 Mechanisms of Lysosomul Glycolipid Degrudution

459

Figure 2. Model for the GM2-activator stimulated degradation of ganglioside GM2 by human 0hexosaminidase A (modified from [ 121). Water-soluble 0-hexosaminidase A does not degrade membrane-bound ganglioside GM2, which has a short carbohydrate chain, in the absence of GM2activator or appropriate detergents. The GM2-activator binds one molecule of ganglioside GM2 and lifts it out of the membrane. The activator-lipid complex can be recognized by water-soluble phexosaminidase A which cleaves the lipid substrate.

mologous proteins, the saposins or SAP-A, -B, -C, and -D (reviewed in [15]). The four SAPS have similar properties, but differ in their specificity and their mechanism of action. The mechanisms by which these membrane-active activator proteins can stimulate glycolipid hydrolysis can be reduced to two events: glycolipid presentation and/or activation of the hydrolytic enzyme. The GM2-activator is required for degradation of ganglioside GM2, the saposins SAP-A to SAP-D operate at different steps of the GSL catabolic pathway [ 151.

The GMZactivator and its role in lysosomal digestion The in vivo degradation of ganglioside GM2 requires the enzyme P-hexosaminidase A and a lysosomal ganglioside binding protein, the GM2 activator [16]. In contrast to the membrane-bound glycosyltransferases involved in GSL biosynthesis, phexosaminidase A is an enzyme resident in the lumen of the lysosomes, the lysosol (Figure 2). It cleaves ganglioside GM2 into GM3 and N-acetylgalactosamine. This glycosidase can cleave membrane-bound substrates only if they extend far enough into the aqueous phase. Its mode of action can be represented as a kind of razor blade sliding over a membrane surface. GSL-substrates with oligosaccharide head groups too short to be reached by the enzymes active site cannot be degraded. Their degradation requires the membrane-active GM2-activator as an additional factor. The membrane-active GM2 activator binds to the membrane and lifts a single GM2 molecule from a lipid aggregate or lipid bilayer and forms a water-soluble 1: 1 protein-lipid complex. P-Hexosaminidase A binds to this complex, cleaves the

460

27 Lysosomal Degradation of Glycolipids

GM2 molecule to yield ganglioside GM3, and releases the activator-lipid complex. The activator in turn may release GM3 into the membrane and bind another GM2 molecule. Formation of the ternary complex presumably involves also a protein-protein interaction between the GM2 activator and the glycosidase, Phexosaminidase A [ 171. SAP-A to SAP-D The first activator protein was discovered in 1964 as a protein required for the hydrolytic degradation of sulfatides by lysosomal arylsulfatase A [ 181. This sulfatideactivator or SAP-B (saposin B), binds GSLs, but with broader specificity than the GM2 activator. In vitro it behaves similarly to the GM2 activator in some aspects, i.e. it can present GSLs to water-soluble enzymes as substrates [ 191. Together with three other activator proteins, SAP-A, -C and -D, SAP-B is derived from a common precursor protein, pSAP or prosaposin. The four SAPs show homology to each other and have similar properties, but differ in their specificity and their mechanism of action. Functions are also attributed to the precursor protein, which is found in extracellular fluids (review: [15]). The physiological role of the SAPs A-D in glycolipid and sphingolipid degradation is only partially clarified to date. Most of our knowledge on this has emerged from studies on patients with atypical lipid storage diseases. While the inherited deficiency of SAP-B leads to an atypical form of metachromatic leukodystrophy (arylsulfatase A-deficiency) with accumulation of sulfatide and other sphingolipids [20], SAP-C deficiency causes an atypical form of Gaucher’s disease, where glucosylceramide accumulates (reviewed in [ 151). In one patient with a complete deficiency of the whole SAP-precursor protein due to a homoallelic mutation within the start codon [21] there is simultaneous storage of sphingolipids and GSLs with short oligosaccharide chains, including ceramide, glucosylceramide, lactosylceramide, ganglioside GM3, galactosylceramide, sulfatides, digalactosylceramide and globotriaosylceramide [22]. Cells from these patient have been used to elucidate the in vivo function of SAP-D [23]. When SAP-D is fed to the cells of the patient with pSAP-deficiency, ceramide accumulation is prevented. In contrast to the main effect of the GM2 activator and SAP-B, the mechanism of action of SAP-C involves also a direct activation of the enzyme glucosylceramide P-glucosidase [24].

27.3.4 Lateral Pressure According to the topological considerations, glycolipid degradation presumably occurs on the surface of intralysosomal vesicles. Besides the activator protein requirement in many cases, additional factors influence the properties of these degrading systems. Among them are the lateral pressure of the lipid bilayers, their lipid composition, and curvature. Physicochemical measurements at lipid monolayers revealed that the GM2-activator is a membrane-active lipid. It is able to insert into membranes as measured by an increase of the lateral pressure of monolayers composed of phosphatidylcholine and ganglioside GM2. Also microcalori-

27.3 Mechanisms of Lysosomal Glycolipid Degradation

46 1

metrical data indicate the GM2 activator to be a membrane-active protein. However, the membrane activity is only observed when the lateral pressure of the monolayer is below 25 mN/m [25]. The lateral pressure of biological membranes is significantly above this value. It can be excluded from these data that the G M 2 activator inserts into the limiting membrane of the lysosomes. Although there is no experimental basis for this hypothesis to date, it can be expected that the combination of size and special lipid composition of intralysosomal vesicles is sufficient to lower their lateral pressure beyond the value of 25 mN/m. At least in the case of the GM2 activator, this would explain why ganglioside G M 2 present on the membrane surface of intralysosomal vesicles can be presented to b-hexosaminidase A by the activator while the same ganglioside should be refractory to degradation when it is present on the inner leaflet of the lysosomal perimeter membrane.

27.3.5 Lipid Composition The acidic compartments of the cell are unique in their lipid composition. This should influence membrane structure and the properties of the lipid-degrading system. Especially negatively charged lysosomal lipids turned out to be important for the degradation of model substances in detergent-free liposomal systems. Bis(monoacylglycero)phosphate (BMP; Figure 3) is synthesized from phosphatidylglycerol (PG) within the acidic compartments [26] and therefore specifically found in the lysosomes and in endosomes [27]. PG required for BMP biosynthesis might be derived from digested mitochondria in the lumen of the lysosomes where BMP is generated on intralysosomal membrane structures. BMP stimulates the enzymatic hydrolysis of several sphingolipids like ceramide, sphingomyelin, and ganglioside G M 2 [ 551. Immunohistochemical techniques have localized BMP to intraendosoma1 structures in normal cultured fibroblasts [27]. In contrast to most other phospholipids, BMP contains an unusual snl, srz1'-configuration. It can be assumed that this renders the lipid less sensitive to phospholipases and therefore more stable within the hostile lysosomal environment. High concentrations of this lipid occur in pulmonary alveolar macrophages [26]. Also phosphatidylinositol and dolichol-

H,C-wP--O-CH,

1

HO-C-H

8

1

HO-C-H

I

HO HO &-CH~ OH

R-C--0-CH2

H.C-@C--R

8

8

BMP

NBu-DNJ

Figure 3. Left: Structure of s n f ,snf '-(Bismonoacylg1ycero)phosphate (BMP). The acyl side chains are derived from oleic acid. Acyl migration to the free hydroxyl groups can occur. Right: Structure of N-Butyldeoxynojirimycin, an inhibitor of glucosylceramide-synthase, which is currently under investigation for the treatment of sphingolipidoses within a substrate deprivation approach ( N Bu-DNJ).

462

27 Lysosomal Degradation o j Glycolipids

phosphate are found in the lysosomes, together with degradation products like fatty acids and dolichol. 27.3.6 Membrane Curvature The curvature of intralysosomal vesicles can be mimicked in vitro by experiments using unilamellar vesicles of a size comparable to those observed in vivo, i.e. with a diameter in the range 50-100 nm [32].

27.4 Degradation of Selected Lipids 27.4.1 Ganglioside GM2 Three different gene products are involved in the degradation of ganglioside GM2 [28, 491: the a-subunit of the dimeric enzymes P-hexosaminidases A (ap) and S (aa); the P subunit of P-hexosaminidases A ( a p ) and B (PP); and the GM2 activator protein. The hexosaminidases differ in their substrate specificity: Phexosaminidase A cleaves terminal P-glycosidically linked N-acetylglucosamine and N-acetylgalactosamine residues from negatively charged (using the active site on the a-subunit) and uncharged glycoconjugates, especially ganglioside GM2 which is negatively charged under physiological conditions. Uncharged substrates like glycolipid GA2 as well as oligosaccharides with terminal N-acetylhexosamine residues can also be cleaved by P-hexosaminidase B. P-Hexosaminidase A can cleave substrates of membrane surfaces only if they extend far enough into the aqueous phase. Its mode of action can be represented as a kind of razor blade sliding over a membrane surface. On the other hand, GSL-substrates with oligosaccharide head groups too short to be reached by the enzymes active site cannot be degraded. Their degradation requires a second component, the GM2-activator. The GM2 gangliosidoses [28] are a group of inherited disorders caused by intralysosomal accumulation of ganglioside GM2 and related glycolipids, particularly in neuronal cells. Usually the patients die in early childhood. Three forms of the disease are distinguished: the B-variant (a-chain deficiency) with its infantile form usually called Tay-Sachs disease, characterized by deficient activity of hexosaminidases A and S but normal Hex B; the 0-variant (P-chain deficiency, Sandhoff disease), resulting from deficient activity of both P-hexosaminidases A and B; and GM2 activator deficiency (AB variant), due to mutation of the GM2-activator gene and characterized by normal Hex A and Hex B. The gross pathology is very similar in B-variant, 0-variant, and GM2 activator deficiency, except that visceral organ involvement is evident in 0-variant. A complete loss of the activator function leads to a variant form of GM2 gangliosidosis with infantile onsets. This indicates that no significant GM2 degradation occurs in the absence of the activator (Figure 4). This is in contrast to the deficiency of other sphingolipid activator proteins.

I

NeuAc

YeuAc

~

Sphingosine

I

acid Ceramidase SAP-C and SAP-D

Ceramide

p-Galactosylceramidase SAP-A or SAP-C

gtbcpCsylceramidase

Glc(b1-1)-Cer Glucosylceramide I

1

Fabryl * a.~alactosldase A

SAP-B GalCer-0-galactosidase GMI -p-galactosidase SAP-B and SAP-C

Gal(f3-4)-Glc-Cer Lactosylceramide

/~aucherj

pizGzZ

Sialidase SAP-B

/Slaldoslsl

p-Hexosaminidase A B GM2-Activator

1

00,S-3-Gal-Cer Sulfatide

I

Gal(p1-I)-Cer Galactosylceramide

l~abryl ; z r t o s i d a s e

I

1

Gal(al-3)-Gal-Cer Digalactosylceramide

Gal(al4)-Gal-Glc-Cer Globotriaosylceramide

GalNAc(p1-3)-Gal-Gal-Glc-Cei Globoside

A

a-N-Acetylgalactosaminidase

Forssman lipid (not in humans)

(ul -3)GalNAc-GalNAc-Gal-GaCGlc-Cer

protein.

Figure 4. Degradation of selected sphingolipids in the lysosomes of the cells [49]. The eponyms of individual inherited diseases (in frame) are given. Those activator proteins which are required for the respective degradation step in vivo are indicated. Variant AB, AB variant of GM2 gdngliosidosis (deficiency of GM2-activator protein); SAP, Sphingolipid-activator

Sphingomyelin

GM3

(a23)-NeuAc-Gal(pl4)-Glc-Cer

J

GMI-0-galactosidase

GalNAc(p1-4)-Gal-Glc-Cei GA2

/GMI-GangliosidosisI

Gal(p1-3)GalNAc-Gal-Glc-Cer GAI

Sphingomyelinase

GMI-p-galactosldase

GalNAc(pl-4)-Gal-Glc-Cer GM2

GMI-Gangliosidosis

Gal(p1-3)GalNAc-GaCGlc-Cer GMI

E

P

&

Fa

e

h

-?A

Y

b

464

27 Lysosomal Degradation of Glycolipids

27.4.2 Lactosylceramide

Two P-galactosidases, GM 1-P-galactosidase and galactocerebrosidase, are involved in the lysosomal degradation of lactosylceramide. In addition, two activator proteins, SAP-B and SAP-C, contribute to this process. SAP-B predominantly accelerates the reaction catalyzed by GM 1-P-galactosidase while SAP-C preferentially stimulates lactosylceramide hydrolysis by galactocerebrosidase [29]. Kinetic and dilution experiments indicated that SAP-B forms a water-soluble complex with lactosylceramide that is recognized by the enzyme as optimal substrate. Dilution experiments indicated that degradation of lactosylceramide by galactocerebrosidase proceeds almost exclusively on liposomal surfaces. Presumably because of the redundant degradation system, no human disease is known which is due to the isolated accumulation of lactosylceramide. 27.4.3 Glucosylceramide Glucosylceramide hydrolysis is catalyzed by a lysosomal acid P-glucosidase (glucocerebrosidase) [30]. It can be stimulated by a sphingolipid activator protein, SAP-C [24], or by negatively charged phospholipids in vitro [31]. A detailed analysis that mimicked lysosomal GlcCer degradation very closely revealed that a membrane curvature similar to that of intralysosomal vesicles, the presumable site of glycolipid degradation, is not sufficient for the degradation of glucosylceramide by glucocerebrosidase, even in the presence of SAP-C. The presence of negatively charged lipids, that are actually present in the intraendosomal and intralysosomal membranes, e.g. bis(monoacylglycero)phosphate, is required for degradation; high concentrations lead to degradation of glucosylceramide even in the absence of SAP-C [32]. Apparently, the combination of appropriate curvature, lipid composition, and lateral pressure can replace most of the action of the activator protein. Inherited glucocerebrosidase deficiency leads to glucosylceramide accumulation and Gaucher disease 1301. Their most abundant form, the adult type I, is characterized by enlargement of liver and spleen; the nervous system is not affected. Complete enzyme deficiency, however, is lethal for human patients and genetically engineered model animals. Although the enzyme activity is reduced in all cells of the adult type I, the phenotype is predominantly expressed in macrophages of the reticuloendothelial system. Due to phagocytosis of erythrocytes, these cells have do degrade large amounts of glycolipids which in turn explains, why defects in GlcCer degradation primarily affects these cells. Two cases of Gaucher disease are known that are not caused by glucocerebrosidase deficiency, but by the absence of a sphingolipid activator protein, SAP-C [ 151. The adult form of Gaucher disease (type I) is currently the only sphingolipid storage disease for which a causal therapy is available [33]. The patients are treated with a modified glucocerebrosidase from human placenta or a recombinant sample. The protein carbohydrates contain the targeting information for the mannose receptor on macrophages.

27.4 Degradation of Selected Lipids

465

27.4.4 Ceramide In the lysosomes, ceramide is cleaved by acid ceramidase into sphingosine and fatty acid [34]. Acid ceramidase needs the assistance of SAP-D [23] or SAP-C [56] for the in vivo degradation of ceramide. Inheritable acid ceramidase deficiency causes the accumulation of ceramide in different tissues in Farber disease [35]. Because of the prevalent symptom, the occurrence of subcutaneous nodules with the accumulation of lipid-loaded macrophages, it is also called lipogranulomatosis. A subtype of the disease, Type 7, is due to a deficiency of the SAP-precursor protein. Therefore, the activator proteins required for ceramide degradation, SAP-D and also SAP-C are missing [21]. As the first molecular defect in the gene of a Farber patient a homoallelic point mutation leading to a Thr222Lys transversion in the P-subunit of the enzyme has been identified [34].

27.4.5 Sphingomyelin Within the lysosome, sphingomyelin is degraded by acid sphingomyelinase [ 361. The enzyme shows a modular structure including a SAP-like domain [37]. In vitro results indicate that acid sphingomyelinase can be stimulated by sphingolipid activator proteins, but this stimulation is not necessary for the in vivo degradation of sphingomyelin by this enzyme [22]. Acid sphingomyelinase deficiency leads to Niemann-Pick disease which is characterized by enlargement of liver and spleen and the occurrence of characteristic storage cells in the bone marrow [38]. In patients of the severe form of the disease (type A), also the function of the nervous system is impaired. Since the identity of the sphingomyelinases involved in the signal transduction process through the sphingomyelin cycle is ambiguous, cells from Niemann-Pick patients are of interest to clarify this process.

27.4.6 Sulfatide Sulfatide, which is particularly abundant in the white matter of the brain, is degraded by arylsulfatase A. This enzyme needs the assistance of a protein cofactor, SAP-B [ 181. Deficiency of this enzyme leads to metachromatic leukodystrophy, a disease which is due to sulfatide accumulation in various organs and which primarily affects the white matter of the brain [39]. Also the inherited deficiency of SAP-B leads to a lysosomal storage disease which resembles metachromatic leukodystrophy. However unlike typical metachromatic leukodystrophy not only sulfatide but also additional glycolipids, e.g. globotriaosylceramide, accumulate due to defects in several points of the pathway of GSL degradation [20]. In another disease, multiple sulfatase deficiency (MSD) or Austin disease, the activities of all known sulfatases are strongly reduced. The phenotype of the patients is clinically characterized as a combination of symptoms of metachromatic

466

2 7 Lysosomal Degradation of Glycolipids

leukodystrophy and a mucopolysaccharidosis. The underlying defect has been identified as an erroneous posttranslational modification which appears to be necessary for sulfate ester hydrolysis. Transformation of a cysteine into a fonnylglycine residue is deficient in cells from MSD patients [40].This modification is required for the catalytic activity of the sulfatases. 27.4.7 Galactosylceramide

Galactosylceramide is degraded by the enzyme galactocerebrosidase. The reaction can be stimulated by SAP-C and also SAP-A in living cells [41]. Inherited galactocerebrosidase deficiency leads to Krabbe disease in which the white matter of central and peripheral nerves is primarily affected [42]. In contrast to other sphingolipidoses, the enzyme deficiency is not accompanied by substrate accumulation due to the rapid destruction of the galactosylceramide synthesizing cells, the myelinating oligodendrocytes. This can be attributed to the action of accumulating galactosylsphingosine, a toxic metabolite.

27.5 Pathobiochemistry Inherited defects within the pathway of glycolipid degradation give rise to a class of human diseases, the sphingolipidoses (see also Chapter 57). With the exception of Fabry disease [43] they exhibit an autosomal recessive mode of inheritance. The majority of enzymes and cofactors deficient in the sphingolipidoses have been characterized at the nucleic acid and the protein level. Many mutations have been identified [44] and animal models of most sphingolipidoses have been created by targeted disruption of the respective genes in mice [48, 491. The symptoms and courses of these diseases are varying widely between forms with onset in early childhood and death within the first years of life on the one hand, and chronic forms on the other hand. The brain is affected in many of these disorders, the function of the skin or visceral organs can be impaired in others. Their pathogenesis is poorly understood and is largely governed by two factors [ 5 ] . Due to the cell type specific expression of GSL, lipid storage in lipidosis patients is observed especially in those cells and organs, in which the lipid substrates of the corresponding deficient enzyme are prevalently synthesized or by which they are taken up by phagocytosis. The other factor is the residual enzymatic activity of the gene product in the lysosomes [45]. A complete deficiency of a lysosomal enzyme leads to an early onset and a severe course of the disease. Many mutations, however, only lead to a partial loss of enzymatic activity. A residual activity in the lysosomes of only a few per cent can be sufficient to delay the onset of the disease and cause a mild course of the disease. Only if the residual enzyme activity decreases beyond a critical value, an accumulation of nondegradable enzyme substrate is observed. This has been quantitatively correlated within the so called threshold theory [46] which has been confirmed for

27.5 Puthobiochemistry

467

most sphingolipidoses (reviewed in [ 5 ] ) .Also toxic substances like psychosine in Krabbe disease (see below) or morphological active compounds may accumulate in sphingolipidoses and contribute to their pathogenesis. Not only the deficiency of a hydrolytic enzyme, but also of other proteins required for sphingolipid degradation can cause a sphingolipid storage disease. Besides deficiencies of activator proteins, this is the case in galactosialidosis. This disease is characterized by the secondary deficiency of b-galactosidase and sialidase activity. The primary defect is due to mutations within the protective protein, which forms a stable complex with the G M 1-b-galactosidase and the lysosomal sialidase [47].

27.5.1 Animal Models for Sphingolipidoses Animal models are valuable means for the study of pathogenesis and approaches towards therapy of human diseases. Besides naturally occurring animal models with defects in sphingolipid catabolism, genetically engineered mice models for several sphingolipidoses have been developed in the recent past by targeted disruption of the respective genes in mice [48]. In animal models of the GM2 gangliosidoses [49], differences in the sphingolipid degradation pathways between mice and human have been discovered and a further physiological function of a group of degrading enzymes, the hexosaminidases, has been elucidated. Also an alternative approach towards therapy has been established in the animal model of one of these diseases [50]. The animal model of Tay-Sachs disease was generated by targeted disruption of the gene of the a-chains of 0-hexosaminidase A in murine embryonic stem cells [511. While in the phenotype, both variants of GM2 gangliosidoses mentioned above are only slightly different in human, the animal models show drastic differences in severity and course of the disease. Mice with deficient P-hexosaminidase A (the Tay-Sachs mice) are phenotypically almost normal. They show a mild accumulation of GM2 in the central nervous system, but do not express the neurological symptoms characteristic for the human case of Tay-Sachs disease. In contrast, mice with deficient b-hexosaminidase B (Sandhoff mice) develop severe neurological disorders, the life span of the animals is strongly reduced. The reason for this is the specificity of sialidase which is different between mouse and human [52] (Figure 5). Mouse sialidase accepts GM2 as a substrate and converts it slowly to GA2. In the human case, this metabolic pathway is of no significance. GA2 can be degraded by the still intact P-hexosaminidase B, so that in spite of a complete loss of bhexosaminidase A the metabolic barrier is in part circumvented. Only the loss of both isoenzymes, hexosaminidases A and B. leads to the symptomatology corresponding to the human Sandhoff disease. Although the mouse sialidase can still convert GM2 to GA2, GA2 cannot be further degraded, since the responsible enzyme, P-hexosaminidase B, is also deficient. Beside their properties as tools for investigation and treatment of human diseases, animal models are also of interest for more detailed insights into metabolic pathways. Besides the animals deficient in biosynthetic enzymes mentioned in the introduction, this is also the case in a double knock out mouse lacking the genes for both the u- and b-chain of b-hexosaminidase. These mice have been developed without

468

27 Lysosomal Degradation of Glycolipids r.

-

b

'0

P

474

28 Lysosomul Degradution qf Glycoprotrins

regulated in both time and space by active turnover. Cells in response to changing physiological demands adjust the required ‘steady state’ concentration of a specific glycoprotein by counterbalancing the rates of its synthesis and degradation. Gene transcription, together with ribosomal synthesis of its polypeptide and co- and posttranslational glycosylation and processing, raise the level of a specific glycoprotein, while proteolytic enzymes joined by glycosidases lower its concentration through hydrolysis. Historically, the intricacies of macromolecular biosynthesis have been more attractive to scientists as an area of experimental study than has been the catabolic process (e.g. see Part 1). It is important to note, however, that metabolic degradation of a large molecule also can be regulated with precision in order to form crucial metabolites. In this chapter, for example, we will focus on degradation of the glycoprotein prohormone thyroglobulin which leads to the controlled production by the thyroid of the small-molecule hormones T3 and T4 [lL4]. Similarly, the biosynthesis of carnitine, an essential metabolite for fatty acid oxidation, requires proteolytic release of trimethyllysines from proteins and/or glycoproteins [ 5-61, and the vital immunological system of processing exogenous protein/glycoprotein polypeptides as antigens for presentation on class I1 MHC molecules involves endosome/lysosome proteolysis [ 7, 81.

28.3 Roles of Lysosomes The major intracellular location of glycoprotein digestion during their turnover is the lysosome, the organelle that possesses a complete set of the necessary hydrolytic enzymes for complete breakdown to amino acids and monosaccharides [9, 101. The degradative process is substantially regulated by segregation of the hydrolytic enzymes in this compartment of the cell. Most glycoproteins enter the lysosome via various endocytic mechanisms and pathways, and may include extracellular (heterophagy) and intracellular (autophagy) sources of the substrates. Lysosomes are bounded by a membrane which not only can undergo the specific endocytic fusion events for substrate uptake [ 111, but also encloses a proper hydrolytic reaction environment that includes an acidic pH and a high concentration of a complete set of proteases and glycosidases. Estimates of the concentrations of these hydrolases in the lysosomes are in the millimolar range [ 121. A very important third feature of the lysosomal membrane is its very narrow specificity in terms of the products that can be released after degradation. Only single amino acids and sugars can be transported out of these organelles. This size limit defines the molecular cause of the many lysosomal storage diseases in which partially digested fragments as small as disaccharides cause extreme pathology in humans whose lysosomes have deficient hydrolytic capacity (see Chapter 57). Most lysosomal storage disorders result from genetic deficiency of an individual glycosidase, but sometimes a deficient membrane transporter of the lysosomes is responsible for pathological accumulation of a normal amino acid or monosaccharide final product, such as cystine in cystinosis [I41 and sialic acid in Salla disease [ 131. Over 20 lysosomal storage diseases involving faulty glycoprotein or glycolipid catabolism are known [ 151.

410

27 Lysosomul Degrudution o j Glycol@&

[54] and may also become valuable tools for the treatment of gangliosidoses and related diseases.

27.6 Future Directions The selective degradation of glycolipids within the lysosome is far from clear. Morphological studies as well as the development of appropriate biochemical approaches for the investigation and reconstitution of intralysosomal membrane structures are necessary for the understanding of this process. Animal models of sphingolipidoses appear to be valuable tools for the study of defects within the glycolipid degradation pathway and for the development of novel therapeutic strategies.

References 1. 2. 3. 4. 5. 6. 7. 8. 9.

10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21.

L. Riboni, P. Viani, R. Bassi, A. Prinetti, G. Tettamanti, Proy. Lipid Res. 1997, 36, 153-195. K. A. Karlsson, Annu. Rev. Biochem. 1989, 58, 309-350. K. Sandhoff, T. Kolter, Trends Cell Biol. 1996, 6, 98-103. P. W. Wertz, B. van den Bergh, Chem. Phys. Lipids 1998, 91, 85-96. T. Kolter, K. Sandhoff, Brain Path. 1998, 8, 79-100. T. Hiesberger, S. Huttler, A. Rohlmann, W. Schneider, K. Sandhoff, J. Herz, EMBO J. 1998, I?, 4617-4625. G. Legler in: Y. Chapleur (Hrsg.), Carbohydrate Mimics, Wiley-VCH, Weinheim 1998, 463490. B. Liessem, G. J. Glombitza, F. Knoll, J. Lehmann, J. Kellermann, F. Lottspeich, K. Sandhoff, J. Bid. Chem. 1995,270, 23693-23699. S. Miao, J. D. McCarter, M. E. Grace, G. A. Grabowski, R. Aebersold, S. G. Withers, J. B i d Chem. 1994,260, 10975-10978. S. Knapp, D. Vocadlo, Z. Gao, B. Kirk, J. Lou, S. G. Withers, J. Am. Chem. Sue. 1996, 118, 6804-6805. C. Peters, K. von Figura, FEBS Lett. 1994, 346, 108-1 14. W. Furst, K. Sandhoff, Biochim. Biophys. Acta 1992, 1126, 1-16. J. K. Burkhardt, S. Huttler, A. Klein, W. Mobius, A. Habermann, G. Griffiths, K. Sandhoff, Eur. J. Biochem. 1997, 73, 10- 18. H. Klima, A. Tanaka, D. Schnabel, T. Nakano, M. Schroder, K. Suzuki, K. Sandhoff, FEBS Lett. 1991, 289, 260-264. K. Sandhoff, T. Kolter, K. Harzer, in C. Scriver, A. L. Beaudet, W. S. Sly, D. Valle (Eds.), The Metabolic and Molecular Bases of Inherited Disease, 8th Edition, David McCraw Hill, New York, in press. E. M. Meier, G. Schwarzmann, W. Furst, K. Sandhoff, J. Biol. Chem. 1991,266, 1879-1887. H. J . Kytzia, K. Sandhoff, J. Biol. Chem. 1985, 260, 7568-7572. E. Mehl, H. Jatzkewitz, Hoppe-Seyler’s Z. Physiol. Chem. 1964,339, 260- 276. S. C. Li, S. Sonnino, G. Tettamanti, Y. T. Li, J. Biol. Chem. 1988, 263, 6588--65Y1. W. Schlote, K. Harzer, B. C. Paton, B. Kustermann-Kuhn, B. Schmid, J. Seeger, U. Beudt, I. Schuster, U. Langenbeck, Eur. J. Pediatr. 1991, 150, 584-591. D. Schnabel, M. Schroder, W. Furst, A. Klein, R. Hunvitz, T. Zenk, J. Weber, K. Harzer, B. C. Paton, A. Poulos, K. Suzuki, K. Sandhoff, J. Biol. Chem. 1992,267; 3312-3315.

References

47 1

22. V. Bradova, F. Smid, B. Ulrich-Bott, W. Roggendorf, B. C. Paton, K. Harzer, Hum. Genet. 1993, 92, 143-152. 23. A. Klein, M. Henseler, C. Klein, K. Suzuki, K. Harzer, K. Sandhoff, Biochem. Biophys. Rrs. Commun. 1994, 200, 1440-1448. 24. M. W. Ho, J. S. O’Brien, Proc. Nut1 Acud. Sci. USA 1971, 68, 2810-2813. 25. A. Giehl, T. Lemm, D. Bartelsen, K. Sandhoff, A. Blume, Eur. J. Biochetiz. 1999,261, 650-658. 26. B. Amidon, A. Brown, M. Waite, Biochemistry 1996, 35, 13995-124002. 27. T. Kobayashi, E. Stang, K. S. Fang, P. de Moerloose, R. G. Parton, J. Gruenberg, Nuture 1998, 392, 193-197. 28. R. A. Gravel, J. T. R. Clarke, M. M. Kaback, D. Mahuran, K. Sandhoff. K. Suzuki in C. Scriver, A. L. Beaudet, W. S. Sly, D. Valle (Eds.), The Metabolic and Molecular Basis of Inherited Disease, 1995, 7th Edition, David Mc Graw Hill, New York, TI, Chapter 92, 28392879. 29. A. Zschoche, W. Fiirst, G. Schwarzmann, K . Sandhoff, Eur. J. Biochem. 1994, 222, 83-90. 30. E. Beutler, G. A. Grabowski, in C. Scriver, A. L. Beaudet, W. S. Sly, D. Valle (Eds.), The Metabolic and Molecular Basis of Inherited Disease, 1995, 7th Edition, David McGraw Hill, New York, 11, Chapter 86, 2641-2670. 31. A . M. Vaccaro, M. Tatti, F. Ciaffoni, R. Salvioli, A. Barca, C. Scerch, J. Riol. Chenz. 1997, 272, 16862-16867. 32. G. Wilkening, T. Linke, K. Sandhoff, J. Biol. Chem. 1998, 273, 30271-30278. 33. N. W. Barton, R. 0. Brady, J. M. Dambrosia, A. M. DiBisceglie, S. H. Doppelt, S. C. Hill, H. J. Mankin, G. J. Murray, R. I. Parker, C. E. Argoff, R. P. Grewal, K. T. Yu, Neiv. Enql. J. Med. 1991,324, 1464-1470. 34. J . Koch, S. Giirtner, C. M. Li, L. E. Quintern, K. Bernardo, 0. Levran, D. Schnabel, R. J. Desnick, E. H. Schuchman, K. Sandhoff, J. Bid. Chem. 1996,271, 331 10- 331 IS. 35. H. W. Moser in C. Scriver, A. L. Beaudet, W. S. Sly, D. Valle (Eds.), The Metabolic and Molecular Basis of Inherited Disease, 1995, 7th Edition, David McGraw Hill, New York, 11, 1995, 2589-2599. 36. L. E. Quintern, E. H. Schuchmann, 0. Levran, M. Suchi, K. Ferlinz, H. Reinke, K. Sandhoff, R. J. Desnick, EMBO J. 1989, 8, 2469-2473. 37. R. S. Munford, P. 0. Sheppard, P. J. O’Hara, J. Lipid Res. 1995, 36, 1653-1663. 38. E. H. Schuchman, R. J. Desnick in C. Scriver, A . L. Beaudet, W. S. Sly, D. Valle (Eds.), The Metabolic and Molecular Basis of Inherited Disease, 1995, 7th Edition, David McGraw Hill, New York, 11, Chapter 84, 2601-2624. 39. E. H. Kolodny, A. L. Fluharty in C. Scriver, A. L. Beaudet, W. S. Sly, D. Valle (Eds.), The Metabolic and Molecular Basis of Inherited Disease, 1995, 7th Edition, David McGraw Hill, New York, 11, Chapter 88. 2693-2739. 40. B. Schmidt, T. Selmer, A. Ingendoh, K. von Figura, Cell 1995, 82, 271L278. 41. K. Harzer. B. C. Paton, H . Christomanou, M. Chatelut, T. Levade, M. Hiraiwa, J. S. O’Brien, FEBS Lett. 1997, 417, 270b274. 42. K. Suzuki, Y. Suzuki, K. Suzuki, in C. Scriver, A. L. Beaudet, W. S. Sly, D. Valle (Eds.), The Metabolic and Molecular Basis of Inherited Disease: 1995, 7th Edition, David McGraw Hill, New York, 11, 2671L2692. 43. R . J. Desnick, Y. A. Ioannou, C. M. Eng in C. Scriver, A. L. Beaudet, W. S. Sly, D. Valle (Eds.), The Metabolic and Molecular Basis of Inherited Disease, 1995, 7th Edition, David McGraw Hill, New York, 11, Chapter 89, 2741-2784. 44. V. Gieselmann. Biochim. Biophys. Actu 1995, 1270, 103-136. 45. P. Leinekugel, S. Michel, E. Conzelmann, K. Sandhoff, Hum. Genet. 1992, 88, 513-523. 46. E. Conzelmann, K . Sandhoff, Dev. Neurosci. 1983/84, 6, 58-71. 47. A. d’Azzo, G. Andria, P. Strisciuglio, H. Galjaard in C. Scriver, A. L. Beaudet, W. S. Sly, D. Valle (Eds.), The Metabolic and Molecular Basis of Inherited Disease, 1995, 7th Edition, David McGraw Hill, New York, 11, Chapter 9 I , 2825 -2837. 48. K. Suzuki, R. L. Proia, K. Suzuki, Bruin. Puthol. 1998, 8, 195-215. 49. T. Kolter, K. Sandhoff, J. Inher. Metah. Dfs. 1998, 21, 548--563. 50. F. M. Platt, G. R. Neises, G. Reinkensmeier, M. J. Townsend, V. H. Perry, R. L. Proia, B. Winchester, R. A. Dwek, T. D. Butters, Science 1997, 276, 428-431.

472

27 Lysosomal Degradation of Glycolipids

51. S. Yamanaka, M. D. Johnson, A. Grinberg, H. Westphal, J. N. Crawley, M. Taniike, K. Suzuki, R. L. Proia, Proc. Natl Acad. Sci. USA 1994, 91, 9975-9979. 52. K. Sango, S. Yamanaka, A. Hoffmann, Y. Okuda, A. Grinberg, H. Westphal, M. P. McDonald, J. N. Crawley, K. Sandhoff, K. Suzuki, R. L. Proia, Nature Genetics 1995, 11, 170-176. 53. K. Sango, M. P. McDonald, J. N. Crawley, M. L. Mack, C. J. Tifft, E. Skop, C. M. Starr, A. Hoffmann, K. Sandhoff, K. Suzuki, R. L. Proia, Nature Genetics 1996, 14, 348-352. 54. T. Kolter, K. Sandhoff, Angew. Chem. 1999, 111, 1633-1670, Angew. Chem. Int. Ed. Engl. 1999,38, 1532-1 568. 55. T. Linke et al., unpublished. 56. T. Linke, K. Sandhoff, unpublished. 57. F. Platt, personal communication.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

28 Lysosomal Degradation of Glycoproteins Nathan N. Aronson, Jr.

28.1 Summary Glycoproteins, like all other biological macromolecules, undergo catabolic turnover, a process that helps to regulate their concentration in order to meet changing physiological demands for their function. The lysosome is the only organelle that contains a complete set of glycosidases, proteases and glycosylasparaginase necessary for complete breakdown of Asn-linked glycoproteins to their constituent sugars and amino acids. These monomeric products are essential for being able to be transported out of the lysosomes. They also are the only final products that can be metabolically reused for production of free energy or resynthesis into polymers or other important metabolites. The lysosomal hydrolases involved in Asn-linked glycoprotein degradation are organized in a two-part, bidirectional metabolic pathway. Sugars are removed sequentially from the non-reducing ends of the carbohydrate chains by exoglycosidases, while degradation of the protein and protein-tocarbohydrate region occurs as an ordered series of reactions delimited by unique substrate specificities of the enzymes involved. Besides general turnover of glycoproteins in the lysosomes, there are now examples of final digestive products that are essential for the overall metabolic well-being of the organism. In humans, for example, the lysosomal degradative process includes the production of antigens and the formation of thyroid hormone. In addition, lysosomes take part in a degradative pathway for polymannose oligosaccharides that appears to be important in the quality control of N-linked glycoprotein biosynthesis. In this chapter these various aspects of N-linked glycoprotein catabolism by lysosomes will be discussed.

28.2 Introduction Glycoproteins have many roles in biology from being membrane receptors to acting as hormones. To operate efficiently in living systems their concentration must be

474

28 Lysosomul Degradution qf Glycoprotrins

regulated in both time and space by active turnover. Cells in response to changing physiological demands adjust the required ‘steady state’ concentration of a specific glycoprotein by counterbalancing the rates of its synthesis and degradation. Gene transcription, together with ribosomal synthesis of its polypeptide and co- and posttranslational glycosylation and processing, raise the level of a specific glycoprotein, while proteolytic enzymes joined by glycosidases lower its concentration through hydrolysis. Historically, the intricacies of macromolecular biosynthesis have been more attractive to scientists as an area of experimental study than has been the catabolic process (e.g. see Part 1). It is important to note, however, that metabolic degradation of a large molecule also can be regulated with precision in order to form crucial metabolites. In this chapter, for example, we will focus on degradation of the glycoprotein prohormone thyroglobulin which leads to the controlled production by the thyroid of the small-molecule hormones T3 and T4 [lL4]. Similarly, the biosynthesis of carnitine, an essential metabolite for fatty acid oxidation, requires proteolytic release of trimethyllysines from proteins and/or glycoproteins [ 5-61, and the vital immunological system of processing exogenous protein/glycoprotein polypeptides as antigens for presentation on class I1 MHC molecules involves endosome/lysosome proteolysis [ 7, 81.

28.3 Roles of Lysosomes The major intracellular location of glycoprotein digestion during their turnover is the lysosome, the organelle that possesses a complete set of the necessary hydrolytic enzymes for complete breakdown to amino acids and monosaccharides [9, 101. The degradative process is substantially regulated by segregation of the hydrolytic enzymes in this compartment of the cell. Most glycoproteins enter the lysosome via various endocytic mechanisms and pathways, and may include extracellular (heterophagy) and intracellular (autophagy) sources of the substrates. Lysosomes are bounded by a membrane which not only can undergo the specific endocytic fusion events for substrate uptake [ 111, but also encloses a proper hydrolytic reaction environment that includes an acidic pH and a high concentration of a complete set of proteases and glycosidases. Estimates of the concentrations of these hydrolases in the lysosomes are in the millimolar range [ 121. A very important third feature of the lysosomal membrane is its very narrow specificity in terms of the products that can be released after degradation. Only single amino acids and sugars can be transported out of these organelles. This size limit defines the molecular cause of the many lysosomal storage diseases in which partially digested fragments as small as disaccharides cause extreme pathology in humans whose lysosomes have deficient hydrolytic capacity (see Chapter 57). Most lysosomal storage disorders result from genetic deficiency of an individual glycosidase, but sometimes a deficient membrane transporter of the lysosomes is responsible for pathological accumulation of a normal amino acid or monosaccharide final product, such as cystine in cystinosis [I41 and sialic acid in Salla disease [ 131. Over 20 lysosomal storage diseases involving faulty glycoprotein or glycolipid catabolism are known [ 151.

28.4 Lysosomcil Deyrudation of’ N-Linked Glycoproteins

415

28.4 Lysosomal Degradation of N-Linked Glycoproteins 28.4.1 General Features

Degradation of N-linked glycoproteins completely to amino acids and monosaccharides is necessary to avoid the storage diseases mentioned above that have severe consequences for cells and tissues. This catabolism has been determined to occur in two major parts 19, lo]:

1) sequential hydrolysis of each sugar from the non-reducing ends of the oligosacchride chains; and 2) ordered disassembly of the protein and protein-to-carbohydrate linkage region. The overall two-part catabolism that is depicted in Figure 1 mostly results from four types of experimental evidences:

1) the chemical structures of accumulated fragments in human and animal forms of lysosomal storage diseases [ 16-25]; 2) the types of products and their rate of production during in vitro digestion of specific glycoproteins by extracts of purified rat liver lysosomes [26]; 3 ) the chemical nature of products formed from endocytosed asialoglycoprotein substrates by a perfused rat liver used in conjunction with metabolic inhibitors of the lysosomal glycosidases [27-291 and proteases [30, 311; and

-

Carbohydrate Chains ( 1 5,4a)

0 0 -

NAN

6Gal-

NAN

6Gal-

4GlcNAc

0 0

4GlcNAc

-

2Mana

Protein / Linkage Region [I IV]

-

Figure 1. Bidirectional disassembly of N-linked glycoproteins in lysosomes [9, 101. The complex carbohydrate chains arc cleaved by exoglycosidases (enzymes 1-5) that sequentially remove nonreducing end sugars. The protein/linkage region is digested beginning with the protein by cathepsins (enzymes 1) followed by a-L-fucosidase (I1 ), glycosylasparaginase (TI1) and reducing-end chitobiase (IV). The rates of the two parts of overall catabolism depend on the structure of each glycoprotein substrate and its susceptibility to denaturation within the lysosomes. The di-N-acetylchitobiosc core can be cleaved either by P-D-hexosaminidase (enzyme 3) or chitobiase (enzyme I V ) .

476

28 Lysosomal Degradation of’ Glycoproteins

4) biochemical characterization of the reactions catalyzed by purified lysosomal glycosidases and proteases involved in the degradation [ 32-35]. As a model N-linked glycoprotein substrate serum a1 -acid glycoprotein (orosomucoid) was used to characterize the lysosomal digestive pathway. Both rat and human orosomucoid, or their asialo-derivatives, were used in vitro with extracts of purified lysosomes [26] or in situ within a perfused rat liver [27-311. In the latter organ system radioactively labeled asialo-orosomucoid was rapidly captured within hepatocyte lysosomes via the well-characterized asialoglycoprotein receptor unique to these cells [36]. Based on the known structures of N-linked glycoproteins and the number and kinds of proteases and glycosidases present in lysosomes, it was learned that complete digestion requires a set of six or seven glycosidases, a relatively small group of proteinases (cathepsins A, B, C, D, H, L and S) and a specific glycosylasparaginase for the hydrolysis of the protein-to-carbohydrate linkage. Overall the process has been termed bidirectional disassembly [9, 101 which like other metabolic pathways utilizes relatively few enzymes that catalyze their reactions in ordered steps.

28.4.2 Carbohydrate Digestion Hydrolysis of each oligosaccharide chain works at the non-reducing end beginning with removal of terminal sialic acid by neuraminidase to expose penultimate p-Dgalactosyl units. Following P-D-galactosidase as the second step are, in order, Nacetyl p-D-glucosaminidase, a-D-mannosidase and finally p-D-mannosidase. All five of these exoglycosidases have been associated with lysosomal storage diseases in humans (see Chapter 57). It is likely that different a-D-mannosidases (steps 4 and 4a, Figure 1) can be involved in cleaving the two a-D-mannoses that form the branch at the core of the oligosaccharide (see below). The non-reducing end exoglycosidase mechanism of each of the hydrolases (steps 1-5, Figure 1) establishes the reaction order to be the same as occurrence of their sugar substrate along the oligosaccharide chain.

28.4.3 Protein and Linkage Hydrolysis The second part of the catabolic pathway begins with breakdown of the polypeptide which is catalyzed by a set of cathepsins [37]. Some of these are exoproteinases that act at either the amino or carboxyl ends of the protein, while others are endopeptidases that cleave at interior positions along the polypeptide. Concerted action of these three kinds of cathepsins leads to the production of free amino acids as final lysosomal products. Glycopeptides are intermediate products which become substrates that define the reaction pathway for the linkage region. Partial inhibition of proteolysis in the prefused rat liver using the thiol protease inhibitor leupeptin substantially blocked hydrolysis of fucose from endocytosed asialo-orosomucoid [9]. Fucose, when present, forms a single-unit branch on the reducing-end GlcNAc

28.5 Formution

of‘ Thyroid Hormone

477

residue joined to Asn, and any residual peptide nearby must sterically interfere with lysosomal a-L-fucosidase required for its hydrolysis. Cleavage of the protein-tocarbohydrate linkage (Asn-GlcNAc) by glycosylasparaginase is another biochemically limited reaction that defines the order of steps. Glycosylasparaginase requires a free a-amino and a-carboxyl group on the asparagine portion of its substrates (321. Proteolysis must therefore be completed where each oligosaccharide joins the protein prior to Asn-GlcNAc hydrolysis. The amidase also is limited by certain aspects of the oligosaccharide structure. First, any remaining fucose on the GlcNAc joined to Asn blocks glycosylasparaginase action ( 331. Second, an a-u-mannosidase specific for the (1+6)-branched a-n-mannose appears to require prior hydrolysis of the Asn-GlcNAc amide bond before it can hydrolyze this mannose [34]. The a(1-6) mannose, however, can also be hydrolyzed by the major lysosomal CX-Dmannosidase [45]. These specificities for substrate structure explain the various fragments noted to accumulate in human patients suffering deficiencies of lysosomal a-D-mannosidase (a-mannosidosis) (24, 251, a-L-fucosidase (fucosidosis) ( 16, 1S] and glycosylasparaginase (aspartylglycosaminuria) [21-231 (see Chapter 57). Both parts of the N-linked catabolic pathway depicted in Figure 1 likely occur simultaneously with synergism, since potential steric inhibition by carbohydrate and protein may be lost with each glycosidic or peptide bond that is hydrolyzed. The rates of the two different parts would vary depending on the particular structure of the glycoprotein and its sensitivity to denaturation and unfolding in the lysosomal milieu. Strecker et al. [38] using lysosomal extracts studied the relative rate of carbohydrate chain digestion by exoglycosidases (steps 1-5) to that of linkage region hydrolysis by glycosylasparaginase and chitobiase (steps I11 and IV, Figure 1). They concluded that release of Asn and the reducing-end GlcNAc preceded release of monosaccharides from the non-reducing-ends of the oligosacccharides.

28.5 Formation of Thyroid Hormone via Lysosomal Degradation of Thyroglobulin 28.5.1 Synthesis of Thyroid Hormone Thyroid hormones act via binding to nuclear receptors to regulate gene transcription. The thyroid hormones T3 (3,3’, 5-triiodo-~-thyronine)and T4 (3,3’,5,5’tetraiodo-L-thyronine) are formed posttranslationally on the glycoprotein thyroglobulin (391. Release of the peptide bound T3 and T4 occurs via digestion of this very large glycoprotein (a dimer, mol. wt. 660,000) in lysosomes of thyroid epithelial cells. T4 is the major lysosomal product, but this isomer lacks significant biological activity and peripheral tissues. mainly liver, convert T4 by specific deiodination to the active T3 hormone. There are over 120 tyrosines in a thyroglobulin dimer and many of them undergo iodination to become mono- and diiodotyrosine residues on the glycoprotein. The enzyme that catalyzes iodination is iodide peroxidase which is bound to the membrane facing the follicular lumen where thyroglobulin is secreted and stored as a hormone precursor. The same membrane

28 Lysosomal Degrudution o j Glycoproteins

478

?-< 510 465

1345 1329

f

J

1696 1754 2275 2230

I

I

L D

I1

BL

2562 w

Lysosomal Glycosidases and Cathepsins 2 749 amino acids

1

I 1

DL

COOH

908 H20

+ 160 sugars + T4, T 3

Figure 2. Structure of human thyroglobulin and its degradation in thyroid lysosomes. A single thyroglobulin molecule is depicted schematically with positions of its N-linked oligosaccharides [41]: 0 , complex; o high-mannose; W , complex hybrid or complex. or high-mannose; 0, Tyrosines Y5 and Y130 are the major residues involved in the formation of hormones T4 and T3 [ 11. Principal sites of proteolytic cleavage by lysosomal cathepsins B, D, and L are shown as vertical lines [ 3 ] .

enzyme also catalyzes the intramolecular coupling of iodinated tyrosines leading mostly to peptide bound T4. The iodinated tyrosines involved in coupling are very limited with the major reaction being donation of an iodinated phenyl group from Tyrl30 to the acceptor residue, iodinated Tyr5 [l]. As a result of the coupling reaction T3 or T4 is formed from Tyr5 while a dehydroalanine is left at position 130. Finally, lysosomal degradation of the prohormone yields T4 and T3 (Figure 2), as well as mono- and diiodotyrosines. The latter digestive products are rapidly deiodinated so that their iodide can be reused to iodinate other thyroglobulin molecules. Thus, a large number of metabolic steps requiring considerable cellular energy and genetic information are used to synthesis the major thyroid hormone T3. This process represents one of the best examples in biology of an essential metabolic function of degradation of N-linked glycoproteins by lysosomes. A defect in the pathway causes one of the most striking storage diseases, p-mannosidosis in Nubian goats [40] in which there is massive engorgement of the thyroid organelles with the carbohydrate fragment Man p( 1-4) GlcNAc p( 1+4) GlcNAc that results in almost immediate death to the newborn animals. Turnover of the carbohydrate chains of thyroglobulin during hormone synthesis must be a major source of these undigested trisaccharides which cause the dramatic pathology.

28.5.2 Carbohydrate Degradation Lysosomal degradation of the carbohydrate chains of thyroglobulin has not been directly studied. The human prohormone glycoprotein has 20 consensus asparagines for N-linked glycosylation of which 16 have oligosaccharides attached [41]. These sites contain eight complex-type chains, five high-mannose chains, two hybrid

28.6 Degradation of Free Polymannose-Type Oligosaccharides

479

(or complex) chains, and one complex (or high mannose) chain. The complex and hybrid chains are expected to be digested via the bidirectional pathway depicted in Figure 1. The high mannose chains are cleaved at the non-reducing ends by the major lysosomal am-mannosidase, whose cDNA and gene have been cloned and sequenced recently by several laboratories [42-441. This glycosidase is capable of cleaving all three types of mannosyl linkages [a(I +2), a( 1-3) and a( 1+6)] in the high mannose oligosaccharide structure [45]. A composite ordered pathway for release of the various mannoses has been summarized by Daniel et al. [45]. In human mannosidosis where this lysosomal a-D-mannosidase is deficient, the major oligosaccharide fragment that accumulates retains only the a(1-3) mannose branch of the core [25]. This observation led to the discovery of a second lysosomal a-D-mannosidase that is specific for the a( 1 1 6 ) units. As mentioned above this glycosidase requires an oligosaccharide substrate from which both the Asn and reducing-end GlcNAc must first be removed by lysosomal glycosylasparaginase and di-N-acetylchitobiase [ 341.

28.5.3 Proteolysis Thyroglobulin is one of the few examples of an N-linked glycoprotein where the lysosomal digestion of the polypeptide has been studied in vitro in some detail [241. Figure 2 shows the major cleavage sites by purified cathepsins B, D and L. In late stages of proteolysis cathepsin B combined with lysosomal dipeptidase I appear to concertedly release T4/T3 from position 5 of the glycoprotein [2]. In general, the polypeptide of each N-linked glycoprotein is likely to undergo a somewhat ordered, specific pathway of proteolysis in the lysosomes depending on its tertiary, secondary and primary structure, and its susceptibility to unfolding within the lysosomal environment. For example, many antigens are poor substrates for lysosomal processing unless they are denatured by disulfide reduction either in endosomes or lysosomes [ 71. Defined proteolytic pathways have been observed during lysosomal digestion of other glycoproteins. Proteolysis of '251-labeled asialofetuin was examined in a perfused rat liver and was rapidly digested TI,^ = 14 min). This degra= 2h) by pretreatment of the dation was dramatically slowed almost ten-fold liver with the cysteine proteinase inhibitor leupeptin [31]. In vitro digestion of asialofetuin with extracts of purified rat liver lysosome yielded specific, rate-limiting proteolytic intermediates, and the pattern was completely changed by adding leupeptin [9]. The effect of leupeptin on the digestion of 125 I-asialoorosomucoid by a perfused rat liver was even more dramatic, virtually stopping protein breakdown while partial digestion of carbohydrate chains still occurred [9, 301.

28.6 Degradation of Free Polymannose-Type Oligosaccharides Derived from N-Linked Glycoproteins During Biosynthesis Over the past 15 years a number of laboratories have noted that during N-linked glycoprotein biosynthesis in the endoplasmic reticulum (ER) free polymannose-type

480

28 Lysosoinul Degrudution of' Glycoproteins

oligosaccharides are produced [47--531. The oligosaccharides seem to be released concurrently or soon after N-glycosylation and at least four possible mechanisms for their formation have been proposed [58].This process could be involved either in controlling glycoprotein synthesis by regulating the concentrations of dolichol pyrophosphate oligosaccharide precursors and/or excess nascent glycoproteins, or in carrying out biosynthetic quality control by degrading misfolded or mutant glycoproteins [ 511. The residual carbohydrate-free protein component may be degraded in the ER [54], but it has been found that deglycosylated proteins most likely are transferred to the cytoplasm and degraded by proteasomes after being ubiquinated [ 591. The free oligosaccharides, however, ultimately reach the lysosomes where final degradation to monosacchrides occurs, with other cell compartments playing a part in their initial trafficking and overall catabolism. Thus, two transfer steps are required (ER to cytosol and cytosol to lysosomes) in addition to the catalytic activity of a number of specific non-lysosomal glycosidases in the ER or cytosol. As shown in Figure 3 , an endoplasmic reticulum peptide: N-glycosidase can initially form free oligosaccharide intermediates in the lumen which contain both core GlcNAc residues. It is also possible that a hydrolytic activity of the oligosaccharyltransferase itself immediately releases the oligosaccharide during biosynthesis [60]. The oligosaccharides are then transferred from the lumen of the ER to the cytosol by an ATP-and Ca2+-dependent pathway [54]. This transfer process is inhibited by mannose or mannose derivatives modified at the glycosyl carbon (e.g. a-methyl mannoside), but not by monosaccharides that have an equatorial C2 hydroxyl, and not by GlcNAc or di-N-acetylchitobiose. Once in the cytosol, the reducing-end GlcNAc is first hydrolyzed by a neutral chitobiase (unique from lysosomal chitobiase) [56], and then the mannose component is trimmed rapidly by a neutral cytosolic mannosidase to a definite M5N product (Figure 3): Man a( I i 2 ) M a n a( 1 i 2 ) Man a(l+3) [Man a(l+6)] Man p(1+4) GlcNAc [57].This mannosidase is unable to cleave oligosaccharides with a di-N-acetylchitobiose reducing-end terminus. The same M j N substrate for lysosomal transport can also be derived from phosphodiester degradation of dolichol pyrophosphate oligosaccharides ( D o I P ~ N ~ M which ~) is exposed on the cytoplasmic side of the ER. In this case it is not yet known which enzyme cleaves the remaining phosphate group from the initial MjN2PI intermediate product [53]. Limited information is currently available about the capture by the lysosomes of the final cytosolic intermediate hexasaccharide MjN I . If glucose is still present, the subsequent transfer of the oligosaccharide (G(nlM~N1)into the lysosomes is blocked. Thus, cells preincubated with the a-u-glucosidase inhibitor castanospermine accumulated G3M5N in the cytosol [48]. There is no information about glucosidases that may be involved in the cytoplasmic release of any glucose units from the oligosaccharides if present. A recent study repeated use of castanospermine, but in addition tested the effect of a simultaneous inhibition of protein synthesis [50]. With both metabolic processes blocked G ~ M ~ accumulated Nz in the lumen of vesicles (ER?) rather than in the cytoplasm. This result might indicate that all three, or at least two, non-reducing end Glc units and/or the reducing-end GlcNAc, must be cleaved in order for the intermediate oligosaccharide to be transported from ER to cytosol. It is also possible that a key protein in this trans-

References

48 1

- 36

- 1M

idase M\s MB-N M-MM-M

4

/3'1 I

*

ATP!

: mannosidase t 5 M + N + MsNj a$-

cytoso/ Lysosome

Figure 3. Degradation of polymannose oligosaccharides derived from biosynthetic processes in the endoplasmic reticulum. Polymannose oligosaccharides are released in the ER possibly form newly formed N-linked glycoproteins that are improperly folded or made in excess. Shown is the initial loss of the three capping glucose units (G) and one mannose (M) via biosynthetic ER processing glucosidases and mannosidase. This is followed by release of the free oligosaccharide via a lumenal PNGase [49] to yield a polymannose structure with an intact di-N-acetylchitobiose core (MgN2). An endoglucosaminidase may instead form the same oligosaccharide with a single reducing-end GlcNAc 1581. These free oligosaccharides are then transferred from the ER into the cytosol via an ATP and Ca*+-dependent process [34]. In the cytosol a neutral chitobiase [56]cleaves the reducing end GlcNAc to form M ~ N which I is trimmed to MjNl via cytosolic a-D-mannosidase [57].The final cytosolic MsN oligosaccharide is then transferred into lysosomes (ATP-dependent [ 551) and completely hydrolyzed to monosaccharides by 1ysosomal a- and P-D-mannosidases.

port process is rapidly turned over and its concentration falls to an ineffective concentration when protein synthesis is blocked. It is clear that a significant metabolic function of the lysosomes is to degrade the cytosolic oligosaccharide intermediate M5N 1 which is derived from the N-linked glycoprotein biosynthetic pathway in the ER. This aspect of N-linked glycoprotein turnover was recently reviewed [51, 581, and further study of this interesting process is needed in order to understand more details about its mechanism and function. References 1. Dunn, A.D.; Corsi, C.M.; Myers, H.E., Dunn, J.T. Tyrosine 130 is an Important Outer Ring Donor for Thyroxine Formation in Thyroglobulin. J. Biol. Chrrn. 1998, 273, 25223-25229.

482

28 Lysosomal Degradation o j Glycoproteins

2. Dunn, A.D.; Myers, H.E.; Dunn, J.T. The Combined Action of Two Thyroidal Proteases Releases T4 From the Dominant Hormone-Forming Site of Thyroglobulin. Endocrinology 1996, 137, 3279-3285. 3. Dunn, A.D.; Crutchfield, H.E.; Dunn, J.T. Thyroglobulin Processing by Thyroidal Proteases. Major Sites of Cleavage by Cathepsins B, D and L. J. Biol. Clzem. 1991, 266, 20198-20204. 4. Dunn, A.D.; Crutchfield, H.E.; Dunn, J.T. Proteolytic Processing of Thyroglobulin by Extracts of Thyroid Lysosomes. Endocrinology 1991, 128, 3073-3080. 5. LaBadie, J . ; Dunn, W.A.; Aronson, N.N. Jr. Hepatic Synthesis of Carnitine from ProteinBound Trimethyl-Lysine. Lysosomal Digestion of Methyl-Lysine-Labelled Asialo-Fetuin. Biochem. J. 1976, 160, 85-95. 6. Dunn, W.A.; Englard, S. Carnitine Biosynthesis by the Perfused Rat Liver from Exogenous Protein-Bound Trimethyllysine. Metabolism of Methylated Lysine Derivatives Arising from the Degradation of 6-N-[Methyl-'H] Lysine-Labeled Glycoproteins. J. Biol. Chem. 1981, 256, 12437-12444. 7. Watts, C.; Antoniou, A.; Manoury, B.; Hewitt, E.W.; Mckay, L.M.; Grayson, L.; Fairweather, N.F.; Emsley, P.; Isaacs, N.; Simitsek, P.D. Modulation by Epitope-Specific Antibodies of Class I1 MHC-Restricted Presentation of the Tetanus Toxin Antigen. Immunol. Rev. 1998, 164, 11-16. 8. Geuze, H.J. The Role of Endosomes and Lysosomes in MHC Class I1 Functioning. Zmmunol. Today 1998 19, 282-287. 9. Kuranda, M.J.: Aronson, N.N. Jr. Receptor-Mediated Endocytosis and Lysosomal Degradation of Asialoglycoproteins by the Liver. In: Lysosomes: Their Role in Protein Breakdown; Glaumann, H.; Ballard F.J., eds.. Academic Press, Inc., London, 1987; pp 241-282. 10. Aronson, N.N. Jr.; Kuranda, M.J. Lysosomal Degradation of Asn-Linked Glycoproteins. FASEB J. 1989, 3,2615-2622. 11. Dunn, W.A.; Hubbard, A.L.; Aronson, N.N., Jr. Low Temperature Selectively Inhibits Fusion Between Pinocytic Vesicles and Lysosomes During Heterophagy of '251-Asialofetuinby the Perfused Rat Liver. J. Biol. Chem. 1980, 255. 5971-5978. 12. Dean, R.T.; Barrett, A.J. Lysosomes. In: Essays in Biochemistry, The Biochemical Society; Campbell, P.N.; Aldridge, W.N. eds., London. 1976, pp. 1-40. 13. Mendla, K.; Baumkotter, J.; Rosenau, C.; Ulrich-Bott, B.; Cantz, M. Defective Lysosomal Release of Glycoprotein-Derived Sialic Acid in Fibroblasts from Patients with Sialic Acid Storage Disease. Biochem. J. 1988, 250, 261-267. 14. Gahl, W.A. Disorders of Lysosomal Membrane Transport-Cystinosis and Salla Disease. Enzyme 1987, 38, 154-160. 15. Cantz, M.; Ulrich-Bott, B. Disorders of Glycoprotein Degradation. J. Inherit. Metah. Dis. 1990, 13, 523-537. 16. Lundblad, A,; Lundsten, J.; Norden, N.E.; Sjoblad, S.; Svensson, S.; Ockerman, P.A.; Gehlhoff, M. Urinary Abnormalities in Fucosidosis. Characterization of a Disaccharide and Two Glycoasparagines. Eur. J. Biochem. 1978, 83, 513-521. 17. Lunblad, A,; Sjoblad, S.; Svensson, S. Characterization of a Penta and an Octasaccharide from Urine of a Patient with Juvenile GMl-Gangliosidosis. Arch. Biochem. Biophys. 1978, 188, 130136. 18. Strecker, G.; Fournet. B.; Montreuil, J. Structure of the Three Major Fucosyl-Glycoasparagines Accumulating in the Urine of a Patient with Fucosidosis. Biochimie 1978, 60, 725-734. 19. Strecker, G.; Herlant-Peers, M.C.; Fournet, B.; Montreuil, J. Structure of Seven Oligosaccharides Excreted in the Urine of a Patient with Sandhoff's Disease ( G MGangliosidosis~ Varient 0).Eur. J. Biochem. 1977, 81, 165-171. 20. Dorland, L.; Haverkamp, J.; Viliegenthart, J.F.; Strecker, G.; Michalski, J.C.; Fournet, B.; Spik, C.; Montreuil, J. 360-MHz ' H Nuclear-Magnetic-Resonance Spectroscopy of SialylOligosaccharides from Patients with Sialidosis (Mucolipidosis I and 11). Eur. J. Biochem. 1978; 151323-329. 21. Kaartinen, V.; Mononen, I. Assays of Aspartylglycosylaminase by High-Performance Liquid Chromatography. Anal. Biochem. 1990, 190, 98-1 01. 22. Maury. P. Accumulation of Two Glycoasparagines in the Liver in Aspartylglycosaminuria. J. Biol. Chem. 1979, 10, 1513-1515.

References

483

23. Lundblad, A,; Masson, P.K.; Norden, N.E. Structural Determination of Three Glycoasparagines Isolated from the Urine of a Patient with Aspartylglycosaminuria. Eur. J. Biochem. 1976, I , 209-214. 24. Norden, N.E.; Lundblad, A.; Svensson, S.; Autio, S. Characterization of Two MannoseContaining Oligosaccharides Isolated from the Urine of Patients with Mannosidosis. Biochemistry 1974, 13, 871-874. 25. Norden, N.E.; Lundblad, A,; Svensson, S.; Ockerman, P.A.; Autio, S. A Mannose-Containing Trisaccharide Isolated from Urines of Three Patients with Mannosidosis. J. Bid. Chem. 1973, 10, 6210-6215. 26. Aronson, N.N. Jr.; DeDuve, C. Digestive Activity of Lysosomes. Ii. The Digestion of Macromolecular Carbohydrates by Extracts of Rat Liver Lysosomes. J. Bid. Chem. 1968, 243, 45644573. 27. Docherty, P.A.; Aronson, N.N. Jr. u-D-Mannopyranosylmethyl-p-Nitrophenyltriazene Inhibition of Rat Liver a-D-Mannosidases. Biochim. Biophys. Acta 1987, 914, 283-288. 28. Docherty, P.A.; Kuranda, M.J.; Aronson, N.N. Jr., BeMiller, J.N.; Myers, R.W.; Bohn, J.A. Effect of a-D-Mannopyranosylmethyl-p-Nitrophenyltriazeneon Hepatic Degradation and Processing of the N-Linked Oligosaccharide Chains of u 1-Acid Glycoprotein. J. Biol. Chem. 1986, 261, 3457-3463. 29. Kuranda, M.J.; Aronson, N.N. Jr. Use of Active Site-Directed inhibitors to Study In Situ Degradation of Glycoproteins by the Perfused Rat Liver. J. Bid. Chem. 1985, 260, 1858-1866. 30. Dennis, P.A.; Aronson, N.N. Jr. Uptake and Degradation of '25i-Labeled Rat Asialoorosomucoid by the Perfused Rat Liver. Biochim. Biophys. Acta 1984, 798, 14-20. 3 1. Dunn, W.A.; LaBadie, J.H.; Aronson, N.N. Jr. Inhibition of 1251-AsialofetuinCatabolism by Leupeptin in the Perfused Rat Liver and In Vioo. J. Bid. Chem. 1979, 254. 4191-4196. 32. Kaartinen, V; Mononen, T.; Laatikainen, R.; Mononen, I. Substrate Specificity and Reaction Mechanism of Human Glycoasparaginase. The N-Glycosidic Linkage of Various Glycoasparagines is Cleaved Through a Reaction Mechanism Similar to L-Asparaginase. J. Biol. Chem. 1992,267,6855-6858. 33. Noronkoski,T.; Mononen, I. Influence of L-Fucose Attached u-l+6 to the Asparagine-Linked N-Acetylglucosamine on the Hydrolysis of the N-Glycosidic Linkage by Human Glycosylasparaginase. Glycohiology 1997, 7, 217-220. 34. Haeuw, J.F.; Grard, T.; Alonso, C.; Strecker, G; Michalski, J.C. The Core-Specific Lysosomal a( I -6)-Mannosidase Activity Depends on Aspartamidohydrolase Activity. Biochem. J. 1994, 297, 463-466. 35. Aronson, N.N. Jr.; Backes, M.; Kuranda, M.J. Rat Liver Chitobiase: Purification, Properties, and Role in the Lysosomal Degradation of Asn-Linked Glycoproteins. Arch. Biochem. Biophys. 1989, 272, 290-300. 36. Ashwell, G.; Harford, J. Carbohydrate-Specific Receptors of the Liver. Annu. Rev. Biochem. 1982, 51, 531-554. 37. Coffey, J.W.; DeDuve, C. Digestive Activity of Lysosomes. I. The Digestion of Proteins by Extracts of Rat Liver Lysosomes. J. Biol. Chem. 1968, 243, 3255-3263. 38. Strecker, G.; Michalski, J.C.; Montreuil, J. Lysosomal Catabolic Pathway of N-Glycosylprotein Glycans. Biochemie, 1988, 70, 1505-1510. 39. Dunn, J.T. Thyroglobulin, Hormone Synthesis and Thyroid Disease. Eur. J. Endocrinol. 1995, 132, 603-604. 40. Boyer, P.J.; Jones, M.Z.; Nachreiner, R.F.; Refsal, K.R. Common, R.S.; Kelley, J.; Lovell, K.L. Caprine P-Mannosidosis. Abnormal Thyroid Structure and Function in a Lysosomal Storage Disease. Lab. Inuest. 1990, 63, 100- 106. 41. Yang, S.X.; Pollock, H G.; Rawitch, A.B. Glycosylation in Human Thyroglobulin: Location of the N-Linked Oligosaccharide Units and Comparison with Bovine Thyroglobulin. Arch. Biochem. Biophys. 1996, 327, 61-70. 42. Riise, H.M.; Berg, T.; Nilssen, 0.; Romeo, G.; Tollersrud, O.K.; Ceccherini, I. Genomic Structure of the Human Lysosomal u-Mannosidase Gene. (MANB). Genomics 1997, 42, 200207. 43. Nilssen, 0.;Berg, T.; Riise, H.M.; Ramachandran, U.; Evjen, G.; Hansen, G.M.; Malm, D.; Tranebjaerg, L.; Tollersrud, O.K. a-Mannosidosis: Functional Cloning of the Lysosomal a-

484

28 Lysosomul Deyrudution of Glycoproteins

Mannosidase cDNA and Identification of a Mutation in Two Affected Siblings. Hum. Mol. Genet. 1997, 6, 717-726. 44. Liao, Y.F.; Lal, A,; Moremen, K.W. Cloning, Expression, Purification, and Characterization of the Human Broad Specificity Lysosomal Acid a-Mannosidase. J. Biol. Chem. 1996, 271, 28348-28358. 45. a1 Daher, S . ; de Gasperi, R.; Daniel, P.; Hall, N.; Warren, C.D.; Winchester, B. The SubstrateSpecificity of Human Lysosomal a-D-Mannosidase in Relation to Genetic a-Mannosidosis. Biochem. J. 1991; 277, 743-751. 46. Daniel, P.F.; Winchester, B.; Warren, C.D. Mammalian a-Mannosidases-Multiple Forms but a Common Purpose? Glycobiology 1994, 4; 55 1-566. 47. Anumula, K.R.; Spiro, R.G. Release of Glucose-Containing Polymannose Oligosaccharides During Glycoprotein Biosynthesis. Studies with Thyroid Microsomal Enzymes and Slices. J. Bid. Chem. 1983, 258, 15274-15282. 48. Moore, S.E.; Spiro, R.G. Intracellular Compartmentalization and Degradation of Free Polymannose Oligosaccharides Released During Glycoprotein Biosynthesis. J. Biol. Chem. 1994, 269, 12715-1272 1. 49. Weng, S.; Spiro, R.G. Demonstration of a Peptide: N-Glycosidase in the Endoplasmic Reticulum of Rat Liver. Biochem. J. 1997, 322, 655-661. 50. Duvet, S.; Labiau, 0.;Mir, A.M.; Kmiecik, D.; Krag, S.S.; Verbert, A,; Cacan, R. Cytosolic Deglycosylation Process of Newly Synthesized Glycoproteins Generates Oligomannosides Possessing One GlcNAc Residue at the Reducing End. Biochem. J. 1998, 335, 389-396. 51. Cacan, R.; Duvet, S.; Kmiecik, D.; Labiau, 0.;Mir, A.M.; Verbert, A. ‘Glyco-Deglyco’ Processes During the Synthesis of N-Glycoproteins. Bioclzimie 1998, 80, 59-68. 52. Grard, T.; Herman, V.; Saint-Pol, A,; Kmiecik, D.; Labiau, 0.;Mir, A.M.; Alonso, C.; Verbert, A,; Cacan, R.; Michalski, J.C. Oligomannosides or Oligosacchride-Lipids as Potential Substrates for Rat Liver Cytosolic a-D-Mannosidase. Biocheni. J. 1996, 3 16, 787-792. 53. Cacan, R.; Lepers, A,; Belard, M.; Verbert. A. Catabolic Pathway o f OligosaccharideDiphospho-Dolichol. Subcellular Sites o f the Degradation of the Oligomannoside Moiety. ELK J. Biochem. 1989, 185, 173-179. 54. Moore, S.E. Transport of Free Polymannose-Type Oligosaccharides from the Endoplasmic Reticulum into the Cytosol is Inhibited by Mannosides and Requires a Thapsigargin-Sensitive Calcium Store. Glycobiology 1998, 8, 373-38 I . 55. Saint-Pol, A.; Bauvy, C.; Codogno, P.; Moore, S.E. Transfer of Free Polymannose-Type Oligosaccharides from the Cytosol to Lysosomes in Cultured Human Hepatocellular Carcinoma HepG2 Cells. J. Cell Bid. 1997, 136, 45-59. 56. R. Cacan, C. Dengremont, 0. Labiau, D. Kmiecik, A.M. Mir., A. Verbert. Occurrence of a Cytosolic Neutral Chitobiase Activity Involved in Oligoinannoside Degradation: A Study with Madin-Darby Bovine Kidney (MDBK) Cells. Biochem. J. 1996, 313. 597-602. 57. D. Kmiecik, V. Herman, C.J. Stroop, J.C. Michalski, A.M. Mir. 0. Labiau, A. Verbert. R. Cacan. Catabolism of Glycan Moieties of Lipid Intermediates Leads to a Single MansGlcNAc Oligosaccharide Isomer: A Study with Permeabilized CHO Cells. Glycohioloyy 1995, 5:483-494. 58. T. Suzuki, Q. Yan, W.J. Lennarz. Complex, Two-way Traffic of Molecules Across the Membrane of the Endoplasmic Reticulum. J. Biol. Clzrm. 1998, 273:10083- 10086. 59. T. Sommer, D.H. Wolf. Endoplasmic Reticulum Degradation: Reverse Protein Flow of N o Return. F A S E B J . 1997, 11:1227-1233. 60. M.J. Spiro, R.G. Spiro. Potential Regulation of N-Glycosylation Precursor Through Oligosaccharide-Lipid Hydrolase and Glucosyltransferase-Glucosidase Shuttle. J. Biol. Chem. 1991, 26615311-5317.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

29 Sialidases Gavry Taylor, Susan Crennell, Carl Thompson, and Marina Chuenkova

29.1 Abstract Sialidases, also known as neuraminidases, catalyze the removal of terminal sialic acids from a variety of glycoconjugates. In animals, sialic acids are involved in a number of functions related to cell adhesion and cell survival and also the modulation of many cellular processes. The ubiquity and unique terminal position of sialic acids in sugars has provided a range of pathogens with a binding site for cell invasion. In addition, many bacteria have acquired the ability to utilize sialic acid as a carbon and energy source. Bacteria do not synthesize sialic acids, they are a recent invention of nature found in higher organisms, suggesting that those primitive pathogens which have sialidases acquired the genes by horizontal gene transfer. The key role of the enzyme in pathogenesis has made sialidases good targets for structure-based drug design. In this Chapter, we make a comparative study of the three dimensional structures of sialidases from viruses and bacteria. This shows that, despite very low sequence identity, all sialidases share a catalytic domain of the propeller structure first observed in the influenza virus neuraminidase. In many cases however, the enzymes possess additional domains that appear to have sugarbinding functions, suggesting that sialidases have evolved to be exquisitely adapted to their operating environment.

29.2 Introduction Sialic acid, or N-acetylneuraminic acid (NeuSAc, NANA), is often the terminal sugar in a variety of glycoconjugates in which they are linked by their 0 - 2 oxygen to galactose, galactosamine or another sialic acid. Over 40 natural derivatives of sialic acid have been identified whose appearance correlates with cell development, tissue type and species [l]. Modulation of this diversity in the terminal sugar is

486

29 Siuliduses

therefore a powerful means of manipulating cell recognition. Although variations are seen in the 0-4, 0-7, 0-8, 0 - 9 and N-acetyl groups, all sialic acids conserve the negatively charged carboxylic acid group on C-2. Regulation of the sialic acid profile of cells is important in various mammalian functions including apoptosis. Sialidases have been identified in a number of species and a variety of tissues, where they may be membrane-bound, cytosolic or extracellular. Several of the enzymes have been cloned and sequenced including those from brain [2, 31, lysosomes [4] and the major histocornpatability complex (MHC) [5, 61. The lysosomal sialidases are part of a multi-enzyme complex (sialidaset cathepsin A P-galactosidase) and the sialidase is also the same as the MHC enzyme. This suggests a role for the complex in the modulation of cell-surface sialylation seen in T-cell activation [7] and the exciting possibility of using the human sialidase to modulate the inflammatory response via the IL4 pathway 181. A number of pathogens also have sialidases to aid in pathogenesis or nutrition. Orthomyxoviruses (e.g. influenza) and paramyxoviruses (e.g. mumps, parainfluenza, Newcastle disease, Sendai) have sialidases on their surfaces. Surface sialidase activity is also exhibited by parasites such as Tryp(inomma cruzi, T. brucei as well as a number of both pathogenic and non-pathogenic bacteria which may also secrete the enzyme. A comparison of their amino acid sequences reveals a large diversity (Figure 1).

+

29.3 Influenza Virus Neuraminidase The most studied sialidase is the influenza virus neuraminidase, whose structure was reported over 15 years ago [9], and which has been the target of highly successful rational drug design [ 10-121. Influenza type A and B viruses have two surface proteins, hemagglutinin (HA) which attaches the virus to host cells by recognizing sialic PILEUP August 3rd 1999

40%

60%

90%

Influenza virus neuraminidases/Non-viral sialidases/Paramyxovirus Figure 1. Dendrogram of sialidase sequences showing the large sequence variation of this superfamily. HN = hemagglutinin-neuraminidase. Percentage identity is shown to the left.

29.5 Non- Vivul Siuliduses

487

acid and is involved in fusion, and neuraminidase (NA) which removes sialic acid from progeny virions as they bud from infected cells to halt viral self-agglutination. On the surface of the virus, the enzyme forms a tetramer of four identical monomers that is anchored to the viral membrane via the four N-terminal regions of the enzymes. The crystal structures of the catalytic head of several subtypes have been determined [13] including avian N6 and N8 subtypes in our own laboratory [14] (Figure 2). The first report of the sialidase structure revealed that the enzyme folded into six four-stranded anti-parallel a-sheets arranged around a pseudo-six-fold axis as the blades of a propeller. This a-propeller fold is highly conserved over all influenza subtypes, with a strict conservation of the active site (Figure 3a). The 30-yearold lead compound, 2-deoxy-2,3-dehydro-N-acetylneuraminicacid (NeuSAc2en, DANA), that inhibits most sialidases with micromolar affinity, has been used as the basis for the development of novel inhibitors with nanomolar binding affinity through a mixture of structure-based drug design and combinatorial chemistry. These new compounds, most notably Glaxo’s Zanamavir (Ki 10P9M) and Gilead’s GS4071 (Ki 10P9M),exploit two features of the conserved active site: a pair of glutamic acid residues close to 0 - 4 (Zanamavir), and a hydrophobic patch close to the glycerol binding site (GS4071).

-

-

29.4 Paramyxovirus Hemagglutinin-Neuraminidase (HN) Paramyxoviruses have two surface glycoproteins: HN that both recognizes sialic acid for cell attachment and hydrolyzes sialic acid, and the F (fusion) protein which is involved in fusing the viral and host cell membranes. In the influenza virus, the fusion and binding activities are carried on the hemagglutinin molecule. Paramyxoviruses include human parainfluenza viruses (the major cause of childhood croup), Newcastle disease virus (a devastating disease of chickens), mumps virus. murine Sendai virus, and a range of animal distemper and parainfluenza viruses. The H N molecule appears to exist as a mix of dimers, (which are usually disulfidelinked), and tetramers on the surface of the virus. We have been working on the structure of Newcastle disease HN, for which we have diffraction to 1.98 A resolution. Although H N has been widely predicted to have a P-propeller fold similar to that of influenza sialidase, molecular replacement using the influenza structure as a model has not been successful. Structure determination is hampered by extreme non-isomorphism of the crystals, making the production of useful heavy atom derivatives very difficult. Knowledge of the conformation of the active sites will enable the design of drugs against croup and Newcastle disease.

29.5 Non-Viral Sialidases The non-viral sialidases show a large sequence divergence (Figure 1) and a large range of sizes, 40-50 kDa in size, similar to that of the viral enzymes, up to over

488

29 Sialiduses

Figure 2. The sialidase superfamily (a) Influenza virus neuraminidase tetramer, (b) the ‘small’ sialidases from S. typhimurium and (c) M. vividqaciens, ( d ) the ‘side’ view of the large M. virid@ciens sialidase, (e) V. cholerae sialidase. The inhibitor NeuSAcZen is shown bound in several active sites and galactose is also shown in the large M. uiridifhciens sialidase. The isolated grey spheres represent divalent ions that play a key structural role in influenza and V. cholerue sialidase, and are an absolute requirement for activity.

29.5 Noiz- Viral Sialidusa

489

Figure 3. The active sites of (a) influenza virus neuraminidase and (b) S. typhimurium sialidase with NeuSAc2en bound (black bonds). Residues in the ‘0-4 pocket’ are drawn in white, others in grey. Note the arginine-triad (influenza residues R I 1 8, R292, R371) that binds the carboxylic acid group, the glutamic acid that stabilizes the first arginine (E425), the tyrosine (Y406) and glutamic acid (E277) that interact with one another and are close to the C1LC2 bond, the two glutamic acids that interact with the glycerol group (E276, E477). the hydrophobic pocket accommodating the N-acetyl group on the left, and the aspartic acid (D151) above the sugar ring thought to be involved in the hydrolysis [31]. The S. typhimurium sialidase active site (b) is extremely similar, but a major difference is the interaction with 0 - 4 by DlOO and R56 which explains the reduced affinity of the influenza neuraminidase drugs to non-viral sialidases.

490

29 Sialidases

100 kDa. Many pathogenic and non-pathogenic bacteria produce sialidases to excise sialic acid which they metabolize for use as a carbon and energy source. However bacterial sialidases are also involved in the pathogenesis of a range of disease [ 151, for example: gas gangrene (Clostridium), septicemia (Streptococcus, Pneumococcus, Bacteroides), pneumonia (Streptococcus), peritonitis (Clostridium, Bacteroides), meningitis (Streptococcus) and cholera (Vibrio). Comparison of the non-viral sialidase sequences reveals two conserved motifs, the RIP-motif followed by several occurrences of the Asp-box motif (S/T-X-D-[XIG-X-T-W/F) [16]. The RIP-motif, which can be found in the variations FRIP, FRIP, YRIP or DRIP, corresponds to REP in the influenza enzyme, where the arginine is one of the arginine-triad (R118 on Figure 3a) and the glutamic acid is one of the conserved residues exploited in the design of novel drugs. The Asp-box motif is not seen in the viral enzymes.

29.6 Small Sialidases The crystal structures of two representative small sialidases, with molecular weights around 40 kDa, have been determined: one from Salmonella typhimurium [17] and one from Micromonospora viridfaciens [ 181. These reveal the same (3-propeller fold seen in the influenza virus neuraminidase (Figure 2), despite having no sequence similarity to the viral enzyme, and not containing any disulphide bonds in contrast to the seven conserved disulfides in the viral enzyme. These structures revealed the position of the Asp-boxes: they occur at topologically equivalent positions in the fold, explaining the regular separation in the sequences. The Asp-boxes form the turn between the third and fourth strands of each sheet, with the tryptophan packed between sheets, and the aspartic acid exposed to solvent. The active site of these sialidases exhibits a much higher similarity to that of the influenza virus enzyme (Figure 3) than does the overall structure, both having an arginine triad which binds the C-2 carboxylic acid, a hydrophobic pocket to accommodate the N-acetyl, a tyrosine and glutamic acid underneath C-2 and aspartic acid above which are involved in the mechanism of hydrolysis. The tight binding of residues in the S. typhimurium active site to the 0 - 4 of NeuSAc2en excludes those inhibitors designed to exploit the influenza pocket at this position. All non-viral sialidases appear to conserve these two residues, thus restricting the substitution allowed at 0 - 4 and rendering the influenza inhibitors highly specific for the virus enzyme. The bacterial sialidases have a 1000-fold increase in turnover over the influenza enzyme so the broad similarity in active site was unexpected, the rate increase may perhaps be understood by generally weaker binding of the inhibitor’s glycerol group by the bacterial enzyme facilitating product release. Other differences also explain the promiscuity of many bacterial enzymes in their choice of sialic linkages while the influenza enzyme prefers a2+3 linkages over a2+6, and differences in binding affinities.

29.8 T. cruzi Trans-Sialiduse ( T S )

491

29.7 Large Sialidases The crystal structures of two ‘large’ sialidases have been determined: the 83 kDa Vihrio cholerae sialidase [ 171 and the 68 kDa M. viridifaciens sialidase [ 191, both shown in Figure 2. The V. cholerue enzyme revealed a central catalytic P-propeller domain, flanked by two additional domains with similar topologies, resembling that of the lectins. The first lectin domain is at the N-terminus, and the second is inserted between the second and third sheets of the P-propeller. The lectin domains bind on as yet unidentified carbohydrate, but their structure suggests a role for these domains in the small intestine where the secreted sialidase may need to ‘grasp’ the cell surface to perform its catalytic function without being swept away. The large M. viridijiuciens sialidase is shown in Figure 2, another remarkable structure. This bacterium secretes either a small or large sialidase, derived from the same gene, dependent on the food source: in the presence of colominic acid, the small form is secreted; and in the presence of milk casein, the larger form is obtained. The small form contains only the catalytic domain while the large enzyme consists of the catalytic domain followed by a domain with an immunoglobulinfold, and then by a jelly-roll domain that binds galactose. These extra domains are connected to the catalytic domain by two glycine residues which suggests that the galactose domain can move relative to the sialic acid domain. This could provide a mechanism for ‘feeding’ appropriate substrates to the active site. The RIP and Asp-box motifs allow the prediction of the location of the catalytic domain in other large non-viral sialidases (Figure 4). Lack of sequence identity with other domains of known function does not allow prediction of the roles of the additional domains in many other large sialidases, however they too are likely to have carbohydrate-binding functionality.

29.8 T. cruzi Trans-Sialidase (TS) The South American trypanosome, the etiological agent of Chagas’ disease, carries on its surface an unusual trans-sialidase that exists in multiple copies in the parasite genome [20, 2 11. The trans-sialidase, or P-D-galactoside w2,3-truns-sialidase, presents a unique type of sialyl-transferase that is highly selective in its substrate specificity for sialic acid linked w2,3 to P-11-galactose [22, 231. Unable to synthesize sialic acid, T. cruzi uses the trans-sialidase to remove sialic acid exclusively from host sialoglycans and transfer it to water (conventional sialidase activity) or suitable P-D-galactose acceptors. This activity is shown to be most highly expressed by the invasive trypomastigote form of T. cruzi [24] which infects mammalian cells by receptor-mediated endocytosis [25, 261. The sialylated acceptor molecules on the plasma membrane of T. cruzi trypomastigotes are involved in parasite adherence and subsequent entry into host cells [27]. In addition, the enzyme can sialylate host

492

29 Siulidases

Figure 4. A schematic depiction of the sequence alignment of a number of non-viral sialidases. Each sequence is drawn to scale, marked with the RIP and Asp-box motifs, (small open boxes). The secondary structure, either known or predicted from our work, i s superimposed on the sequence. Central to all i s the lightly shaded P-propeller, the only domain in the small sialidases. The galactose-binding domain of M. viridifaciens shares 30Y0identity with the N-terminus of C. septicum, both being coloured black. The open boxes are endoglucanase-like domains seen in C. septicurn and the large C. perfringens although the boundary between the two postulated C. septicum N-terminal domains is unclear. V . c-holerue has two lectin domains (scalloped boxes) and fibronectin type I11 domains (bricked boxes) are predicted at the C-terminus of C. septicurn and T. cruzi trans-sialidase.

cell glycoconjugates to generate receptors used by trypomastigotes to attach to and penetrate target cells [28]. The enzymatically active protein isolated from T. cruzi trypomastigotes through cleavage of the glycophosphoinositol anchor, comprises several distinct domains [28] (Figure 4): i) an N-terminal domain of approximately 380 amino acids which shares up to 30% sequence identity with bacterial sialidases; ii) a region of -150 residues which shows no similarity to any known sequence possibly forming a second domain; iii) a domain containing a fibronectin type 111 (FnIII) motif; and iv) a long C-terminal stretch of 12-amino-acid tandem repeats, which is not required for the enzyme activity but may be involved in oligomerization [29]. The first, N-terminal domain contains two copies of the Asp-box motif and also all of the key residues identified in the bacterial sialidases as being important for sialic

29.9 Conclusion

493

acid hydrolysis, and is therefore likely to adopt a very similar fold to the catalytic domain of these enzymes. However the structural features of T. cruzi TS that support its efficient sugar transfer activity rather than simply hydrolysis are not fully understood. Studies have shown the importance of the additional domains for full TS activity, and pinpoint specific a specific residue in the sialidase-like domain that is essential for TS activity [30]. The recognition of D-D-galactose by TS implies that the enzyme may have two galactose-binding sites: one on the sialidase-like domain, and the other on one of the additional domains. Thus the large M. uiridifaciens sialidase may be a model for the trans-sialidase. Our laboratory, and several others too, are attempting to crystallize the T. cruzi TS which, due to its crucial role in infectivity, presents a possible target for drug design. The role of the TS may not be solely that of a transferase, but the non-catalytic members of the gene family may act as lectins, thereby having a role in cell attachment via sialic acid.

29.9 Conclusion The sialidase superfamily represents a remarkable set of variations on a theme. Each member contains the canonical P-propeller catalytic fold, which provides the framework for a set of highly conserved active-site residues. This high conservation allows active site residues to be predicted within a sialidase domain from sequence alignment. The scaffolding itself shows very high sequence and structural variation (within the confines of the fold) across the superfamily members. Superposition of the catalytic domains shows a wide variation in the twist of the sheets (propeller blades), the positioning of the sheets relative to the pseudo-six-fold axis (propeller axle), and the lengths and conformations of the loops connecting strands within sheets. In general the loops are longer and more variable on the catalytic surface of the molecule where manipulation, by Nature, of the surface topology around the active site appears to be the mechanism used to create specificity for certain sugar linkages. As well as the canonical P-propeller, larger sialidases have extra linked domains which apparently facilitate the primary role of the enzyme by binding other carbohydrates, The most studied, influenza, sialidase contains a single domain but forms a tetramer on the viral surface, presumably increasing stability and valency of the enzyme. Paramyxovirus sialidases similarly occur in dimeric or tetrameric forms. T cruzi trans-sialidase in the infective, trypomastigote stage, is linked to the surface by a GPI anchor and forms a trimer through association of the tandem repeat. Certain bacteriophage also express a trimeric endosialidase, linked to the virus. By contrast bacterial sialidases are normally secreted in a monomeric, soluble form, although that of Streptococcus pneumoniae is tethered to the bacterial membrane. Mammalian sialidases cover the whole spectrum, being membrane-bound, soluble and found in complexes associating with the ‘protective protein’, cathepsin A, for full activity.

494

29 Sialiduses

These structural studies have shed light on the remarkable adaptability of this enzyme to its operating environment. The success of the development of the influenza virus drug, targeted to the sialidase, lays the foundation for similar studies on the range of sialidases, and their associated diseases, discussed in this paper. Acknowledgments We thank Margaret Taylor for expert technical assistance. We also thank our many collaborators around the world; in particular Elspeth Garman, Graeme Laver, Miercio Pereira, Allen Portner, Eric Vimr and Robert Webster. This work was funded by The Wellcome Trust, The National Institutes of Health and The Royal Society.

References I . A. Varki, Glycobiology, 1992, 2 , 25-40. 2. K. Hata, T. Wada, A. Hasegawa, M. Kiso, T. Miyagi, Journal ofBiochemistry. 1998, 123, 899905. 3. E. Bonten, A. vanderspoel, M. Fornerod, G. Grosveld, A. dAzzo, Genes & Development, 1996, 10, 3156-3169. 4. M. Hiraiwa, M. Saitoh, Y. Uda, N. Azuma, B. M. Martin, Y. Kishimoto, J. S. Obrien, Comparative Biochemistry and Physiology B-Biochemistry & Molecular Biology, 1996, 115, 541546. 5. M. B. Carrillo, C . M. Milner, S. T. Ball, M. Snoek, R. D. Campbell, Glycobiology, 1997, 7, 975-986. 6. S. A. Igdoura, C. Gafuik, C. Mertineit, F. Saberi, A. V. Pshezhetsky, M. Potier, J. M. Trader, R. A. Gravel, Human Molecular Genetics, 1998, 7, 115-120. 7. N. Razi, A. Varki, Proceedings of the National Academy qf Sciences o f the United States of America, 1998, 95, 1469-7474. 8. X. P. Chen, E. Y. Enioutina, R . A. Daynes, Journal of Immunology, 1997, 158, 3070-3080. 9. J. N. Varghese, W. G. Laver, P. M. Colman, Nature, 1983, 303, 35-40. 10. M. Vonitzstein, W. Y. Wu, G. B. Kok, M. S. Pegg, J. C. Dyason, B. Jin, T. V. Phan, M. L. Smythe, H. F. White, S. W. Oliver, P. M. Colman, J. N. Varghese, D. M. Ryan, J. M. Woods, R. C . Bethell, V. J. Hotham, J. M. Cameron, C. R. Penn, Nature, 1993, 363, 418-423. 11. J. N. Varghese, P. M. Colman, Journal of Molecular Biology, 1991, 221, 473-486. 12. W. P. Burmeister, R. W. H. Ruigrok, S. Cusack, EMBO Journul, 1992, 11, 49-56. 13. M. N. Janakiraman, C. L. White, W. G. Laver, G. M. Air, M. Luo, Biochemistry, 1994, 33, 8172-8179. 14. G. Taylor, E. Garman, R. Webster, T. Saito, G. Laver, Journal of Moleculur Biology, 1993, 230, 345-348. 15. T. Corfield, Glycobiology, 1992, 2, 509-521. 16. P. Roggentin, R. Schauer, L. L. Hoyer, E. R. Vimr, Molecular Microbiology, 1993, 9, 915-921. 17. S. J. Crennell, E. F. Garman, W. G. Laver, E. R. Vimr, G. L. Taylor, Proceedings o j the Nationul Academy of Sciences of the United States ojilmerica, 1993, 90, 9852-9856. 18. A. Gaskell, S. Crennell, G. Taylor, Structure, 1995, 3, 1197-1205. 19. S. Crennell, E. Garman, G. Laver, E. Vimr, G. Taylor, Structure, 1994, 2, 535-544. 20. C. M. Egima, M. R. S. Briones, L. H. G. Freitas, R. P. F. Schenkman, H. Uemura, S. Schenkman, Molecular and Biochemical Parasitology, 1996, 77, 115-125. 21. F. Vandekerckhove, S. Schenkman, L. P. Decarvalho, S. Tomlinson, M. Kiso, M. Yoshida, A. Hasegawa, V. Nussenzweig, Glycohiology, 1992, 2, 541-548.

Rejerences

495

22. P. Scudder, J. P. Doom, M. Chuenkova, I. D. Manger, M. E. A. Pereira, Journal ofBiological Chemistry. 1993, 268, 9886-9891. 23. R. P. Prioli, J. S. Mejia, M. E. A. Pereira, Journal oflmmunology, 1990, 144, 4384-4391, 24. R. Cavallesco, M. E. A. Pereira, Journal qflmmunology, 1988, 140, 611--625. 25. B. A. Burleigh, N. W. Andrews, Annuul Reuiew ojMicrobiology, 1995, 49, 175- 200. 26. S. Schenkman, D. Eichinger, Parasitology Toduy, 1993, 9, 218-222. 27. S. Schenkman, D. Eichinger, M. E. A. Pereira, V. Nussenzweig, Annuul Reuiew of Microbiology, 1994, 48, 499-523. 28. M. Ming, M. Chuenkova, E. Ortegabarria, M. E. A. Pereira, Molecular and Biocheniicul Purasitology).y,1993, 59, 243-252. 29. M. E. A. Pereira, J. S. Mejia, E. Ortegabarria, D. Matzilevich, R. P. Prioli, Journul ofExperimental Medicine, 1991. 174, 179-191. 30. S. Schenkman, L. B. Chaves, L. C. P. Decarvalho, D. Eichinger, Jo~rizulqf Biologicul Chemistry, 1994, 269, 1970-7915. 31. L. E. Smith, D. Eichinger, G I ~ c ~ b i o l o g1997, j ~ , 7, 445--451. 32. A. K. J. Chong, M. S. Pegg, N. R. Taylor, M. Vonitzstein, European Journal ofBiochemistry, 1992,207, 335-343.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

30 Microbial Glycosidases Kenji Yamamoto, Su-Chen Li, and Yu-Teh Li

Glycosidases are useful for studying the structure, function, biosynthesis and catabolism of the complex carbohydrate chains in glycoconjugates. Glycosidases can be broadly classified into exo-glycosidases and endo-glycosidases according to their specificities. Exo-glycosidases release the monosaccharide residue from the nonreducing end of a sugar chain, whereas endo-glycosidases release oligosaccharides from a sugar chain by cleaving defined sites within the sugar chain. In addition to cleaving glycosidic linkages, glycosidases also catalyze the synthesis of sugar chains using transglycosylation reactions. This Chapter briefly summarizes various microbial glycosidases that are useful in glycobiology.

30.1 Exo-Glycosidases The four important specificities of exo-glycosidases are anomeric-, glycon-, linkageand aglycon-specificities. The anomeric-specificity is the most strict specificity of exo-glycosidases followed by glycon-specificity. The linkage- and aglycon- specificities on the other hand are generally not so strict. Some glycosidases are able to recognize two structurally related glycons. Glycosidases also exhibit varying degrees of preference toward the cleavage of certain glycosidic linkages. Although the aglycon-specificities of exo-glycosidases are the least well understood, they do play important roles in the degradation of glycoconjugates. 30.1.1 a-Glucosidase

a-Glucosidases hydrolyze the a-linked terminal D-glucose from oligo- or polysaccharides. Among the three disaccharides, Glcal+4Glc, Glcctl+2Glc and Glcal+3Glc, the disaccharide Glccll44Glc (maltose) is the preferred substrate for

498

30 Microbial Glycosidases

mold a-glucosidases. They also hydrolyze Glca1+6Glc at a very slow rate. Thus, these enzymes have been referred to as maltases. a-Glucosidases are also able to hydrolyze maltooligosaccharides. The rate of their hydrolysis decreases with increasing size of the substrate. The best studied microbial a-glucosidases are those isolated from Aspergillus niger [ 11 and the yeast Saccharomyces cerevisiue [2]. Transglucosylation activity of the enzyme has been used for glucosylation of various hydrophobic compounds to increase their hydrophilicity [ 3 ] . Since complex carbohydrates seldom contain a-linked glucose residues, a-glucosidases have not been widely used for structural analysis of glycoconjugates. Amylases can be regarded as a special class of a-glucosidases including glucoamylases, a- and P-amylases. Glucoamylases cleave the a-glucosyl residue from the non-reducing end of the a1+4 linked gluco-homopolysaccharides such as starch and dextrin. a-Amylases on the other hand are endo-a-glucosidases that convert starch and dextrin to oligosaccharides. Although P-amylases are exoglycosidases, they remove the maltose unit (Glca1+4Glc) from the non-reducing end of polysaccharide chains such as starch and dextrin. A large number of bacterial and mold amylases are used in large quantities for industrial gelatinization of starch. Distinct from amylases, yeast a-glucosidases cleave sucrose to form glucose and fructose. In contrast to a-amylases, they hydrolyze malto-oligosaccharides at a much slower rate. 30.1.2 P-Glucosidase P-Glucosidases act on the terminal P-linked Glc. Since P-glucosidases hydrolyze the disaccharide cellobiose (GlcP1+4Glc), they have been referred to as cellobiases. These enzymes comprise part of the cellulolytic enzyme systems. One of the most effective enzymes is that obtained from the fungus Trichoderma viride [4]. Cellobiases are involved not only in the degradation of cellulose but also in 0-D-glucosyl transfer reactions leading to the synthesis of P-linked gluco-oligosaccharides. Succesive P-D-glucosyl transfer or disproportionation of cellodextrins, with the formation of cello-oligosaccharides is brought about by the action of cellobiases of cellulolytic microorganisms. Although 0-glucosidases are important for the degradation of cellulose to glucose, they are seldom used for structural analysis of glycoconjugates. 30.1.3 a-Galactosidase a-Galactosyl linkages are widely distributed in glycoproteins, glycolipids and oligosaccharides of the raffinose family. The blood group B antigens contain the terminal a-Gal residue which is susceptible to certain a-galactosidases. Compared to those in higher plants and mammalian tissues, microbial a-galactosidases have not been studied in detail. Most microbial a-galactosidases only hydrolyze simple oligosaccharide substrates. They are not effective in releasing a-linked Gal from glycoproteins and glycolipids. For example, the crystalline a-galactosidase iso-

30.1 Exo-Glycosidases

499

lated from Mortierella vinacea [ 51 only hydrolyzed melibiose, raffinose, stachyose, Galal7-4Gal and Gala1 +4GalPI-glycerol. It was not effective in cleaving a-linked Gal from glycoproteins or glycolipids. The specificity of the a-galactosidase isolated from A . niger [6] was found to be similar to that isolated from M. uinacea. The enzyme prepared from Clostridium sporogenes [7] was the first a-galactosidase reported to convert blood group B antigen to blood group H (0)antigen. Although the linkage specificities of most of the microbial a-galactosidases have not been fully characterized, the enzyme from Xanthomonas manihotis [8] has been found to cleave the Galal +3Gal and Galal+6Glc linkages, but not the Galul-+4Gallinkage. The activity of this enzyme toward the Galal -16Gal linkage remains to be determined.

30.1.4 0-Galactosidase f3-Galactosidaseshave been found in d wide variety of microorganisms 191. The Escherichia coli P-galactosidase is well known due to its association with the discovery of the lactose operon and it is the first 0-galactoside to be isolated in crystalline form 191. Microbial p-galactosidases, synonymous with lactases, have been widely used to improve the digestion of lactose and to reduce lactose content in dairy products. Among the various microbial P-galactosidases, the enzymes from A . niger [6] and Streptococcus pneumoniae [ 101 have been used for structural studies of glycoconjugates. In general, the 1,3-1inked P-galactosides are refractory to microbial 0-galactosidases. In addition to hydrolyzing lactose, the A . niger 0-galactosidase hydrolyzes Gal01 i 4 G l c N A c and GalPl+6GlcNAc linkages. This 0-galactosidase releases 70-75% of the total Gal residues from asialo fetuin and asialo a]-acid glycoprotein 161. Similarly, this enzyme also liberates about 80%1of the Gal residues from sialic acid-free human chorionic gonadotropin 161. Unlike the E. coli f3galactosidase which has a neutral optimal pH, the optimum pH of the enzyme from A. niger is between 3.2-4.0. The p-galactosidase of S. pneumoniae [lo] is a l + 4-linkage specific p-galactosidase which has been used for structural analyses of both glycoproteins and glycolipids. Similarly to the E. coli P-galactosidase, this enzyme also has a neutral optimal pH. Both Aspergillus and Streptococcus P-galactosidases are able to hydrolyze the Gal from biantennary N-linked sugar chains [ 1 I]. However, they are not able to cleave the Gal from lacto-hi-fucopentaose (Galf31i4(Fuculi3)GlcNAc~li3Gal~li4Glc). After removal of the branched fucose residue, the Gal in the fucose-free lacto-N-neotetraose can then be easily cleaved by these enzymes. This indicates that the enzyme is sterically hindered by the branched fucose on the penultimate GlcNAc of the sugar chain. In contrast to other microbial P-galactosidases, the enzyme isolated from X. manihotis [8] has been found to preferentially hydrolyze the GalP 1 +3GlcNAc linkage. The gene encoding this 0-galactosidase showed a high degree of homology to several eukaryotic pgalactosidases [ 121. Recently, a Bacillus circulans P-galactosidase which specifically cleaves the P I 1 3 linkage has been expressed in E. coli 1131. The recombinant enzyme was able to synthesize GalPl i3GlcNAc and GalPl +3GalNAc by transglycosylation [ 131.

500

30 Microbial Glycosidases

30.1.5 a-Mannosidase

Most of the u-mannosidases purified from animal, plant and microbial sources have acidic pH optima and broad linkage and aglycon specificities. Two different microbial u-mannosidases that are specific for cleaving the Manul42Man linkage have been isolated from A . niger [14] and A . saitoi 1151. The extracellular ul-1 2-mannosidase from A . niger [ 141 readily hydrolyzes 2-O-u-r>-mannobiose and mannotriose, but does not hydrolyze p-nitrophenyl u-D-Man, u l 1 3 - , ul +4- or ul+6-linked mannobioses. It is also not able to liberate mannose from yeast mannan, ovalbumin, the u-linked mannoses in ul -acid glycoprotein, fetuin or human chorionic gonadotropin, whereas the A . saitoi u-1+2-mannosidase, a rather heatstable enzyme [ 151, is able to hydrolyze a-1+2-linked manno-oligosaccharides, yeast mannan as well as intact bovine ribonuclease. p-Nitrophenyl u-u-Man, Manal-13Man and Manul+bMan linkages are also refractory to A . saitoi u-l+ 2-specific mannosidase. As a result of its strict linkage specificity, this enzyme is useful for characterization of the high mannose type sugar chains. The molecular cloning and the cDNA sequence encoding this enzyme have been reported [ 161. Another u-mannosidase distinct from the A. niyer u-1,2 linkage specific umannosidase was found to hydrolyze p-nitrophenyl-u-D-Man and the mannose residues linking a-17-4 and u-17-6 to Man or GlcNAc residues [17]. After treating the asialo-ul -acid glycoprotein and asialo-fetuin glycopeptide with P-galactosidase and p-N-acetylglucosaminidase, the exposed mannose residues in the sugar chains are hydrolyzed by this enzyme. The mannose residues in the glycopeptide prepared from human chorionic gonadotropin are also susceptible to this enzyme. Another enzyme which is different from the above a-1+2 specific u-mannosidase has also been prepared from A . saitoi [18]. This enzyme is strongly activated by Ca2+ ions and is not able to hydrolyze p-nitrophenyl a-D-Man. The rates of hydrolysis of different mannobioses by this u-mannosidase are in the order: M a n u l i 3 M a n > Manul+6Man>Manul-t2Man [ 181. Recently, three linkage-specific u-mannosidases have been found in the cellextract of X manihotis [8];one is specific for 1+2- and 1+3-linkages, the second is specific for only the l+6-linkage, and the third is specific for 1-13- and 1+ 6-linkages. The I ,6-specific u-mannosidase was reported to be highly specific in catalyzing the hydrolysis of only the terminal Manul7-6Man linkage which is attached to an unbranched saccharide chain [8]. 30.1.6 fl-Mannosidase

P-Linked D-mannoses are the main constituents of plant mannans. They also occur at the core region of asparagine-linked sugar chains in glycoproteins. As compared with other glycosidases, the sources of P-mannosidases are rather limited. The enzymes that cleave the P-linked D-mannoses have been found mainly in animal and plant tissues. The crude enzyme preparations that are used commercially to degrade plant fibers such as cellulases and hemicellulases are often contaminated with pmannosidase activities. Microbial P-mannosidases have been isolated only from

Pol-vporus suljiureus [19] and A . niger [20]. The enzymes from both sources have acidic optimal pH and hydrolyze p-nitrophenyl-P-mannoside. The enzyme from P. suljiureus hydrolyzes ManPl+4GlcNAc~l+4GlcNAc+Asn, ManPl+4Man, Man@l+6Gal, and ManPIi4GlcNAc [ 191. The A . niyer P-mannosidase hydrolyzes Man01+4GlcNAc, ManPli4GlcNAcPl i4GlcNAc-Asn-Lys and M a n P l i 4GlcNAcPl+4(Fucal +b)GlcNAc-tAsn-peptide. This enzyme also slowly hydrolyzes P1+4 linked mannobiose, mannotriose and guar gum [ZO]. 30.1.7 P-N-Acetylhexosaminidase

N-Acetyl-D-glucosamine (GlcNAc) and N-acetyl-D-galactosamine (GalNAc) are common constituents of both glycoproteins and glycolipids. The exoskeletons of crustaceans and insects consist of chitin which is a P-144-linked homopolysaccharide of GlcNAc. In nature, microbial P-N-acetylhexosaminidases are closely associated with the bio-degradation of chitin. In addition to cleaving P-linked GlcNAc, most P-N-acetylhexosaminidasesalso hydrolyze 0-linked GalNAc at a slower rate. The name “P-N-acetylhexosaminidase” stemmed from this dual glycon specificity. P-N-Acetylhexosaminidases are widely found in the culture broth of microorganisms. Among the microbial P-N-acetylhexosaminidases, the enzyme from S. pizeumoniae has been widely used for structural studies of glycoconjugates 1211. This enzyme readily hydrolyzes GlcNAcP1+2Man, GlcNAcpl i 3 G a l and the GlcNAcP 1 4 6 G a l linkages in asparagine-linked sugar chains. However, the GlcNAcPIi4Man (bisecting GlcNAc) and GlcNAcPl i 6 M a n linkages are refractory to this enzyme. After removal of sialic acid and galactose residues from a1 -acid glycoprotein, the exposed GlcNAc is also susceptible to this enzyme. The enzyme from A. niger [ 61 liberates GlcNAc from chito-oligosaccharides containing 2 to 6 GlcNAc residues. It also hydrolyzes the terminal GlcNAc residues of glycopeptides derived from ovalbumin and the exposed GlcNAc residues in al-acid glycoprotein and human chorionic gonadotropin after treatment with P-galactosidase and sialidase. The transglycosylation activity and the reverse hydrolytic activity of A. oryzue @N-acetylhexosaminidase have been used to synthesize N-acetylchitooligosaccharides PI.

30.1.8 a-N-Acetylgalactosaminidase Both glycoproteins and glycolipids contain a-linked GalN Ac, a linkage between the GalNAc and serine or threonine at the core of mucin type glycoproteins. The blood group A epitope found in glycoproteins and glycolipids, and Forssman’s hapten also contain a-linked GalNAc residues. a-N-Acetylgalactosaminidases, useful for structural analysis of glycoconjugates, have been found in the culture media of A . iziger [23], Acremonitim sp. [24] and Clostridium perfvingens [ 2 5 ] .The enzymes from these sources are capable of cleaving p-nitrophenyl-a-GalNAc and the a-GalNAc linkages in glycoproteins and glycolipids. The pH optima of microbial a-N-

502

30 Microhiul Glycosiduses

acetylgalactosaminidases are around pH 6. The enzyme isolated from C. perfringens is stimulated by Ca2+. The a-N-acetylgalactosaminidases from Acremonium and Clostridium are able to hydrolyze the 0-glycosidic linkage between the GalNAc and the serine in asialo ovine submaxillary mucin. However, this is not the case for the Aspergillus a-N-acetylgalactosaminidase. The Acremonium a-N-acetylgalactosaminidase releases GalNAc from various blood group A active glycoproteins isolated from porcine gastric mucin, horse gastric mucin and human salivary mucin. It also converts type A erythrocytes to type 0 (H) erythrocytes. The enzyme is also able to release the a-linked GalNAc from Forssman hapten and the sialidase-treated bovine submaxillary mucin.

30.1.9 a-L-Fucosidase L-Fucose (Fuc) occurs widely in both glycoproteins and glycolipids. This sugar is also found in various blood group substances, milk oligosaccharides, marine algal polysaccharides and plant gum. Most microbial a-L-fucosidases are a1+2-specific fucosidases. They do not hydrolyze synthetic substrates such as p-nitrophenyl a - ~ Fuc. In contrast, most of the a-L-fucosidases from animal sources readily hydrolyze p-nitrophenyl a - ~ - F u cand have relatively wide substrate specificities towards naturally occurring glycoconjugates. a-L-Fucosidase isolated from Aspergillus niger [26] is highly specific for the Fucal +2Gal linkage. This enzyme hydrolyzes 2-O-a-~-fucosyl-~-Gal and 2’fucosyllactose (Fucal+2Galj31+4Glc). It also hydrolyzes the terminal Fucal+ 2Gal linkage in intact as well as the desialylated canine and porcine submaxillary mucins. Incubation of blood group H (0)substances with this enzyme resulted in the complete loss of blood group H activity. This enzyme, however, does not hydrolyze p-nitrophenyl ~-L-Fuc. The L-fucose residues linked a1+3- and a1+4- to GlcNAc in lacto-N-fucopentaose I1 and 111, respectively, are also refractory to this enzyme. A different a-L-fucosidase isolated from the culture fluid of A. niger [27] was reported to hydrolyze the Fucal+6GlcNAc, but failed to cleave Fucal+ 2GlcNAc, Fucal-3GlcNAc and Fucal +4GlcNAc. This a-L-fucosidase releases Fuc from bovine IgG glycopeptide and asialoagalactofetuin. The Fuc residues in these two glycoconjugates are a 1 1 6 linked to the core GlcNAc moiety of the asparagine-linked oligosaccharide. The a-L-fucosidase from Bacillus circulans isolated from soil can also act on a-l+6-~-fucosyllinkages [28].This enzyme readily hydroc 2’-fucosyllactose,but not blood group H substances. lyzes p-nitrophenyl ~ - L - F uand The substrate specificity of the a-L-fucosidase isolated from B. jiulminans [29] is very similar to that isolated from A. niger. This a-L-fucosidase also specifically hydrolyzes the terminal Fucal i 2 G a l in glycoproteins, glycopeptides and oligosaccharides. Similarly to that from A. niger, the enzyme does not cleave p-nitrophenyl ~ - L - F uor c the a1+3 and a l h 4 linkages in lacto-N-fucopentaose I1 and 111. The only difference between these two enzymes is their optimal pH; unlike the A. niger enzyme which has an acidic optimal pH (3.6-4.0), the optimal pH of the enzyme from B. julminans is pH 6.3-6.6. Thus, a-L-fucosidase from B. fulminans can be used to treat living cells.

30.1 Exo-Glycosidasrs

503

C. perfringens also contains an a1-2 specific L-fucosidase [30]. This enzyme does not cleave simple methyl or nitrophenyl fucosides. Similarly to the enzymes isolated from A. niger and B. fulminans, the most susceptible oligosaccharides are 2’-fucosyllactose and lacto-N-fucopentaose I. This extracellular enzyme readily releases L-FUCfrom the high molecular weight glycoproteins such as blood group Hpositive porcine submaxillary mucin. A novel type of a-L-fucosidases have been prepared from X. manihotis [8] and Streptonzycrs sp. [31]. These two fucosidases exhibit high specificity toward the a1+3 and a-1+4 L-fucosidic linkages [8]. Although the enzymes cannot hydrolyze p nitrophenyl ~x-L-Fuc,they are able to cleave Fucal+4GlcNAc in the sialyl-Lewis X structure. 30.1.10 P-D-Fucosidase

As a result of the structural similarity between D-fucose and D-galactose, most P-D-galactosidases are able to hydrolyze P-u-fucosides. Therefore, it is difficult to place P-D-fucosidases as a separate class of enzymes. The only exception to date is the P-D-fucosidase of A. phoenicis. This enzyme shows high specificity toward p nitrophenyl-P-D-fucoside [32]. However, it does not hydrolyze p-nitrophenyl-P-DGal or p-nitrophenyl-P-D-Glc. 30.1.I I Sialidase

Sialic acids are the best studied monosaccharides found in glycocojugates, because of their wide distribution and well recognized biological functions. Sialidases, the enzymes responsible for releasing sialic acids from various glycoconjugates, have been the subject of intensive studies since the discovery of the receptor-destroying enzyme in Mbrio cholerae [33]. There have been several excellent reviews and monographs dealing with sialic acids and sialidases [34-361. Sialidases are widely distributed in microorganisms and animal tissues [34-361. Several microbial sialidase genes have been cloned and their crystal structures elucidated [34, 371. Among the viral sialidases, those of influenza viruses have been studied most extensively [38]. Due to its strict linkage specificity, the Newcastle disease virus sialidase [39] is the only viral sialidase which has been used for structural analysis of glycoconjugates. Among the sialidases isolated from a wide variety of bacteria and fungi, those from V. cholerae (401, C. perfringens [41], Arthrobucter uregfaciens [42] and Salmonella typhimurium [43] are most frequently used in studying the structure and function of glycoconjugates. These four microbial sialidases and the Newcastle disease virus sialidase are commercially available. With the exception of the sialidase from A . ureafaciens, the viral and bacterial sialidases generally hydrolyze sialic acids that are linked through the a2+3Gal linkage faster than the a 2 1 6 G a l linkage [44]. The NeuSAca243NeuSAc linkage is hydrolyzed more slowly than these two linkages. Sialidases usually hydrolyze the ketosidically linked Neu5Ac faster than the ketosidically linked NeuSGc [45]. In

504

30 Microbial Glycosidases

contrast, the Avthrobucter sialidase hydrolyzes the Neu5Acu2-6Gal linkage faster than the NeuSAca2i3Gal linkage followed by the NeuSAca2-8NeuSAc linkage [42]. Only the Newcastle disease sialidase [39] and S. typhimurium sialidase [43] show much greater preference for cleaving the NeuSAca2+3Gal linkage over the NeuSAca2-6Gal linkage. When using these two enzymes to discriminate the a2+3Gal- and the a2+6Gal-linked sialic acids, extreme caution should be taken to control the incubation conditions. At high enzyme concentrations, both sialidases are also able to cleave the NeuSAca2-6Gal linkage. The NeuSAcu2+8NeuSAc linkage is also susceptible to these two sialidases. The NeuSAca2-6GalNAc linkage found in submaxillary glycoproteins and the cholinergic neuron-specific a-series gangliosides is susceptible to viral as well as bacterial sialidases [46]. To date, only the leech sialidase L has been found to exhibit absolute linkage specificity toward the Neu5Aca2-3Gal linkage [46]. Structural comparisons have shown a close relationship between sialidase L and bacterial sialidases [47]. The sialic acids at the peripheral positions of gangliosides are susceptible to sialidases in the absence of a detergent. However, the sialic acid linked a2-3 to the internal Gal of GMl or GM2 is refractory to sialidases. Therefore, in the absence of a detergent, polysialogangliosides with a gangliotetraosyl chain are converted into GM1 as the final product upon digestion with sialidases. The reason for the resistance of this specific sialic acid to sialidases is still not well understood. It has been suggested that the resistance of NeuSAc in GM 1 or GM2 to enzymatic hydrolysis is caused by “steric hindrance” (for review see [36]). In the presence of a bile salt, the sialic acids in GM1 and GM2 can be hydrolyzed by clostridial sialidase [48]. In the presence of sodium cholate, Arthrohacter sialidase was found to be 100-fold more effective than clostridial sialidase in converting GM1 to asialo GM1 [49]. It is intriguing that the lipid-free sialooligosaccharide derived from GM 1 is resistant to sialidases in the presence or absence of a detergent.

30.1.12 KDNase KDN (2-keto-3-deoxy-~-glycero-~-gufucto-nononic acid), a deaminated neuraminic acid analogue first isolated from the polysialoglycoproteins of rainbow trout eggs [50],has been found to be widely distributed in nature (for review see [35]). Structurally, KDN is very similar to NeuSAc, but KDN-containing glycoconjugates are refractory to bacterial sialidases [51]. KDNase, an enzyme capable of exclusively hydrolyzing various KDN-containing glycoconjugates, has been induced in the bacterium Sphingobacterium multivorum 1521.

30.1.13 a-L-Rhamnosidase L-Rhamnose (6-deoxy-~-mannose)is widely distributed in plant pigments, gum, mucilage, and bacterial heteropolysaccharides. Citrus flavonoid compounds such as naringin and hesperidin also contain L-rhamnose. Although not a common enzyme, a-L-rhamnosidase has been widely used in the food industry. For example, U-L-

rhamnosidase has been used industrially to remove naringin and the main bitter components of citrus juices. a-L-Rhamnosidase can be purified from commercial naringinase and hesperidinase preparations obtained from A. niger [ 53, 541. This enzyme is active towards naringin, hesperidin and rutin, but not quercitrin. The enzyme has been used to release L-rhamnose from geranyl p-D-rutinoside and 2phenylethyl P-D-rutinoside found in grape juice and wine to enhance the aroma.

30.1.14 P-Xylosidase Xylan, the major hemicellulose component of plant cell walls, is the most abundant polysaccharide after cellulose. The backbone of xylan consists of Plt4-linked Dxylose residues substituted with acetyl groups, glucuronosyl- and arabinosyl sidechains. The xylo-oligosaccharides released from xylan by endo-P 1+4-xylanases are further graded by P-xylosidases, enzymes found in various bacteria and fungi. Most bacterial P-xylosidases are intracellular enzymes. whereas fungal P-xylosidases remain cell-associated during the early stage of growth but may be secreted into the medium in the later stages. p-Xylosidases are inducible and subjected to catabolite repression by glucose. The carbon sources known to induce both P-xylosidase and endo-pli4-xylanase in A. niyer are xylan, xylobiose and D-XYIOSC [ 5 5 ] . The best known bacterial p-xylosidase is the intracellular enzyme of B. pumilus [56]. Although the enzymes from both Bucillus and Aspergillus can hydrolyze xylobiose, their ability to hydrolyze the xylosyl pl-0-serine linkage present in the core region of proteoglycans remains to be established.

30.2 Endo-Glycosidases Endoglycosidases cleave specific endo-glycosidic linkages within the sugar chains. Endoglycosidases can recognize the,glycon moiety containing more than one sugar residue, a property quite different from the glycon specificity of exoglycosidases.

30.2.1 Endo-P-N-acetylglucosaminidase Endo-P-N-acetylglucosaminidasesare the best studied endoglycosidases that cleave the N , N'-diacetylchitobiosyl linkage in the core region of asparagine-linked sugar chains in various glycoproteins. The following microbial endo-P-N-acetylglucosaminidases have been widely used for studying the structure and function of N-linked sugar chains: endo D from Streptococcus pneumoniue (formerly Diplococcus pneumoniue) [57],endo H from Streptornyces plicutus (formerly Streptomyces yriseus) [58], endo CI and CII from Clostridium perfrinyens [59], and endo F from Fluvobacterium meninyosepticurn [60]. Endo H, D, CI and CII cleave the N,N'-diacetylchitobiosyl linkage in certain high mannose type and hybrid type sugar chains.

506

30 Microbial Glycosiduses

The glycon requirement for both endo D and CI was found to be Manal-, 3Man~li4GlcNAcP-[61, 621. Endo H was shown to recognize a tetrasaccharide structure, Manal-+3Manal+6Man~l+4GlcNAcP- while the glycon requirement of endo CII was found to be a branched pentasaccharide, Manal-+3Manal+ 6(Manal+3)Man~1+4GlcNAcP- [61, 621. Endo F was a preparation initially found to cleave the N , N’-diacetylchitobiosyl linkage in both the high mannose type and the complex type sugar chains [60]. This preparation was subsequently shown to contain three distinct endoglycosidases, Endo FI, F2 and F3 [63]. While the specificity of Endo F1 is similar to that of endo H, endo F2 preferentially hydrolyzes the biantennary complex type, and also high-mannose type sugar chains. Endo-F3 hydrolyzes both the bi- and the triantennary complex type oligosaccharides [ 641. In addition to its hydrolytic action, endo-P-N-acetylglucosaminidases have also been shown to catalyze transglycosylation reactions [ 65-67].

30.2.2 Peptide-N-glycanase F After the discovery of the presence of endo F in an enzyme preparation isolated from F. meningosepticum [60], this preparation was shown to also contain a unique deglycosylating amidase, identified as peptide-N4-(N-acetyl-P-~-glucosaminyl) asparagine amidase, also known as peptide N-glycanase (PNGase F) [68]. PNGase F acts on the glycosylamine linkages of a broad range of asparagine-linked high mannose, hybrid and complex type sugar chains. Thus, this versatile microbial enzyme has been widely used for detaching the N-linked glycans from glycoproteins and glycopeptides.

30.2.3 Endo-a-N-acetylgalactosaminidase The endo-a-N-acetylgalactosaminidasereleases the unsubstituted disaccharide GalPl+3GalNAc linked to serine or threonine of glycoproteins or glycopeptides. This enzyme is also called 0-glycanase, and was first found in the culture fluid of C. perfringens [69]. A similar enzyme was subsequently found in S. (D.) pneumoniue [70, 711. The enzyme liberates GalPl+3GalNAc from asialofetuin, porcine submaxillary asialomucin, asialoglycophorin and anti-freeze glycoprotein [72]. A similar enzyme has also been induced in Alculiyenes sp. [73]. A novel type of endoa-N-acetylgalactosaminidase induced in Streptomyces sp. was reported to release tetrasaccharide Gal~l+3(Gal~l-+4GlcNAc~1+6)GalNAc from bovine asialofetuin [ 741. The specificity of this interesting endoglycosidase remains to be established. The Diplococcus enzyme was found to also catalyze transglycosylation and show reversed hydrolysis activity [75]. The enzyme has been shown to transfer the disaccharide from an asialoglycoprotein substrate to glycerol to form GalP 1+ 3 G a l N A c a l j 1(3)-glycerol. The enzyme was also found to transfer the free GalPl-+3GalNAc to glycerol, D-Glc, D-Gal, p-nitophenol and threonine etc. by reversed hydrolytic reaction.

30.2 Endo-Gl)~cosiduses

507

30.2.4 Endo-P-galactosidase Endo-P-galactosidases hydrolyze the internal GalP 1+4GlcNAc or GalPl+4Glc linkages. The endo-P-galactosidases isolated from Escherichiu jieundii [76], E kerutolyticus [77] and Bucteroides fvugilis [78] are able to cleave the endo-P-galactosyl linkages (XPl+3GalPl+4Y ) in polylactosaminoglycans. The sugar residue X is usually GlcNAc, but Y can be GlcNAc or Glc. Thus, in addition to depolymerizing polylactosaminoglycans, they can also hydrolyze the GalP 1 4 4 Glc linkage at the reducing end of lacto-N-tetraosyl chain. For example, this type of endoglycosidases can convert Gal~l-.4GlcNAc~l+3Gal~1+4GlcCer to GalPl i4 Glc N A c P l + 3Gal and GlcCer by cleaving the endo-P-galactosyl linkage of the lactosyl core. The lactosyl cores in glycosphingolipids with ganglio-series and globo-series sugar chains are not hydrolyzed by this enzyme. The endo-P-galactosidase isolated from E. Jieundii also depolymerizes the sugar chains in polyglycosylceramides [76]. The hydrolysis of glycosphingolipids by endo-P-galactosidases is also facilitated by the presence of a detergent such as bile salt. In general, sulfation or attachment of a sugar unit at C-6 of the Gal residue renders the endo-$galactosyl linkage refractory to this type of endo-P-galactosidase. A similar endo-P-galactosidase has been induced in Pseudomonas sp. [79]. This endo-P-galactosidase, however, was not able to cleave the endo-P-galactosyl linkage in lacto-N-tetraose. Endo-P-galactosidases have been extensively used for structural characterization of polylactosaminoglycans and related glycoconjugates. An interesting endo-P-galactosidase capable of releasing the blood group A or B trisaccharide linked P1+4 to the GlcNAc of lacto-N-triaosyl chain has been detected in the culture filtrate of S. (D.)pneumoniue [go]. Another unusual endo-P-galactosidase which releases Galal +3Gal from Galal +3GalP 1+ 4GlcNAc~1+3GalP1+4GlcCer has been isolated from the culture supernatant of C. perfiingens [811. The substrate specificities and possible biological functions of these two endo-P-galactosidases remain to be established.

30.2.5 Endoglycoceramidase Endoglycoceramidases (EGCases ) are special endoglycosidases that act exclusively on glycosphingolipids at the junction between the sugar chain and the ceramide. A different name, “ceramide glycanase”, has been proposed for a similar enzyme isolated from annelids [82]. The first microbial EGCase was induced in Rhodococcus using bovine brain gangliosides as the inducer [83]. A mutant strain of Rhodococcus sp. [84] was found to produce three molecular species of EGCases (I, I1 and 111) without inducer. EGCase I is able to detach the intact sugar chains from neutral glycosphingolipids and gangliosides with globo-, ganglio- and lacto- series sugar chains. The glycosphingolipids with globo-type sugar chains are refractory to EGCase 11. Among these three EGCases, only EGCase 111 is able to detach sugar chains from glycosphingolipids with gal-series sugar chains. The hydrolysis of glycosphingolipid substrates by EGCases is also enhanced by detergents such as sodium cholate or Triton X- 100. Interestingly, the culture filtrate of Rhodococcus

508

30 Microbial Glycosidases

sp. M777 was found to contain two activator proteins capable of stimulating the hydrolysis of glycosphingolipids by EGCase [ 851. The two activator proteins were named activator I (minor species) and 11 (major species) and stimulated EGCase I and 11, respectively, to hydrolyze glycosphingolipids. A membrane-bound EGCase has also been prepared from Corynebacterium sp. [86]. The substrate specificity of this enzyme is very similar to that of Rhodococcus enzyme. This enzyme also catalyzes the glycosyltransfer reaction [ 871.

References 1. J.H. Pazur, Glucoamylase from Aspergillus niger. Methods Enzymol., 1972, 28, 931-934. 2. H. Halvorson, a-Glucosidase from yeast. Methods Enzymol., 1966, 8, 559-562. 3 . H. Murase, R. Yamauchi, K. Kato. T. Kunieda and J. Terao, Synthesis of a novel vitamin E derivative, 2-(a-D-glucopyranosyl)methyl-2,5,7,8-tetramethylchroman-6-ol, by a-glucosidasecatalyzed transglycosylation. Lipi~is,1997, 32, 73-78. 4. D.R. Whitaker, Cellulases. The Enzynzes, 1971, 5 , 273-290. 5. H. Suzuki, S.-C. Li and Y.-T. Li, a-Galactosidase from Mortierella tiinaceu, crystallization and properties. J. Biol. Chem., 1970. 245, 781 -786. 6. O.P. Bahl and K.M.L. Agrawal, a-Galactosidase, P-galactosidase, and P-N-acetylglucosaminidase from Aspergillus nicger. 12.lrthod.s Enzymol., 1972. 28, 128-734. 7. S. Iseki, K. Furukawa and S. Yarnamoto, B substance-decomposing enzyme produced by an anaerobic bacterium. Proc. J. Acad., 1959, S S , 507 -517. 8. S.T. Wong-Madden and D. Landry, Purification and characterization of novel glycosidases from the bacterial genus Xanthomonas. Glycohiology, 1995, 5, 19-28. 9. K. Wallenfels and R. Weil, 0-Galactosidase. The Enzymes, 1972, 7, 617-663. 10. K. Kojima, M. Iwamori, S.I. Takasaki, K. Kubushiro, S. Nozawa, R. lizuka and Y. Nagai, Diplococcdl P-galactosidase with a specific reacting to p-1-4 linkage but not to p-1-3 linkage as a useful exoglycosidase for the structural elucidation of glycolipids. Anal. Biochem., 1987, 16.5, 465-469. 11. R. Zeleny, F. Altmann and W. Praznik, A capillary electrophoretic study on the specificity of P-galactosidases from Aspergillus oryzue, Excherichia coli, Streptococcus pneumoniae, and Canuvulia ens;jormis (jack bean). Anal. Biochem., 1997, 246, 96-101. 12. C.H. Taron, J.S. Benner, L.J. Hornstra and E.P. Guthrie. A novel P-galactosidase gene isolated from the bacterium Xunthomonus nzanihotis exhibits strong homology to several eukaryotic Pgalactosidase, Glycohiology, 1995, 5, 603--610. 13. H. Fujimoto, M. Miyasato, Y. Ito, T. Sasaki and K. Ajisaka, Purification and properties of recombinant 0-galactosidase from Bacillus circulun. Glycomnj. J., 1998, IS, 155- 160. 14. N. Swaminathan, K.L. Matta and O.P. Bahl, 1,2-a-D-Mannosidase from Aspergillus niger. Methods Enzymol., 1972, 28, 744-749. 15. E. Ichishima, M. Arai, Y. Shigematsu, H. Kumagai and R.S. Tanaka, Purification of an acidic a-D-mannosidase from Aspergillus saitoi and specific cleavage of I ,2-a-D-mannosidic linkage in yeast mannan. Biochim. Biophys Actu, 1981, 658, 45-53. 16. T. Inoue, T. Yoshida and E. Ichishima, Molecular cloning and nucleotide sequence of the 1,2a-D-mannosidase gene, msdS, from Aspergillus suitoi and expression of the gene in yeast cells. Biochim. Biophys. Actu, 1995, 1253, 141-145. 17. K.L. Matta and O.P. Bahl, a-Mannosidase from Aspergillus rziger. Methuh Enzymol., 1972,28, 749%755. 18. J. Amano and A. Kobata, Purification and characterization of a novel a-mannosidase from Aspergillus saitoi. J. Biochem. (Tokyo), 1986, 99, 1645-1 654. 19. C.C. Wan, J.E. Muldrey, S.-C. Li and Y.-T. Li, P-Mannosidase from the mushroom Polyporus sulfureus. J. Biol. Chem., 1976, 251, 4384-4388. 20. A.D. Elbein, S. Adya and Y.C. Lee, Purification and properties of P-mannosidase from Aspergillus niger. J. Biol. Chem., 1997, 252, 2026-2031.

References

509

21. K. Yamashita, T. Ohkura, H. Yoshima and A. Kobata, Substrate specificity of diplococcal pN-acetylhexosaminidase, a useful enzyme for the structural studies of complex type asparaginelinked sugar chain. Biochem. Biophys. Rex Comniun., 1981, 100, 226-232. Packwood, C.J. Samuel, P. Critchley and D.H. Crout, Glycosidase-catalysed oligosaccharide synthesis: preparation of N-acetylchitooligosaccharides using the p-N-acetylhexosaminidase of Asperyillus oryzae. Curbohydr. Res., 1995, 279, 293--305. 23. M.J. McDonald and O.P. Bahl. a-N-Acetylgalactosaminidase from Aspergillus niyer. Methods Enzymol.. 1972, 28, 734--738. 24. S. Kadowaki, T. Ueda, K . Yamamoto, H. Kumagai and T. Tochikura, Isolation and characterization of a blood group A substance-degrading a-N-acetylgalactosaminidase from an Acremonium sp. Ayric. Biol. Cheni., 1989, 53, 11 1 -120. 25. G.N. Levy and D. AminofT, Purification and properties of a-N-acetylgalactosaminidase from Clostridium perfringens. J. Biol. Chem., 1980, 255, 11737-1 1742. 26. O.P. Bahl, 1,2-a-L-Fucosidase from A.spergil1u.s nigrr. Metliods Etzzymol., 1972, 28, 738-743. 27. S. Yazawa, R. Madiyalakan, R.P. Chawda and K.L. Matta, a-L-Fucosidase from Aspergillus niyer: demonstration of a novel a-L-( I +6)-fiicosidase acting on glycopeptides. Bioclzeni. Bioplivs. Res. Comrnun., 1986, 136. 563-569. 28. Y. Tsuji, K. Yamamoto and T. Tochikura. Purification and some properties of a novel a-Lfucosidase capable of acting on a-(1 +6)-L-fucosidic linkages from Bacillus circulans M28. J. Biochem. ( T o k y o ) , 1990, 108, 235-240. 29. N. Kochibe, Purification and properties of a-L-fucosidase from Bucillusfulminuns. J. Biochem. ( T o k y o ) , 1973, 74, 1141-1149. 30. D. Aminoff, 1,2-a-L-Fucosiddse from Clostridium perpingens, Met1iod.s.Enzymol., 1972, 28, 763-769. 3 I . M. Sano, K. Hayakawa and I. Kato, Purification and characterization of a-L-fucosidase from Streptomyces species. J. Biol. Chem., 1992. 267, 1522-1527. 32. Y.-C. Zeng, Y.-T. Li, Y.-J. Gu and S.-Z. Zhang, Purification and characterirdtion of a strictly specific P-D-fucosidase from A.spergi1lus plioenicis. Arch. Biochem. Biophys., 1992,298, 226-230. 33. F.M. Burnet and J.D. Stone, The receptor-destroying enzyme of Vibrio chokerue. Austr. J. Exp. Biol. and Med Sci. 1947, 25, 227-233. 34. R. Schauer and J.P. Kamerling, Chemistry, biochemistry and biology of sialic acids. in “Glycoproteins II” (J. Montreuil, J.F.G. Vliegenthart and H. Schachter, editors), 1997, Elsevier Science B.V.; Amsterdam, pp 243-402. 35. A. Gottschalk and A.S. Bhargava, Neuraminidase. The Enzymes, 1971, 5, 321-342. 36. M. Saito and R.K. Yu, Biochemistry and function of sialidases. in “Bioloyy o f t h e siulic ucids” (A. Rosenberg, editor). 1995, Plenum Press, New York, pp 261-313. 37. E.R. Vimr, Microbial sialidases: does bigger always mean better? Trend.s in Microbiology, 1994, 2. 271-277. 38. G.M. Air and W.G. Laver, The neuraminidase of influenza virus, Proteins, 1989, 6, 341-356. 39. J.C. Paulson, J. Weinstein, L. Dorland, H. van Halbeek and J.F. Vliegenthart, Newcastle disease virus contains a linkage-specific glycoprotein sialidase. J. Bid. Chern., 1982, 257, 1273412738. 40. G.L. Add, E.L. French and P.E. Lind, Purification and properties of neuraminidase from Vibrio cholerae. J. Gen. Microhiol., 1961. 24, 409 -421. 41. J.T. Cassidy, G.W. Jourdian and S. Roseman, The sialic acid. VI. Purification and properties of sialidase from Clostridiurn perfringens. J. Bid. Chem., 1965, 240, 3501-3506. 42. Y. Uchida, Y. Tsukada and T. Sugimori, Distribution of neuraminidase in Arthrobacter and the purification by affinity chromatography. J. Biochem. ( T o k y o ) , 1977, 82, 1425-1433. 43. L.L. Hoyer, P. Roggentin, R. Schauer and E.R. Vimr, Purification and properties of cloned Salmonella typhimurium LT2 sialidase with virus-typical kinetic preference for sialyl a 2,3 linkages. J. Biochem. ( T o k y o ) , 1991, 110, 462-467. 44. A.P. Corfield, H. Hiad, J.C. Paulson and R. Schauer. The sDecificitv of viral and bacterial sialidases for a(2-3) and a(2-6)-linked sialic acids in glycoproteins. B i h i m . Biophys. Acta, 1983, 744, 121-126. 45. A.P. Corfield, R.W. Veh, M. Wember, J.C. Michalski and R. Schauer, The release of N-acetyl and N-glycolloyl-neuraminic acid from soluble complex carbohydrates and erythrocytes by bacterial, viral and mammalian sialidases. Bbchem. J . , 1981, 197, 293-299.

5 10

30 Microbial Glycosidases

46. M.-Y. Chou, S.-C. Li, M. Kiso, A. Hasegawa and Y.-T. Li, Purification and characterization of sialidase L, a NeuAc a(2-3)Gal-specific sialidase. J. Biol. Chem., 1994, 269, 18821-18826. 47. Y. Luo, S.-C. Li, M.-Y. Chou, Y.-T. Li and M. Luo, The crystal structure of an intramolecular trans-sialidase with a NeuAcct(2-3)Gal-specificity. Structure, 1998, 6, 521-530. 48. D.A. Wenger and S. Wardell, Action of neuraminidase from Clostridium perfringens on brain gangliosides in the presence of bile salts. J. Neurochem. 1973, 20, 607-612. 49. M. Saito, K. Sugano and Y. Nagai, Action of Arthrobucter ureafiiciens on sialoglycolipid substrates. J. Biol. Chen?. 1979, 254, 7845-7854. 50. D. Nadano, M. Iwasaki, S. Endo, K. Kitajima, S. Inoue and Y. Inoue, A naturally occurring deaminated neuraminic acid, 3-deoxy-D-yl~~cero-D-gu/ucto-nonulosonic acid (KDN). J. Biol. Chem., 1986,261, 1 1550- 1 1557. 51. S. Kitazume, K. Kitajima, S. Inoue and Y. Inoue, Detection, isolation, and characterization of oligo/poly(sialic acid) and oligo/poly(deaminoneuraminic acid) units in glycoconjugates. Anal. Biochem., 1992,202, 25-34. 52. K. Kitajima, H. Kuroyanagi, S. Inoue, J. Ye, F.A. Troy and Y. Inoue, Discovery of a new type of sialidase, “KDNase” which specifically hydrolyzes deaminoneuraminyl (3-deoxy u-glycero~-ga/ucto-2-nonulosonicacid) but not N-acetylneuraniinyl linkages. J. Bid. Chem. 1994, 269, 21415-21419. 53. Y. Kurosawa, K. Ikeda and F. Egami, a-L-Rhamnosidases of the liver of Turbo curnutus and Aspergillus niger. J. Biochem. (Tokyo), 1973, 73, 31-31. 54. P. Manzanares, L.H. de Graaff and J. Visser, Purification and characterization of an a-Lrhamnosidase from Aspergillus niger. FEMS Microhiol. Lett., 1997, 157, 279-283. 55. N.N.M.E. van Peij, J. Brinkmann, M. Vrsanska, J. Visser and L.H, de Graaff, P-Xylosidase activity, encoded by xlnD, is essential for complete hydrolysis of xylan by Aspergillus niger but not for induction of the xylanolytic enzyme spectrum. Eur. J. Biochem., 1997,245, 164-173. 56. W. Panbangred, 0. Kawaguchi, T. Tomita, A. Shinmyo and H. Okada, Isolation of two Pxylosidase genes of Bucillus pumilus and comparison of their gene products. Eur. J. Biochern., 1984,138, 267-273. 57. N. Koide and T.Muramatsu, Endo-b-N-acetylglucosaminidase acting on carbohydrate moieties of glycoproteins. Purification and properties of the enzyme from Diplococcus pneumoniue. J. Biol. Chern., 1974,249, 4897-4904. 58. A.L. Tarentino and F. Maley, Purification and properties of an endo-P-N-acetylglucosaminidase from Streptomyces griseus. J. Biol. Chem., 1974, 249, 81 1-817. 59. S. Ito, T. Muramatsu and A. Kobata, Endo-P-N-acetylglucovaminidasesacting on carbohydrate moieties of glycoproteins. Purification and properties of two enzymes with different specificities from Clostridiurn perfringens. Arch. Biochern. Biophys., 1975, 171, 78-86. 60. J.H. Elder and S. Alexander, Endo-13-N-acetylglucosaminidase F: Endoglycosidase from Fluvobucteriuni meninyosepticurn that cleaves both high-mannose and complex glycoproteins. Prac. Nutl Acad. Sci. USA, 1982, 79, 4540-4544. 61. A. Kobata, Use of endo- and exoglycosidases for structural studies of glycoconjugates. Anal. Biochem., 1979, 100, 1-14. 62. F. Maley, R.B. Trimble, A.L. Tarentino and T.H. Plummer, Characterization of glycoproteins and their associated oligosaccharides through the use of endoglycosidases. Anal. Biochern., 1989, 180, 195-204. 63. R.B. Trimble and A.L.Tarentino, Identification of distinct endoglycosidase activities in Fluvohucterium meningosepticurn Endo F1, Endo F2, and Endo F3. J. Bid. Chern., 1991, 266, 1646-1651. 64. T.H. Plummer and A.L. Tarentino, Purification of the oligosaccharide-cleaving enzymes of Fluvobacterium meningosepticurn. Glycobiology, 1991, I , 257-263. 65. R.B. Trimble, P.H. Atkinson, A.L. Tarentino, T.H. Plummer, F. Maley and K.B. Tomer, Transfer of glycerol by end0-P- N-acetylglucosaminidase F to oligosaccharides during chitobiose core cleavage. J. Biol. Chem., 1986,261, 12000- 12005. 66. K. Takegawa, M. Tabuchi, S. Yamaguchi, A. Kondo, I. Kato and S. Iwahara, Synthesis of neoglycoproteins using oligosaccharide-transfer activity with endo-P-N-acetylglucosaminidase. J. Bid. Chem., 1995, 270, 3094-3099. 67. K. Yamamoto, S. Kadowaki, J. Watanabe and H. Kumagai, Transglycosylation activity of Mucor hiernah endo-P-N-acetylglucosaminidasewhich transfers complex oligosaccharides to

References

5 11

the N-acetylglucosamine moieties of peptides. Biochern. Biophys. Rex Con?mzrn. 1994, 203. 244-252. 68. A.L. Tarentino, C.M. Gomez and T.H. Plummer, Deglycosylation of asparagine-linked glycans by peptide: N-glycanase F. Biochemistry, 1985, 24, 4665-4671. 69. C.C. Huang and D. Aminoff, Enzymes that destroy blood group specificity. J. Biol. Cheni., 1972,247, 6737-6742. 70. V.P. Bhavanandan, J. Umemoto and E.A. Davidson, Characterization of an endo-a-Nacetylgalactosaminidase from Diplococcus pneumoniue. Biochern. Biophys. Rex Commun., 1976, 70, 738-745. 71. Y. Endo and A. Kobata, Partial purification and characterization of an endo-a-Nacetylgalactosaminidase from the culture medium of Diplococcus pneumoniae. J. Biochem.. 1976, 80, 1-8. 72. J. Umemoto, V.P. Bhavanandan and E.A. Davidson, Purification and properties of an endo-aN-acetylgalactosaminidase from Diploc~oc~cus pneurnoniue. J. B i d . Chem.. 1977, 252, 86098614. 73. J.-Q. Fan, K. Kanatani, N. Andoh, K. Yamamoto, H. Kumagai and T. Tochikura. Release of T-antigen, a carcinoma marker from native cells, by endo-a-N-acetylgalactosaminidase of Akuliyenes sp. Biochem. Biophys. Res. Cornmun., 1990, 172, 341 --347. 74. 1. Ishii-Karakasa, H. Iwase, K. Hotta, Y. Tanaka and S. Omura, Partial purification and characterization of an endo-a-N-acetylgalactosaminidase from the culture medium of Streptovzyces sp. OH-1 1242. Biocheni. J., 1992, 288, 475 482. 75. R.M. Bardales and V.P. Bhavanandan, Transglycosylation and transfer reaction activities of endo-a-N-acetyl-D-galactosaminidasefrom D ~ l o c o c c u s(Streptocvccus) pneumoniae. J. Bid. C/ierv., 1989, 264, 19893-19897. 76. Y.-T. Li, H. Nakagawa, M. Kitamikado and S.-C. Li, Endo-P-galactosidase from Escherichia freundii. Methods Enzymol., 1982, 83, 610-619. 77. M. Kitamikado, M. Ito and Y .-T. Li, Endo-0-galactosidase from Flauobucteriunz kerutolyticus. Methods Enzvmol.,1982, 83, 619-625. 78. P. Scudder, K.-I. Uemura, J. Dolby, M.N. Fukuda and T. Feizi, Isolation and characterization Biochem. J., 1983, 213, 485-494. of an endo-a-galactosidase from Bacteroi~~~,.s,Grigili.s. 79. K. Nakazawa and S. Suzuki; Purification of keratan sulfate-endogalactosidase and its action on keratan sulfates of different origin. J. Bid. C'hem., 1975, 250, 912-917. 80. S. Takasaki and A. Kobata, Purification and characterization of an endo-P-galactosidase produced by Diplococcus pneumoniae. J. Biol. Chern., 1976, 251, 3603-3609. 81. N. Fushuku, H. Muramatsu, M.M. Uezono and T. Muramatsu, A new endo-P-galactosidase releasing Galal,3Gal from carbohydrate moieties of glycoproteins and from a glycolipid. J. Biol. Chenz., 1987, 262, 10086-10092. 82. S.-C. Li, R. DeGasperi, J.E. Muldrey and Y.-T. Li, A unique glycosphingolipid-splitting enzyme (ceramide-glycanase from leech) cleaves the linkage between the oligosaccharide and the ceramide. Biochem. Biophys. Rex Commun., 1986, 141, 346-352. 83. M. Ito and T. Yamagata, A novel glycosphingolipid-degrading enzyme cleaves of the linkage between the oligosaccharide and ceramide of neutral and acidic glycosphingolipids. J. Bid. Chem., 1986,261, 14278-14282. 84. M. Ito and T. Yamagata, Purification and characterization of glycosphingolipid-specific endoglycosidases (endoglycoceramidases) from a mutant strain of Rhodococcus sp. J. B i d Chew., 1989,264, 9510-9519. 85. M. Ito, Y. Ikegami and T. Yamagata, Activator proteins for glycosphingolipid hydrolysis by endoglycoceramidases. J . Biol. Chem., 1991,266, 7919-7926. 86. H. Ashida, K. Yamamoto, H. Kumagai and T. Tochikura, Purification and characterization of membrane-bound endoglycoceramid& from Corynehacterium sp. Eur. J. Biochem., 1992.205, 729-735. 87. H. Ashida, Y. Tsuji, K. Yamamoto, H. Kumagai and T. Tochikura, Transglycosylation activity of endoglycoceramidase from C'orynehucturium sp. Arch. Biochem. Biophys., 1993,305, 559562.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

31 Glycoprotein Processing Inhibitors Ahdelwzjid Ossor and Alan D. Elbein

31.1 Introduction Polyhydroxylated alkaloids, a biologically significant class of molecules, have been isolated from natural sources and also synthesized chemically. The alkaloids discussed in this Chapter are characterized by having: 1) a high degree of hydroxylation; 2) a “sugar-like’’ ring structure with a nitrogen replacing the ring oxygen; and 3) the ability to inhibit various glycosidases, including glycoprotein processing glycosidases.

In general, these alkaloids are analogs of monosaccharides in which the ring oxygen has been replaced by a nitrogen (imino-sugars) (See Figure 1 for structures). The glycosidase inhibition is effected by the fact that the inhibitor mimics the pyranosyl or furanosyl ring structure as well as the chirality of the hydroxyl groups of the specific sugar substrate for that enzyme. This group of alkaloids aroused considerable interest because of their ability to inhibit various processing glycosidases and thereby cause changes in the structure of the N-linked oligosaccharide chains of biologically important glycoproteins. As a result, they affect the function of many glycoproteins. The best known and most highly characterized of the naturally occurring alkaloids are swainsonine [I], castanospermine [2], deoxynojirimycin [3], and calystegine Bl [4]. The structures of the various glucosidase (A) and mannosidase (B) inhibitors are shown in Figure 1 (A and B). These compounds are extremely potent and specific inhibitors of the aryl-glycosidases and processing glycosidases, especially amannosidase, a- and P-glucosidase, and a-galactosidase. These particular alkaloids have become the standards by which new and novel naturally occurring and synthetic glycosidase inhibitors are currently evaluated. In addition, such inhibitory

31 Glycoprotein Processing Inhibitors

5 14

Avstraline

Castanospermine

1.4 - Dideaxy- I,4 -

Deoxymannojirimycin

I mino- D- Mannitol

HWH2

I:G$=o HO

HO,,

PH

HO

D M C H z O H N HOH2C

He

C+ H A 2

HO

HO 2 , 5 - Dihydraaymethyl3,4- Dihydroaypyrrolidine

H NH2 G H @ 3

H H

Kifunensine

OH

HO H

H OH

MannosfatinA

MDL 25,637

Swainsonine Lentiginorine

HoH%

A3

c1

Figure 1. Structures of members of the different classes of glycosidase inhibitors. A. Glucosiddse Inhibitors; B. Mannosidase Inhibitors; C. Calystegines.

activities have attracted considerable attention because of the possibility that these inhibitors could be used for chemotherapy in such devastating diseases as cancer, diabetes, AIDS and various viral diseases. It is not possible in this chapter to cover all of the naturally occurring and synthetic polyhydroxylated alkaloids. Therefore for the most part, we will concentrate our attention on the naturally occurring alkaloids, and focus on their glycosidase inhibitory activity, especially as it affects N-linked oligosaccharide processing.

31.2 Structural Clussijicution

5 15

31.2 Structural Classification Polyhydroxylated alkaloids can be classified according to the ring system that they contain. In contrast to most classes of alkaloids, these compounds have certain common features: that is they all have a high degree of hydroxylation and they all resemble monosaccharides with pyranoside or furanoside ring structures, but with a nitrogen replacing the ring oxygen. Five classes are encompassed by these properties and each of these classes is shown in Figure 1 (A, B, C) by some of the known glucosidase or mannosidase inhibitors. The various classes of naturally occurring alkaloids are as follows:

1) Class 1 are the pyrrolidine alkaloids having five-membered ring structures, as shown by DMDP, i.e. 2,5-dihydroxymethyl-3,4-dihydroxypyrrolidine [ 51 (Figure 1A). 2) Class 2 alkaloids are the piperidines with six-membered rings, shown by deoxynojirimycin [ 3 ] and deoxymannojirimycin [6] (DMJ; Figure l B ) . There are currently nine known members of this class. 3 ) Class 3 alkaloids are the pyrrolizidines as represented by australine [7] (Figure 1A). In a formal sense, these alkaloids can be considered to be the result of a fusion of two pyrrolidine ring systems with the common nitrogen atom at the bridgehead. 4) Class 4 compounds are the indolizidines as indicated by two alkaloids very widely used to inhibit glycoprotein processing, i.e. swainsonine [ 11 (Figure 1B) and castanospermine [2] (Figure 1A). The indolizidine group can also be considered to be a product of the fusion of a pyrrolidine ring and a piperidine ring to give a bicyclic 5/6 ring system. 5 ) Class 5 compounds, represented by the nortropanes, are a relatively new addition to those alkaloids having glycosidase inhibitory activity. The nortropanes and C1 [9] (Figure 1C). In a are represented by calystegine A3 [8], B1, BZ [4], sense, the nortropane ring system can also be considered to result from the fusion of a five-membered pyrrolidine ring with a six-membered piperidine ring, but in contrast to the indolizidine ring, the fusion points are c1 to the nitrogen of each monocyclic system. This nortropane alkaloid group now consists of more individual members than any of the other classes of polyhydroxy-alkaloids described above.

31.3 Distribution of Glycosidase Inhibitors in the Plant Kingdom Most classes of alkaloids have a well established distribution such that they can be used for taxonomic characterization. However, the polyhydroxylated alkaloids have so far been found to be widely distributed, suggesting that, either they have a

5 16

31 Glycopvotein Processing Inhibitors

quite different function in plants than other groups of alkaloids, or that their pathway of biosynthesis is widespread in plant species and is distinct from that of other alkaloids. For example, DMDP (Figure 1A) and swainsonine (Figure 1B) have been found in many different species of plants and microorganisms [lo, 111. In addition to their widespread distribution in plants, the polyhydroxylated alkaloids are also widely distributed in bacteria, fungi, insects, and perhaps various animal species that accumulate them as a result of eating alkaloid-containing plants.

31.4 Isolation and Structural Determination Glycosidase inhibitors are highly hydroxylated and readily dissolve in water, or in mixtures of water and methanol, or water and ethanol. Usually, the plant material is ground in a food processor and then treated with hot, 75% methanol to extract the alkaloid. The extract is then filtered through cheese cloth and then through Whatman 3MM paper. The resulting clear solution is taken to dryness and the residue is suspended in water, and initially purified by ion-exchange chromatography on a column of Dowex-50-H+. The alkaloids bind to the column by virtue of their positive charge, and can be eluted with a gradient of ammonium hydroxide. Further purification and separation from the basic amino acids is achieved using a weak ion-exchanger, such as Amberlite CG50. Final purification is usually obtained using radial and thin layer chromatography (tlc), and/or paper chromatography [ l l , 121. These compounds are fairly difficult to purify to homogeneity from plant extracts because most of these extracts contain a number of closely related structures having similar molecular weights and similar charges. Chromatography on columns of silica gel and aluminum oxide may also be helpful for purification of these compounds, as are various HPLC techniques. The alkaloids can be detected on tlc plates with ninhydrin, but those compounds with glycosidase inhibitory activity are more specifically detected by virtue of their inhibitory activity against individual glycosidases. The most sensitive technique for the general detection of the alkaloids, and for determining their general structure, is GC-MS of their trimethylsilated derivatives. There is a fairly extensive database available that lists the retention times and fragmentation patterns of the known polyhydroxylated alkaloids. All of these techniques have recently been reviewed in detail [ 131. Once the alkaloid is purified, the absolute structure and stereochemistry can be determined by a combination of spectroscopic methods including H and I3CNMR spectroscopy, and high resolution mass spectrometry. All three methods should be in agreement on one structure for the compound in question. Additional structural data and confirmation can be obtained from the analysis of the trimethylsilyl ether derivative by GC-MS as indicated above [13]. The absolute stereochemistry can be determined by X-ray crystallography, which can be used whenever well-defined crystal structure is obtained, either from the alkaloid itself or from its hydrochloride salt.

'

31.5 Glyrosidusr Inhibitory Activity

5 17

31.5 Glycosidase Inhibitory Activity Almost all of the polyhydroxylated alkaloids discovered to date have been found to be glycosidase inhibitors, usually causing a competitive type of inhibition. Generally, these compounds have been initially identified, and then isolated and purified, based on their ability to inhibit one of the common aryl-glycosidases. For example, swainsonine was first identified based on its ability to inhibit lysosomal, and jack bean, aryl-a-mannosidase activity [ 141. Some of the alkaloid inhibitors have a rather broad spectrum of activity against a number of related glycosidases, whereas others are much more specific. Thus, castanospermine inhibits many different glucosidases and was first identified as a glycosidase inhibitor because of its effect on lysosomal, and almond emulsin, P-glucosidase [ 151. Later, this alkaloid was found to also inhibit fungal amyloglucosidase, intestinal maltase, sucrase, and trehalase [ 161, as well as the glycoprotein processing enzymes, glucosidase I and glucosidase I1 [17]. It should be emphasized, however, that the most useful inhibitors for biological studies would be those with very limited and specific inhibitory activity that only affect the enzyme of interest, or at least have a fairly limited spectrum of inhibitory activity. That is, it would be of considerable interest and importance to have a glucosidase inhibitor that only affected glucosidase I and not glucosidase 11, or sucrase or other a-glucosidases. Perhaps even more exciting would be a glucosidase I1 inhibitor that only blocked the removal of the final glucose, i.e. the sugar on high mannose chains that is recognized by calnexin [ 181. The difficulty is in identifying such specific glycosidase inhibitors. That is, it is relatively easy to screen extracts, made from plants or other organisms, against the commercially-available glycosidases because these enzymes are readily available, and because they use the y nitrophenyl-sugar derivatives as substrates. As a result, the assays used to detect glycosidase activity are very rapid colorimetric determinations that measure the release of p-nitrophenol from the glycoside, and its inhibition in the presence of the appropriate alkaloid. On the other hand, most of the enzymes involved in glycoprotein processing, such as glucosidase I or mannosidase I, do not cleave these aryl-glycosides, but require oligosaccharide substrates, i.e. Glc3MangGlcNAc for glucosidase I, or Man9GlcNAc for mannosidase I. These oligosaccharide substrates are difficult to make and the assays to measure their hydrolysis by processing enzymes are much more tedious [19]. As a result, most of the inhibitors that have been isolated or synthesized to date are of a more general nature. And some compounds that might actually have been specific for glucosidase I or ER mannosidase, etc., have probably been discarded because the wrong assays were used and many were missed. That is to say, the more specific the enzyme of interest, the more specific the assay must be.

5 18

31 Glycoprotcin Processing Inhibitors

31.6 Structure-Activity Relationships As might be anticipated, the first studies exploring the relationship between the structure of a polyhydroxylated alkaloid and its glycosidase activity demonstrated that the glycosidase to be inhbited could be predicted on the basis of the number, position and configuraton of the hydroxyl groups of the alkaloid in question. Thus, nojirimycin superimposes perfectly with the glucopyranoside molecule, and nojirimycin is a good inhibitor of both a- and (3-glucosidase 1201. Likewise, swainsonine was viewed as an aza-analog of mannose lacking the hydroxymethine group at C-4 but having the same disposition of the remaining hydroxyl groups, and this compound proved to be an excellent inhibitor of a-mannosidase [ 141. Castanospermine [211 and other structurally related compounds, i.e. deoxynojirimycin [4] and calystegine B2 [22],fit well with D-glucose in the pyranose configuration, and they are all good inhibitors of various glucosidases. Unfortunately, there are a number of cases where an alkaloid appears to show very close similarity to the substrate of a given glycosidase and is proposed to be an inhibitor of that enzyme, and experimentation proves that prediction to be erroneous. For example, molecular modeling as well as molecular orbital calculations 123, 241 indicate that 6-epi-castanospermine is closely related structurally to mannose, but this alkaloid inhibits a-glucosidase rather than a- or (3-mannosidase. A similar situation exists with regard to the polyhydroxylated alkaloid, calystegine B3, which appears to be closely related structurally to galactose, but this compound does not inhibit either a- or P-galactosidase 1251. These studies indicate that structure-activity correlations are more complex then initially believed, and will probably require more sophisticated molecular modeling in order to be able to give reliable predictions. Since at the current time there are not more comprehensive molecular models available, the results of inhibition studies have been rationalized using generally accepted models for glycosidase inhibition. Such an approach has been most effectively used for the calystegine inhibitors which provide a comprehensive series of structurally related and naturally occurring polyhydroxylated alkaloids. For (3glucosidase inhibition, the model suggests the presence of two carboxylic acid groups in the active site of the enzyme, one of which is responsible for generation and one for stabilization of the glycosyl intermediate which forms a hydrogen bond with the hydroxyl group 19, 261. In terms of the calystegines B1 and C1 (see Figure 1C for structures), it has been suggested that the exo-hydroxyl group at the 6position is protonated by the acidic group responsible for catalytic activity within the active site, in an analogous manner to conduritol B epoxide 1271. Thus, the interaction of inhibitory calystegines with glycosidases can be envisioned as binding to the sites determining specificity, and to the catalytic center, through specific hydroxyl groups and through the imino group [28].

31.7 N-Linked Olignsaccharide ProcesJing

519

31.7 N-Linked Oligosaccharide Processing Following the transfer of the N-linked oligosaccharide chain, i.e. Glc3MangGlcNAcz, from the lipid carrier to the protein, the oligosaccharide begins to undergo a number of trimming or processing reactions that result in the sequential removal of all three glucoses and up to six of the mannose residues [29]. The early processing reactions begin in the endoplasmic reticulum of the cell and continue in the cis and medial Golgi apparatus [30]. This processing pathway is outlined in Figure 2. The first two processing enzymes are glucosidases that are located in the ER and remove all three glucose residues [31]. Glucosidase I is a transmembrane protein that removes the terminal al,2-glucose [32], and then glucosidase 11, a luminal ER enzyme, releases the next two al,3-linked glucoses 1331. The action of these two enzymes results in the formation of a protein having a ManY(G1cNAc)Z oligosaccharide. In some cases, a single al,2-linked mannose may be removed while the protein is still in the ER by a unique and specific a1,Zmannosidase [34]. Actually several ER al,2-mannosidases have been reported which differ in their sensitivity to DMJ, and in the Mans(G1cNAc)z oligosaccharide that they produce 135). The role of these different enzymes is not known. The resulting Mans(G1cNAc)z-protein is then transported to the ci.eGolgi apparatus, where further trimming of mannose

PROCESSING OF N-LINKED GLYCOPROTEINS

I

M-M,M, M-M’ M-N-N-Ain G-G-G-M-M-~ Glc I

4 4 Glc II

M-M. M-M++N-N-A~~

-,HIGH-MAN CHAINS

M-M-M’

Figure 2. Glycoprotein Processing Pathway. As described in the text, this pathway is initiated by transfer of Glc-iMang(G1cNAc)l from lipid to protein in the ER. The initial removal of glucose and one mannose occurs in the ER. but removal of other mannose residues proceeds in the cis and medial Golgi apparatus by a number of alphamannosidase activities. Other processing reactions that go on in the medial and trans Golgi involve the addition of various sugars by individual and specific glycosyltransferases.

BRID AINS

M,

*

4 GlcNAc Tronr.1

,M-N-N-A& GlcNAc-M J MonII M.

,M-N-N-Asn GlcNAc-M

SA-Gal-GlcNAc-M, SA-G.~-GI~NA~-M’

M-N-N-A+

520

31 Glycoprotein Processing Inhibitors

units occurs. Table 1 lists (some of) the various mannosidase activities in the ER and Golgi and some of their properties. In the Golgi, there are at least two a1,2-mannosidase activities, referred to as mannosidase IA and IB, that remove the remaining a1,2-linked mannoses to give a Mans(G1cNAc)z structure [36]. Mannosidase IA and IB are Ca++ requiring transmembrane proteins that are called Class 1 enzymes [37]. These enzymes are called inverting glycosidases because they cause an inversion in the anomeric configuration upon hydrolysis. There may also be other mannosidase activities that have not been completely elucidated and whose function may be to remove selected mannose residues either for targeting or other functions. After removal of the four a1,2-linked mannose units, the Mans(G1cNAc)z oligosaccharide is the acceptor for a GlcNAc, the first sugar that initiates formation of complex chains. This sugar is transferred from UDP-GlcNAc by GlcNAc transferase I to that mannose that is linked a1,3 to the P-linked mannose [38]. GlcNAc transferase I is a Golgi enzyme that is probably located in the medial Golgi stacks. This enzyme was purified to homogeneity and shown to be a type I1 membrane protein that is specific for the Mana1,3Man~1,4-GlcNAc,and that transfers a GlcNAc in pl,2-linkage to the terminal a1,3-linked mannose [39]. The addition of this GlcNAc is essential for the activity of the next processing enzyme, mannosidase I1 [40]. Once the GlcNAc has been added to form GlcNAcMan5(GlcNAc)2-protein, mannosidase 11, a more general a-mannosidase located in the medial Golgi stacks, removes the a1,3- and a1,6-linked mannose residues from that mannose on the 6branch to give a GlcNAc-Man3 (G1cNAc)z-protein. Mannosidase 11 belongs to the Class 2 mannosidases which are called retaining glycosidases, because the anomeric configuration of the released mannose is retained in the a-configuration. Mannosidase I1 has been purified to homogeneity from animal [41] and plant [42] sources. The animal enzyme is a type IT transmembrane glycoprotein with a cytoplasmic domain of five amino acids, a single transmembrane domain, and a luminally oriented catalytic domain [43]. Mannosidase I1 has been found in all mammalian tissues examined, although its activity is low in brain [44]. Interestingly enough, brain and perhaps other tissues, have been shown to have other mannosidases of rather broad specificity that have al,2-, al,3-, and al,6-mannosidase activities [45-47]. These mannosidases trim Mans(G1cNAc)z structures down to Man3(GlcNAc)zstructures without any assistance from other enzymes, and are clearly distinct from other known animal amannosidases in terms of pH optima, substrate specificity and susceptibility to various mannosidase inhibitors [48]. Some animal cells contain an enzyme that initiates an alternate processing pathway that avoids the need for glucosidase I and glucosidase 11. This enzyme in an endo-al,2-mannosidase that cleaves the glucose branch, i.e. Glc3Man, Glc2Man or Glcl Man from the Glc3-lMan~(GlcNAc)z,to produce a Mang(GlcNAc)2-protein [49]. The specific function of this enzyme is not known, but its presence in certain cells renders them resistant to the effects of various glucosidase inhibitors, i.e. castanospermine, deoxynojirimycin or australine [ 501. Following the action of the various glycosidases on the N-linked oligosaccharides in the trimming part of the pathway, a number of glycosyltransferases further

~~

ND = Not Determined

ER Mannosidase Mg N Mannosidase Mannosidase IA (Golgi) Mannosidase I (Mung Bean) Mannosidase I1 (Rat Liver) Mannosidase I1

~~~

Enzyme

0.09 pM

0.2 pM

-

-

~

-

Swainsonine

4 PM

ND 0.1 pM

-~

ND 0.09 pM

0.02-0.05 pM --

ND ND

40-50 pM

O.S--l pM

Mannoamidrazone

ND ND ND

Mannostatin

ND ND

Kifunensine

5-7 pM 1-2 pM

Deoxymannojirim ycin

Table 1. Effect of processing inhibitors on various a-mannosidases.

522

31 Glycoprotein Processing Inhibitors

modify the GlcNAc-Man3(GlcNAc)*-protein to produce a wide array of complex types of oligosaccharides. The specific structures produced are species and cell specific and depend on the types and specificities of the transferases contained within that cell type [51]. In general, in the trans Golgi stacks, there are a number of GlcNAc transferases, galactosyltransferases, sialyltransferases, fucosyltransferases, sulfotransferases, etc., that can add various sugars or other substitutions in specific glycosidic linkages to give a great diversity of oligosaccharides having biantennary, triantennary or tetraantennary chains [52].Many of these enzymes have been wellcharacterized, and the genes for some of them have been cloned [ 5 3 ] .At present, there are no known inhibitors that can be used in vivo for any of the glycosyltransferases, but the isolation of such compounds from natural sources, or their chemical synthesis, will be an important addition to the compendium of valuable biological tools and perhaps useful chemotherapeutic agents.

31.8 Inhibitors of N-Linked Oligosaccharide Processing There are now a number of low molecular weight compounds that have been isolated from natural sources and found to be potent and relatively specific inhibitors of exoglycosidases. Many of these compounds have become valuable tools for biological studies because they are inhibitors of the glycoprotein processing enzymes. In addition, since these compounds are small molecules they are able to permeate most cells, and therefore they can be used with intact cells and tissues to study “in vivo” situations. Another important use of these inhibitors is to distinguish the various processing enzymes from each other, based on their susceptibility to selected inhibitors. The best example of the efficacy of this approach is shown in Tables 1 and 2 which lists many of the well characterized animal a-mannosidases, and their susceptibility to the different mannosidase inhibitors. It is clear that the various amannosidases have quite different sensitivities to these different inhibitors, and therefore they can be distinguished on the basis of their sensitivities. The remaining sections of this chapter discuss the chemistry and biological activity of the various alkaloids that function as inhibitors of glycoprotein processing. 31.8.1 Glucosidase Inhibitors Castanospermine, as indicated above, is an indolizidine alkaloid that was first isolated from the seeds of the Australian tree, Castanospermum australe [2]. Initial studies indicated that this compound inhibited lysosomal and almond emulsin P-glucosidase [15], but later studies showed that it was a better inhibitor of aglucosidases and affected a variety of different a-glucosidase activities, including the animal processing glucosidases, glucosidase I and glucosidase I1 [ 16, 171. Because of the inhibition of the intestinal glucosidases, seeds that contain castanospermine are quite toxic to animals and cause severe diarrhea and other gastrointestinal upsets

Man~,(GlcNAc)?

=

NK

Not Known; Mans Nz

N-ManiN2 ManiNz

N-ManSN2 Man9-4Nz

Golgi NK

=

MangNz MangNz Man3N2 Man~N2 MansN~

Man9Nl Man9Nz Man9-5Nl Mang-N1 Man9Nz

ER ER NK Golg1 NK

Final Product

ER-Cytosolic Man-ase Mg-Man-ase a l , 2; a l , 3; ul, 6 Man-ase Man-ase IA Phospholipid-Dependent Man-ase Man-ase I1 Brain Man-ase

Oligosaccharide Substrate

Localization of Enzyme

Enzyme Identification (Mannosidases)

Table 2. Properties of various a-mannosidases.

5-6 6.0

6.0 6.0 6 1-6 5 6.0 5-6

pH Optima

~

-

Ca++

-

COT+

Ca++

-

Cation Requirement

124 kDa, Yes NK, Yes

107 kDa, No 49-56 kDa, No 110 kDa, NK 57 kDa, Yes 52 kDa, Yes

Mol. Wt. of Protein: Is it a Glyco-Protein:

wl h, w

cg

3

2

0

2

R

5.

s-

23

G' 8

9

2L

3 ??

s

2

$ 5

h

P

34 s.

30

'1

111

524

31 Glycoprotein Processing Inlz ihitors

[54]. Furthermore, when castanospermine is fed to mice over 4-5 days, it inhibits lysosomal a-glucosidase and causes the accumulation of partially degraded glycogen within the lysosomal structures [55]. This accumulation resembles the accumulation of glycogen particles in the lysosomal storage disease, Pompe’s disease, where individuals lack the lysosomal a-glucosidase [ 561. When animal cells are grown in the presence of castanospermine, the processing of the N-linked oligosaccharides is blocked at the glucosidase I step, and the oligosaccharide on the asparagine-linked glycoproteins are mostly of the Glc3Manpg(GlcNAc)Z structures [ 171. As indicated earlier, some cells have an endomannosidase activity and are able to process their N-linked oligosaccharides even in the presence of castanospermine or other glucosidase I inhibitors [49]. There are other known glucosidase inhibitors that also affect glucosidase I, such as deoxynojirimycin [ 571 and DMDP (2,5-dihydroxymethyl-3,4-dihydroxypyrrolidine) [ 581 (see Figure 1A for structures). These inhibitors also give rise to glycoproteins having GlqMan7-9 (G1cNAc)Z structures, but the amount of inhibitor required for a given amount of activity varies greatly with the inhibitor in question. While the above glucosidase inhibitors show some differences in their relative activities towards glucosidase I and glucosidase 11, none of them show the kind of specificity that would be most useful for biological purposes, i.e. inhibition of glucosidase I but not 11, or inhibition of glucosidase I1 but not I. However, some more recently discovered inhibitors come closer to achieving that goal. although they are also not as specific, or as potent, as one would like. Thus, australine (Figure 1A) is a tetrahydroxypyrrolizidine alkaloid that occurs along with castanospermine in Castanospermum australe seeds. This alkaloid is a reasonable inhibitor of glucosidase I and a poor inhibitor of glucosidase I1 [59]. Australine is the first inhibitor to differentiate these two processing glucosidases. Another compound, trehazolin, was isolated as an inhibitor of insect trehalase [60]. This compound also proved to be a reasonable inhibitor of glucosidase I, but had very low inhibitory activity against glucosidase I1 [ 6 11. Another glucosidase inhibitor of considerable interest is 2,6-diamino-2,6-imino-70-(p-D-glucopyranosy1)-D-glycero-L-guloheptitol (MDL25,637) (see Figure 1 A for structure). This compound was synthesized chemically to resemble a disaccharide as a transition state analog of sucrose [62]. It does inhibit intestinal sucrase, maltase, isomaltase and trehalase at micro-molar concentrations [63]. More interesting from the standpoint of this review is the observation that this inhibitor also shows a marked difference in its ability to inhibit glucosidase I and glucosidase 11, but in the opposite way to australine. That is, MDL25,637 was much more effective against glucosidase I1 than it was towards glucosidase I [64]. When this compound was added to cultured cells, it caused the accumulation of glycoproteins having mostly Glc;?Man7-9(GlcNAc)2structures [64]. Inhibiting the removal of glucose from the N-linked oligosaccharides prevents additional processing reactions and causes dramatic effects on the function and/or targeting of many glycoproteins [65].For example, when Hep-G2 cells are incubated for various times in the presence of deoxynojirimycin (DNJ), the rate of secretion of some serum glycoproteins, such as a1 -antitrypin, is greatly reduced, while the rate of secretion of others such as ceruloplasmin is only slightly affected [66]. This re-

31.8 Inhibitors of N-Linked Oligosuccharide Processing

525

duction in the secretion of antitrypsin was not due to an inhibition in the synthesis of the protein, but was shown to be due to a change in the structure of the oligosaccharide chains (see below). Cell fractionation studies indicated that DNJ caused an accumulation of a1 -antitrypsin in the ER, suggesting that the presence of glucose on the N-linked oligosaccharide might retard the transfer of the glycoprotein from the ER to the Golgi apparatus [66]. Similar results were obtained with regard to the synthesis and targeting of the low density lipoprotein receptor in smooth muscle cells grown in the presence of castanospermine [67]. and in the synthesis and targeting of the insulin receptor in the presence of DNJ or castanospermine [68]. Assembly of the AIDS virus was also inhibited by glucosidase inhibitors because these inhibitors blocked transfer of the envelope glycoprotein from the ER to the Golgi [69]. Recently, the explanation for these results has become apparant. Some elegant studies have demonstrated that the ER has a quality control system that functions to ensure proper folding and oligomerization of newly synthesized glycoproteins, and also ensures that only properly folded proteins are transported to the Golgi apparatus [70]. Thus, the ER contains two chaperones, called calnexin and calreticulin, that recognize and bind to high-mannose oligosaccharides that have a single terminal glucose residue [70-731. Thus, once glucosidase I and glucosidase I1 have removed the two terminal glucoses from the high-mannose structure, these chaperones can bind to the single terminal glucose on unfolded proteins and help these proteins fold properly [ 741. Furthermore, the ER also has a “safety valve” for those glycoproteins whose glucoses have all been removed by the two glucosidases, but that have still not folded into the proper conformation. Thus, there is a glucosyltransferase in the ER, referred to as UDP-g1ucose:glycoprotein glucosyltransferase that can add a single glucose in al,3-linkage to the terminal mannose on the 3-branch of unfolded glycoproteins [75]. Interestingly enough, this transferase will not glucosylate a highmannose chain on a properly folded glycoprotein even if it is available, but requires denatured or unfolded glycoproteins as substrates [ 761. These studies indicate that one important role of high-mannose oligosaccharides, and perhaps more specifically of the glucose residue(s) on these oligosaccharides, is to expedite the folding of some proteins in the ER. Apparantly there are many other glycoproteins, such as the influenza viral hemagglutinin [17, 64, 771, that do not require the assistence of these chaperones to fold. Thus, the hemagglutinin (as well as other proteins) is transported out of the ER and packaged into mature virus particles at the same rate in the presence or absence of glucosidase inhibitors. 31.8.2 Mannosidase Inhibitors

A number of mannosidase inhibitors have been isolated from various plants and fungi, and the specificity and activity of many of these inhibitors has been outlined in Table 1. The first glycoprotein processing inhibitor to be identified was the indolizidine alkaloid, swainsonine whose structure is presented in Figure 1B. Swainsonine was shown to specifically inhibit plant and animal mannosidase I1 (781. This alkaloid also inhibits lysosomal a-mannosidase and jack bean a-mannosidase, but

526

31 Glycopvotein Processing Inhibitors

has no effect on the other processing mannosidases such as the ER a-mannosidase or the Golgi mannosidases, IA or IB [36, 411. The first studies on the effects of swainsonine on glycoprotein processing showed that this inhibitor caused a reduction in the size of the N-linked oligosaccharide chains and prevented the formation of complex types of oligosaccharides [79]. Later studies indicated that it caused the formation of hybrid types of structures in the influenza viral hemagglutinin [SO], and that it specifically inhibited mannosidase 11, and had no effect on the processing mannosidase I [Sl]. Swainsonine has been found in various wild plants that are toxic to animals. Swainsonine does cause significant neurological problems in animals that eat these plants, but it is not clear whether these symptoms are due to its inhibition of glycoprotein processing or because of other as yet unknown effects [821. Following the isolation of DNJ and the demonstration that it was an inhibitor of glycoprotein processing and glucosidase I, the 2-epimer of DNJ, i.e. deoxymannojirimycin (DMJ), was synthesized chemically, and found to be an inhibitor of the Golgi processing mannosidase I [83]. DMJ, however, did not inhibit either lysosomal a-mannosidase or jack bean a-mannosidase. These results on the selective activity of DMJ emphasize the point made earlier in this chapter, i.e. it is “dangerous” to screen for new glycoprotein processing inhibitors using arylglycosidases such as aryl-a- or 0-mannosidase as the enzymes to test for inhibitory activity. DMJ has more recently been found to be naturally occurring in various plants [84], but might have been missed if it were not already known from the chemical synthesis. In animal cells, DMJ inhibited Golgi mannosidase IA/B and caused the accumulation of glycoproteins with mostly Mans (G1cNAc)z structures [85].However, in contrast to the effect of DNJ which prevented the secretion of IgM and IgD by cultured cells, DMJ had no effect on the synthesis or secretion of these glycoproteins [ 861. DMJ was used to demonstrate that the ER mannosidase was only involved in trimming one mannose from HMGCoA reductase. The reductase is located in the ER of UT-1 cells and is an N-linked glycoprotein having Man&lcNAc2 and MansGlcNAc2 structures [87]. In the presence of DMJ which does not inhibit the ER mannosidase, the HMGCoA reductase had mostly MansGlcNAc2 structures suggesting that the ER mannosidase was involved in removing only the first mannose, but the other mannoses must be removed by Golgi mannosidases [SS]. DMJ was also used in an interesting way to demonstrate that some membrane glycoproteins could be recycled through the Golgi apparatus during endocytosis. In this study, the transferrin receptor was synthesized in CHO cells in the presence of DMJ to block mannose trimming, and [2-3H]mannoseto label the N-linked oligosaccharides. After an appropriate period of incubation, the media was changed to remove label and inhibitor, and the cells were reincubated in fresh media for various times, and under conditions that would cause the transferrin receptor to undergo endocytosis. At each time, the transferrin receptor was isolated and its oligosaccharide structure was examined. Initially, the structure was of the high mannose type, but with increasing times a small amount of the [ 3H]-receptor apparantly underwent processing to become complex chains. These studies suggest that some of the endocytosed glycoproteins can undergo additional processing reactions [89].

31.8 Inhibitors

of N-Linked Oligosaccharide Procrssing

527

Figure 1B) is another inhibitor that DIM (1,4-dideoxy-l,4-imino-~-mannitol; was synthesized from benzyl-u-D-mannopyranoside, and shown to be a good inhibitor of jack bean a-mannosidase [901. It also inhibited glycoprotein processing in MDCK cells and caused the accumulation of glycoproteins having mostly Man9(GlcNAc)z structures indicating that it was a n inhibitor of Golgi a-mannosidase. Thus, DIM, and mannonolactam amidrazone, another synthetic mannosidase inhibitor [91], are somewhat unusual in that they inhibit both aryl-u-mannosidases as well as the Golgi mannosidase I. On the other hand, most of the other known mannosidase inhibitors that affect Golgi mannosidase I do not inhibit mannosidase I1 or aryl-a-mannosidase. These inhibitory effects are summarized in Table 1 . Kifunensine (see Figure 1B for structure) is an alkaloid produced by the actinomycete Kitusutosporiu kijiinense that corresponds in structure to the cyclic oxamide derivative of l-amino-DMJ [92]. This alkaloid is a very weak inhibitor of jack bean a-mannosidase, as is DMJ, but it is a potent inhibitor of Golgi mannosidase I (ZC50 = 2-5 x lop8 M). This level of inhibition is about 100-fold higher than inhibition of mannosidase I by DMJ. Interestingly, kifunensine did not inhibit either the ER mannosidase or mannosidase I1 [93]. Influenza virus raised in MDCK cells in the presence of kifunensine had N-linked oligosaccharides mostly with Man9(GlcNAc)2 structures in their envelope glycoproteins, rather than the typical complex chains [93]. This is the same effect as previously demonstrated with DMJ, except that kifunensine is much more potent and required only about 1/50 of the amount of material as required with DMJ. As mentioned above, swainsonine was found to be a potent inhibitor of mannosidase I1 and caused the formation of hybrid types of oligosaccharides in the N linked glycoproteins of MDCK cells, or in influenza virus grown in those cells [79]. In most studies where swainsonine was used to determine the effect of changes in oligosaccharide structure on glycoprotein function, this inhibitor had little effect on functional aspects of proteins in general. The inhibitor did prevent receptormediated uptake of mannose-terminated glycoproteins by macrophages, probably because the formation of hybrid chains on the macrophage surface allowed them to interact with their own receptors, and tie up these mannose receptors [94]. Swainsonine has been useful as a tool to determine the sequence of sugar addition during assembly of the N-linked oligosaccharides. Thus, addition of L-fucose or of sulfate to the N-linked oligosaccharides on the influenza viral hemagglutinin was studied in cell culture, in the presence of various processing inhibitors. When the glycoproteins were produced in the presence of castanospermine or DMJ, there was no [ 3H]fucose[95] or [ 35S]~ulfate [77] associated with these glycoproteins, suggesting that addition of fucose and sulfate required an oligosaccharide structure that did not contain glucose and had been partially trimmed by mannosidase I (see Figure 2). However, inhibition of mannosidase I1 by swainsonine did not prevent the addition of fucose or sulfate to the glycoproteins, indicating that the transferases that added these groups required the addition of GlcNAc by GlcNAc transferase I, but not the removal of mannoses by mannosidase 11. Swainsonine may also prove to be an effective chemotherapeutic agent against certain types of cancers. Thus, swainsonine has been found to reduce tumor cell metastasis, enhance cellular immune responses and resuce solid tumor growth in

528

31 Glycoprotein Processing Inhibitors

mice [96]. At least part of the reason for these varied effects is because swainsonine prevents the formation of the complex chain on the 6-branch of the N-linked oligosaccharides. This is because swainsonine inhibits mannosidase 11, therefore preventing the removal of the a1,3- and a1,6-mannoses from the a1,6-branched mannose, and blocking the addition of the P1,6-GlcNAc [79]. This GlcNAc is apparantly a regulatory step in the formation of polylactosamine chains which are important in tumor cell growth. Another inhibitor of mannosidase 11, called mannostatin, was isolated from the fungus, Streptoverticillium verticillus [97]. This compound is quite unusual because it has an exocyclic nitrogen rather than a nitrogen in the ring. Mannostatin does have a five-membered ring structure but without a bridge nitrogen, and a thiomethyl group, both of which are also uncommon in glycosidase inhibitors. Nevertheless, mannostatin is a fairly good mannosidase 11 inhibitor (I& = 100 nM), and causes the formation of hybrid chains in cell cultures, or in enveloped viruses [97]. Of interest is the observation that acetylation of the amino group of mannostatin resulted in loss of inhibitory activity. While mannostatin does not offer any advantage over swainsonine as an inhibitor, it is an important structure since it adds significant information to our knowledge of what is required for a compound to be a glycosidase inhibitor.

31.9 Summary and Future Prospects There are currently a number of known glucosidase and mannosidase inhibitors that are reasonably active against the glycoprotein processing glycosidases, and that have been valuable tools to assess the functions of specific carbohydrate structures in the function of various glycoproteins. Some of these inhibitors, especially those that affect the processing mannosidases, have also been useful tools to distinguish the different types of mannosidase activities, and to identify specific mannosidases in tissue or cell fractions. Thus, one might assume that there are already enough of these types of compounds available to fulfill the needs for processing inhibitors. However, there are still good reasons to either look for new naturally-occurring inhibitors, or to make new and different inhibitors chemically. For one thing, most of the currently-available compounds do not have the substrate specificity to only affect one of the processing enzymes, or to not affect some of the aryl-glycosidases as well. Secondly, many of these compounds do not have the high degree of potency that one would desire, and therefore it will be important to increase the effectiveness of such inhibitors. Thirdly, new glycosidases, especially those in the ER and Golgi are continually being discovered, and therefore new inhibitors will be useful to differentiate these enzymes from other already known activities. Another group of inhibitors that could be very important tools in Glycobiology are glycosyltransferase inhibitors that are able to permeate intact cells and that are specific for individual enzymes such as GlcNAc transferase I or GlcNAc transferase

Rejkvences

529

11, individual fucosyltransferases, and so on. There are many oligosaccharide structures whose function depends on the addition of one or several terminal sugars to the structure [98, 991 and thus compounds that can prevent specific additions of sugars, in cells or tissues, can have valuable uses as biochemical tools and may also have a place in chemotherapy.

References I . Colegate, S.M., Dorling, P.R. and Huxtable, C.R. (1979) Ausf. J. Chem. 32, 225772264, 2. Hohenschutz, L.D., Bell, E.A., Jewess, P.J., Leworthy, D.P., Pryce, R.J., Arnold, E. and Clardy, J. (1981) Phytochemistry 20, 81 1-814. 3. Kite, G.C., Horn, J.M., Romeo, J.T., Fellows, L.E., Lees, D.C., Scofield, A.M., and Smith, N.G. (1990) Phytochemistry 29, 103-105. 4. Goldman, A,, Milat, M., Ducrot, P.H., Lallemand, J.Y., Maill, M., Lepingle, A., Charpin, 1. and Tepfer, D. (1990) Plzyfochemistry 29, 2115- 2127. 5. Evans,.E.V., Fellows, L:E., Shing, T.K.M., Fleet, G.W.J. (1985) Phytochemistry 24, 19531955. 6. Legler, G. and Julich, E. (1984) Carbohyd. Res. 128, 61-72. 7. Molyneux, R.J., Benson, M., Wong, R.Y., Tropea, J.E. and Elbein, A.D. (1988) J. Nut. Prod. 51. 1198-1206. 8. Molyneux, R.J., Pan, Y.T., Goldmann, A., Tepfer, D.A. and Elbein, A.D. (1993) Arch. Biochew. Biophys. 304, 81 -88. 9. Asano, N., Kato, A., Oseki, K., Kim, H. and Matsui, K. (1995) Eur. .I. Biochem. 229, 369-376. 10. Evans, E.V., Fellows, L.E., Shing, T.K. and Fleet G.W.J. (1985) Phytochenzistry 24, 19531955. 11. Fellows, L., Doherty, C.H., Horn, J.M., Kite, G.C., Nash, R.J., Romeo, J.T., Simmonds, M.S.J. and Scofield, A.M. (1%9) in “Swainsonine cind Related Glycosidase Inhibitors”, pp. 396416, (James, L.F., Elbein, A.D., Molyneux, R.J. and Warren, C.D., eds.), Iowa St. Univ. Press, Ames, 10. 12. Stahr, H.M. and Martin, P.J. (1989) in “Swainsonine and Related Glycosidase Inhibitors”, pp. 118- 124, Iowa St. Univ. Press, Ames, 1 0 . 13. Molyneux, R.J. (1993) Meth. Plant Bioclzem. Vol 8, 511-530 (Waterman, P.G., Ed), Academic Press, NY) 14. Dorling, P.R., Huxtable, C.R. and Colegate, S.M. (1980) Biochen?. J. 191, 649-652. 15. Saul, R., Chambers, J.P., Molyneux, R.J., and Elbein, A.D. (1983) Arch. Bioclzem. Biophys. 221, 593-597. 16. Saul, R., Molyneux, R.J. and Elbein, A.D. (1984) Arch. Biochem. Biophys. 230, 668-675. 17. Pan, Y.T., Hori, H., Saul, R., Sanford, B.A., Molyneux, R.J. and Elbein, A.D. (1983) Biochemistry 22, 3975--3984. 18. Trombetta, E.S., Simons, J.F. and Helenius, A. (1996) J. Biol. Chem. 271, 27509%27516. 19. Szumilo, T. and Elbein, A.D. (1985) Anal. Biocliem. 151, 32-40. 20. Stutz, A,, (1999) in “Iminosugurs us Gtycmidasa Inhibitor.s”, pp. 158-187. (Stutz, A., ed), Wiley-VCH, Weinheim, FRG. 2 I . Tyler. P.C. and Winchester, B.O. (1999) in “Iminosugurs as Glycosidase Inhibitors”, pp. 125156. (Stutz, A,, ed.) Wiley-VCH, Weinheim, FRG. 22. Ducrot, P-H. and Lallemand, J.Y. (1990) Tetruhedron Lett. 31: 3879-3882. 23. Winkler, D.A. and Holm, G. (1989) J. M e d Cliem. 32, 2084-2089. 24. Winkler, D.A. (1996) J. Med. Cham. 39, 4332- 4334. 25. Asano, N., Kato, A. Miyauchi, M., Kim, H., Tomimori, T., Matsui, K.: Nash, R.J. and Molyneux, R.J. (1997) Eur. J. Biochern. 248, 296-~303. 26. Ebner, M., Ekhart, C.E. and Stutz, A . (1997) in “Electronic Conference on Heterocyclic Chemistry 96”, (Rzepa, H.S., Snyder, J. and Leach, C., eds.), Royal Society of Chemistry, London.

530 27. 28. 29. 30.

31 Glycoprotein Processing Inhibitors

Legler, G. (1973) Cell. Biochem. 2, 31-38. Dale, M.P., Ensley, H.E., Kern, K., Sastry, K. and Byers, L.D. (1985) Biochemistry 24, 3530. Hubbard, S.C. and Ivatt, R.J. (1981) Annu. Rev. Biochem. 50, 555-583. Snider, M. (1984) in “Bioloqy of Carbohydrates”, (Ginsberg, V and Robbins, P.W., eds.) Vol. 2, pp. 163-198, Wiley and Sons, NY. 31. Moreman, K.W., Trimble, R.B. and Herscovics, A. (1994) Glycobiology 4, 113-126. 32. Katz-Fuller, B., Bieberich, E. and Bause, E. (1995) Eur. J. Biochem. 231. 344-351. 33. Brada, D., Kerjaschki, D. and Roth, J. (1990) J Cell. Bid. 110, 309-318. 34. Bischoff, J., Liscum, L. and Kornfeld, R. (1986) J. Biol. Chem. 261, 4766-4774. 35. Lal, A., Pang, P., Kalelkar, S., Romero, P.A., Herscovics, A. and Moreman, K.W. (1998) Glycobiology 8, 981-995. 36. Tulsiani, D.P.R. and Touster, 0. (1988) J. Biol. Chenz. 254, 5408-5417. 37. Bieberich, E., Tremi, K., Volker, C., Rolfs, A,, Katz-Fuller, B. and Bause, E. (1997) Eur. J. Biochem. 246, 68 1-689. 38. Schachter, H. (1986) Biochem. Cell. Biochem. 64,163-181. 39. Yip, B., Chen, S.H., Mulder, H., Hoppner, J.M. and Schachter, H. (1997) Biochem. J. 321, 465-472. 40. Herscovics, A. (1999) in “ComprehensiveNatural Products Chemistry”, (Pinto, B.M., ed.), Vol. 3, Chapter 1. Elsevier Press, NY. 41. Moreman, K.W., Touster, 0. and Robbins, P.W. (1991) J. Bid. Chem. 252, 3227-3233. 42. Kaushal, G.P., Szumilo, T., Pastuszak, I. and Elbein; A.D. (1990) Biochemistry 29, 21682176. 43. Misago, M., Liao, Y.-F., Kudo, S., Eto, S., Mattei, M-G., Moreman, K.W. and Fukuda, M.N. (1995) Proc. Natl Acud Sci. USA 92, 11766- 1 1770. 44. Tulsiani, D.P.R. and Touster, 0. (1985) J. Biol. Chem. 260, 13081-13087. 45. Monis, E., Bonay, P. and Hughes, R.C. (1987) Eur. J. Biochern. 168,287-294. 46. Bonay, P. and Hughes, R.C. (1991) Eur. J. Biochem. 197, 229- 238. 47. Tulsiani, D.P.R., Skudlarek, M.D. and Orgebin-Crist, M.C. (1989) J. Cell Biol. 109, 12571267. 48. Herscovics, A. (1999) Biochim. Biophys. Acta 1426, 275-285. 49. Spiro, R.G. (1994) in “Guidebook to the Secretory Pathwlay” (Rothblatt, J., Novick, P. and Stevens, T., eds.) pp. 188-189, Oxford press. 50. Karaivanova, V.K., Luan, P. and Spiro, R.G. (1998) Glycobiology 6, 861-868. 51. Kornfeld, R. and Kornfeld, S. (1985) Annu. Rev. Biochem. 54, 631-664. 52. Beyer, T.A., Sadler, J.E., Rearick, J.L., Paulson, J.C. and Hill, R.L. (1981) Adv. Enzymol. 52, 23-175. 53. Fukuda, M., Bierhuizen, M.F.A. and Nakayama, J. (1996) Glycobiology 6, 683-689. 54. Rhinehart, B.L., Liu, P.S. and Robinson, K.M. (1989) in “Swainsonineand Related Glycosidase Inhibitors” (James, L.F., Elbein, A D . , Molyneux, R.J. and Warren, C.D., eds.) pp. 279-290, Iowa St. Univ. Press, Ames. 10. 55. Saul, R., Ghidoni, J.J., Molyneux, R.J. and Elbein, A.D. (1985) Proc. Natl Acad. Sci. U S A 82, 93-97. 56. Chen, Y.T. and Burchell, A. (1995) in “The Metubolic and Molecular Basis of Inherited Diseases”, (Scriver, C.R., Beaudet, A.L., Sly, W.S. and Valle, D., eds.) McGraw-Hill, Inc. NY. 57. Romero, P.A., Friedlander, P. and Herscovics, A. (1985) FEBS Lett. 83, 29-32. 58. Elbein, A.D., Mitchell, M., Sanford, B.A., Fellows, L.E. and Evans, S.V. (1984) J. Biol. Chem. 259, 12409-1241 5. 59. Tropea, J.E., Molyneux, R.J., Kaushal, G.P., Pan, Y.T., Mitchell, M. and Elbein, A.D. (1989) Biochemistry 28, 2027-2034. 60. Ando, O., Satake, H., Itoi, K., Sato, A,, Nakajima, M., Takahashi, S., Haruyama, H., Ohkuma, Y., Kinoshita, T. and Enokita, R. (1991) J. Antibiot. 44,1165-1 168. 61. Zeng, Y-C. and Elbein, A.D. (1998) Arch. Biochem. Biophys. 355, 26-34. 62. Liu, P.S. (1987) J. Org. Chem. 52, 4717-4721. 63. Rhinehart, B.L., Robinson, K.M., Liu, P.S., Payne, A.J., Wheatley, M.E. and Wagner, S.R. (1987) J. Pharmucol. E-xp. Ther. 241, 915-921.

64. Kaushal, G.P., Pan, Y.T., Tropea. J.E.; Mitchell, M., Liu, P. and Elbein, A.D. (1988) J. Bid. Chem. 263, 17278- 17283. 65. Elbein, A.D. (1991) FASEB J. 5, 3055-3063. 66. Lodish, H. and Kong, N . (1984) J. Cell Biol. 98, 1720-1729. 67. Edwards, E.H., Sprague, E.A., Kelley, J.L., Kerbacher, J.J., Schwartz, C.J. and Elbein, A.D. (1989) Biochemistry 28, 7679-7687. 68. Arakai, R.F., Hedo, J.A., Collier, E. and Gordon, P. (1987) J. Biol. Chem. 262, 11886-11892. 69. Montefiori, D.C., Robinson, W.E., Jr. and Mitchell, W.M. (1988) Proc. Nut1 Acad, Sci. USA 85, 9248-9252. 70. Hammond, C., Braakman, I. and Helenius, A. (1994) Proc. Nut/ Acad. Sci. U S A 91, 913-917. 71. Van Leeuwen, J.E.M. and Kearse, K.P. (1997) J. B i d . Chcw7. 272, 4179-4186. 72. Ware, F.E., Vassilakos, A,, Peterson, P.A., Jackson, M.R., Lehrman, M.A. and Williams, D.B. (1995) J. Biol. Chem. 270, 4697-4704. 73. Hebert, D.N., Foellmer, B. and Helenius, A. (1995) Cell81, 425-433. 74. Hammond, C. and Helenius, A. (1995) Curr. Opiiz. Cell Bid. 4, 523-529. 75. Trombettd, S.E. and Parodi, A.J. (1992) J. Bid. Chem. 267, 9236-9240. 76. Fernandez, F.. Jannatipour; M., Hellman. U., Rokeach, L.A. and Parodi, A. (1996) EMBU J. 15, 705-713. 77. Merkle, R.K., Elbein, A.D. and Heifetz, A. (1985) J. Bid. Chem. 260, 1083-1089. 78. Kang, M.S. and Elbein, A.D. (1983) Plant Physiol. 71, 551-554. 79. Elbein, A.D., Solf, R., Dorling, P.R. and Vosbeck, K. (1981) Priic. Nut/ Accid. Sci. U S A 78, 7393-1397. 80. Elbein, A.D., Vosbeck, K., Dorling, P. and Horisberger, M. (1982) J. B i d . Chem. 257, 15731576. 81. Tulsiani, D.P.R., Broquist, H, and Touster, 0. (1985) Arch. Biochem. Biophys. 236, 427-434. 82. Stegelmeier, B.L.. Molyneux, R.J., Elbein, A.D. and James, L.P. (1995) Vet. Pathol. 32, 289298. 83. Fuhrmann, V., Bause, E., Legler, G. and Ploegh, H. (1984) Nuture 307, 755-758. 84. Fellows, L.E., Bell, E.A., Lynn, D.G., Pilkiewicz, F., Miura, 1. and Nakanishi, J. (1979) J. Chem. Soc. Chem. Commun. 22, 977-978. 85. Elbein, A.D., Legler, G. Tlusty, A,, McDowell, W. and Schwarz, R. (1984) Arch. Biochem. Biophys. 235, 579-588. 86. Gross, V., Steube, K., Tran-Thi, A. McDowell, W. and Schwarz, R. (1985) Eur. J. Biochem. 150, 41L46. 87. Chin, D.J., Luskey, K.L., Anderson, R.G., Faust, J.R., Goldstein, J.L. and Brown, M.S. (1982) Proc Nut1 Acud. Sci. U S A 79, 1185-1 189. 88. Bischoff, J., Liscum, L. and Kornfeld, R. (1986) J. B i d . Chem. 261, 4766-4774. 89. Snider, M.D. and Rogers, O.C. (1986) J. Cell Biol. 103, 265-275. 90. Palamarczyk, G., Mitchell, M., Smith, P.W., Fleet, G.W.J. and Elbein, A.D. (1985) Arch. Biochem. Biouhvs. 243. 35-45. 91. Pan, Y.T., Kaushal. G.P., Ganem, B. and Elbein, A.D. (1992) J. Bid. Chem. 267, 83138318. 92. Kayakiri, H., Takese, S., Shibata, T., Okamoto, M., Terano, H., Hashimoto. M., Tada, T. and Koda, S. (1989) J. Orq. Chem. 54, 4015-4016. 93. Elbein, A.D., Tropes.-J.E.. Mitchell. M. and Kaushal, G.P. (1990) J Bid. Chem. 265, 1559915605. 94. Chung, K.M., Shepard, V.L. and Stahl, P. (1984) J. Biol. Chem. 259, 14637-14641. 95. Schwartz, P. and Elbein, A.D. (1985) J. Bid. Chem. 260, 14452-14458. 96. Goss, P.E., Reid, C.L., Bailey, D. and Dennis, J.W. (1997) Clin. Cancer Rex 3, 1077-1086. 97. Tropea, J.E, Kaushal, G.P., Pastuszak, I., Mitchell, M., Aoyagi, T., Molyneux, R.J. and Elbein, A.D. (1990) Biochemistry 29, 10062 ~10069. 98. Varki, A. (1997) Glycohiology 3, 97-130. 99. Stanley, P. (1992) Glycobiology 2, 99-107.

Part I1 Volume 4

111

Lectins

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

32 Plant Lectins Murilynn E. Etzler

32.1 Summary Although united by their defining feature of carbohydrate binding activity, plant lectins show great diversity in carbohydrate specificity, structure and function. Some of these proteins also exhibit other activities such as enzyme activity or the ability to bind plant hormones. This Chapter provides an overview of our present state of knowledge of the activities and structures of these lectins and discusses the current state of our information and hypotheses on the roles these proteins may play in the plant.

32.2 Introduction Since their discovery in the late 18OOs, lectins have been found in a wide variety of plant species representing almost every main taxonomical classification in the plant kingdom [I]. The wide range of carbohydrate specificities found among these carbohydrate binding proteins has enabled them to be used as tools for a great variety of purposes, ranging from glycoprotein isolation and characterization to cell sorting, drug targeting and various biomedical diagnostic assays [2-41. Although lectins are also found in animals and microorganisms, it is the plant lectins that have been primarily used for such applications because of their solubilities in aqueous solvents and ready availability. In fact, over 60 plant lectins are now available commercially, and this number represents only a small percentage of the vast number of plant lectins that have been described to date. In recent years substantial progress has been made in achieving an understanding of the structural attributes that contribute to the carbohydrate binding properties of these proteins. Indeed the crystal structures of approximately 20 plant lectins have

536

32 Plant Lectins

now been determined, many of them in the presence of their carbohydrate ligands [4]. Yet, despite the wealth of information available on these proteins, at this point in time we know little of their biological role(s) in the plant. This chapter provides an overview of our current state of information on plant lectin structure and function and reviews newly emerging information that is promising to lead to an understanding of the biological roles of this interesting group of plant proteins.

32.3 Carbohydrate Specificity Extensive carbohydrate specificity studies have been conducted on a large number of plant lectins [l, 31. It should be noted that most of these studies originate from the fortuitous finding that a particular lectin agglutinates a specific type of erythrocyte or precipitates a certain glycoconjugate. The carbohydrate specificity of the lectin is then determined by comparing the abilities of a wide range of monosaccharides and oligosaccharides to inhibit the interaction of the lectin with such cells or glycoconjugates. Although considerable information has been obtained on the specificities of plant lectins by this approach, at present we have no information on the physiological ligands for any of these plant lectins and there is always a possibility that the lectin may combine with other ligands that may not immediately be predicted to interact with the protein based on the previous specificity studies. For example, the presence of an aromatic aglycon has been found to substantially increase the affinity of some legume lectins for a glycoside and at times even overrule the anomeric preference established in previous specificity studies of these lectins using methyl glycosides. Such an effect is due to a hydrophobic pocket close to the carbohydrate-binding sites of these lectins [ l , 41. Another example is the recent finding that concanavalin A, which is probably the most widely used plant lectin, interacts with peptides that have a Tyr-Pro-Tyr motif with an affinity similar to its affinity for methyl a-mannoside [5, 61. The peptide was found to mimic the carbohydrate ligand by binding to the carbohydrate binding site of the lectin [7]. Such findings emphasize the need for caution when using lectins to characterize the structures of glycoconjugates. Most plant lectins can bind to simple sugars and are usually categorized on the basis of their monosaccharide specificities as demonstrated in Table 1. Although this table lists only a small number of plant lectins, these representative lectins can be used as examples to demonstrate a number of general features of the carbohydrate specificities of plant lectins. It should first be pointed out that although a lectin may be grouped into a particular monosaccharide specificity category, the lectin may also be able to bind with other monosaccharides. For example, both concanavalin A and the pea lectin can also bind to N-acetylglucosamine' due to their toleration of variations at the C-2 position of the pyranose ring [8, 91. The wheat germ agglutinin has a weak affinity for N-acetylneuraminic acid [lo]; this interacAll sugars discussed in this chapter are in the D-configuration and the pyranose ring form unless specified otherwise.

Ricinus communis agglutinin (RCAI)

Peanut (PNA)

ECorL

Soybean agglutinin (SBA)

Galactose/CalNAc specific Dolichos bijlorus (DBL)

GlcNAc specific Wheat germ agglutinin (WGA) GSII

Snow drop (GNA)

Pea (PSL)

Mannose/Glucose specific Concanavalin A (Con A)

Lectin Name (abbreviation)

Glycine mux (soybean) seeds Erythrina corallodendron (Coral tree) seeds Arachis hypogaea (Peanut) seeds Ricinus communis (castor bean) seeds

D. bijlorus seeds

Mannose

Mannose

Monosaccharide preference

50

6-7

Gal p1,3GalNAc

Gal 0 1,4Clc

P

Euphorbeaceae Galactoseg

Leguminosae

B

30-50

5

63

Galactoseg

GalNAc al,3GalNAc P1,3Gal al,4Gal pl,4Clc GalNAc a1,3Gal P1,6Glc

3,000

28

12

>I30

Gal P1,4GlcNAc

GalNAc E Gal

N-acetylgalactosamine a

N-acetylgalactosamine' a

a

N-acetylglucosamine

(GlcNAc p1,4)3

Man al,6(Man al,3)Man a-0-Methyl

Glc al,2Glc

Man al,6(Man a1,3)Man

R A ~

None

Leguminosae

Leguminosae

Leguminosae

None

a

U

U

Anomeric' Preferred preference oligosaccharide

N-acetylglucosamine

Amaryllidaceae Mannose"

Leguminosae

Leguminosae

Plant family

Triticum vulgare (wheat) Graminaceae grain Griffoniu simplicijblia Leguminosae seeds

Cunavaliu ensiformis (jackbean) seeds Pisum satiuum (pea) seeds Galanthus nivalis (snow drop) bulbs

Sourceb

Table 1. Carbohydrate specificities of representative plant lectinsa.

5 .1"

Q

2$

$

B

5-

ci

L4

lu

k

Sourceb

Plant family

Solanurn tuberosum (potato) tubers

Phaseolus tdgaris (red kidney bean) seeds

S. nigra (elderberry) bark

Solanaceae

Leguminoseae

Caprifohaceae

None

None

a

Gal pl,4GlcNAc 81,2 (GlcNAc p1,4)2-5

Gal p 1,4GlcNAc p 1,6

NeuSAc a2,6Gal

, , Man

81,6R L-FUCal.6GlcNAc

a

Fucose

Gal

L-FUCal,2Gal fi1,4GlcNAc

a

Anomeric" Preferred preference oligosaccharide

Fucose

Monosaccharide preference

1600

900

RAd

" A limitation on number of references allowed for this chapter does not enable the citation of the original references for the information included in this table. The reader is referred to references [ 1, 3,4] for this information. Common name of plant in parentheses. Based on ability to react with the a or p glycoside. Relative affinity compared to that of the preferred monosaccharide. Unlike the other lectins listed in this category, the GNA Iectin does not bind to Glc or GlcNAc. Has a very pronounced preference for GalNAc. Does not bind to GalNAc. Does not bind to free NeuSAc but does bind weakly to Gal.

Potato

Oligosaccharide only specific L-PHA

Sialic acid specific Sarnbucus nigra I (SNA)

Ulex europaeus (Gorse) Leguminosae seeds Lotus tetragonolobus (LTA) L. tetragonolobus Leguminosae (asparagus pea) seeds

L-Fucose specific UEAI

Lectin Name (abbreviation)

Table 1 (continued)

L

5

"cr

L-

b

LJ

00

w

ul

32.4 Other Activities

539

tion is due to the similarities at positions C-2 (acetamide group) and C-3 (hydroxyl group) of the pyranose rings of these monosaccharides and the major role of the substituents at these two positions in binding to the carbohydrate-binding site of the lectin [ l l , 121. Secondly, it must be acknowledged that all lectins within a particular specificity category may not bind to all of the monosaccharides of that category or may show a marked preference for one of the sugars. For example, the GNA lectin in the mannose/glucose category does not bind glucose 1131; within the galactose/ N-acetylgalactosamine category the DBA lectin shows a marked preference for Nacetylgalactosamine over galactose [ 141, whereas the ECorL lectin binds to both monosaccharides with similar affinity [15]. As shown in Table 1, lectins within an individual specificity category can also differ from one another in anomeric preference and show marked differences in specificities toward particular oligosaccharides. Although the association constants of plant lectins for their individual monosaccharides are quite low, usually in the millimolar range, these lectins often show a much higher affinity for particular oligosaccharides (Table 1). Lectins within a particular category often show marked differences from one another in oligosaccharide specificity and can thus be used to discriminate among various types of oligosaccharides. The specificities of most plant lectins are directed toward the nonreducing ends of the oligosaccharides; however some lectins can bind to residues in the interior of a carbohydrate chain. For example, concanavalin A can bind to internal a-mannosyl residues glycosidically linked at the C-2 position and thus can bind exceptionally well to many glycoproteins [ 11. Some plant lectins do not recognize simple sugars at all and bind only to oligosaccharides (Table 1). Most plant lectins are multivalent. As initially shown with antibodies [16], such multivalence can enhance the affinity of a protein for multivalent ligands by several orders of magnitude. Many plant lectins therefore bind strongly to cell surfaces and highly branched glycoconjugates containing ligands that are recognized by their carbohydrate-binding sites. Recent studies have established that such interactions between some multivalent lectins with multivalent oligosaccharides or glycoproteins results in the formation of highly ordered cross-linked lattices 1171. When the valence of either the lectin or ligand is greater than two, it has been found that separate homogeneous complexes can be formed upon the addition of the lectin to a mixture of oligosaccharides. X-ray crystallographic studies of such complexes have established that differences in structures of the cross-linking carbohydrates determine the different lattice structures [ 181. Such crystal-packing interactions that occur in these multivalent complexes provide an extra dimension of specificity and provide a mechanism for propagating long range order in a system.

32.4 Other Activities Although plant lectins are distinguished by their carbohydrate binding abilities, a number of these proteins have additional activities. Some lectins, such as ricin,

540

32 Plant Lectins

abrin and modeccin are heterodimers composed of a carbohydrate binding chain (B chain) linked by disulfide bond to a toxic chain (A chain) [ l , 31. The A chain is an enzyme that cleaves the N-glycosidic linkage of a specific adenine of 28s ribosomal RNA, thus inhibiting protein synthesis [ 191. Binding of the B chains of these lectins to carbohydrate on the cell surface initiates the internalization of the lectin and results in cell death. These lectins are extremely toxic and it has been estimated that a cell can be killed by a single molecule [4]. Several conventional legume lectins have been found to contain high affinity hydrophobic sites that bind to adenine and derivatives of adenine with association constants in the micromolar range, a value two or three orders of magnitude higher than the association constants of these same lectins for their monosaccharide ligands [20]. No intcraction has bcen dctectcd between thcse adcninc binding sites and the carbohydrate binding sites of these lectins. The hydrophobic sites are located in the interface between the lectin subunits and are dependent on the quaternary structures of these lectins [21]. A new class of legume lectin with no significant homology to any lectin described to date has recently been identified in the roots of the legume, Dolichos blJorus [22]; homologs of this lectin have also been found in other legumes [23]. Members of this class of lectins are distinguished by their abilities to hydrolyze the phosphoanhydride bonds of nucleoside di- and triphosphates, an activity that places them in the apyrase category of enzymes. This category of lectin has been named LNP to designate the jectin and gucleotide phosphohydrolase activities of these proteins. The apyrase activity of the D. blJorGs LNP is increased in the presence of its carbohydrate ligands, suggesting that there may be an interaction between the enzymatic and carbohydrate-binding sites of this lectin [22]. The possibility that other plant lectins may have enzyme activity is suggested by the finding of a cDNA from Arabidopsis thaliana that encodes a putative legume seed lectin-like domain as well as a receptor-like serine/threonine kinase [24]. Such combinations of carbohydratebinding and enzyme activity within a single protein may have considerable ramifications on the functions of these lectins as discussed later in this chapter.

32.5 Structure A vast amount of information is presently available on the structures of plant lectins, including the complete amino acid sequences of many of these proteins and the crystal structures of almost 20 different lectins, many in complex with their carbohydrate ligands. Several excellent reviews on these structures have appeared within the past two years [4, 25, 261 to which the reader is referred for more detailed considerations than can be provided in the present chapter. In general, the structures of lectins obtained from species within the same phylogenetic class of plants are quite similar to one another but differ markedly from the structures of lectins from plants of other classes. The largest amount of information is available on the lectins from the seeds of legumes which can comprise up to 10% of the total seed protein. The

32.5 Structure

541

Figure 1. Ribbon representations of monomeric units of three different types of plant lectins. A) A typical legume seed lectin monomer with the position of the carbohydrate binding site indicated with an asterisk. The metal ions are designated by the small spheres. The structure shown is that of the Dolichos hiforus seed lectin, PDB entry 1 LU. B) Wheat germ agglutinin monomer, PDB entry 9WGA. The sulfurs of the disulfide bonds are shown as light colored spheres. C ) Guhnthus niualis monomer, PDB entry INIV, with three mannoses bound. The figures were drawn with the MOLSCRIPT program [28].

abundance of these legume seed lectins, their solubilities and the wide range of carbohydrate specificities found among them have made this group of lectins the most widely used and intensively studied group of carbohydrate binding proteins. There is a marked conservation in the primary, secondary and tertiary structures of the monomeric units of all of the legume seed lectins. The monomeric unit (Figure 1A) has a compact structure consisting of three anti-parallel P-sheets: a sixstranded back sheet, a seven-stranded front sheet and a five-stranded sheet which helps to hold the other two sheets together [27]. About 50% of the amino acid residues reside in the loops and bends connecting the peptide strands in the sheets. The carbohydrate binding site is located in a shallow depression on the perimeter of the front sheet and is in close proximity to sites for Ca2+ and a transition metal ion, which must be occupied for Carbohydrate binding activity. Although these metal ions do not directly interact with the carbohydrate ligand, they are in coordination with conserved asparagine and aspartate residues and help to position these amino acids for the formation of hydrogen bonds with the ligand. Other essential residues for carbohydrate binding include a main chain glycine or arginine that forms a hydrogen bond with the ligand and a hydrophobic residue that stabilizes the sugar by stacking interactions [26]. The differences in monosaccharide specificity among the legume lectins are primarily due to sequence variations within a peptide loop located at the perimeter of the carbohydrate binding site [25, 291. Variations in sequence among the lectins also contribute to the presence of subsites adjacent to the monosaccharide-binding site that bind to other substituents on the monosaccharide or to portions of an oligosaccharide and thus help to define the fine specificity of the lectin [26]. The legume seed lectins thus exhibit a wide range of carbohydrate specificities while maintaining a highly conserved structural framework. Although the monomeric units of different legume lectins are extremely similar with one another, marked variations have been found in their quaternary structures.

542

32 Plant Lectins

B

C

D

Figure 2. Several different types of legume lectin oligomers. A) The canonical dimer is represented by the concanavalin A dimer, PDB entry 1APN. B) The concanavalin A tetramer, PDB entry 10NA. C ) A new type of dimer found in DB58, the D. bgorus stem and leaf lectin, PDB entry 1LUL. D) The D. bijlorus seed lectin tetramer, PDB entry 1BJQ. This type of tetramer is also characteristic of the soybean agglutinin and the PHA-L lectin from Phaseolus vulgaris. The position of the hydrophobic binding sites are denoted by asterisks. The figures were drawn with the MOLSCRIPT program [28]. Two other types of dimers found with the GSIV and ECorL lectins and another type of tetramer found with the peanut agglutinin are not shown.

Four different dimeric and three different tetrameric structures have been described to date. In the most common dimer the back sheets of the lectin monomers associate to form a continuous 12-stranded P-sheet along the length of the dimer (Figure 2A). This dimer, called the “canonical legume lectin dimer” has been found in most legume lectins studied [26]. In some lectins these dimers combine to form tetramers of which there are two types. In the case of concanavalin A, the dimers interface at the central parts of their back sheets to form a structure as shown in Figure 2B. In the case of such lectins as PHA-L, the soybean agglutinin and the D. bijorus seed lectin, the dimers interface at the outermost strands of the 12-stranded P-sheets to

32.6 Biological Roles

543

form a large channel in the middle of the tetramer (Figure 2D). It is of interest that only lectins with this latter type of quaternary structure have been found to bind adenine [26]. A recent crystallographic study of the D. bzjlorus lectin in complex with adenine has shown that the two hydrophobic sites per tetramer for this ligand are formed by the subunit interfaces at opposite ends of this cavity as designated by the asterisks in Figure 2D 1211. A closely related lectin (DB58) found in the stems and leaves of this plant exhibits a unique dimeric structure (Figure 2C), which essentially resembles the top half of the tetrameric structure of the seed lectin 1211. Indeed, this unique dimer has been found to have a single hydrophobic binding site 1301. In addition to the above quaternary structures, two additional types of legume lectin dimers have been found in the GS-IV and ECorL lectins, which are both dimeric proteins 1261. The peanut lectin, which is a tetramer, has a unique structure composed of two GS-IV type dimers 1271. Such variation in quaternary structure of the legume lectins may be of significance to the role(s) of these lectins in the plant. In comparison to the legume lectins, the barley and rice lectins, wheat germ agglutinin and other cereal lectins which are found in members of the Graminaceae family of plants, contain no metal ions, are very rich in cysteine and have similar carbohydrate specificities [ 1, 41. The crystal structure of the wheat germ agglutinin has been determined 1311. Each of the monomers composing this homodimer consists of four homologous subdomains held together by four disulfide linkages (Figure 1B). These monomers associate with one another in opposite directions to form a dimer composed of four domains, each composed of a subdomain from each monomer. The lectin has multiple carbohydrate-binding sites located at the interface between these subunits [4]. The crystal structures of lectins from other classes of plants, such as the snow drop lectin (Figure 1C) from the Amaryllidaceae family, jacalin from the Moraceae family and ricin from the Euphorbiaceae family are distinctly different from one another and from the legume and wheat germ agglutinin structures described above 141. Although detailed structural information on relatively few of these nonleguminous plant lectins is available, no similarity in three-dimensional structure is evident among lectins from different classes of plants. The high degree of sequence homology of lectins from the bulbs of plants in the amaryllis, orchid and garlic families predicts, however, that the lectins in these three families will have common structures [32, 331. The extent to which the presence of lectins in different classes of plants may represent divergent versus convergent evolution remains to be determined.

32.6 Biological Roles For many years the biological role of plant lectins has been an intriguing but elusive subject that has generated the proposal of a wide variety of hypotheses, but little substantiation of function [20, 34, 351. Recent developments in lectin research and

544

32 PlantLectins

plant biology are providing new insights on this subject and are helping us to begin to decipher the roles of these proteins. The extensive structural differences found in comparisons of lectins from plants of various phylogenetic classes and the different localizations of these lectins in the plant suggest that these proteins may play different physiological roles, some of which may be related to specialized activities associated with the individual classes of plants. Different lectins have also been found within a single plant species. In some cases these lectins have been found to be encoded by separate genes that are differentially expressed, both spatially and temporally [20]. It is thus clear that there are many different types of plant lectins and that these lectins may serve different functions in the plant. It is also clear that there is a wide variety of oligosaccharides in the plant that may be potential physiological ligands for these lectins. Recent work has established that plants utilize oligosaccharide signaling mechanisms to initiate various responses during development, defense and interactions of the plant with the environment [36-381. Some of these oligosaccharides are produced by agents in the plant environment whereas others are endogenous to the plant itself. It is highly probable that some lectins may serve as receptors for these signals or possibly be involved in their transport or presentation to the receptors. Indeed, as discussed later in this section, at least one lectin has been identified that may function in this capacity. The identification and characterization of these signals is proceeding rapidly and is expected to yield a variety of physiological ligands that will be invaluable in identifying other such lectins. Having considered such general principles, let us now consider some specific roles in which lectins have been implicated. For many years it has been proposed that lectins may play a role in plant defense. This idea originated almost 50 years ago with the recognition of some similarities in properties between lectins and antibodies and a proposal that lectins may function as components of a plant immune system [34]. Although it was soon recognized that plants do not have an immune system comparable to that of animals, circumstantial evidence began to accumulate that lectins may be involved in the defense response. A variety of lectins are presently implicated or envisioned to be involved in one or more of at least three roles related to plant defense. One such defense role for some lectins may be in the recognition of oligosaccharide signals produced by the breakdown of cell wall components of the plant or pathogen upon contact with the plant. Such signals, called elicitors, stimulate the plant to release low molecular weight isoflavonoids, called phytoalexins, that function as antimicrobial agents [39]. A second type of defense role may involve a direct interaction of a lectin with the infectious agent. Some chitin-binding lectins, such as the wheat germ agglutinin, have been found to inhibit the growth of fungi, insects and other organisms that contain chitin, possibly by interfering with chitin synthesis or deposition in the cell walls of these organisms [40,41],whereas other lectins have been implicated in the encapsulation of a variety of other pathogenic agents [34]. A third defense-type role for which there is considerable support is the role some lectins play in protecting the plant against animal predators [41]. As discussed earlier in this chapter, some lectins, such as ricin, consist of a carbohydrate binding

32.6 Biological Roles

545

subunit coupled to a highly toxic subunit. Ingestion of very small amounts of such lectins can have deadly effects on animal predators. The consumption of high concentrations of other lectins, such as some of the legume seed and bark lectins, has been found to promote severe gastrointestiiial disturbances. These lectins often account for up to 10% of the total protein in these legume tissues and are thought to play a role in warding off potential predators [41]. In this context it is noteworthy that in wild lines of the legume, Phaseolus uulgaris, a genetic correlation has been found between the presence of a protein, called arcelin, and the resistance of the seeds of this plant to insect predators [42]. The amino acid sequence and crystal structure of arcelin have established that this protein is a member of the family of legume seed lectins but is missing the metal ion binding loop that is essential for carbohydrate binding activity [43]. This finding suggests that the ability of the legume seed lectins to function in this defense role is not dependent on their carbohydrate binding activity but may instead be a secondary function conferred during the early evolution of the legumes by the acquisition of the ability of an ancestral seed lectin gene to be expressed in high concentrations. Although the protective effect of such high concentrations of lectins in legume seeds may have provided an evolutionary selective advantage to the plant and helped glycobiology by furnishing an abundant supply of lectins, the high concentrations of these lectins in the seeds may be misleading in attempts to determine the primary role of such lectins in the plant. Several factors suggest that the type of lectins found in legume seeds may indeed play some fundamental role in plants other than staving off predators or serving as a storage protein for food reserves for the plant. First, active carbohydrate binding sites of these lectins have been preserved during the evolution of the legumes. Second, lectins with structures very similar to the seed lectins have been found in vegetative tissues of the plant [20, 341. The gene encoding such a lectin in the Dolichos hiJoyus plant has been characterized and found to be located within 3 kb of the gene encoding the seed lectin; the linkage of these genes and their similar transcriptional orientations in the genome suggest that they arose by a gene duplication event [44]. The preservation of the genes encoding such vegetative tissue lectins with conventional legume seed lectin-like structure during evolution and their expression at very low concentrations in their respective tissues suggest that these lectins fulfill a role that is not dependent on high concentrations of lectin. Third, recent evidence suggests that these lectins may be related to the ERGIC-53 and VIP36 lectins in animals [45], thus indicating that this type of lectin fulfills a role that has been conserved throughout evolution. It is of interest that these animal proteins have been implicated in glycoprotein trafficking events in the secretory system [46, 471. Most considerations of plant lectin function revolve around the carbohydrate binding activities of these proteins. Yet, as discussed in a previous section of this chapter, additional activities are associated with some of these lectins. These activities must also be taken into account in considering the possible roles of these proteins. For example, a major class of plant hormones, the cytokinins, are derivatives of adenine that bind with high affinity to the hydrophobic binding sites found on some of the legume seed lectins and at least one seed-like vegetative tissue lectin [20, 301. It is of interest that the specificity of this site, when present, appears to have

546

32 Plant Lectins

been maintained during evolution despite the variations that have occurred in carbohydrate specificity and tissue distribution of these lectins. It is possible that this hormone binding site may play a regulatory role in the function of these lectins or perhaps even play a primary role that is regulated by the carbohydrate binding sites of these proteins. Additional activities would also be anticipated to be associated with lectins that may function in oligosaccharide signaling processes. In addition to recognizing the oligosaccharide ligands, such lectins would be expected to participate directly or indirectly in the initiation of the activation of downstream events involved in the signaling pathway. One such signaling process is the signaling event that occurs in the initiation of the establishment of the nitrogen-fixing rhizobium-legume symbiosis. Lipochitooligosaccharidic signals, called Nod factors, that are produced by the rhizobia elicit a plant response that includes the differentiation of cells in the root cortex to form a new organ, the nodule, and the deformation of root hairs [37]. The rhizobia adhere to these root hairs and travel to the nodule along an infection thread produced by the plant [48]. For many years lectins have been proposed to play a role in this symbiosis, however attention was focused on the conventional legume seed lectin-like proteins [34]. Although transgenic studies [49, 501 have suggested that such conventional legume lectins may participate in the adhesion of the rhizobium to the root, these lectins do not bind Nod factors and have not been found in the roots of all legumes. Recent studies have shown that the LNP lectin (Db-LNP) from the legume, D. bijlorus, binds to Nod factors produced by rhizobial strains that symbiose with this plant [22]. Members of this new category of legume lectins, which have no homology to the conventional legume seed lectins and have apyrase activity in addition to their carbohydrate binding activity, have been found in all legumes examined to date and are present in the roots of the plants [23]. Although it is not yet known whether Db-LNP exhibits the rhizobial strain specificity characteristic of this symbiosis, this lectin is present on the surface of the root hairs in the zone of nodulation of the root [22] and undergoes a redistribution to the tips of the root hairs in response only to those rhizobial strains that nodulate the plant [51]. Treatment of the roots with antibodies to this lectin prior to rhizobial infection blocks root hair deformation as well as nodule formation [22]. Members of this new category of lectins are thus promising candidates for a role in this oligosaccharide signaling process. Although we are just beginning to tap the surface with respect to information on the roles of plant lectins, a foundation has been laid and it is anticipated that rapid progress will be made in this area in the next several years. The acquisition of more information about plant oligosaccharide signaling processes, the identification of such signals and the use of genetic and transgenic approaches for assessing the role of specific lectins in particular functions should be of particular value in these studies. Acknowledgments

The author thanks Dr. David Wilson and Samantha Barling-Silva for the preparation of the computer drawings.

References

547

References 1. I.J. Goldstein, R.D. Poretz in The Lectins. Properties, Functions, and Applications in Biology and Medicine. (Eds.: I.E. Liener, N. Sharon, I.J. Goldstein), Academic Press, New York, 1986, pp. 33-247. 2. H. Lis, N. Sharon in The Lectins. Properties, Functions, und Applications in Biology and Medicine. (Eds.: I.E. Liener, N. Sharon, I.J. Goldstein), Academic Press, New York, 1986, pp. 293-

370. 3. I.J. Goldstein, H.C. Winter, R.D. Poretz in Glycoproteirzs II. (Eds.: J. Montreuil, J.F.G. Vliegenthart, H. Schachter), Elsevier, New York, 1997, pp. 403-474. 4. H. Lis, N. Sharon, Chem. Rev. 1998, 98, 637-674. 5. K.R. Oldenburg, D. Loganathan, I.J. Goldstein, P.G. Schultz, M.A. Gallop, Proc. Natl Acad. Sci. ( U S A ) 1992, 89, 5393-5397. 6. J.K. Scott, D. Loganathan, R.B. Easley, X. Gong, I.J. Goldstein, Proc. Nail Acad. Sci. ( U S A ) 1992,89, 5398-5402. 7. K.J. Kaur, S. Khurana, D.M. Salunke, J. Biol. Chem. 1997, 272, 5539-5543. 8. L.L. So, I.J. Goldstein, J. Immunol. 1967, 99, 158-163. 9. J.-P. Van Wauwe, F.G. Loontiens, C.K. DeBruyne, Biochim. Biopliys. Acta 1975, 379, 456461. 10. P.J. Greenaway, D. LeVine, Nature (London), New Biol. 1973,241, 191-192. 1 1 . V.P. Bhavanandan, A.W. Katlic, J. Bid. Chem. 1979, 254, 4000-4008. 12. B.P. Peters, S. Ebisu, I.J. Goldstein, M. Flashner, Biochemistry 1979, 18, 5505-5511. 13. N. Shibuya, I.J. Goldstein, E.J.M. Van Damme, W.J. Peumans, J. Biol. Chem. 1988,263, 728734. 14. S. Hammarstrom, L.A. Murphy, I.J. Goldstein, M.E. Etzler, Biochemistry 1977, 16, 27502755. 15. N. Gilboa-Garber, L. Mizrahi, Cun. J. Biochem. 1981,59, 315-320. 16. C.L. Hornik, F. Karush, Immunochrmistrj) 1972, 9, 325-340. 17. C.F. Brewer, ClTemtracts-Biochem. Mol. Bid. 1996, 6, 165-179. 18. L.R. Olsen, A. Dessen, D. Gupta, S. Sabesan, J.C. Sacchettini, C.F. Brewer, Biochemistry 1997, 36, 15073-15080. 19. Y . Endo, K. Tsurugi, J. Bid. Chem. 1987,262, 8128W3130. 20. M.E. Etzler, Trends Glycosci. Glycotech. 1998, 10, 247-255. 21. T.W. Hamelryck, R. Loris, J. Bouckaert, M-H. Dao-Thi, G. Strecker, A. Imberty, E. Fernandez, L. Wyns, M.E. Etzler, J. Mol. Bid. 1999, 286, 1161-1177. 22. M.E. Etzler, G. Kalsi, N.N. Ewing, N.J. Roberts, R.B. Day, J.B. Murphy, Proc. Natl Acud. Sci. ( U S A ) 1999, 96, 5856-5861. 23. N.J. Roberts, J. Brigham, B. Wu, J.B. Murphy, H. Volpin, D.A. Phillips, M.E. Etzler, Mol. Gm. Genet. 1999,262, 261-267. 24. C. Herve. P. Dabos, J.-P. Galaud, P. Rouge, B. Lescure, J. Mol. Biol. 1996, 258, 778-788. 25. V. Sharma, A. Surolia, J. Mol. Biol. 1997, 267, 433-445. 26 R . Loris, T. Hamelryck, J. Bouckaert, L. Wyns, Biochim. Biophys. Acta 1998, 1383, 9-36. 27 R. Banerjee, K. Das, R. Ravishankar, K. Suguna, A. Surolia, M. Vijayan, J. Mol. Biol. 1996, 259. 28 1-296. 28 , P. Kraulis, J. Applied Crystallog., 1991, 24, 946-950. 29 , N.M. Young, R.P. Oomen, J. Mol. Biol. 1992,228, 924-934. 30. C.V. Gegg, D.D. Roberts, I.H. Segel, M.E. Etzler, Biochemistry 1992, 31, 6938-6942. 31. C.S. Wright, J. Mol. Biol. 1989, 209, 475-487. 32. E.J.M Van Damme, 1.J. Goldstein, W.J. Peumans, Phytochemistry 1991, 30, 509-514. 33. E.J.M. Van Damme, J. Balzarini, K. Smeets: F. van Leuven, W.J. Peumans, Glycoconj. J. 1994, 11, 321-332. 34. M.E. Etzler in The Lectins. Properties, Functions, and Applications in Biology and Medicine. (Eds.: I.E. Liener, N. Sharon, I.J. Goldstein), Academic Press, New York, 1986, pp. 371-435. 35. E. Van Driessche in Advances in Lectin Research, Vol. 1. (Ed.: H. Franz), VEB Verlag Volk und Gesundheit, Berlin, 1988, pp. 73-1 34.

548

32 Plant Lectins

36. A. Darvill, C. Augur, C. Bergmann, R.W. Carlson, J-J. Cheong, S. Eberhard, M.G. Hahn, V-M. Lo, V. Marfa, B. Meyer, D. Mohnen, M.A. O’Neill, M.D. Spiro, H. van Halbeek, W.S. York, P. Albersheirn, Glycobiology 1992,2, 181-198. 37. J. Denarie, F. Debelle, J.-C. Prome, Annu. Rev. Biochem. 1996, 65, 503-535. 38. M.E. Etzler, J. Cell. Biochem. 1998, Suppl. 30131, 123-128. 39. R.A. Dixon, N.L. Paiva, Plant Cell 1995, 7, 1085-1097. 40. M.J. Chrispeels, N.V. Raikhel, Plant Cell 1991, 3, 1-9. 41. W.J. Peumans, E.J.M. Van Damme, Plant Physiol. 1995, 109, 347-352. 42. T.C. Osborn, T. Blake, P. Gepts, F.A. Bliss, Theor. Appl. Genet. 1986, 71, 847-855. 43. T.W. Hamelryck, F. Poortmans, A. Goossens, G. Angenon, M. Van Montagu, L. Wyns, R. Loris, J. Bid. Chem. 1996,271, 32796-32802. 44. J.J. Harada, J.P. Spadoro-Tank, J.C. Maxwell, D.J. Schnell, M.E. Etzler, J. Biol. Chem. 1990, 265, 4997-5001. 45. K. Fiedler, K. Simons, Cell 1994, 77, 625-626. 46. K. Fiedler, K. Simons, J. Cell Sci. 1995, 109, 211-276. 47. B.L. Tang, S.H. Low, H-P. Hauri, W. Hong, Eur. J. Cell Bid. 1995,68, 398-410. 48. P. Mylona, D. Pawlowski, T. Bisseling, Plant Cell 1995, 7, 869-885. 49. C.L. Diaz, L.S. Melchers, P.J.J. Hooykaas, B.J.J Lugtenberg, J.W. Kijne, Nature 1989, 338, 579-58 1. 50. P. van Rhijn, R.B. Goldberg, A.M. Hirsch, Plant Cell 1998, 10, 1233-1249. 51. G. Kalsi, M.E. Etzler, submitted for publication.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

33 Interactions of Oligosaccharides and Glycopeptides with Hepatic Carbohydrate Receptors Yuan C. Lee und Reiko T. Lee

33.1 Summary Mammalian hepatic lectin (HL), also known as asialoglycoprotein receptor (ASGPR), is a type I1 transmembrane glycoprotein present both on the plasma membrane and internal membranes of hepatocytes. The cell surface receptor binds asialoglycoproteins that carry many exposed Gal/GalNAc residues and internalizes them. The subunit of HL is composed of, from the N-terminal end, a cytosolic tail, trans-membrane segment, a neck region and a C-type lectin domain at the Cterminal end. The C-type lectin domain is a polypeptide composed of 135-150 amino acid residues that binds sugar in a calcium-dependent manner. The lectin domain of HL recognizes Gal/GalNAc, while the N-terminal cytosolic tail interacts with clathrin-coated pits which leads first to internalization of the receptor-ligand complex, followed by parting of ligand from receptor in the endosome. The separated ligdnd will be degraded in lysosomes, while the receptor cycles back to the surface to be reutilized. There are two types of subunit for mammalian HLs, which are products of two genes and share a great deal of sequence homology. The shorter one is designated as HL-1 and the longer one, HL-2. The studies with human, rat, and mouse HLs all showed that both types of subunit are necessary to express the ASGP-R that has high affinity and internalizes ligands efficiently. It was found that on hepatocyte cell surfaces large proportions of HL-1 and HL-2 exist in close proximity to each other. ASGP-R on the hepatocyte surface exhibits a very high order of glycoside cluster effect; i.e., a ligand carrying two and three terminal Gal residues with proper interresidue spacing will have an affinity increment of 1000-100,000-fold from the monovalent galactosides that have KD in millimolar ranges. Photoaffinity labeling experiments on isolated rat hepatocytes showed that such affinity enhancement is made possible due to specific and rigid organization of HL-1 and HL-2 to form a unique 3-site binding area which readily accommodates three terminal Gal residues

550

33 Interactions of Oligosaccharides and Glycopeptides

of the predominant form of non-sialylated triantennary complex-type, N-glycans. On hepatocyte surfaces, this basic binding unit, which most likely is composed of two HL-1 and one HL-2, is further clustered to form receptor patches. When HL is solubilized in Triton X-100 and affinity-purified, it appears to be present in aggregates of various sizes, of which the hexamer is predominant. The soluble HL exhibits much weaker cluster effects of -60- and 100-fold affinity increase for the best di- and trivalent ligands. Based on the knowledge of spacial requirements for the terminal Gal residues, a number of high-affinity ligands with two and three terminal Gal/GalNAc residues were synthesized. These synthetic ligands typically contain only mono- or disaccharide units, since HL does not recognize internal sugar residues on the oligosaccharide chain. Because GalNAc binds more strongly to HLs than Gal, cluster glycosides containing GalNAc give larger degrees of affinity enhancement. These high-affinity, synthetic glycosides as well as poly-glycopeptides (based on poly-Llysine) were used in a number of studies to probe the properties and functions of the receptor, and also used as carriers of genes and drugs for delivery to the liver and in other medically oriented applications.

33.2 Introduction Discovery of the asialoglycoprotein receptor in mammalian liver by Morel1 and Ashwell [ l , 21 marks the beginning of modern glycobiology. The fortuitous discovery was made when they were studying the metabolic fate of ceruloplasmin by labeling it via modification of its sugar chains. The removal of sialic acid followed by specific oxidation of exposed galactose residues with galactose oxidase and incorporation of tritium with sodium borotritide resulted in a form of ceruloplasmin that was rapidly cleared from circulation. Subsequent studies showed that a transmembrane receptor exists on the parenchymal cells of liver, which binds, internalizes, and degrades galactose-exposed glycoproteins in a calcium-dependent manner. This receptor was named hepatic lectin (HL) or asialoglycoprotein receptor (ASGP-R). To study HLs in vitro, they are typically solubilized in a buffer containing non-ionic detergent, such as Triton X-100, and purified by affinity chromatography on a gel matrix containing desialylated serum glycoprotein, such as al-acid glycoprotein or fetuin. The functional entity was shown to be hexamer by cross-linking reaction [3] and mild SDS-PAGE separation [4]. Interestingly as shown in Table 1, hepatic lectins from different animal families express different carbohydrate specificities. Therefore, the designation of “asialoglycoprotein receptor” for HL does not apply for non-mammalian species. It is also to be noted that in addition to parenchymal cells (hepatocytes), Kupffer cells and endothelial cells of the liver have receptors for other sugars, such as Man and GlcNAc. Lectins identical or very similar to HLs were also found to exist in the extrahepatic tissues such as testis/sperm [5-8], kidney [9], mesangial cells [ 101 and

33.3 Molecular Characteristics of Hepatic Lectins

55 1

Table 1. Sugar specificities of HL Animals

Specificities

References

Human Pig Rabbit Rat Chicken Fish Alligator

Gal, GalNAc Gal, GalNAc Gal, GalNAc Gal, GalNAc GlcNAc Man Man/Fuc

[601 [611 [2, 621 I631 [64-661

1671 1681

human intestinal epithelial cell line [ 11, 121. Tracer studies also indicated that extrahepatic tissues can accumulate Gal- or GalNAc-containing compounds [ 13, 141.

33.3 Molecular Characteristics of Hepatic Lectins Hepatic lectins belong to a group of lectins known as C-type, so named because they require calcium for ligand binding, and it is now known that they share a structurally homologous carbohydrate-recognition domain [ 151. All mammalian hepatic lectins so far examined are composed of two types of subunit, HL-1 and HL-2, which share a considerable homology in amino acid sequence and the same polypeptide domain construct. They are composed of, from the N-terminal end: 1) 2) 3) 4)

cytosolic domain; trans-membrane domain; a stalk region composed of varying numbers of heptad; and C-terminal C-type lectin domain.

Of the two HL gene products, HL-1 is more abundant and is smaller in molecular size. In the case of rat HL, there are two kinds of HL-2 products, RHL-2 and RHL-3, which differ in molecular size due to glycosylation variance. The estimated molecular sizes of RHL-I, RHL-2, and RHL-3 are approximately 42, 54 and 62 kDa, respectively. Since the primary structure of RHL-2/3 is only longer than RHL-1 by 18 amino acid residues, the bulk of the size difference must come from the glycosylation difference, RHL-3 containing the largest amount of carbohydrates. The post-translational modification of these subunits includes, in addition to glycosylation, phosphorylation [ 16, 171 and fatty acylation [ 181. The glycosylation sites are in the extracellular domain in the stalk region, while both phosphorylation and fatty acylation (mostly palmitate and stearate) occur in the cytosolic domain. Phosphorylation occurs mainly on serine residue, but is also observed on threonine and tyrosine, while fatty acylation is on a cysteine residue. Both of these modifications appear to be dynamic processes.

552

33 Interactions of Oligosuccharides and Glycopeptides

33.4 Cellular Aspects of HL As mentioned in the introduction, HL or ASGP-R is a cell surface receptor. For a comprehensive treatise on cell biological aspects of this receptor, readers should consult relevant review articles [19-211. Here we limit our discussion to the most fundamental cell biological aspects. The distribution of HL studied using intact and digitonin-permeabilized hepatocytes showed that more than 2/3 of the total HL actually resides inside the cells. Even on the cell surface, hepatocytes typically express nearly 500,000 HL molecules, making this one of the most abundant receptor molecules. After HL binds its ligand, the complex is internalized via the coated-pit pathway, and the ligand dissociates from the receptor in a prelysosomal compartment called CURL due to a lowering of pH (<6). The receptor is then recycled back to the surface, while the ligand is transported to the lysosome where it is degraded. The actual endocytotic events are not as simple as originally portrayed. There appear to be two populations of receptor, called state 1 and state 2 receptors, which have different dissociation rates, different sensitivities to various chemicals, and operate independently of each other [21]. The fact that two types of polypeptide for HL are strictly conserved in mammalian species suggests that hetero-oligomers have some sort of distinct advantage over homo-oligomers. In fact both human and rat ASGP-Rs appear to form preferentially hetero-oligomers on the hepatocyte cell surface, although homooligomers may also exist [22, 231. When RHL-1 alone is expressed on cell surface, it does not exhibit typical high affinity binding and endocytosis of ASGPs [24], although the receptor function could still be demonstrated with higher affinity polylysine derivatives that bear large numbers of Gal residues [25].The major subunit of human HL (Hl) can bind triantennary glycopeptide efficiently, but only when expressed in excessive amounts in COS-1 cells [26]. The in vivo experiments on mice with disrupted MHL-2 gene showed much reduced expression of MHL-1 on hepatocellular surface, and what expressed was not capable of clearing typical asialoglycoprotein ligands [27, 281. These results suggest that the HL-2 gene product may help stabilize HL-1 or assist in the transport of HL-1 to the cell surface (and vice versa). The hepatic lectin is reported to be nearly absent in mammalian fetal liver but rises to adult levels during the early postpartum period. Low levels of asialoglycoprotein receptor in 18-day-old fetal liver correlated to a three-fold higher level of total asialoglycoproteins in fetal serum compared with those of adults. The serum asialoglycoproteins decreased to adult values as more hepatic lectin is generated later in development [29]. What is the physiological role of ASGP-R? The receptor is abundantly present in liver, and its function is probably related to its ability to remove efficiently Gal/ GalNAc-exposed glycoconjugates from circulation. However, at the moment there is no consensus opinion on its function [30]. Initially the receptor was postulated to function for the removal of senescent serum glycoproteins. It was found later that ASGP-R may not be involved in the normal turnover of serum glycoproteins, although the receptor may have a significant role to play when there is a large influx

33.5 Binding Spec$city

553

of Gal/GalNAc-glycoconjugates under pathological conditions. More recently, it was hypothesized that the hepatic ASGP-R, together with macrophage receptor and the particulate receptor of Kupffer cell of the same sugar specificity, is involved in the surveillance and removal of Gal/GalNAc-glycoconjugates which are injurious to self in adult mammals [211. The interaction of GallGalNAc-glycoconjugates with the corresponding cellular lectins (e.g., galectins) is thought to be involved in the differentiation and development processes, and it is argued that once the course of development is complete, Gal/GalNAc-glycoconjugates would have a harmful effect, so their level in circulation has to be kept in check. However, this hypothesis is not compatible with the fact that the specificity of avian HL is for GlcNAc (Table 1) even though chicken galectins appear to be involved in similar functions as in mammalian species. In addition, there are some intriguing phenomena associated with this receptor, which suggest that the true physiological function of ASGP-R may be more complex and multi-faceted. For instance, ASGP-R activity increases greatly during pregnancy and returns to normal post-partum [31]. Also, ASGP-R activity is downregulated when the glucose level is in the range of hypoglycemic conditions [30]. Also as mentioned in Introduction, the receptor or its close analog has been found in a number of organs besides liver, and appears to carry out function unrelated to asialoglycoprotein clearance.

33.5 Binding Specificity It was soon realized that the valency of Gal/GalNAc residues on a given ligand molecule is an important determinant for its affinity, since desialylated glycoproteins with typical biantennary N-glycans such as transferrin have much slower clearance rate than those having triantennary N-glycans such as a-1-acid glycoprotein and fetuin. Using neoglycoproteins, i.e., proteins modified with defined sugar derivatives, an amazing effect of valency on binding became apparent. A linear increase in the sugar density on a given type of neoglycoprotein (e.g., Gal-AI-BSA [ 321) produced virtually logarithmic increase in the binding efficacy (expressed as relative inhibitory power) [33]. A series of synthetic analogs representing partial structures of typical N-glycans (Figure I ) played an important role in further delineating the importance of valency as well as spacial arrangement of terminal Gal residues in determining the binding efficacy. Figure 2 demonstrates a dramatic effect of valency upon the inhibitory potency of oligosaccharides in the isolated rat hepatocyte binding assay. As summarized in Table 2, an increase in valency by one unit had as high as a 1000-fold increase in affinity up to the triantennary structure. The increase in affinity going from triantennary to tetraantennary was not nearly as dramatic. Compounds within the same valency group (e.g., divalent structures) exhibited a considerable range of affinity, suggesting that the inter-Gal distances are also important determinants. Since the intrinsic affinity of a single Gal residue in terms of KD is about 0.3 mM,

554

33 Interactions of Oligosaccharides and Glycopeptides

5W

n

II

L Galp4GlcNAcp4/

Galp4Glcp6 Galp4GlcNAcpZMan

/

\

Galp4GlcNAcpZManOL PENTA-2,4 Galp4GlcNAcp4

Galp4GlcNAcpZMana6

LACTOS-Z,6

GaIP4GlcpZMan

Gal~4GlcNAc~2Man Galp4GlcNAcp4

I

NONA I1

3

/

PENTA-Z94-0L

\

Galp4GlcNAcpZMana3

’

Galp4GlcNAcpZMana6

\

Manb4GlcNAcb4GlcNAc-R

BI-GP

Manb4GlcNAcb4GlcNAc-R

TRI-GP

Manb4GlcNAcb4GlcNAc-R

fet-TRI-GP

Galp4GIcNAcp4

GaIp4GlcNAcpZMana6

\

Galp4GlcNAepZMana3 /

/

Galp3GlcNAcp4

,Pep1 R =Am, Pep2

Figure 1. Structures of some synthetic oligosaccharides and natural glycopeptides of complex type N-glycans.

the KD of 0.2 pM for the best biantennary inhibitor, PENTA-2,4, represents close to the maximal possible affinity enhancement. This suggests that the conformation of PENTA-2,4 is complementary to the binding site configuration. Similarly, a further 100-fold increase in affinity by adding the third branch in the NONA I structure suggests that this triantennary structure is quite complementary to the arrangement of three binding sites of the receptor complex. In the preferred conformation of NONA I structure obtained from NMR studies, the three terminal

555

33.5 Binding SpeciJicity

B::.-':,": I

Figure 2. Dependency of binding affinity on valency of branched N-glycans. Curves marked Di, Tri, and Tetra refer to the results obtained with synthetic oligosaccharides PENTA-2,4, NONA I, and UNDECA in Figure 1.

66 33 0

-

Tetra I

Tri

-

I

I

I

I

I

I

Gal residues are located in a triangular arrangement in space, as shown in Figure 3 [34]. The importance of such inter-Gal distance and their relative orientation in determining the binding affinity was amply demonstrated with a triantennary glycopeptide (fet-TRI-GP) which differs from TRI-GP (Figure 1) at a single Gal linkage position. This unusual TRI-GP had KI ca. 100-fold higher than that of normal TRI-GP 1351. The strong preference for a defined configuration for the terminal Gal residues implies that a reciprocal requirement exists for the binding site arrangement of the receptor complex. Receptor subunits must be held together in a rigid configuration so as to produce a complementary arrangement of binding sites. The specific orientation of three binding sites was clearly demonstrated later by photoaffinity labeling studies (See next section). A number of synthetic di- and tri-valent glycopeptides having only the terminal sugar residue proved to be surprisingly strong inhibitors [36, 371. Structures of some of these synthetic glycopeptides are shown in Figure 4, and their inhibitory potencies are listed in Table 3 . It was shown that the structures that allow maximal inter-

Table 2. Binding affinity of oligosaccharides shown in Figure 1. Compounds

Binding of '"I-Tyr-Tri-GP

Gal-GlcNAc-Man H EPTA LACTOS-2,6 PENTA-2,6 PENTA-2,COL PENTA-2,4 BI-GP NONA I1 Tyr-TRI-GP NONA-I UNDECA TRI-GP

82 1 20.4 3.6 2.8 1.9 0.25 1.8 0.111 0.0062 0.002 0.0013 0.0043

283 13.2 2.3 1 1.75 1.33 0.168 1.13 0.08 13 0.00345 0.00185 0.000877 0.0024

4.79 6.58 7.61 1.77 7.93 9.14 8.05 13.6 15.4 15.8 16.3 15.7

556

33 Interactions of Oligosaccharides and Glycopeptides Ala(Asn)

df

,.".:....*%

P,r

8.

.% ' Figure 3. Spatial arrangement of Gal residues in triantennary N-glycan. The inter-Gal distances are between the C-4, and the distances between Ala and Gal are those of naphthyl group on amino group of Ala and dansyl group on C-6 of Gal.

A

sugar separation of ca. 30 give the highest affinity enhancement; structures with longer separation do not improve further on the enhancement and the ones with much shorter separation do not give much enhancement effect. Because as a monovalent ligand, GalNAc glycoside is ca. 40-fold better inhibitor than the corresponding Gal (or Lac) glycoside, the glycoside cluster effect is also much larger for the GalNAc-containing cluster ligands as compared with Lac-containing cluster ligands (Table 3). Triton-solubilized, affintiy-purified HL still expresses glycoside cluster effect, albeit to a much lesser extent than the HL on the hepatocyte surface. The best di-valent oligosaccharide, PENTA-2,4, had 20-fold higher affinity than the monoantennary structure, as compared to 1000-fold for the hepatocyte surface receptors, while the best triantennary oligosaccharide, NONA I, had 700-fold higher affinity over the monovalent structure, as compared to 100,000-fold enhacement for the hepatocyte system. Presumably organization of subunits in the soluble form is somewhat relaxed compared to the cell surface HLs.

YE E(ah-Glyc)3

(Trivalent)

CONH -CHpCON H(CHz)@Glyc

NACTYPCONH-

P

HCONHCHZCONH(CHZ)~OGIYC

NAcYD(G-tl h-Glyc)z

(Divalent)

Figure 4. Synthetic di- and trivalent glycopeptides using Asp or Glu for branching.

33.6 Photoujinity Lubeliny

557

Table 3. Inhibitory potency of synthetic ligands toward HL on hepatocyte surface Monovalent

150,

Lac-ah” GalNAc-ah

190 5

PM

Divalent

NAc[D(ah- lac)^] NAc[D(G-ah-GaINAc)z]

I 0.03

Trivalent

YEE(ah-Lac)3 Y EE(ah-GalNAc)i a

3 0.004

“ah” denotes “6-aminohexyl”

33.6 Photoaffinity Labeling Probing the binding requirements at the sub-structural level of galactosyl residue showed that 3- and 4-OH groups are absolutely needed, the exocyclic methylene group contributes to the binding energy, and C-6 can be substituted with a large group [38]. Based on the last finding, we developed a useful reaction scheme for the C-6 substitution of galactosyl residues as shown in Figure 5 [39]. Once an amino group is generated at the C-6 position, there are a number of reagents available that allow attachment of photoactivating and fluorescent groups. Using a triantennary glycopeptide preparation with photoactivatable group(s) attached randomly on the terminal Gal residues, we showed that the minor subunits (RHL-2/3) were photoaffinity labeled more strongly than the major subunit (RHL-1) on the isolated rat hepatocyte [40]. Similar phenomenon was also observed in the cell surface iodination of ASGP-R with lactoperoxidase [41, 421. These phenomena suggest that ASGP-R subunits are tightly packed and RHL-1 is somewhat cryptic for large molecules. More sophisticated photoaffinity labeling studies using three isomeric TRI-GP derivatives having one photoactivatable group on one of the Gal residues (Figure 6) showed exquisite specificity in the binding of TRI-GP to rat hepatocyte ASGP-R [43]. Two of the TRI-GP derivatives photo-labeled exclusively RHL-1, while the third isomer photo-labeled only RHL-2/3. The same methodology of glycopeptide modification was used for determination of conformation of triantennary glycopeptides. Instead of the photoaffinity label, dansyl group (acceptor) was attached to the terminal Gal residue and the naphthylacetyl group (donor) was attached to Ala residue of Ala-Asn(G1ycan) and energy transfer between them was determined [44, 451. The results are summarized in Figure 3.

ow-

558

33 Interactions of Oligosacchavides and Glycopeptides

HO

-* BnNH2

% o

Gal. oxidase

& 'OR

NaCNBH3

HO

HO

/

+ //

D~~sNH(CH~)~NH~

PdlC

Pyr.BH3

RO&' HO

e.g.

k

NHCH2C

I

H2

R=

2

NHCOR X = ONo>

HO

0

HO

Figure 5. Chemoenzymatic scheme for conversion of 6-OH into 6-NH2.

33.7 Subunit Organization on Rat Hepatocyte Surface From the accumulated experimental evidences, one can envisage the high-affinity ASGP-R on rat hepatocyte surfaces. One probable model is shown in Figure 7. The basic binding unit of ASGP-R is likely to be composed of two molecules of RHL-1 6'

Ala

Gal~4GlcNAcpZMnna6 6

I

\

Man~GlcNAc~4GlcNAc~Asn

Galp4GlcNAcpZMana3 / 8 Galp4GlcNAcp4

/

PCON" /NO2

ANBA- =

Figure 6. Photoaffinity labeling reagents derived from a triantennary N-glycan

33.8 Applications

1

559

1

Sugar combining Figure 7. Subunit organization of rat hepatic ASGP-receptor.

site

and one molecule of either RHL-2 or -3. Two such trimeric units associate to form a hexamer which exists as stable entity in the Triton-solubilized form. Each RHL-I of the trimer is likely to provide one Gal/GalNAc binding site, and RHL-2 or -3 provides the third site. These three sites are located at the apices of a triangle whose three sides are approximately 15, 20, and 25 A in length. Many such binding units are probably packed tightly together to form receptor patches on the cell surface, since it was shown that the binding of a high-affinity macromolecular ligand, asialo a1-acid glycoprotein, can eclipse many potential binding sites [46].Tight packing of subunits makes RHL-1 somewhat hindered, possibly due to a more extended stalk of RHL-2/3 because of their higher sugar content and probably longer sugar chains.

33.8 Applications The abundance of H L on the hepatocyte surface induced many applications for delivery and targeting. The more prominent targeting application was in the delivery of nucleotides, utilizing charge-mediated association to poly-lysine which had been conjugated to ASOR [47-501. Nucleotides [511 or antisense phosphononucleotides [ 14, 521 coupled to the synthetic trivalent derivative, YEE(ah-GalNAc)3 (Figure 4), were found to be efficiently delivered to liver. Rice and coworkers developed a technique by which N-glycans released from glycopeptides are transformed into glycosylamine which was then derivatized with L-tyrosine [53]. The use of tyrosine for the acylation of glycosylamine bestows UVdetectability as well as possibility of radio-iodination. The presence of a hydrophobic group also helps separation of oligosaccharide mixtures on RP-HPLC. The destination of rediolabeled tyrosinamidated Gal-exposed oligosaccharide thus prepared in the whole animal showed that the TRI-GP derivative is almost exclusively taken up by the liver, whereas the BI-GP derivative is less specifically localized in the liver [ 541. Human serum albumin modified with technetium (Tc-99m)-labeled galactose [55] has been successfully used for imaging of human liver as well as for diagnosis of acute liver diseases [56, 571.

560

33 Interactions of Oligosaccharides and Glycopeptides

Animal experiments in BALB/c-mice and in DBA/2-mice confirmed that the blockade of HL with D-Gal-containing compounds can inhibit metastatic spread into liver. In a clinical trial of colorectal carcinoma patients (UICC stages 1-111) with D-Gal treatment (1.5 g/kg body weight per day), the patients with stage 111 carcinoma showed some indication for an overall benefit in survival [58]. Similar D-Gal treatment of stomach adenocarcinoma patients also showed significantly reduced hepatic metastases and significantly improved overall survival for patients in the treatment group [59]. Modification of antibodies with Gal did not cause loss of antigen binding, complement-mediated cytotoxicity, or antibody-dependent cell-mediated cytotoxicity. Thus, galactose-conjugated antibodies appear to have potential for diverse applications in regional or systemic immunotherapy [ 131.

References 1. Morell, A. G.; Irvine, R. A.; Sternliev, I.; Scheinberg, I. H.; Ashwell, G. J. Biol. Chem. 1968, 243, 155-1 59. 2. Hudgin, R. L.; Pricer, W. E.; Ashwell, G.: Stockert, R. J.; Morell, A. G. J. Biol. Chem. 1974, 249, 5536-5543. 3. Halberg, D. F.; Wager, R. E.; Farrell, D. C.; Hildreth, J., IV; Quesenberry, M. S.; Loeb, J. A,; Holland, E. C.; Drickamer, K. J. Biol. Chem. 1987, 262, 9828-9836. 4. Kawasaki, T.; Ashwell, G. J. Biol. Chem. 1976, 251, 1296-1302. 5. Abdullah, M.; Kierszenbaum, A. L. J. Cell Biol. 1989, 108, 367-375. 6. Monroe, R. S.; Huber, B. E. Gene (Amsterdam) 1994, 148, 237-244. 7. Goluboff, E. T.; Mertz, J. R.; Tres, L. L.; Kierszenbaum, A. L. Mol. Reprod. Dev. 1995, 40, 460-466. 8. Mertz, J. R.; Banda, P. W.; Kierszenbaum, A. L. Mol. Reprod. Dev. 1995, 41, 374-383. 9. Mu, J. Z.; Tang, L. H.; Alpers, D. H. Amer. J. Physiol. 1993, 264, G752-762. 10. Gomez-Guerrero, C.; Duque, N.; Egido, J. J. Amer. Soc. Nephrology 1998, 9, 568-576. 11. Mu, J. Z.; Fallon, R. J.; Swanson, P. E.; Carroll, S. B.; Danaher, M.; Alpers, D. H. Biochim. Biophys. Acta 1994, 1222, 483-491. 12. Mu, J. Z.; Gordon, M.; Shao, J. S.; Alpers, D. H. Gastroenterology 1997, 113, 1501-1509. 13. Ong, G. L.; Ettenson, D.; Sharkey, R. M.; Marks, A,; Baumal, R.; Goldenberg, D. M.; Mattes, M. J. Cancer Res. 1991, 51, 1619-1626. 14. Hangeland, J. J.; Flesher, J. E.; Deamond, S. F.; Lee, Y. C.; Ts’o, P. 0. P.; Frost, J. J. Antisense & Nucleic Acid Drug Dev. 1997: 7. 141-149. 15. Drickamer, K. Prog. Nucleic Acid Rex Mol. Biol. 1996, 45, 207-232. 16. Schwartz, A. L. Biochem. J. 1984,223, 481-486. 17. Haynes, P. A,; Oka, H. A,; Weigel, P. H. J. Biol. Chem. 1994, 269, 33146-33151. 18. Zeng, F.-Y.; Weigel, P. H. J . Biol. Chem. 1996, 271, 32454-32460. 19. Ashwell, G.; Harford, J. Annu. Rev. Biochem. 1982, 51, 531-554. 20. Breitfeld, P. P.; Simmons, C. F., Jr.; Strous, G. J. A. M.; Geuze, H. J.; Schwartz, A. L. Zntl Rev. Cytol. 1985, 97, 47-95. 21. Weigel, P. H. In Endocytic Components: Identification and Characterization; Bergeron, J. J. M., Harris. J. R.. Eds.; Plenum Press: New York, NY, 1993; Vol. 19, p 125-161. 22. Oka, J. A,; Herzig, M. C. S.; Weigel, P. H. Biochem. Biophys. Ris. Comm. 1990, 170, 13081313. 23. Henis, Y. I.; Katzir, Z.; Shia, M. A.; Lodish, H. F. J. Cell Bid. 1990, I l l , 1409-1418. 24. McPhaul, M.; Berg, P. Proc. Nut1 Acad. Sci. ( U S A ) 1986, 83, 8863-8867. 25. Braiterman, L. T.; Chance, S.; Porter, W. R.; Townsend, R. R.; Lee, Y. C.; Hubbard, A. L. J. Biol. Chem. 1989,264, 1682-1688. 26. Bider, M. D.; Cescato, R.; Jeno, P.; Spiess, M. Eur. J. Biochem. 1995, 230, 207-212.

Rrj2rence.r

56 1

27. Braun, J. R.; Willnow. T. E.; Ishibashi, S.; Ashwell. G.; Herz, J. J. Bid. Clzem. 1996, 271, 21 160-21166. 28. Ishibashi, S.; Hammer, R. E.; Herz, J. J. Biol. Chern. 1994, 269, 27803-27806. 29. Bujanover, Y.; Lebenthal, E.; Petell, J. K. Biol. Neonate 1991, 60, 45-51. 30. Ashwell, G. Zentralhlatt j u r Bakteriologie Supplement 1994, 25, 26-36. 31. Collins, J. C.; Stockert, R. J.; Morell, A. G. Hepatoloyy 1984, 4, 80-83. 32. Lee, Y. C.; Lee, R. T. In The Glycoconju(gates; Horowitz, M. I., Ed.; Academic Press: New York, 1982; Vol. IV, p 57-83. 33. Lee, R. T. In Liver Diseases. Targeted diagnosis and therapy using specific. receptors and ligunds.; Wu, G. Y., Wu, C., Eds.; Marcel Dekker, Inc.: New York, 1991, p 65-86. 34. Bock, K.; Arnarp, J.; Loenngren, J. Eur. J. Biochem. 1982, 129, 171-178. 35. Townsend, R. R.; Hardy, M. R.; Wong, T. C.; Lee, Y. C. Biochemistry 1986,25, 5716-5725. 36. Lee, R. T.; Lin, P.; Lee, Y. C. Biochemistry 1984, 23, 4255--4261. 37. Lee, R. T.; Lee; Y. C. Glycoconj. J. 1987, 4, 317-328. 38. Lee, R. T.; Myers, R. W.; Lee, Y. C. Biochemistry 1982, 24, 6292-6298. 39. Lee, R. T.; Lee, Y. C. Biochemistry 1986, 25, 6835--6841. 40. Lee, R. T.; Lee, Y. C. Biochemistry 1987,26, 6320-6329. 41. Schwartz, A. L.; Marshak-Rothstein, A,; Rup, D.; Lodish, H. L. Proc. Nut1 Acud. Sci ( U S A ) 1981, 78, 3348-3352. 42. Warren, R.; Doyle, D. J. B i d . Chem. 1981, 256, 1346-1355. 43. Rice, K. G.; Weisz, 0. A.; Barthel, T.; Lee, R. T.; Lee, Y. C. J. Biol. Chem. 1990, 265, 1842918434. 44. Rice, K. G.; Wu, P.; Brand, L.; Lee, Y. C. Biochemistry 1991, 30, 6646-6655. 45. Rice, K. G.; Lee, Y. C. Adu. Enzymol. Rel. Areas 1993, 66, 41-83. 46. Hardy, M. R.; Townsend, R. R.; Parkhurst, S. M.; Lee, Y. C. Biochemistry 1985, 24, 22-28. 47. Lee-Young, A. W.; Wu, G. Y.; Wu, C. H. In Neoglycoconjugates: Preparation and Applications; Lee, Y. C., Lee, R. T., Eds.; Academic Press: San Diego, CA, 1994, p 51 1-537. 48. Wu, G. Y.; Wu, C. H. J. Bid. Chem. 1992,267, 12436-12439. 49. Wu, G. Y.; Wu, C. H. In Liver Diseases Targeted diagnosis and therapy using specijiic receptors and ligands; Wu, G. Y., Wu, C. H., Eds.; Marcel-Dekker: New York. NY, 1991, p 127-150. 50. Wu, G. Y.; Wilson, J. M.; Shalaby, F.: Grossman, M.; Shafritz, D. A,; Wu, C. H. J. B i d . Chem. 1991,266, 14338-14342. 51, Merwin, J. R.; Noell, G. S.; Thomas, W. L.; Chiou, H. C.; DeRome, M. E.; McKee, T. D.; Spitalny, G. L.; Findeis, M. A. Bioconj. Chem. 1994, 5, 612-620. 52. Hangeland, J. J.; Levis, J. T.; Lee, Y. C.; Ts’o. P. 0. P. Bioconj. J. 1995, 6, 695-701. 53. Tamura, T.; Wadhwa, M. S.; Rice, K. G. Anal. Biochem. 1994,216, 335-344. 54. Chiu, M. H.; Thomas, V. H.; Stubbs, H. J.; Rice, K. G. J. Biol. Chem. 1995,270,24024-24031. 55. Kudo, M.; Vera, D. M.; Stadalanik, R. C. In Neoglycoconjugates; Lee, Y. C., Lee, R. T., Eds.; Academic Press: San Diego, CA, 1994, p 373-403. 56. Kudo, M.; Ikekubo, K.; Todo, A.; Mimura, J.; Okabe, Y.; Kashida, H.; Hirasa, M.; Ibuki, Y.; Tomita, S.; Komori, H.; et al. Nippon Shokakihyo Gakkai Zasshi--Jap. J. Gustroent. 1992, 89, 61 6--626. 57. Torizuka, K.; Ha-Kawa, S. K.; Ikekubo, K.; Suga, Y.; Tanaka, Y.; Hino, M.; Ito, H.; Yamamoto, K.; Yonekura, Y . Kaku Zgaku-Jap. J. Nuclear Med. 1991,28, 1321-1331. 58. Isenberg, J.; Stoffel, B.; Stuetzer, H.; Oette, K.; Beuth, J. Anticancer Res. 1997, 17, 3767-3772. 59. Kosik, J.; Gil, J.; Szmigielski, S.; Beuth, J.; Pulverer, G. Anticancer Res. 1997, 17, 1411-1415. 60. Baenziger, J. U.; Maynard, Y. J. B i d . Chem. 1980, 255, 4607--4613. 61. Vera, D. R.; Woodle, E. S.; Stadalnik, R. C. J. Nuclear Med. 1989, 30, 1519--1530. 62. Lee, R. T.; Lee, Y. C. Biochem. Biophys. Res. Cornrnun. 1988, 155, 1444--1451. 63. Harford, J.; Ashwell, G. G. Proc. Natl Acad. Sci. ( U S A ) 1981, 78, 1557-1561. 64. Kawasaki, T.; Ashwell, G. J. Biol. Chern. 1977, 252, 6536-6543. 65. Lunney, J.; Ashwell, G. Proc. Nut1 Acud. Sci. ( U S A ) 1976, 73, 341-343. 66. Lee, R. T.; Rice, K. G.; Rao, N. B. N.; Ichikawa, Y.; Barthel, T.; Piskarev, V.; Lee, Y. C. Biochemistry 1989, 28, 8351-8358. 67. Fritzvold, R.; Dannervig, B. H.; Berg, T. Fish Physiol. Biochem. 1989, 6, 367-375. 68. Lee, R. T.; Yang, G. C.; Kiang, J.; Bingham, J. B.; Golgher, D.; Lee, Y. C. J. Biol. Chem. 1994,269, 19617-19625.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

34 P-Type Lectins and Lysosomal Enzyme Trafficking Patriciu G. Murron-Terada and Nancy M. Duhms

34.1 Introduction The P-type lectins derived their name from their ability to bind phosphorylated mannose residues. There are only two members of this lectin family: the 300 kDa cation-independent mannose 6-phosphate receptor (CI-MPR) and the 46 kDa cation-dependent mannose 6-phosphate receptor (CD-MPR), both of which are involved in targeting newly synthesized soluble acid hydrolases to the lysosome. Lysosomal enzymes acquire mannose 6-phosphate (Man-6-P) residues on their N linked oligosaccharides which function as a recognition marker for the MPRs. The MPRs are found ubiquitously in higher eukaryotes, and most cell types express both receptors. The CI-MPR is often referred to as the mannose 6-phosphate/ insulin-like growth factor 11 receptor to reflect the bifunctional nature of the mammalian receptor which is capable of binding Man-6-P-containing ligands as well as the polypeptide hormone, insulin-like growth factor I1 (IGF-11). However, recent studies showing that the CI-MPR binds two other ligands, retinoic acid [ 11 and the urokinase-type plasminogen activator receptor [2], demonstrate that the CI-MPR is truly a multifunctional receptor capable of recognizing a number of different ligands via distinct binding sites. In addition to soluble lysosomal enzymes, Man-6-P residues have been identified on a number of other proteins, including epidermal growth factor receptor, Herpes simplex virus glycoprotein D, thyroglobulin, transforming growth factor-p precursor, proliferin, leukemia inhibitory factor, and DNase I. The ability of these proteins to be recognized by the MPRs suggests that the receptors are involved in a number of diverse cellular events due to their carbohydrate binding activities. This Chapter will focus on the carbohydrate binding specificities of the MPRs and will summarize the information gained from the recent crystal structure of the CD-MPR which has provided the first look at the carbohydrate-recognition domain (CRD) of this lectin family. The reader is referred to reviews [3, 41 for a more complete summary of earlier work on the biosynthesis and intracellular trafficking of the MPRs and lysosomal enzymes.

564

34 P-Type Lectins und Lysosomul Enzyme Truficking

34.2 Intracellular Trafficking of the MPRs Lysosomes are acidified organelles filled with numerous acid-dependent hydrolases that are responsible for the degradation of both internalized and endogenous macromolecules [5]. Soluble lysosomal enzymes, like secretory proteins, are synthesized on membrane-bound ribosomes as precursors that contain an N-terminal signal sequence and undergo co-translational glycosylation at selected Asn residues (Figure 1). However, lysosomal enzymes become differentiated from other proteins by the acquisition of Man-6-P residues in early Golgi compartments through the concerted action of two enzymes. UDP-N-acetylg1ucosamine:lysosomalenzyme N acetylglucosamine-1-phosphotransferase transfers N-acetylglucosamine 1-phosphate to one or more mannose residues to give rise to a phosphodiester intermediate [3]. This selective phosphorylation of N-linked oligosaccharides on lysosomal enzymes is achieved by the ability of the phosphotransferase to recognize a conserved threedimensional polypeptide determinant which is enriched in one or more lysine residues [6]. Genetic defects in the phosphotransferase cause the autosomal recessive lysosomal storage disorders, I-cell disease (mucolipidosis 11) and pseudo-Hurler

Figure 1. Targeting of lysosomal enzymes to lysosomes. The movement of lysosomal enzymes and the MPRs between the various intracellular compartments and the cell surface are shown. TrafLysosomal enzymes, ficking from the cell surface and TGN occurs via clathrin-coat pits (\\\////). which are released from the MPRs in the late endosomal compartment, become packaged into lysosomes where dephosphorylation of their N-linked oligosaccharides (f ) may occur. Occupied and unoccupied Man-6-P binding sites of the CI-MPR and CD-MPR are shown. Lysosomal enzymes that are not phosphorylated ( O ) , contain a phosphodiester (0-P-W) with N-acetylglucosamine (m), or contain a phosphomonoester ( 0 - P ) are shown.

(m)

(n)

34.2 Intvucellulur Tuufickiny of the MPRs

565

polydystrophy (mucolipidosis 111) [ 71. The second enzyme, N-acetylglucosaminel-phosphodiester a-N-acetylglucosaminidase, removes the N-acetylglucosamine residue to generate the phosphomonoester [ 31. Oligosaccharides isolated from lysosoma1 enzymes are quite heterogeneous as they can be of the high mannose or hybrid-type and contain one or two phosphomannosyl residues that can be located at five different positions in the oligosaccharide chain. Once tagged with Man-6-P, lysosomal enzymes are recognized by the MPRs and are subsequently sorted from secretory proteins in the trans Golgi network (TGN). The CI-MPR and CD-MPR traverse similar intracellular trafficking pathways (Figure 1). Immunolocalization and biochemical analyses reveal that at steady state the majority of the MPRs are located in endosomes and TGN, with -5-1O'Yi of the receptor molecules on the cell surface and virtually none in the lysosome [3, 41. The MPRs interact with newly synthesized lysosomal enzymes in the Golgi and deliver these ligands to a late endosomal compartment where the low pH of this compartment causes the receptor:enzyme complex to dissociate. The receptors either recycle back to the Golgi or move to the cell surface whereas the lysosomal enzymes are packaged into lysosomes by a mechanism which is currently unknown. The CIMPR, unlike the CD-MPR, is able to bind and internalize extracellular ligands from the cell surface for delivery to the lysosome via the endocytic pathway. Measurements of the number and half-life of MPRs and the rate of ligand internalization indicate that the MPRs are reutilized and can undergo many rounds of ligand delivery. In addition, studies using antibodies to label receptors on the cell surface indicate that all MPRs in the cell are in rapid equilibrium [8]. This suggests there is only one pool of receptor and that a single MPR functions in both the endocytic and biosynthetic pathways. Although studies have shown that the routing of the MPRs occurs constitutively, the rate of internalization of the CI-MPR has been shown to be influenced by ligand occupancy, as indicated by the observation that a multivalent, but not a monovalent, ligand rapidly enhances the internalization of the receptor at the cell surface [9]. The identification of the components of the cellular machinery which specifically recognize the MPRs and mediate their intracellular movements between the TGN, plasma membrane, and endosomal compartments has been actively pursued in recent years and several proteins, including the adaptor proteins (AP-1 and AP-2) [lo], Rab9 [ l l ] , and a novel protein called TIP47 [12] have now been implicated at various sites along the pathway. In addition, both receptors utilize clathrin-coated vesicles for export from the TGN and for endocytosis from the cell surface. The signals required for endocytosis and for efficient intracellular sorting of lysosomal enzymes have been mapped to the cytoplasmic region of the MPRs (Figure 2). Endocytosis of the CI-MPR is directed by the sequence YKYSKV [13] whereas the CD-MPR contains three sequences, a phenylalanine-containing sequence (FPHLAF), a YRGV sequence, and a di-leucine motif located within the C-terminal seven residues [ 141, which can function in the internalization of the receptor from the cell surface. Efficient endosomal sorting of the CI-MPR involves a Cterminal di-leucine motif (LLHV), which includes an upstream acidic cluster [ 131. The CD-MPR utilizes a similar di-leucine motif (HLLPM), which also contains an upstream acidic cluster. However, another signal that is required for the efficient

566

34 P-Type Lectins and Lysosomal Enzyme Truflcking

CI-MPR

6P

CD-MPR

M

IGF-II

OHLLPM

Figure 2. Diagram of the CI-MPR and CD-MPR proteins. The three-dimensional structure of the extracytoplasmic domain of the CD-MPR [ S O ] is depicted using a space filling model. The 15 repeating domains of the CI-MPR (numbering starts at the amino terminus) are depicted using the same space filling model as the CD-MPR based on the assumption that each of these repeating units adopts a similar polypeptide fold as the CD-MPR. The locations of the Man-6-P (M6P) binding sites (X) and the potential N-linked glycosylation sites ( 0 )are indicated. The IGF-11 binding site in domain I 1 is indicated. The fibronectin type I1 sequence ( 0 ) is shown which functions to enhance the affinity of the IGF-I1 binding site. The serine phosphorylation (PO4) and palmitoylation ( 3 ) sites in the cytoplasmic region are indicated. The sequences in the cytoplasmic domain of the bovine MPRs, some of which contain a region of acidic residues (O), that are important for internalization and lysosomal enzyme trafficking are listed.

targeting of lysosomal enzymes by the CD-MPR is a di-aromatic signal (FW) which is modulated by palmitoylation [ 151.

34.3 Primary Structure and Biosynthesis of the MPRs 34.3.1 CI-MPR The CI-MPR is a 300 kDa type I integral membrane protein that is highly conserved among mammalian species, with the amino acid identity ranging over 77-93% be-

34.3 Prirnury Structure und Biosynthesis ofthe MPRs

567

tween bovine, human, murine, and rat receptors. However, only -60%) sequence identity is observed between mammalian and chicken [ 161 receptors, which, in part, reflects the differences observed in their ability to bind IGF-11 [4]. The bovine CIMPR is composed of four structural domains: a 44-residue amino terminal signal sequence, a 2269-residue extracytoplasmic region, a single 23-residue transmembrane region, and a 163-residue carboxy-terminal cytoplasmic domain [ 171 (Figure 2). The extracytoplasmic region has a repetitive structure resulting from the presence of 15 contiguous domains. Each domain is -147 residues in length and when compared to each other, their amino acid identity ranges over 16-380/0. A striking similarity shared among the 15 domains is the conserved positioning of cysteine residues which participate in the formation of intrachain disulfide bonds [18]. The overall similarity between the 15 domains and the distinct pattern of cysteine residues in each domain suggests the occurrence of a gene duplication event and that each domain has a similar tertiary structure. Only one of the 15 domains differs significantly from the others: the 13th domain contains a 43-residue insert which is similar to the type I1 sequence found in fibronectin and represents the only homology observed between the MPRs and other known proteins. This sequence has recently been implicated in enhancing the affinity of the mammalian CI-MPR for IGF-I1 [19]. The Cl-MPR undergoes a number of co- and post-translational modifications during its synthesis and maturation (Figure 2). The bovine CI-MPR contains 19 potential N-linked glycosylation sites, and a number of these sites are modified with high-mannose oligosaccharides which are converted predominantly to complextype structures [ 181. However, glycosylation is not required for either IGF-I1 [20] or phosphomannosyl [ 181 binding. The CI-MPR also undergoes palmitoylation [211 and phosphorylation of specific Ser residues within the cytoplasmic region (221. However, the functional significance of these modifications is not known. An important question concerning ligand binding valency centers on the oligomeric state of the CI-MPR. It is clear that the receptor does not exist as a disulfidebonded higher order complex and, when solubilized in detergent, the receptor is a monomer [9]. However, chemical cross-linking experiments [23] and gel filtration studies performed in the presence of a bound lysosomal enzyme [9] indicate a dimeric subunit composition. Thus, the receptor either:

1) cycles between a monomeric and a dimeric state, with dimerization being absolutely dependent upon the presence of a bound multivalent ligand, such as a lysosomal enzyme; or 2) exists as a “loosely” associated dimer which is further stabilized by the addition of a multivalent ligand. 34.3.2 CD-MPR The CD-MPR is a 46 kDa type I transmembrane protein that has been cloned from bovine, human, and murine sources. As observed with the CI-MPR, the mammalian CD-MPRs share significant amino acid identity (290%). The CD-MPR is

568

34 P-Type Lectins and Lysosomal Enzyme TrufJicking

similar to the CI-MPR in that it is composed of four structural domains: a 28residue amino terminal signal sequence, a 159-residue extracytoplasmic domain, a single 25-residue transmembrane region, and a 67-residue carboxy-terminal cytoplasmic domain [24] (Figure 2). Although there are no significant primary sequence similarities between the two MPRs’ signal sequences, transmembrane regions, or their cytoplasmic domains, the extracytoplasmic region of the CD-MPR shares sequence identity (14-28%) with each of the 15 repeating domains of the CI-MPR [17].In addition, the six cysteine residues in the extracytoplasmic region of the CDMPR, which are required for the folding and proper assembly of the CD-MPR during its biosynthesis [25], align with the conserved positions of the CI-MPR cysteine residues, suggesting the two MPRs have a similar tertiary structure. Similarities between the MPRs are also observed at the genomic level. Although the genes for the two MPRs map to different chromosomes (the CI-MPR has been localized to human chromosome 6 and contains 48 exons, whereas the CD-MPR has been mapped to human chromosome 12 and contains seven exons), analysis of their genomic structure has revealed that the position of the intron/exon splice junctions is conserved between several repeating domains of the CT-MPR and the extracytoplasmic region of the CD-MPR [26]. However, there is no obvious correlation between the exon boundaries and the structural or functional protein domains of the receptors. The CD-MPR also undergoes several co- and post-translational modifications (Figure 2). The CD-MPR contains five potential N-linked glycosylation sites, four of which are utilized [24] and consist of both high mannose- and complex-type oligosaccharides [27]. Numerous studies have addressed the role of N-linked carbohydrates in the functioning of the receptor. Analysis of glycosylation-deficient mutants generated by site-directed mutagenesis indicate that oligosaccharides facilitate the proper folding of the protein and aid in the stabilization of the ligand binding conformation [28]. Equilibrium binding experiments have shown that a truncated glycosylation-deficient CD-MPR retains a similar affinity for the lysosomal enzyme, 0-glucuronidase, as the fully glycosylated CD-MPR [29]. These results demonstrate that the N-linked oligosaccharides of the CD-MPR are not required for high affinity binding of ligand. However, studies by Li and Jourdian [30] have provided evidence that the type of carbohydrates on the CD-MPR can affect the affinity of the receptor for phosphomannosyl-containing ligands. Like the CI-MPR, the CD-MPR undergoes palmitoylation. Two cysteine residues in the cytoplasmic tail have been identified which are palmitoylated in a reversible manner via a thioester linkage. The addition of a fatty acid to the CD-MPR, which has been shown to be essential for the normal trafficking of the receptor, has been implicated in altering the conformation of the cytoplasmic region by serving as an attachment point to the lipid bilayer [311. The CD-MPR undergoes reversible phosphorylation at a single serine residue in its cytoplasmic region, which does not appear to influence the ability of the receptor to target lysosomal enzymes to the lysosome [32]. The predominant form of the CD-MPR in membranes is a dimer, although trimeric and tetrameric species have also been detected [33, 341. Further studies have shown that dimerization does not require the transmembrane region since a truncated CD-MPR containing only the extracytoplasmic domain is dimeric when ex-

34.4 Lysosomul Enzyme Recoynition by the MPRs

569

pressed in mammalian [35] or insect cells [29]. Thus, both MPRs can assemble into oligomeric structures.

34.4 Lysosomal Enzyme Recognition by the MPRs Soluble acid hydrolases constitute a heterogeneous population of > 40 enzymes that differ in size, oligomeric state, number of N-linked oligosaccharides, extent of phosphorylation, and the position of the Man-6-P moiety and its linkage to the penultimate mannose residue in the oligosaccharide chain. Due to the presence of both MPRs in most cell types, several experimental approaches have been undertaken to evaluate the relative contribution of each MPR to the targeting of this diverse population of enzymes to the lysosome. Analysis of cell lines that are deficient in either the CI-MPR or the CD-MPR have demonstrated that both receptors are necessary for the efficient sorting of all lysosomal enzymes as neither MPR can fully compensate for the other [36]. These results suggest that the two receptors recognize distinct subsets of lysosomal enzymes. This hypothesis was tested by twodimensional gel electrophoresis analyses of the lysosomal enzymes secreted by cells expressing either the CD-MPR or the CI-MPR. The results showed that three different pools of lysosomal enzymes can be identified: the first interacts preferentially with the CD-MPR, the second interacts preferentially with the CI-MPR, and the third interacts equally with both MPRs [36]. Furthermore, studies using either MPR-deficient mice [37] or cell lines [38] indicate that the CI-MPR is more efficient than the CD-MPR in the targeting of lysosomal enzymes. Consistent with these results are binding studies performed in vitro which demonstrate that the CI-MPR has a higher affinity for and recognizes a broader spectrum of individual lysosomal enzymes than does the CD-MPR [39, 401 (Table 1). Analyses of the binding properties of the two receptors have provided insight into the molecular basis for the observed differences in lysosomal enzyme sorting. The CI-MPR and CD-MPR have been shown to exhibit a number of similarities in their ability to recognize phosphorylated high mannose-type oligosaccharides. Both M) [41, 421. The MPRs bind Man-6-P with essentially the same affinity (7-8 x specificity for binding is determined by the 2-hydroxyl group and the 6-phosphate monoester group based on the observation that fructose I-phosphate is a competitive inhibitor whereas mannose or glucose 6-phosphate is not [42] (Table 1). Inhibition studies using chemically synthesized oligomannosides or neoglycoproteins demonstrated that the presence of the phosphomonoester Man-6-P at a terminal position is the major determinant of receptor binding. In addition, linear mannose sequences which contained a terminal Man-6-P linked a1,2 to the penultimate mannose were shown to be the most potent inhibitors [43, 441, suggesting that the MPRs bind an extended oligosaccharide structure which includes the Man-6Pal ,2Man sequence. In contrast to these similarities, the two MPRs also display a number of differences in their binding properties which include pH dependence, cation dependence,

570

34 P-Type Lectins and Lysosomul Enzyme Truficking

Table 1. Ligand binding constants for the MPRs. Compound

CI-MPR

K,

Kd

CD-MPR Kd

References

K,

(MI

Mannose 6phosphate Pentamannose phosphate (- MnCI2) Pentamannose phosphate (+ MnC12) Mannose Glucose 6-phosphate 2-Deoxy glucose 6-phosphate Fructose 1-phosphate GlcNAc-P-aMMb Diphosphorylated oligosaccharide P-Galactosidase P-Glucuronidase

Ix

P-Hexosaminidase Cathepsin D

3 x 10-9 2 x 10-9

a

10-6

8x

141,421

6x

25 x

141,421

ND"

6x

~41,421

1-8 x lo-* 1-8 x 1-8 x

1-5 x lo-' 1-5 x 1-5 x

[41, 421 [41, 421 [41,42]

I x 10-6

1x

[41, 421

1 x 10-4 2 x 10-9

> 4 x lo-' 2 10-7

2-3 10-9 2 x 10-9

3 x 10-7 3 x IO-'O or 4-5 x 10-9 ND" 7 x 10-9

[41,421 [53-55] ~561 [401

not determined N-Acetylglucosamine1'-(a-D-methylmannopyranose 6-monophosphate)

recognition of phosphodiesters, and recognition of diphosphorylated oligosaccharides. The two MPRs display optimal ligand binding at -pH 6.3 and no detectable binding below pH 5 [41-431, which is consistent with their function of releasing ligands in the acidic environment of the endosome. The CI-MPR retains phosphomannosyl binding capabilities at neutral pH which explains the ability of this receptor to bind and internalize lysosomal enzymes at the cell surface and may also provide a partial explanation for the increased efficiency of lysosomal sorting observed for this receptor. In contrast, the ligand binding of the CD-MPR is dramatically reduced at a pH >6.3. This loss of binding activity at neutral pH is the likely cause for the observed inability of the CD-MPR to bind and internalize ligands from the cell surface. The presence of cations has no influence on the binding affinity of the CI-MPR, whereas for the CD-MPR, at least in some species, the presence of cations enhances its affinity towards phosphomannosyl residues [39, 42, 451 (Table 1). Unlike the CD-MPR, the CI-MPR interacts with phosphodiesters [39, 421 (Table 1). Additional studies have shown that diphosphorylated oligo-

34.5 Structurul Determinants oj Mun-6-P Recognition

571

saccharides bind the MPRs with an affinity greater than Man-6-P, with the CIMPR exhibiting a significantly greater affinity than the CD-MPR (Table 1). These in vitro binding studies are consistent with the observed glycosylation state of the lysosomal enzymes that are bound by each MPR in vivo: the CI-MPR preferentially binds acid hydrolases enriched in oligosaccharides containing two phosphomonoesters while the CD-MPR preferentially interacts with acid hydrolases containing oligosaccharides which bear only a single phosphomonoester [ 361.

34.5 Structural Determinants of Man-6-P Recognition 34.5.1 Expression of Mutant Forms of the MPRs The generation of various truncated forms of the MPRs has greatly facilitated the localization of the CRDs within the extracytoplasmic regions of the MPRs. Equilibrium dialysis experiments demonstrate that the CD-MPR binds 1 mol Man-6-P [42, 431 and 0.5 mol diphosphorylated high-mannose oligosaccharide per monomeric subunit [42]. Since the CD-MPR exists predominantly as a dimer, the CDMPR contains two Man-6-P binding sites in its functional form in the membrane. Expression of the extracytoplasmic domain alone, which was shown to retain the ability to bind Man-6-P [33, 351, demonstrated that sequences within the transmembrane and cytoplasmic regions do not contribute to carbohydrate recognition. The observation that the extracytoplasmic domain when expressed as a monomer was functional in binding Man-6-P indicates that each polypeptide of the CD-MPR homodimer is capable of folding into an independent CRD [33]. Although the CIMPR contains 15 repeating units which are homologous to the CD-MPR, equilibrium dialysis experiments have demonstrated that this receptor binds only 2 mol Man-6-P [41, 431 or 1 mol diphosphorylated oligosaccharide per monomer [41], thus suggesting only two domains function in the binding of Man-6-P. N-terminal sequencing of proteolytic fragments of the CI-MPR purified by pentamannosyl phosphate-agarose affinity chromatography plus an analysis of truncated forms of the receptor have localized the CRDs of the CI-MPR to domains 1-3 and 7-9 [46, 471. Domain 3 and domain 9 are each predicted to contain a Man-6-P binding site since these domains share high sequence identity with the CD-MPR [ 171 and sequences within these domains are required for carbohydrate recognition [47]. Site-directed mutagenesis experiments have identified His105 and Arglll of the CD-MPR as essential components of Man-6-P recognition [47, 481. Comparison of the bovine, human, and mouse CD-MPR sequences reveal that Arglll is absolutely conserved and is also conserved in domain 3 and domain 9 of the CI-MPR. Replacement of this conserved arginine in domain 3 (Arg435) and domain 9 (Arg1334) with alanine or lysine, in a construct encoding domains 1-3 or domains 7-9, respectively, resulted in the complete loss of ligand binding as assessed by affinity chromatography on pentamannosyl phosphate-agarose columns [47].Taken together, these mutagenesis studies suggest that the CRDs of the CD-MPR and CI-

572

34 P-Type Lectins and Lysosomal Enzyme Traficking

MPR are comparable in structure and utilize similar amino acids for the recognition of phosphomannosyl residues. In contrast to the CD-MPR, which contains two identical CRDs in the homodimer, recent studies have demonstrated that the two CRDs of the CI-MPR monomer are not functionally equivalent [49]. It is possible that during evolution the CI-MPR has optimized its ligand binding ability by incorporating two different Man-6-P binding sites in order to recognize a diverse population of lysosomal enzymes. 34.5.2 Crystal Structure of the CD-MPR The first view of the CRD of the P-type lectins has been provided by the threedimensional structure of the extracytoplasmic region of the bovine CD-MPR [50]. The CD-MPR folds into a compact domain which contains one a helix located near the N-terminus followed by nine P-strands that form two P-sheets which are positioned orthogonal to one another (Figure 3). The polypeptide fold of the CD-MPR bears no resemblance to other lectins for which structural information is available, but surprisingly, contains a fold topologically similar to that of the biotin binding protein, avidin. The significance of this structural similarity is not clear, but it is intriguing to speculate that the CRD of the MPRs utilizes the same polypeptide fold as avidin in order to facilitate high-affinity ligand binding. All six cysteine residues of the extracytoplasmic domain of the CD-MPR are involved in disulfide bond formation (Cys6-Cys52, CyslO6-Cysl41, and Cysl19Cysl53). This disulfide bond pairing was previously predicted based on the sequence alignment with the 15 domains of the CI-MPR [17] and on mutational studies [51].The disulfide bonds have been shown to be important for the functional integrity of the receptor since a complete loss of binding activity occurs upon reduction of the CD-MPR with dithiothreitol [34]. The structure confirms that disulfide bonds are important for ligand binding since the linkage between Cys 106 and Cysl41 is critical for the stabilization and/or orientation of the loop involved in Man-6-P binding (see below). The CD-MPR crystallizes as a dimer, which is consistent with its dimeric state in membranes [33, 341 (Figure 3). Each monomer contains a single Man-6-P molecule

Figure 3. Ribbon diagram of the bovine CD-MPR dimer. Each monomer (light grey or dark grey) of the dimer is depicted along with the ligand, Man-6-P (ball-and-stick model). The amino terminus (N) and cdrboxyl terminus (C) of one monomer are labeled.

34.5 Structural Deterrninants of Man-6-P Recognition

573

Figure 4. Schematic representation of the potential interactions between Man-6-P (M6P) and amino acids of the CD-MPR. Potential hydrogen bond distances are shown in angstroms. The black and white balls of the amino acids represent nitrogen and oxygen atoms, respectively. The water molecule (WAT) and Mn'+ found in the binding pocket are also shown.

which interacts with a number of main chain and side chain atoms, including the two residues (His105 and Argll 1) that were previously mutated and shown to be important for Man-6-P recognition [47, 481. These residues are located in loop regions as well as in p strands from both p sheets. Unlike many other lectins [52], the recognition of carbohydrate by the CD-MPR does not involve van der Waals interactions or the stacking of aromatic residues with the sugar ring. Biochemical studies using sugar analogs of Man-6-P have demonstrated the importance of the phosphate group and the axial 2-hydroxyl group of mannose for specificity of ligand binding [42] which is reflected in the number of the interactions these moieties have with the protein (Figure 4). The phosphate group interacts with a number of amino acids including the main chain atoms of Aspl03, Asnl04, and HislO5, the imine nitrogen of HislO5, as well as the divalent cation, Mn+*. These residues are located within a loop between p strands six and seven which is stabilized by a disulfide bond (CyslO6-Cysl41). Argl l l , although not in close proximity for hydrogen bonding with the phosphate oxygens, may play a role in ligand stabilization by providing a positively charged environment for the phosphate moiety of Man-6-P. A number of residues are involved in forming hydrogen bonds with the hydroxyl groups of mannose including Arg135 (C-4 hydroxyl), Glu133 (C-3 and C-4 hydroxyls), Tyr143 (C-2 and C-3 hydroxyls), Argl11 (C-2 hydroxyl), Gln66 (C-2 and C-3 hydroxyls), and Tyr45 (C-1 hydroxyl) (Figure 4). These residues, along with Aspl03, Asnl04, and HislOS, are absolutely conserved among bovine, human, and murine CD-MPRs. Site-directed mutagenesis studies have shown that amino acid

514

34 P-Type Lectins and Lysosomal Enzyme Truficking

substitutions at position 66, 105, 111, 133, 135, or 143 exhibit a dramatic loss in the ability to bind to a pentamannosyl phosphate-agarose affinity column [48] (Olson, unpublished data), thus confirming the importance of these residues in Man-6-P binding. The crystal structure also reveals that -65% of the Man-6-P surface is buried by the receptor, with the phosphate moiety being quite concealed. This observation provides an explanation for the inability of the CD-MPR to bind oligosaccharides containing Man-6-P residues in a phosphodiester linkage with either Nacetylglucosamine or a methyl group [39, 42, 431 since the buried nature of the phosphate group leaves no room in the structure to accommodate a phosphodiester. The bovine CD-MPR exhibits optimal ligand binding in the presence of divalent cations [27]. Analysis of the three-dimensional structure reveals that the Mn+2 cation, which coordinates one of the phosphate oxygen atoms of the Man-6-P ligand, is coordinated in the Man-6-P binding pocket by Asp103 and a water molecule (Figure 4). It is predicted from the structure that the absence of a cation would result in Asp103 directly interacting with the phosphate oxygen. However, this negative charge interaction is likely to interfere with ligand binding. Thus, the presence of a positive ion may act as a “shield” between Asp103 and the phosphate oxygens, resulting in an increased affinity for Man-6-P.

34.6 Conclusions The MPRs have been studied extensively with respect to their role in targeting lysosomal enzymes to the lysosome. However, the cellular components involved in regulating the vesicular transport of the MPRs and their ligands remain to be clarified. In addition to soluble acid hydrolases, a growing number of proteins which bear no resemblance in structure or function to the acid hydrolases have been identified that display phosphomannosyl residues. The interaction of these various proteins with the MPRs points to an involvement of the MPRS in diverse cellular processes and adds to the biological complexity of these receptors. The lack of obvious sequence homology between the MPRs and other known proteins had left many questions about their structure unresolved. As the sole members of the P-type family of lectins, it was unclear whether the CI-MPR and CD-MPR contained structural features reminiscent of other animal or plant lectins. The recent crystal structure of the ligand binding domain of the CD-MPR has now addressed this question by revealing the presence of a unique structural motif for carbohydrate recognition and has provided a basis for speculation concerning how the CI-MPR may interact with its carbohydrate and non-carbohydrate ligands. Acknowledgments This work was supported by National Institutes of Health grant DK42667. This work was done during the tenure of an Established Investigatorship from the American Heart Association to N.M.D.

References

575

References I . Kang, J. X., Y. Li, and A. Leaf. 1997. Mannose-6-phosphate/insulin-like growth factor-I1 receptor is a receptor for retinoic acid. Proc Natl Acad Sci U S A. 94 (25):13671-13676. 2. Nykjaer, A., E. I. Christensen, H. Vorum, H. Hager, C. M. Petersen, H. Roigaard, H. Y. Min, F. Vilhardt, L. B. Moller, S. Kornfeld, and J. Gliemdnn. 1998. Mannose 6-phosphate/insulinlike growth factor-I1 receptor targets the urokinase receptor to lysosomes via a novel binding interaction. J Cell Biol. 141 (3):815-828. 3. von Figura, K., and A. Hasilik. 1986. Lysosomal enzymes and their receptors. Annu Reo Biochem. 55:167- 193. 4. Kornfeld, S. 1992. Structure and function of the mannose 6-phosphate/insulinlike growth factor I1 receptors. Annu Rev Biochem. 61:307--330. 5. de Duve, C. 1963. The Lysosome Concept. Churchill, London. 1-35 pp. 6. Cuozzo, J. W., K. Tao, M. Cygler, J. S. Mort, and G. G. Sahagian. 1998. Lysine-based structure responsible for selective mannose phosphorylation of cathepsin D and cathepsin L defines a common structural motif for lysosomal enzyme targeting. J Biol Chem. 273 /33):2106721076. 7. Neufeld, E. F. 1991. Lysosomal storage diseases. Annu Reo Biochem. 60:257-280. 8. Sahagian, G. G. 1984. The mannose 6-phosphate receptor: function, biosynthesis and translocation. Biol Cell. 51 (2):207-214. 9. York, S. J., L. S. Arneson, W. T. Gregory, N. M. Dahms, and S. Kornfeld. 1999. The rate of internalization of the mannose 6-phosphate/insulin-like growth factor I1 receptor is enhanced by multivalent ligand binding. J Biol Chem. 274 (2) :1164- 1 171. 10. Glickman, J. N., E. Conibear, and B. M. Pearse. 1989. Specificity of binding of clathrin adaptors to signals on the mannose-6-phosphate/insulin-likegrowth factor I1 receptor. Emho J . 8 (4) :1041- 1047. 1 1 . Riederer, M. A,, T. Soldati, A. D. Shapiro, J. Lin, and S. R. Pfeffer. 1994. Lysosome biogenesis requires Rab9 function and receptor recycling from endosomes to the trans-Golgi network. J Cell Biol. 125 (3):573-582. 12. Diaz, E., and S. R. Pfeffer. 1998. TIP47: a cargo selection device for mannose 6-phosphate receptor trafficking. Cell. 93 (3):433-443. 13. Chen. H. J., J. Yuan, and P. Lobel. 1997. Systematic mutational analysis of the cationindependent mannose 6-phosphate/insulin-like growth factor I1 receptor cytoplasmic domain. An acidic cluster containing a key aspartate is important for function in lysosomal enzyme sorting. J Biol Chem. 272 (ll/:7003-7012. 14. Denzer, K., B. Weber, A. Hille-Rehfeld, K. V. Figura, and R. Pohlmann. 1997. Identification of three internalization sequences in the cytoplasmic tail of the 46 kDa mannose 6-phosphate receptor. Biochem J . 326 ( P t 2):497-505. 15. Schweizer, A,, S. Kornfeld, and J. Rohrer. 1997. Proper sorting of the cation-dependent mannose 6-phosphate receptor in endosomes depends on a pair of aromatic amino acids in its cytoplasmic tail. Proc Natl Acad Sci U S A . 94 (26):14471-14476. 16. Zhou, M., Z. Ma, and W. S. Sly. 1995. Cloning and expression of the cDNA of chicken cationindependent mannose-6-phosphate receptor. Proc Natl Acad Sci USA. 92 (21) :9762-9766. 17. Lobel, P., N. M. Dahms, and S. Kornfeld. 1988. Cloning and sequence analysis of the cationindependent mannose 6-phosphate receptor. J Biol Chem. 263 (5):2563-2570. 18. Sahagian, G. G., and E. F. Neufeld. 1983. Biosynthesis and turnover of the mannose 6phosphate receptor in cultured Chinese hamster ovary cells. J Biol Chem. 258 (11):71217128. 19. Devi, G. R., J. C. Byrd, D. H. Slentz, and R. G. MacDonald. 1998. An insulin-like growth factor I1 (IGF-11) affinity-enhancing domain localized within extracytoplasmic repeat 13 of the IGF-II/mannose 6-phosphate receptor. Mol Endocrinol. 12 (11):1661-1612. 20. Kiess, W., L. A. Greenstein, L. Lee, C. Thomas, and S. P. Nissley. 1991. Biosynthesis of the insulin-like growth factor-I1 (IGF-II)/mannose-6-phosphate receptor in rat C6 glial cells: the role of N-linked glycosylation in binding of IGF-I1 to the receptor. Mol Endocrinol. 5 (2):281291.

516

34 P-Type Lectins and Lysosomul Enzyme Trufickiny

21. Westcott, K. R., and L. H. Rome. 1988. Cation-independent mannose 6-phosphate receptor contains covalently bound fatty acid. J Cell Biochem. 38 (1):23-33. 22. Meresse, S., T. Ludwig, R. Frank, and B. Hoflack. 1990. Phosphorylation of the cytoplasmic domain of the bovine cation-independent mannose 6-phosphate receptor. Serines 242 1 and 2492 are the targets of a casein kinase I1 associated to the Golgi-derived HA1 adaptor complex. J Biol Chem. 265 (31):18833-18842. 23. Stein, M., H. E. Meyer, A. Hasilik, and K. von Figura. 1987. 46-kDa mannose 6-phosphatespecific receptor: purification, subunit composition, chemical modification. Bid Chem HoppeSeyler. 368 (8):927-936. 24. Dahms, N. M., P. Lobel, J. Breitmeyer, J. M. Chirgwin, and S. Kornfeld. 1987. 46 kd mannose 6-phosphate receptor: cloning, expression, and homology to the 2 15 kd mannose 6-phosphate receptor. Cell. 50 (2):181-192. 25. Wendland, M., K. von Figura, and R. Pohlmann. 1991. Mutational analysis of disulfide bridges in the Mr 46,000 mannose 6-phosphate receptor. Localization and role for ligand binding. J Biol Chem. 266 (llj:7132-7136. 26. Szebenyi, G., and P. Rotwein. 1994. The mouse insulin-like growth factor II/cationindependent mannose 6-phosphate (IGF-II/MPR) receptor gene: molecular cloning and genomic organization. Genomics. 19 (1):120-129. 27. Hoflack, B., and S. Kornfeld. 1985. Purification and characterization of a cation-dependent mannose 6-phosphate receptor from murine P388D1 macrophages and bovine liver. J Bid Chem. 260 (22):12008-12014. 28. Wendland, M., A. Waheed, B. Schmidt, A. Hille, G. Nagel, K. von Figura, and R. Pohlmann. 1991. Glycosylation of the Mr 46,000 mannose 6-phosphate receptor. Effect on ligand binding, stability, and conformation. J Biol Chem. 266 (7):4598-4604. 29. Marron-Terada, P. G., K. E. Bollinger, and N. M. Dahms. 1998. Characterization of truncated and glycosylation-deficient forms of the cation-dependent mannose 6-phosphate receptor expressed in baculovirus- infected insect cells. Biochemistry. 37 (49): 17223-1 7229. 30. Li, M. M., and G. W. Jourdian. 1991. Isolation and characterization of the two glycosylation isoforms of low molecular weight mannose 6-phosphate receptor from bovine testis. Effect of carbohydrate components on ligand binding. J Biol Chem. 266 (26):17621--17630. 31. Schweizer, A,, S. Kornfeld, and J. Rohrer. 1996. Cysteine34 of the cytoplasmic tail of the cation-dependent mannose 6-phosphate receptor is reversibly palmitoylated and required for normal trafficking and lysosomal enzyme sorting. J Cell Bid. 132 (4):577-584. 32. Hemer, F., C. Korner, and T. Braulke. 1993. Phosphorylation of the human 46-kDa mannose 6-phosphate receptor in the cytoplasmic domain at serine 56. J Biol Chem. 268 (23):17108171 13. 33. Dahms, N. M., and S. Kornfeld. 1989. The cation-dependent mannose 6-phosphate receptor. Structural requirements for mannose 6-phosphate binding and oligomerization. J Biol Chem. 264 ( I 9): 1 1458-1 1467. 34. Li, M., J. J. Distler, and G. W. Jourdian. 1990. The aggregation and dissociation properties of a low molecular weight mannose 6-phosphate receptor from bovine testis. Arch Bioch Biophys. 283 (1):150-157. 35. Wendland, M., A. Hille, G. Nagel, A. Waheed, K. von Figura, and R. Pohlmann. 1989. Synthesis of a truncated Mr 46,000 mannose 6-phosphate receptor that is secreted and retains ligand binding. Biochem J . 260 (1):201-206. 36. Munier-Lehmann, H., F. Mauxion, U. Bauer, P. Lobel, and B. Hoflack. 1996. Re-expression of the mannose 6-phosphate receptors in receptor- deficient fibroblasts. Complementary function of the two mannose 6-phosphate receptors in lysosomal enzyme targeting. J Biol Chem. 271 (25): 15166-15 174. 37. Sohar, I., D. Sleat, C. Gong Liu, T. Ludwig, and P. Lobel. 1998. Mouse mutants lacking the cation-independent mannose 6-phosphate/insulin-like growth factor I1 receptor are impaired in lysosomal enzyme transport: comparison of cation-independent and cation-dependent mannose 6-phosphate receptor-deficient mice. Biochem J . 330 (Pi2) :903-908. 38. Stein, M., J. E. Zijderhand-Bleekemolen, H. Geuze, A . Hasilik, and K. von Figura. 1987. Mr 46,000 mannose 6-phosphate specific receptor: its role in targeting of lysosomal enzymes. EMBO J . 6 (9):2677-2681.

References

511

39. Hoflack, B., K. Fujimoto, and S. Kornfeld. 1987. The interaction of phosphorylated oligosaccharides and lysosomal enzymes with bovine liver cation-dependent mannose 6-phosphate receptor. J Biol Chem. 262 (1):123-129. 40. Sleat, D. E., and P. Lobel. 1997. Ligand binding specificities of the two mannose 6-phosphate receptors. J Biol Chem. 272 (2/:731-738. 41, Tong, P. Y., W. Gregory, and S. Kornfeld. 1989. Ligand interactions of the cation-independent mannose 6-phosphate receptor. The stoichiometry of mannose 6-phosphate binding. J Biol Chem. 264 (14):7962-7969. 42. Tong, P. Y., and S. Kornfeld. 1989. Ligand interactions of the cation-dependent mannose 6-phosphate receptor. Comparison with the cation-independent mannose 6-phosphate receptor. J Biol Chem. 264 (14):7970-7975. 43. Distler, J. J., J. F. Guo, G. W. Jourdian, 0. P. Srivastava, and 0. Hindsgaul. 1991. The binding specificity of high and low molecular weight phosphomannosyl receptors from bovine testes. Inhibition studies with chemically synthesized 6-0-phosphorylated oligomannosides. J Biol Chem. 266 (32):21687-21692. 44. Tomoda, H., Y. Ohsnmi, Y. Ichikawa, 0. P. Srivastava, Y. Kishimoto, and Y. C. Lee. 1991. Binding specificity of D-mannose 6-phosphate receptor of rabbit alveolar macrophages. Curb Res. 213:37-46. 45. Junghans, U., A. Waheed, and K. von Figura. 1988. The ‘cation-dependent’ mannose 6phosphate receptor binds ligands in the absence of divalent cations. FEBS Let?.237 11-2/:8184. 46. Westlund, B., N. M. Dahms, and S. Kornfeld. 1991. The bovine mannose 6-phosphate/insulinlike growth factor I1 receptor. Localization of mannose 6-phosphate binding sites to domains 1-3 and 7-11 of the extracytoplasmic region. J Biol Chem. 266 (34):23233-23239. 47. Dahms, N. M., P. A. Rose, J. D. Molkentin, Y. Zhang, and M. A. Brzycki. 1993. The bovine mannose 6-phosphate/insulin-like growth factor I1 receptor. The role of arginine residues in mannose 6-phosphate binding. J Biol Chem. 268 (8):5457-5463. 48. Wendland, M., A. Waheed, K. von Figura, and R. Pohlmann. 1991. Mr 46,000 mannose 6phosphate receptor. The role of histidine and arginine residues for binding of ligand. J Biol Chem. 266 (5):2917-2923. 49. Marron-Terada, P. G., M. A. Brzycki-Wessell, and N. M. Dahms. 1998. The two mannose 6phosphate binding sites of the insulin-like growth factor-II/mannose 6-phosphate receptor display different ligand binding properties. J Biol Chem. 273 (35):22358-22366. 50. Roberts, D. L., D. J. Weix, N. M. Dahms, and J. J. Kim. 1998. Molecular basis of lysosomal enzyme recognition: three-dimensional structure of the cation-dependent mannose 6-phosphate receptor. Cell. 93 (4):639-648. 51. Wendland, M., K. von Figura, and R. Pohlmann. 1991. Mutational analysis of disulfide bridges in the Mr 46,000 mannose 6-phosphate receptor. Localization and role for ligand binding. J Biol Chem. 266 (11):7132-7136. 52. Weis, W. I., and K. Drickamer. 1996. Structural basis of lectin-carbohydrate recognition. Annu Rez: Biochem. 65:441-473. 53. Fischer, H. D., A. Gonzalez-Noriega, and W. S. Sly. 1980. Beta-glucuronidase binding to human fibroblast membrane receptors. J Bid Chem. 255 (11):5069-5074. 54. Watanabe, H., J. H. Grubb, and W. S. Sly. 1990. The overexpressed human 46-kDa mannose 6-phosphate receptor mediates endocytosis and sorting of beta-glucuronidase. Proc Nut1 Acud Sci USA. 87 (20):8036-8040. 55. Ma, Z. M., J. H. Grubb, and W. S. Sly. 1991. Cloning, sequencing, and functional characterization of the murine 46-kDa mannose 6-phosphate receptor. J Biol Chem. 266 (16):1058910595. 56. Fischer, H. D., K. E. Creek, and W. S. Sly. 1982. Binding of phosphorylated ohgosaccharides to immobilized phosphomannosyl receptors. J Biol Chem. 257 (17):9938-9943.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

35 The Siglec Family of I-Type Lectins Paul R. Crocker and Soerge Kelm

35.1 Introduction I-type or immunoglobulin-type lectins represent a relatively recently discovered subset of animal lectins. The best-characterised I-type lectins are the siglecs (sialic acid binding Ig-like lectins) which appear to be involved in specialised forms of adhesion and cell signalling. This chapter discusses the structure and function of known siglecs, with a particular emphasis on the molecular basis for proteincarbohydrate interactions.

35.2 The Immunoglobulin Superfamily and Carbohydrate Recognition The immunoglobulin (Ig) fold consists of around 100 amino acids arranged into a sandwich of two P-sheets made up of 7-10 anti-parallel P-strands designated A-G, each strand consisting of 5-10 amino acids (Figure 1). This structure provides a platform on which large sequence variation can be displayed due to differences in surface exposed residues on P-strands and in loops between the strands. Ig domains are conventionally grouped into ‘sets’ either V, C1, C2 or I depending on sequence and structure. The Ig fold is one of the best-represented protein structures at the cell surface where it is involved in a multitude of functions relating to cell recognition and signal transduction. This functional diversity is consistent with predictions made by Williams and colleagues over a decade ago, around the time when the Ig superfamily of proteins (IgSF) was first defined [l]. It was suggested that the Ig fold evolved to mediate primitive forms of self-self recognition and later, through a process of gene duplication and divergence, to mediate more complex forms of immune cell recognition.

580

35 The Siglec Family of I-Type Lectins

m

D E B A

C F G

n

G F C C' C"

A B E D

Domain 1 V set

Domain 2 C2 set

-

Disulphide bonds: intrasheet interdomain intersheet

Figure 1. Unusual disulfides in the NH2-terminal two domains of members of the siglec family. Three cysteines exist in each of the V-set and adjacent C-2 set domains of siglec family members, resulting in a conserved pattern of inter-sheet, intra-sheet and inter-domain disulfide bonds.

For leukocytes, it has recently been estimated that around one third of all cell surface proteins contain at least one Ig domain [2]. The evolutionary success of the Ig fold is likely to be due to a combination of factors such as its resistance to extracellular proteases and the fact that Ig domains are usually encoded by exons with phase 1 splice junctions close to the domain boundary that are permissive to exon duplication and domain shuffling. Such processes would be fundamental in allowing the generation and selection of the wide variety of IgSF molecules that are now known to exist [2]. Although antibody molecules and T cell antigen receptors are able to bind carbohydrate structures [3, 41, the ability of invariant cell surface IgSF members to mediate carbohydrate recognition has only recently been recognised. In their seminal review [l], Williams and Barclay stated that for ligand recognition by IgSF proteins: 'The determinants involved are likely to be mostly protein in nature, but there is always the possibility that the chemical entities recognised are carbohydrate structures'. This Chapter discusses the recent work carried out in our laboratories and others which validates this prediction. We will focus on the siglec (sialic acid binding Ig-like lectins) family of the IgSF and review progress on the structural and functional characterization of these proteins, with a particular emphasis on the molecular interactions between siglecs and their carbohydrate ligands.

35.3 Siglecs as a Family of Sialic Acid Binding Proteins I-type, or immunoglobulin-type lectins was a term proposed in 1995 to describe members of the IgSF that could bind oligosaccharide structures [5]. To date, the only well-characterized I-type lectins are those that recognise sialic acid as their

35.4 Biology of Siglecs

581

dominant ligands. It has been reported that the IgSF protein NCAM can recognize mannosylated oligosaccharides [6] but this has not been substantiated and the binding activity of NCAM appears to be mediated by homophilic and heterophilic protein-protein dependent interactions (reviewed in [7]). It is possible that other well-characterized members of the IgSF mediate biologically important interactions with carbohydrates [5], but direct evidence for this is lacking. For the family of sialic acid binding I-type lectins, a group of interested scientists recently proposed the term ‘siglecs’ as an acceptable nomenclature, since it encompasses the essential features, namely ‘sialic acid binding’, ‘Ig-like’ and ‘lectin’ [8]. As discussed further below, siglecs share obvious evolutionary origins and appear to use a common mechanism for sialic acid recognition, further justifying the use of a generic term to describe them as a family (Figures 1 and 2). There was no proposal to change the names of existing members, which had mostly been characterized several years previously as either antigenic structures (CD22, CD33) or cell surface glycoproteins (sialoadhesin (Sn), myelin associated glycoprotein (MAG) and Schwann cell myelin protein (SMP)),although it was agreed that for classification purposes Sn should be designated as siglec-1, CD22 as siglec-2, CD33 as siglec-3, MAG as siglec-4a and SMP as siglec-4b. Any new, anonymous members would then be designated a siglec number (e.g. siglec-5 etc.) based on the chronological order of their description in the scientific literature. The discovery of siglecs came about through independent studies on Sn, a wellcharacterized sialic acid binding receptor of mouse macrophages (reviewed in [9]), and CD22, a B cell-restricted antigen (reviewed in [ 5 ] ) .Although a CD22 cDNA sequence was reported in 1990 [lo], it only became apparent several years later that this protein could function as a sialic acid binding receptor when expressed on monkey COS cells or as an isolated recombinant protein [ l l , 121. The contemporaneous molecular cloning of Sn [ 131 revealed striking sequence similarity with CD22 and raised the possibility that certain other related and well-characterized members of the Ig superfamily, namely CD33, MAG and SMP, were also sialic acid binding proteins. This was subsequently shown to be the case [ 14, 151 and thus established the existence of a multigene family of sialic acid binding proteins, each with a distinct specificity for sialylated glycans (Table 1). Recently, siglec-5 [ 161 and at least one further siglec-like sequence (CD33L) have been identified [ 171 as a result of random sequencing of cDNA libraries. The number of siglecs is likely to increase further but the final tally will only be known on completion of the human genome sequencing project.

35.4 Biology of Siglecs Apart from MAG (siglec-4a) and SMP (siglec4b), found exclusively in the nervous system, all siglecs described so far are expressed in a highly regulated fashion within the hemopoietic and immune systems (Table 1). Thus, Sn (siglec-1) is expressed by subsets of macrophages, CD22 (siglec-2) is uniquely found on B lymphocytes,

Table 1. Features of the siglec family.

NcuSAm NcuSGc, Neu5,9Ac2 ND

2,3 3 2,6

2,3 = 2,6

' Only mammalian siglecs are included. SMPjsiglec-4b) has so Car been identified in avian species only. 'Mouse CD22 binds Ncu5Gc niorc strongly thau NcuSAc whereus human CD22 binds both equally.

Neulrophils, monocytes

4

19q13.1

Siglec-S

Oligodendrocytcs Schwann cclls

s

19q13.1

MA.G (siglec-4.4)

Neutrophil cellular iiitcraction rnolecitk? Signalling?

my din-axon interactions Kcgulalion ol' ncurik gruwlh

Myclomonocytic diITerentiahn?

ND

2,3 > 2,6

Immature rnyeloid cclls Monocytes

2

19ql3.4

CD33 (siglcc-3)

Negative regulator of B cell activation Bone miirrow homing receptor

'NeuSGc = Ncu S Ac >> Neu5,9Ac2

2,6 >>> 2 , 3

B lymphocytes

7

19q13.1

CD22 (siglec-2)

Macrophage cell interaction molecule

Neu5Ac>> NeuSGc, NeuS,9Ac2

2,3 3 2,h

Macrophage subsets

17

2Opl3

Sialoadhesin (siglec-1 )

Proposed functions

Sialic acid type

Sialic acid linkage

~~

Cellular expression

Ig-like domains

Chromosomal localisation (human)

SigIec

~~~~~~~

G-

3.

e

'rl

3 m

N

00

wl

35.5 Sialic Acids in Cellulur Recognition

583

CD33 (siglec-3) is on immature myeloid cells and siglec-5 is on mature myeloid cells. These expression patterns indicate a specialized role for each protein in hemopoietic cell biology. However, clear insights into the biological functions of siglecs have only been obtained recently, due largely to the production of ‘gene knockout’ mice for both MAG and CD22. In the case of CD22, the overall phenotype was found to be consistent with previous biochemical data that CD22 functions as a negative regulator of B cell activation (reviewed in [ 181 and discussed further below). However, the importance of sialic acid recognition in the signaling functions of CD22 is currently unclear and is an important area for future research. For MAG, the phenotype (or lack of it) of mutant mice was surprising since this molecule had been widely thought of as being involved in the process of myelin formation. However, mice deficient in MAG were found to develop essentially normally [19, 201, although defects in myelin-axon interactions became apparent in older animals [2l]. These mice have also provided an experimental system to evaluate the relevance of MAG as an inhibitor of neurite outgrowth in vivo [22, 231. Clearly, a potential drawback of interpreting results from knock-out mice is the possibility of functional compensation by homologous genes. In the case of MAG, for example, it is possible that a protein like SMP, so far only found in birds [24], functionally compensates for MAG-deficiency in myelin formation. The recent description of a novel siglec (siglec-5) [16] and identification of CD33L as a siglec-like sequence [ 171 suggests that the siglecs may comprise a large gene family encoding proteins that perform both distinct as well as partially overlapping functions. Only when the full repertoire of siglec genes is known can the issue of functional redundancy be properly addressed in siglec-deficient animals.

35.5 Sialic Acids in Cellular Recognition Sialic acids are abundantly expressed on cell surfaces, secreted glycoproteins and in the extracellular matrix where they are usually found at the exposed non-reducing termini of oligosaccharide chains. There are more than 40 natural forms existing in nature and they can be attached in a variety of linkages to other sugars (including themselves) on N-glycans, 0-glycans and glycolipids, thus providing a considerable degree of molecular diversity and specificity (reviewed in [25]). In principle, therefore, sialic acids are well-suited to mediating recognition functions at the cell surface, a property that has been exploited by a range of pathogenic viruses, bacteria and protozoa (reviewed in [25]). Up until relatively recently, however, few mammalian lectins had been identified that specifically bind sialic acids. Selectins are important in mediating leukocyte-endothelial cell interactions in inflammatory and immune reactions by binding sialylated, fucosylated and sometimes sulphated oligosaccharides. However, recent structural findings indicate that selectins do not make direct, molecular contact with sialic acid [26]. Rather, sialic acid may promote specific interactions of selectins with fucose through a non-specific electrostatic effect involving a cluster of positively charged amino acids. In contrast, siglecs, as

584

35 The Siglec Family of I-Type Lectins

discussed further below, are true sialic acid binding proteins which interact specifically with the substituents of sialic acid, namely the carboxylate, the N-acyl group and the glycerol side chain (Figure 4).

35.6 Mode of Carbohydrate Recognition by Siglecs Since all of the experimentally measurable binding activity of siglecs is sialic aciddependent [14, 151, a significant component of the biological functions of these molecules is likely to involve sialic acid recognition. This is expected to be particularly the case for siglecs like Sn that seem to be predominantly involved in cell-cell interactions rather than in signaling (discussed below). Over the last few years, a priority in several laboratories has been to define the molecular basis of sialic acid recognition by siglecs. The extracellular region of each siglec is made up of an Nterminal V-set domain followed by varying numbers of C-2 set domains (Figure 2). The V-set domain and adjacent C-2 set domain display an unusual arrangement of three cysteine residues in each domain that is a unique feature of siglecs. As illustrated in Figure 1, these are predicted to result in an intra-(3-sheet disulphide bond in the V-set domain, a disulfide bond between the two domains and a canonical inter-(3-sheet disulfide in the C-2 set domain. The interdomain disulfide is likely to have a major influence on the orientation of the N-terminal two domains [27]. However, its significance for ligand recognition is unclear because the sialic acid binding site is contained exclusively within the N-terminal V-set domain [27-301. Alignment of this domain for different siglecs reveals only limited regions of homology and does not provide obvious clues as to how siglecs recognise sialic acids (Figure 3). A crystal structure of Sn complexed to 3’-sialyllactose has provided a detailed understanding of how Sn mediates carbohydrate recognition [27] (Figures 2 and 4). Most of the molecular contacts are seen to occur with sialic acid rather than the attached sugars, as depicted in Figure 4. Whereas the galactose and glucose residues are almost completely surrounded by solvent, most of the sialic acid is in close contact with amino acids from the A, G and F (3-strands, forming a binding site which seems to be very well adapted to sialic acid binding. The relatively high specificity of siglecs for particular types of sialic acid (Table 1) can be explained by the observation that the protein-carbohydrate interactions involve most of the structural features of sialic acid, i.e. the carboxylic group, the N-acyl substituent and the glycerol side chain (Figure 4B). X-ray crystallography, site directed mutagenesis, hapten inhibition assays and NMR experiments have provided complementary data on the molecular contacts involved and their contributions to the binding strength. Whereas the most complete information is available for Sn, experiments with MAG and CD22 have suggested that overall the binding sites of siglecs for sialic acids are very similar. As illustrated in Figure 4B, a highly conserved arginine residue (Arg97in Sn, see Figure 3) forms a bidentate salt bridge with the carboxylate of sialic acid. This interaction is essential for binding, as shown by site directed mutagenesis of the arginine residue

35.6 Mode of Carbohydrate Recognition by Siglecs

Sialoadhesin

CD22

MAG

Siglec-5

585

CD33

Figure 2. Diagram illustrating the domain organisation and expression pattern of mammalian siglecs. Each siglec is made up of an extraccllular region comprised of an amino-terminal V-set domain followed by varying numbers of C-2 set domains. Greatest sequence similarity between siglecs occurs in the N-terminal two domains which are bridged by a conserved disulphide bond (bar). The cytoplasmic tails vary between siglecs and; apart from Sn, contain tyrosine residues embedded within ITIM-like motifs (boxed). Also shown is a ribbon diagram of the 3D structure of the V-set domain of Sn complexed with 3’sialyllactose, as deduced by X-ray crystallography. For further information see [27].

[29-321, by chemical modification of the sialic acid [33] and in binding studies using the anomer of sialic acid [34, 351. Results of site-directed mutagenesis studies carried out with Sn [27, 291, CD22 [30] and MAG ([31] and own unpublished observations) have demonstrated that even a conservative substitution with lysine leads to -ten-fold loss in binding affinity for Sn [35] and unmeasurable binding in the case of CD22 [30] or MAG. All siglecs investigated so far require an intact glycerol side chain of sialic acid for binding [33, 34, 36-38]. For Sn and MAG it has been shown that the hydroxyls at

a

586

35 The Siglec Family of I-Type Lectins A

B strand

strand

1

10

-

- so C strand

30

20

C' strand

40

60

Sn MAG SMP

...

CD22 CD33 Siglec-5 CD33L

REP

...

... VHP

* D strand E strand

Sn MAG SMP CD22 CD33

......... ......... .........

SELYLSKCG .... .ETQG Siglec-5 .ETCG CD33L .... . E T R G

....

70

80

F strand

Dn GL TG I.

SR

RR

G strand 110

100

90

*

KGIIWIV T EHSVLDIV EHAEL ... I H L N V . .. GS.TRYSYKSECLSVHV GRDVKYSYCCNKLNLEV KW.MHYGYISSKLSVRV

*

Figure 3. Alignment of the N-terminal V-set domain of the siglec family. The amino acid sequences shown are from the murine (Sn, CD22), quail (SMP) or human (MAG, CD33, siglec-5, CD33L) homologs. For CD22, which is polymorphic, the Balb/c allele is shown. Amino acids that are identical to Sn are shown in grey boxes. Residues which when mutated led to reduced sialic aciddependent binding are in black boxes. Residues whose side groups are seen to mediate specific contacts with sialic acid in the crystal structure of the Sn-3' sialyllactose complex are marked by an asterisk. The predicted p-strands are marked with bars above the sequences.

C-8 and at C-9 form hydrogen bonds necessary for binding [34, 381. According to the crystal structure of Sn, these are to the amide and carbonyl groups of Leu'07 respectively (Figure 4B). In addition, a conserved aromatic residue at position 106 is close enough to form a van der Waals contact with the hydrophobic part of the side chain. Interestingly, the interactions with the substituent at C-5 of sialic acid appear to be different amongst the individual siglecs [ 14, 30, 34, 38, 391. A conserved aromatic group (Trp2 for Sn) is involved in hydrophobic interactions with methyl group of the N-acetyl moiety of N-acetylneuraminic acid, as demonstrated by site-directed mutagenesis [27] and NMR [35]. Despite the conservation, NMR binding studies suggest that this contact may not be used to such a large extent in MAG or CD22 [35]. In these siglecs, other amino acids are likely to contribute to the binding strength, since hydroxylation of the N-acetyl moiety (NeuSGc) or its replacement with a hydroxyl group (Kdn) can improve binding of CD22 or MAG respectively, whereas both modifications interfere with Sn binding [34, 36, 381. The crystallographic data also suggest a contribution of a hydrogen bond between the nitrogen amide of the N-acetyl moiety and the backbone carbonyl of Arglo5 (Figure 4B). For MAG binding this may be especially important, since increasing the H-donor capacity of the nitrogen by halogenation increases the affinity significantly [34]. Interactions of Sn with the adjacent galactose and glucose moieties are far more limited. The only amino acids deduced from the crystal structure to be close enough for interactions are LeuIo7with a potential van der Waals contact to the apolar face

35.6 Mode o j Carbohydrate Recognition by Siglecs

587 A

B

11061

' I

I

1

I

1.n

Hydrophobic Interaction HydrogenBond

Figure 4. Interactions at the 3' sialyllactose binding site. A. Spacefilling models of Sn and MAG showing the regions of the V-set domain of each protein that form the binding site for 3' sialyllactose. The model for Sn is based on the crystal structure whereas the model for MAG has been built using the Sn model as a template (A. May, personal communication), followed by changing the conformation of several amino acids to prevent clashes with other amino acid side chains. Shown are amino acids 1-5 (A strand), 40-44 (CjC' strands), 97-112 (C-terminal part of F strand and N-terminal part of G strand) of Sn, shaded as indicated, and the bound 3'-sialyllactose (light). The same scheme is used for the MAG model. B. Diagram depicting the interactions between amino acids in the V-set domain of Sn and sialic acid deduced from the crystal structure of the V-set domain complexed with 3' sialyllactose. The sialic acid is shown in thick black lines.

588

35 The Siglec Family of I-Type Lectins

of galactose and a hydrogen bond between the hydroxyl group of Tyr44and the 6hydroxyl of galactose. However, the latter is unlikely to be important, since mutation of Tyr44 to alanine [27] or the removal of the 6-hydroxyl [38] had no effect on binding of Sn to human RBC. No obvious contacts between the protein and the glucose residue could be deduced from the crystal structure 1271 or from hapten inhibition assays with synthetic sialosides [ 381. In conclusion, the structural studies of Sn have provided a template for sialic acid binding that is likely to be applicable to the other siglecs. However, further structural studies are needed to provide insight into the molecular basis for the sialic acid type and linkage preferences exhibited by siglecs.

35.7 Importance of Multivalent Binding Most siglecs have a clear preference for either 2,3 or 2,6 linked sialic acids (see Table I), with the exception of siglec-5 which appears to bind both linkages equally [16]. Where studied, methyl sialosides have been found to bind almost as well as sialyllactose with affinities, in the case of Sn, at around M [35]. Although the difference in affinities between simple oligosaccharides with sialic acid in either linkage is relatively small as determined by NMR [35] or hapten inhibition assays [38], the high valency of sialic acids on glycoconjugates or cell surfaces can amplify these small differences and result in a marked specificity for one or the other linkage. For Sn, Leulo7may be an important determining factor leading to a preference for 2,3-linked sialic acids. However, in view of the limited contacts between the galactose and the protein, it is also plausible that steric hindrance plays an important role in determining the linkage specificity. As with other lectins, high avidity binding of Sn depends on clustering of both receptor and ligand. Direct experimental evidence for the importance of ligand valency in binding has been obtained in experiments using complexes of biotinylated albumin and streptavidin coupled to GT1b oligosaccharides. It was shown that a 140-fold increase in oligosaccharide valency led to a 3,500-fold increase in inhibitory activity for binding of the streptavidin complex to Sn expressed on Chinese Hamster Ovary (CHO) cells [40]. The very low affinity of the Sn binding site could be important in its ability to mediate cell-cell interactions in plasma [41], since the poorly clustered glycans of plasma glycoproteins would not be expected to compete efficiently with the highly clustered sialic acids presented on cell surfaces.

35.8 Sialic Acid Recognition by the Immunoglobulin FoldEvolutionary Considerations Since carbohydrate recognition by the Ig fold has been well-characterized for antibodies, an obvious issue is whether the V-set domains in siglecs evolved from car-

35.9 Role of cis Interactions in Modulutiny Adhesion to Other Cells in trans

589

bohydrate-binding V-set domains in immunoglobulins. This seems unlikely for a number of reasons. Firstly, the mode of carbohydrate recognition by siglecs and immunoglobulins is different. Whereas immunoglobulins bind carbohydrate antigens via residues located on the hypervariable interstrand loop regions, siglecs bind sialylated glycans using residues located predominantly on the A, F and G (3-strands, in a manner more similar to that seen for other members of the IgSF involved in protein-protein interactions (reviewed in [42]). Secondly, sequence comparisons performed with the ALIGN program show that of various V-set domains analysed, the Sn V-set domain is much more similar to the Po adhesion molecule than to IgVK and IgVh sequences [43]. Po is a single V-set containing transmembrane protein that has been proposed to represent a typical primordial Ig-like domain involved in homophilic protein-protein interactions (reviewed in [ 11). Interestingly, structural comparisons of the Sn V-set domain showed that Po currently provides the closest match in main chain conformation [27].This includes the presence of an extended B-C loop which is a distinctive feature for both Sn and Po. For Po, this loop region presents a tryptophan residue that is important for adhesive function, but for Sn the loop region does not appear to be involved in ligand binding and is probably a vestige of evolution [27]. IgSF domains have been found in some prokaryotes [44, 451 and primitive multicellular organisms like Cuenorhuhditi.r elegans [46], whereas sialic acids appear to have evolved later. Apart from certain pathogenic microorganisms, they are generally present only in higher animals from echinoderms onwards (reviewed in [47]). It is therefore likely that the siglec family evolved from an ancestral ‘Po-like’ IgSF gene that mediated protein- protein interactions and then, through mutation and selection, acquired the ability to interact with sialic acids. If so, siglecs would be expected to exhibit certain key differences that reflect their adaptation from a protein recognition module to one that binds sialic acid. From the structural studies of Sn, the most striking of these relates to the exposure of the key interacting aromatic amino acids on the A and G p-strands (Trp2 and Trp’06) which would normally be buried in the hydrophobic core of the domain [27].The exposure of these residues is probably a consequence of widening of the f3 sandwich by -2A as a result of the loss of the canonical disulphide bond that is normally present between the B and F (3-strands of apposing (3 sheets, which, in the case of siglecs, has been replaced by an intra-(3 sheet disulphide between cysteines on the B and E P-strands (Figure 1). The two exposed aromatic groups, together with the highly conserved arginine on the F strand, provide the basic template for sialic acid recognition shared between all siglecs. Further mutations in other exposed residues would be expected to lead to the observed differences in fine specificity of individual siglecs.

35.9 Role of cis Interactions in Modulating Adhesion to Other Cells in trans

Given the ubiquity of sialic acids on cell surfaces and in the extracellular environment, it might be anticipated that siglecs are involved in a multitude of cellular in-

590

35 The Siglec Family of I-Type Lectins

teractions. However, whilst as purified proteins all of the siglecs so far studied can mediate sialic acid-dependent binding to cells, there are striking differences in their ability to do this when expressed on cell surfaces. This was first noted for CD33, which when expressed transiently on COS cells, required sialidase treatment to promote binding to cells carrying sialylated ligands 1151. Similar findings were made with MAG and SMP expressed in CHO cells 1481 and siglec-5 in COS cells 1161. When CD22 was expressed in COS or CHO cells, high levels of cellular binding were observed [lo, 141. However, these cells lack ST6Gal1, a sialyltransferase that creates ligands recognised by CD22 1491. When ST6GalI was coexpressed with CD22 in both CHO and COS cells, all CD22-dependent binding activity was lost but could be restored following sialidase-treatment of the transfected cells or a soluble secreted form of CD22 [50, 511. These observations raise the question of whether CD22 on B lymphocytes can mediate cell-cell interactions since these cells can express high levels of ST6GalI which is important for normal B cell activation 1491. Recently, it was shown using human peripheral blood B lymphocytes that the Sia binding activity of CD22 was undetectable, but after activation of these cells, CD22-dependent binding of a2,6 sialylated ligands could be demonstrated 1521. The unmasking of CD22 binding activity could be due to reduced cell surface a2,6linked sialic acids resulting from either increased activity of sialidase or decreased expression of ST6GalI. Interestingly, recent studies have revealed a potential role of unmasked CD22 on mature recirculating B cells as a bone marrow homing receptor that interacts with sialylated ligands on bone marrow endothelium [ 531. While cis-inhibition by endogenous sialic acids is likely to be important for such masking effects, sialic acids are a major carrier of negative charge at the cell surface and can reduce cell-cell adhesion non-specifically through a charge repulsion effect. It is likely that a significant component of the increased siglec-dependent adhesion that results from sialidase treatment is a consequence of this, as shown for MAG expressed in CHO cells [ 3 3 ] .

35.10 Sialoadhesin as a Macrophage Adhesion Molecule The observations with CD33, MAG, SMP, siglec-5 and CD22 provide a possible explanation for why Sn has evolved such a large number of Ig domains (17), namely to extend the sialic acid binding site away from the plasma membrane, thus avoiding the inhibitory environment of cis-interacting sialic acids (Figure 5). This model has been tested for Sn by introducing membrane-proximal deletions and stably expressing the truncated molecules on CHO cells. Despite high levels of expression, surface-expressed forms of Sn containing less than six Ig domains are unable to mediate sialic acid-dependent binding unless treated with sialidase [67]. Although the length of Sn extracellular region is clearly important in its ability to mediate sialic acid dependent binding to other cells, early observations with macrophages [41] and recent experiments with transfected COS cells 1541 have shown that even the binding of full-length Sn can be modulated by co-expressed sialic acids. This

35. I I Signalling Versus Adhesion Mediuted by SigErcs

59 1

i

Sn Cis interactions

-

Trans interactions

++

CD22 +I+I-

CD33

++

Figure 5. Potential cis-inhibition of siglec binding sites. Diagram illustrating how extension of the sialic acid binding site of siglecs away from the plasma membrane could be important in regulation of cell-cell interactions. See text for further discussion.

may have important implications for the adhesion-competence of Sn expressed naturally on different macrophage populations in vivo. The molecular features and binding properties of Sn suggest that it evolved as an accessory molecule to promote interactions of macrophages with host cells and extracellular matrix displaying appropriate sialylated ligands. Characterization of cells that bind to Sn has revealed high binding of granulocytes [%I, lymphoma cells [56] and breast cancer cell lines [57],with lower levels of binding to other cells tested. Macrophage-host cell interactions that could be influenced by Sn include antigen presentation to primed T cells [58] and recognition and clearance of apoptotic cells, effete erythrocytes and opsonized cells (eg in malaria and certain autoimmune diseases). Each of these involves a variety of well-characterized macrophage receptors, but the potential contribution of Sn in either increasing or decreasing specific recognition is unknown.

35.1 1 Signalling Versus Adhesion Mediated by Siglecs An emerging theme for siglecs is that they are involved in cellular signaling cascades. This involves key tyrosine residues in their cytoplasmic tails (Figure 2), which, following phosphorylation, can interact with certain SH2-containing downstream effector molecules. So far the obvious exception is Sn which lacks cytoplasmic tyrosine residues but has several serine and threonine phosphorylation sites [ 131. In contrast to the extracellular region, the cytoplasmic tail of Sn is poorly conserved between human and mouse Sn (unpublished observations), further supporting the notion that Sn evolved primarily to mediate extracellular functions rather than cytoplasmic signaling. To date, CD22 is the best-characterized siglec in the context of cell signaling.

592

35 The Siglec Family of I-Type Lectins

Extensive biochemical and genetic studies have established that tyrosine residues in three of five so-called immune receptor tyrosine based inhibitory motifs (ITIMs) within the cytoplasmic region of CD22 become phosphorylated by the src-related kinase, lyn, following B cell activation [59].This leads to recruitment and activation of the SHP- 1 tyrosine phosphatase which suppresses B cell activation through poorly understood mechanisms. In this way, CD22 is thought to be important in establishing the threshold levels for B cell activation (reviewed in [18]).When the threshold levels are altered, as occurs in CD22-deficient mice, this can lead to dysregulated B cell activation and inappropriate generation of autoantibodies [60]. Although less well-characterized, one form of MAG, L-MAG, contains tyrosine residues in the cytoplasmic tail, of which one may be phosphorylated, leading to association with phospholipase Cy and the src-like tyrosine kinase, fyn, the latter becoming activated [61, 621. CD33 (siglec-3), siglec-5 and the siglec-like CD33L share striking sequence similarity in their cytoplasmic tails, particularly involving two regions centred around tyrosine residues in ITIM-like motifs (Figure 6). The carboxyl terminal ITIM-like motif is also present in L-MAG (Figure 6) and corresponds to the previously characterized tyrosine phosphorylation site (Tyr620)[62]. Following tyrosine phosphorylation by appropriate kinases, ITIMs are generally considered to mediate negative regulatory functions via recruitment and activation of the tyrosine phosphatases SHP-1 and SHP-2 or the inositol5’ phosphatase, SHIP (reviewed in [63]). As shown in Figure 6, the ITIM-like motifs of CD33, siglec-5

Siglec-5 CD33L1 CD33 BGPl MAG

Siglec-5 CD33Ll CD33 BGPl MAG

463 360 283 456 537 512

406 331 487 587

ITIM Consensus = I/V/LXYXXL Figure 6. Potential tyrosine-based signalling motifs in siglecs. The amino acids shown correspond to the cytoplasmic regions of human siglec-5, CD33L1, CD33, BGP-I and MAG. BGP-1 is a characterized signalling IgSF molecule and is included for comparison. CD22 is not included since a poor alignment was obtained. Amino acids that are identical in three or more sequences are boxed in black; those that are similar are boxed in grey. Two regions of homology are seen between all proteins that correspond to the ITIM-like motifs (underlined). Only the first of these is needed for SHP-1 and SHP-2 binding by CD33 whereas both are required for BGP-1 The consensus sequence for an ITIM is shown.

References

593

and CD33L1 align well with those of BGP-1, a carcinoembryonic antigen (CEA) family member which has been shown to mediate growth inhibition of epithelial cells [64] and activation of neutrophils [6S].For BGP-1, recent studies on pervanadate-treated HEK293 cells have shown that both ITIM-like motifs are needed for binding to SHP-1 and SHP-2 [66]. As expected, CD33 can also bind SHP-1 and SHP-2, but in this case only the membrane-proximal ITIM-like region appears to be needed [32]. It is very likely that siglec-S will behave similarly to CD33 but considerably more work is needed to determine the significance of these cytoplasmic tail interactions in myeloid cell biology.

35.12 Conclusions The discovery and molecular characterization of siglecs has uncovered a hitherto unrecognized role for sialic acids as specific recognition molecules potentially involved in the diverse cellular activities of siglecs ranging from myelin-axon interactions to regulation of B cell activation to macrophage adhesive properties. A major challenge for the future will be to elucidate the precise biological functions of siglecs and in each case determine the role of sialic acid recognition, both in cellcell interactions and in signalling functions. Acknowledgments

The authors would like to thank all colleagues, past and present, for their contributions to the work discussed here. The Wellcome Trust, Deutsche Forschungsgemeinschaft and Human Frontiers Science Programme support the research in the authors’ laboratories. References 1. A. F. Williams, A. N. Barclay, Annu Rev Inznwmol6 (1988) 381. 2. A. N. Barclay, M. H. Brown, S. K. A. Law, A. J. McKnight, M. G. Tomlinson, P. A. van der Merwe, The Leucocyte Antigen Facts Book, Academic Press, San Diego 1997. 3. J. A. Speir, U. M. Abdel-Motal, M. Jondal, I. A. Wilson, Immunity 10 (1999) 51. 4. A. Glithero, J. Tormo, J. S. Haurum, G. Arsequell, G. Valencia, J. Edwards, S. Springer, A. Townsend, Y. L. Pao, M. Wormald, R. A. Dwek, E. Y . Jones, T. Elliott, Immunity 10 (1999)

63. 5. L. D. Powell, A. Varki, J Biol Clzem 270 (1995) 14243. 6. R. Horstkorte, M. Schachner, J. P. Magyar, T. Vorherr, €3. Schmitz, J Cell Biol I21 (1993) 1409. 7. L. C. Ronn, B. P. Hartz, E. Bock, E,YP Grrontol33 (1998) 853. 8. P. R. Crocker, E. A. Clark. M. Filbin, S. Gordon, Y. Jones, J. H. Kehrl, S. Kelm, N. Le Douarin, L. Powell, J. Roder, R. L. Schnaar, D. C. Sgroi, K. Stamenkovic, R. Schauer, M. Schachner, T. K. van den Berg, P. A. van der Merwe, S. M. Watt, A. Varki, Glycohiology 8 ( 1 998) v.

594 9. 10. 11. 12. 13.

35 The Siglec Family of I-Type Lectins

P. R. Crocker, A. Hartnell, J. Munday, D. Nath, Glycoconj J 14 (1997) 601. I. Stamenkovic, B. Seed, Nature 345 (1990) 74. L. D. Powell, D. Sgroi, E. R. Sjoberg, I. Stamenkovic, A. Varki, J Biol Chem 268 (1993) 7019. D. Sgroi, A. Varki, S. Braesch-Andersen, I. Stamenkovic, J Biol Chem 268 (1993) 7011. P. R. Crocker, S. Mucklow, V. Bouckson, A. MeWilliam, A. C. Willis, S . Gordon, G. Milon, S. Kelm, P. Bradfield, EMBO J 13 (1994) 4490. 14. S. Kelm, A. Pelz, R. Schauer, M. T. Filbin, S. Tang, M. E. de Bellard, R. L. Schnaar, J. A. Mahoney, A. Hartnell, P. Bradfield, et al., Curr Biol 4 (1994) 965. 15. S. D. Freeman, S. Kelm, E. K. Barber, P. R. Crocker, Blood 85 (1995) 2005. 16. A. L. Cornish, S. Freeman, G. Forbes, J. Ni, M. Zhang, M. Cepeda, R. Gentz, M. Augustus, K. C. Carter, P. R. Crocker, Blood 92 (1998) 2123. 17. Y. Takei, S. Sasaki, T. Fujiwara, E. Takahashi, T. Muto, Y. Nakamura, Cytogenet Cell Genet 78 (1997) 295. 18. J. G . Cyster, C. C. Goodnow, Immunity 6 (1997) 509. 19. C. Li, M. B. Tropak, R. Gerlai, S. Clapoff, W. Abramow-Newerly, B. Trapp, A. Peterson, J. Roder, Nature 369 (1 994) 747. 20. D. Montag, K. P. Giese, U. Bartsch, R. Martini, Y. Lang, H. Bluthmann, J. Karthigasan, D. A. Kirschner, E. S. Wintergerst, K. A. Nave, et al., Neuron 13 (1994) 229. 21. M. Fruttiger, D. Montag, M. Schachner, R. Martini, Eur J Neurosci 7 (1995) 511. 22. U. Bartsch, C. E. Bandtlow, L. Schnell, S. Bartsch, A. A. Spillmann, B. P. Rubin, R. Hillenbrand, D. Montag, M. E. Schwab, M. Schachner, Neuron 15 (1995) 1375. 23. M. Schafer, M. Fruttiger, D. Montag, M. Schachner, R. Martini, Neuron 16 (1996) 1107. 24. C. Dulac, M. B. Tropak, P. Cameron-Curry, J. Rossier, D. R. Marshak, J. Roder, N. M. Le Douarin, Neuron 8 (1992) 323. 25. S. Kelm, R. Schauer, Int Rev Cytoll75 (1997) 137. 26. K. K. Ng, W. I. Weis, Biochemistry 36 (1997) 979. 27. A. P. May, R. C. Robinson, M. Vinson, P. R. Crocker, E. Y. Jones, Mol Cell 1 (1998) 719. 28. D. Nath, P. A. van der Merwe, S. Kelm, P. Bradfield, P. R. Crocker, J Biol Chem 270 (1995) 26 184. 29. M. Vinson, P. A. van der Merwe, S. Kelm, A. May, E. Y. Jones, P. R. Crocker, J Biol Chem 271 (1996) 9267. 30. P. A. van der Menve, P. R. Crocker, M. Vinson, A. N. Barclay, R. Schauer, S. Kelm, J Biol Chem 271 (1996) 9273. 31. S. Tang, Y. J. Shen, M. E. DeBellard, G. Mukhopadhyay, J. L. Salzer, P. R. Crocker, M. T. Filbin, J Cell Bioll38 (1997) 1355. 32. V. C. Taylor, C. D. Buckley, M. Douglas, A. J. Cody, D. L. Simmons, S. D. Freeman, J Biol Chem 274 (1999) 11505. 33. B. E. Collins, L. J. Yang, G. Mukhopadhyay, M. T. Filbin, M. Kiso, A. Hasegawa, R. L. Schnaar, J Biol Chem 272 (1997) 1248. 34. S. Kelm, R. Brossmer, R. Isecke, H. J. Gross, K. Strenge, R. Schauer, Eur J Biochem 255 (1998) 663. 35. P. R. Crocker, M. Vinson, S. Kelm, K. Drickamer, Biochem J. 341 (1999) 355. 36. S. Kelm, R. Schauer, J. C. Manuguerra, H. J. Gross, P. R. Crocker, Glycoconj J 11 (1994) 576. 37. L. D. Powell, A. Varki, J Biol Chem 269 (1994) 10628. 38. K. Strenge, R. Schauer, N. Bovin, A. Hasegawa, H. Ishida, M. Kiso, S . Kelm, Eur J Biochem 258 (1998) 677. 39. E. A. Muchmore, S. Diaz, A. Varki, Am J Phys AnthropoZ 107 (1998) 187. 40. Y. Hashimoto, M. Suzuki, P. R. Crocker, A. Suzuki, J Biochem (Tokyo) 123 (1998) 468. 41. P. R. Crocker, S. Gordon, J Exp Med 164 (1986) 1862. 42. C. Chothia, E. Y. Jones, Annu Rev Biochem 66 (1997) 823. 43. M. Vinson, S. Mucklow, A. P. May, E. Y. Jones, S. Kelm, P. R. Crocker, TGG 9 (1997) 283. 44. A. Holmgren, M. J. Kuehn, C. I. Branden, S. J. Hultgren, Embo J 11 (1992) 1617. 45. A. Bateman, S. R. Eddy, C. Chothia, Protein Sci 5 (1996) 1939. 46. J. A. Zallen, B. A. Yi, C. I. Bargmann, Cell 92 (1998) 217. 47. R. Schauer, Adu Carhohydr Chem Biochem 40 (1982) 131. 48. M. B. Tropak, J. C. Roder, J Neurochem 68 (1997) 1753.

References 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67.

595

T. Hennet, D. Chui, J. C. Paulson, J. D. Marth, Proc Nut1 Acud Sci U S A 95 (1998) 4504. K. Hanasaki, A. Varki, L. D. Powell, J Biol Chem 270 (1995) 7533. S. Braesch-Andersen, I. Stamenkovic, J Biol Clzetn 269 (1994) 11783. N. Razi, A. Varki, Proc Nut1 Acad Sci U S A 95 (1998) 7469. L. Nitschke, H. Floyd, D. J. P. Ferguson, P. R . Crocker, J. Exp. Med. 189 (1999) 1513. Y. C. Barnes, T. P. Skelton, I. Stamenkovic, D. C. Sgroi, Blood 93 (1999) 1245. P. R. Crocker, S. Freeman, S. Gordon, S. Kelm, .I Clin Invest 95 (1995) 635. T. K. van den Berg, J. J. Breve, J. G. Damoiseaux, E. A. Dopp, S. Kelm, P. R. Crocker, C. D. Dijkstra, G. Kraal, J Exp Med 176 ( 1 992) 647. D. Nath, A. Hartnell, L. Happerfield, D. W. Miles; J. Burchell, J. Taylor-Papadimitriou, P. R. Crocker, Inimunoloqy 98 (1999) 21 3. S. Muerksoster, Mr'Rocha, P.'R. Crocker, V. Schirrmacher. V. Umdnsky, Blood 213 (1999) 4315. K. G. C. Smith, D. M. Tarlinton, G. M . Doody, M. L. Hibbs, D. T. Fearon, J E-up Med IN7 (1 998) 807. T. L. O'Keefe, G. T. Williams, F. D. Batista, M. S. Neuberger. J Exp Men 189 (1999) 1307. H. Umemori, S. Sato, T. Yagi; S. Aizawa, T. Yamamoto, Nature 367 (1994) 572. M. L. Jaramillo, D. E. Afar, G. Almazan, J. C. Bell, J Biol Cliem 269 (1994) 27240. F. Vely, E. Vivier, J Immunol 159 (1997) 2075. T. Kunath, C. Ordonez-Garcia, C. Turbide, N . Beauchemin, Oncogene I 1 (1995) 2375. S. C. Stocks, M. A. Kerr, C. Haslett, I. Dransfield, J Leukoc Biol 58 (1995) 40. M. Huber, L. Izzi, P. Grondin, C. Houde, T. Kunath, A. Veillette, N. Beauchemin, J Biol Chem 274 (1999) 335. P. R. Crocker, unpublished observations.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

36 C-Type Lectins and Collectins Russell Wallis

36.1 Summary Animal lectins have evolved to recognize the enormous diversity of carbohydrate structures found in biological systems. Prominent among these proteins are the Ctype lectins that bind to their sugar ligands in a Ca2+-dependent manner. Within the family of C-type lectins are proteins with a diverse range of funtions including endocytosis, cell adhesion and host defense. In each case, sugar binding is mediated by an evolutionarily conserved carbohydrate-recognition domain (CRD), while accessory domains both configure the CRDs in an orientation to bind to their ligands and mediate the additional activities specific for each functional group. Ligand binding by C-type lectins occurs through a common mechanism in which a tertiary complex is formed between the CRD, a calcium ion and the carbohydrate. Binding specificity is in part effected by the arrangement of the amino acid residues that form the binding site. Further specificity is mediated through additional interactions between the CRD and ligand and as a result of multivalent interactions involving multiple CRDs. The collectins are C-type lectins that contain collagen-like regions in addition to CRDs. These proteins bind to carbohydrate structures on the surface of microorganisms and are important components of the innate immune system. One member of the collectin family, serum mannose-binding protein (MBP) mediates clearance of foreign cells through activation of the complement cascade. It also functions directly as an opsinin by binding to specific receptors on phagocytic cells. The relatively broad monosaccharide-binding specificity of MBP and other collectins together with the spatial configuration of multiple CRDs is fundamental to the mechanism by which these lectins discriminate between host and foreign cells. Structural analysis of the C-terminal regions of MBPs indicate that CRDs within each subunit are ideally configured to bind to the surface of microorganisms. The N-terminal domains both mediate association of subunits to form higher order oligomers and confer the ability to fix complement. For reviews on the subject, see [ l , 21.

598

36 C-Type Lectins und Collectins

36.2 Structure and Function of C-Type Animal Lectins C-Type animal lectins are a large family of proteins that bind to their carbohydrate ligands in a Ca2+-dependent manner [l]. These lectins are characterized by the presence of a conserved carbohydrate-recognition domain (CRD) that includes 14 invariant and 18 highly conserved residues distributed over the 115-134 amino acid domain [3]. C-type lectins that recognize endogenous carbohydrate ligands include the asialoglycoprotein receptor and other endocytic receptors that regulate the turnover of glycoproteins in the serum and selectins that mediate the initial adhesion steps during extravasation of leukocytes. Lectins that bind to exogenous sugars include the mannose-binding proteins (MBPs) and endocytic receptors such as the macrophage mannose receptor that mediate the innate immune response against invading microorganisms by interacting with cell surface sugar structures. Based on the primary structures of their CRDs, C-type lectins can be divided into several evolutionary groups (Figure 1) [3]. Proteins within each group have a similar structural organization, suggesting a common evolutionary origin. Accessory domains include membrane anchors, modular domains and collagen-like regions.

GROUP I

GROUP 111 GROUP VII

kT

GROUP II GROUP IV (GROUP V)

GROUP VI

0

COL

GAG

N

SOLUBLE

N

C

C

MEMBRANE-ASSOCIATED

N

Figure 1. Domain organization of C-type lectins. Groups I1 and V are both type I1 transmembrane proteins. Groups IV and VI comprise the selectins and the macrophage mannose receptor family that are both type I transmembrane proteins. Group I includes proteoglycans that form part of the extracellular matrix, and groups 111 and VII comprise the collectin family and proteins consisting of isolated CRDs, respectively.In each case, CRDs are represented by circles.Other domains include: CR, complement-regulatory repeats; EGF, epidermal growth factor-like domains; COL, collagen-like domains; GAG, glycosaminoglycan attachment sites; and HA, hyaluronic acid-binding domains.

36.2 Structure and Function of C-Type Animal Lectins

599

These domains enable the proteins to carry out their diverse biological functions such as endocytosis, cell adhesion and complement fixation. 36.2.1 The Carbohydrate-Recognition Domain

Crystal structures of the CRDs of three collectins: human MBP, rat serum MBP, both free and complexed with a high mannose oligosaccharide and rat liver MBP, in complex with a series of monosaccharides, together with the structure of the CRD of E-selectin have provided insight both into the structural organization of CRDs and the mechanisms by which these domains interact with sugar ligands [481. The CRD fold is shown in Figure 2. The major secondary structural elements consist of 5 P-strands and 2 a-helices arranged to form a globular domain. The loop-out topology, in which the N- and C-termini are close together, provides an explanation for the finding that C-type CRDs can be located at the N- or C-terminal end of a polypeptide chain. Four invariant cysteine residues form two disulfide bonds that link the first a-helix to the last P-strand and the third P-strand to the loop following the fourth 0-strand. The remaining conserved residues that define

1 / '

ASP 206

Figure 2. Structure of the CRD from rat serum MBP. (left) Calcium ions are represented by spheres. Disulfide bonds are shown in white. u-Helices, P-strands and loops are marked as u, fl and L, respectively. (right) Detailed view of the binding bite around Ca2+ site 2 in complex with a mannose-containing ligand. White, grey and black spheres represent carbon, nitrogen and oxygen atoms. The larger grey sphere represents the calcium ion. Coordination bonds are shown by long dashed lines and hydrogen bonds as short dashed lines. Ring positions of the mannose carbon atoms are numbered. Reproduced from [ 5 ] .

600

36 C-Type Lectins and Collectins

the C-type CRD are located within turns or form part of the hydrophobic core or the binding sites for two calcium ions. Both Ca2' binding sites are generally well conserved in C-type lectins, although site 1 is absent in the selectins.

36.2.2 Ligand Binding Crystal structures of MBPs in complex with sugars reveal that the binding sites for carbohydrate ligands are localized around Ca2+ site 2 (Figure 2) 15, 81. Hydroxyl groups corresponding to the 3- and 4-OH of mannose serve as coordination ligands for the Ca2+.Additional coordination ligands are provided by asparagine and glutamic acid residues that also form hydrogen bonds with the 3- and 4-OH groups of the sugar. Relatively few additional interactions contribute to the overall binding energy. The limited nature of the contacts between CRD and ligand provides an explanation for the broad range of monosaccharide sugars recognized by MBP. Thus, sugars such as N-acetylglucosamine and fucose, that contain equatorial hydroxyl groups corresponding to the 3- and 4-OH of mannose, also serve as good ligands [91. Comparison of the sequences of the CRDs of lectins that bind mannose derivatives to those that bind galactose indicates that the former group contain the sequence Glu-Pro-Asn (at the position corresponding to residues 185-187 in MBP) while the latter group contain the sequence Gln-Pro-Asp. Substitution of the galactose-binding residues into the CRD of MBP results in a protein with a marked preference for galactose [lo]. The crystal structure of a mutant MBP CRD with galactose-binding activity reveals that fundamental aspects of binding including the coordination geometry around the Ca2+ site are conserved [ I 11. Thus, galactose binding is introduced simply as a consequence of exchanging the positions of the amide and carboxylate groups (Glu + Gln and Asn 4 Asp) in two of the four residues that coordinate the Ca2+ and form hydrogen bonds with the ligand. Additional selectivity is achieved through relatively subtle changes around the binding site resulting in favorable stacking interactions with galactose and steric exclusion of mannose [ 121. Typically, isolated CRDs interact weakly with monosaccharides with dissociation constants in the order of millimolar 121. However, intact lectins generally bind much more tightly to their physiological ligands and show a much greater specificity for particular oligosaccharides. There appear to be several mechanisms by which these binding properties are achieved. For example, chicken hepatic lectin binds to N acetylglucosamine much more tightly than it binds to mannose or fucose [13]. Sitedirected mutagenesis studies indicate that the enhanced affinity is due to additional interactions between the CRD and the monosaccharide ligand [ 141. Ligand binding by the selectins illustrates another mechanism by which binding specificity can be achieved. In this case, binding to 3'-sialyl-Lewis" and related sugars occurs through additional interactions between the CRD and multiple saccharide moieties of the ligand [ 15-17]. A third mechanism by which both binding specificity and increased binding affinity is achieved by C-type lectins is exemplified by the collectins, the macrophage

36.3 Mannosr-Binding Protein and Collectins

60 1

mannose receptor and other endocytic receptors [ 11. In these lectins, tight binding occurs as a result of clustering of CRDs. Multiple CRDs may be present either on a single polypeptide chain as in the macrophage mannose receptor or on separate polypeptides as in the collectins. In each case, the spatial arrangement of multiple binding sites is configured in a specific geometry to enable recognition of the physiological ligands. Thus, the CRDs of the asialoglycoprotein receptor bind distinct residues of tri-antennary oligosaccharides [ 181, while the CRDs of the macrophage mannose receptor are oriented to interact with sugar structures on the surfaces of microorganisms [ 191. The remainder of this review will focus on the collectins, which like the macrophage mannose receptor, bind to carbohydrate ligands on bacterial, fungal and parasitic cell walls and constitute an integral part of the mammalian innate immune response.

36.3 Mannose-Binding Protein and Collectins Collectins are a family of C-type lectins that mediate host defense against pathogenic micro-organisms. Each collectin polypeptide is characterized by an Nterminal collagen-like region and a C-terminal C-type CRD [ 11. Six members of the collectin family have been identified in vertebrates. Serum MBP, collectin-43 (CL-43) and conglutinin are plasma proteins, liver MBP is found exclusively in the liver and pulmonary surfactant proteins-A and -D (SP-A, SP-D) are associated with the surfactant that lines alveoli in the lung. Collectins can be divided into two sub-groups based on the length of their collagen-like domains. Serum and liver MBPs and SP-A have shorter collagenous regions comprising 18-24 Gly-Xaa-Yaa repeats, while conglutinin, CL-43 and SP-D have up to 59 repeats [ 11. All collectins consist of a basic structural unit comprising a homotrimer of polypeptide chains. Liver MBP and CL-43 each consist of a single trimeric subunit, while other members of each sub-group assemble into higher order oligomers consisting of multiple subunits. Serum MBP and SP-A adopt bouquetlike structures containing 2-6 trimeric subunits and conglutinin and SP-D are cruciform with four associated subunits (Figure 3) [ 11. 36.3.1 Domain Organization

Collectin polypeptides assemble into four distinct domains, an N-terminal domain containing one or more cysteine residues, a collagen-like region, an a-helical coiled coil neck region and a C-terminal CRD (Figure 3) [20].The N-terminal domains of serum and liver MBPs, CL-43, SP-D and conglutinin each contain two cysteines separated by 4-5 amino acid residues [21]. Analysis of the disulfide bonds in MBPs indicates that polypeptides within each subunit are linked by disulfide bonds between the conserved cysteines arranged in an asymmetrical configuration [22, 231. A

602

36 C-Type Lectins and Collectins BOUQUET

CRUCIFORM

Figure 3. Structural organization of collectins. (left) Collectins are assembled from trimeric subunits. (right) Bouquet and cruciform conformations.

similar bonding arrangement occurs in CL-43 [24].In rat serum MBP, an additional cysteine residue forms disulfide bonds that link polypeptides in separate subunits to form the large covalent bouquet-like structures [23]. Despite the similarity in the arrangement of cysteine residues, the disulfide bonding pattern may be different in other members of the collectin family. For example, SP-D is secreted predominantly as tetramers of subunits assembled from covalent trimers of polypeptide chains [25]. On substitution of both N-terminal cysteine residues in SP-D to serine residues, the secreted protein consists of single trimeric subunits. This finding implies that in the native protein some disulfide bonds link polypeptides within separate subunits to form the larger oligomeric forms [25]. The disulfide bonding pattern in SP-A is different to that of other collectins. The primary structure of SP-A predicted from the sequence of the gene contains only a single cysteine residue within the N-terminal domain [26]. However, alternate Nterminal processing of SP-A polypeptides results in a tripeptide (Ile-Lys-Cys) extension to certain chains [27].Native SP-A appears to contain a mixture of isoforms in which the additional cysteine residues form disulfide bonds linking polypeptides during assembly of the protein. The collagen-like domains of collectins resemble vertebrate collagens with respect to their sequence, post-translational modifications and structures [20]. 4Hydroxyproline residues and glucosylgalactosyl-5-hydroxylysineresidues, modifications characteristic of vertebrate collagens, have been identified within the consensus sequences Pro-Gly-Xaa and Lys-Gly-Xaa in MBP [20, 281. Biophysical analysis indicate that collectins are highly asymmetrical in solution, consistent with the collagenous regions forming highly extended structures [22, 231. Measurements of the length of the collagenous domains from images of proteins observed by rotary shadowing electron microscopy are consistent with estimates based on the structure of synthetic collagen peptides [29].

36.3 Munnose-Binding Protein and Collectins

603

In SP-A and MBP, the collagenous domains contain an interruption in the collagen consensus sequence [20, 261. This region is thought to introduce a kink or region of flexibility into the structure that enables the stems to angle away from the core in the bouquet-like conformations. There is also a break in the collagen consensus sequence at the start of the collagenous domain in conglutinin [30]. This region contains a cysteine residue that may stabilize the oligomers through formation of disulfide bonds. The junction between the collagen-like domain and the neck region may represent another region of flexibility in the collectins. Polypeptides within the collagen triple helix are staggered while the a-helices that form the coiled coil neck region are aligned, indicating that the junction between these domains cannot have simple rotational symmetry [22]. Although no direct evidence of flexibility has been demonstrated in collectins, the corresponding junction in class A macrophage scavenger receptors is highly flexible [31]. This region may be important for ligand recognition by the collectins, enabling some degree of adjustment to the alignment of each trimeric cluster of CRDs within a subunit. As in vertebrate collagens, assembly of each collectin subunit occurs in a C- to Nterminal direction [22]. Peptides encompassing the CRDs and neck regions assemble into trimers in the absence of the N-terminal domains [32, 331. Assembly and association of these domains in collectins acts as the nucleation site for formation of the collagen triple helix that in turn enables formation of the disulfide bonds at the N-terminal end of the molecule. 36.3.2 MBPs as Prototype Collectins Serum MBP is the most well characterized member of the collectin family both in terms of its structural organization and its role within the innate immune system. MBP mediates defense against invading pathogenic micro-organisms by activation of the complement pathway and by direct opsinization of foreign particles [34-361. Serum MBP is synthesized in the liver and secreted into the serum. Analysis of the upstream regions of the human gene has revealed the presence of several control elements. Sequences resembling both heat shock and glucocorticoid response elements may regulate MBP gene expression [37]. The human MBP gene contains an upstream acute-phase regulatory element that results in elevated expression leading to increased MBP production in response to trauma [37, 381. Sequence polymorphisms in the promoter region of the MBP gene also influence the levels of MBP in the serum [39]. 36.3.3 Ligand Binding by Serum MBP Serum MBP has been shown to interact with a wide variety of bacteria including strains of Escherichia coli, SaImonella montevideo, Listeria monocytogenes, Streptococci, Neisseria meningitidis, Neisseria cineru and N. subflava [40]. It can also interact with yeasts including Cundida albicans and Cryptococcus neoformans, viruses

604

36 C-Type Lectins and Collectins

such as HIV-1, HIV-2 and strains of influenza A and parasitic protozoa including Leishnzuniu [40]. Biophysical studies combined with structural information have provided insight into the mechanisms by which serum MBP can bind to such a wide variety of sugar structures on the surfaces of micro-organisms and still discriminate between exogenous and endogenous ligands. The CRDs of rat serum MBP interact with sugars such as mannose, fucose and N-acetylglucosamine [9]. These structures rarely occur at the terminal positions of carbohydrates on mammalian glycoproteins and glycolipids. However, they are found in high density arrays on the surfaces of many bacterial, fungal and parasitic cells. Thus, target selectivity is in part provided by the relative scarcity of ligands on host compared to foreign cells. However, the valency of ligand interactions is also critical for determining the binding affinity and hence the biological significance of the interaction. Multiple CRD-ligand contacts formed as a result of clustering of CRDs enable the formation of tight complexes that lead to complement activation [2]. Crystal structures of fragments of rat serum MBP and human MBP encompassing the CRDs and part of the neck region reveal the spatial arrangement of binding sites within a single trimeric subunit (Figure 4) [7, 331. In these structures, the neck region consists of a parallel a-helical coiled coil. A hydrophobic interface between the upper part of the polypeptide forming the neck and the CRD in the adjacent polypeptide maintains the CRDs in a fixed o@entationzelative to the neck such that the sugar-binding sites are separated by 53 A and 45 A in the rat and human proteins, respectively. The binding sites are too far apart for multiple CRDs to interact with a single mammalian high mannose oligosaccharide. However, these sites appear to be ideally configured for multivalent interactions with the arrays of sugars found on the surfaces of bacteria and other micro-organisms [33]. Thus, the number and spatial arrangement of CRDs in MBP and other collectins relative to the organization of the carbohydrate ligands is a major factor in determining the affinity of binding and in providing the ability to discriminate between host and foreign cells. 36.3.4 MBP and Innate Immunity

Serum MBP, like the first component of the classical complement pathway Clq, functions to prevent the invasion of pathogenic microorganisms by fixing complement [34].However, in contrast to Clq, serum MBP binds directly to carbohydrates on the surface of pathogens through interactions involving its CRDs and activates complement in an antibody-independent manner. Since antibodies targeted at a specific pathogens are not required for complement activation) this pathway may be particularly important in the first stages of infection both by neutralizing pathogens directly and by stimulating the host adaptive immune response. During complement fixation, serum MBP interacts with two serine proteases MASP-1 and MASP-2 (MBP-associated serine protease-1 and -2) that are homologs of C l r and C l s of the classical pathway [41-431. On activation, these proteases cleave C-4 and C-2 generating products that assemble into C4b2a complexes with C-3 convertase activity [43]. Subsequent reactions in the complement cascade lead

36.3 Munnose-Binding Protein und Collectins

605

Figure 4. Modelled structure of a trimeric fragment of rat serum MBP in complex with carbohydrate ligands. The model has been created using the coordinates from the crystal structure of a trimeric fragment of rat serum MBP comprising part of the neck region and CRDs and the structure of the isolated CRD in complex with a mannose-containing oligosaccharide. Calcium ions are shown in grey. A third Ca2+ thought to be an artifact of the crystallization conditions is shown in white. Reproduced from [33].

to direct neutralization of pathogens by formation of a host-mediated lytic complex and also generate cleavage products that stimulate the host adaptive immune system [44]. Serum MBPs are heterogeneous in solution, consisting predominantly of multiple covalent subunits. For example, human MBP consists of oligomers ranging from dimers to hexamers of trimeric subunits [29, 451, while rat serum MBP comprises mainly dimers, trimers and tetramers of subunits [46]. Each oligomeric form is able to interact with downstream components of the complement cascade leading to complement fixation, although the larger oligomeric forms are most efficient [23]. Oligomers of rat MBP do not self-associate to form higher order structures, indicating that the total complement-fixation activity is a function of the molecular activities of oligomers and their abundance within the serum. The MASP binding site on MBP is probably located within the collagenous domain. In vitro studies indicate that human MBP can interact and activate Clr2s2

606

36 C-Type Lectins and Collectins

complexes derived from the C-1 complex of the classical complement pathway [29]. Since the collagenous domain is the only structurally related region in Clq and MBP, it seems likely that this domain contains the protease-binding sites in each case. Studies of rat serum and liver MBPs indicate that both proteins are able to fix complement but the serum protein is more efficient [23]. High level complementfixation activity is associated with the N-terminal part of the collagenous domain of serum MBP and the cysteine-rich domain. These regions form a core that links the stems of the bouquet-like structures in images of human MBP [29]. The presence of these domains also correlates with the ability of serum MBP to form high molecular weight oligomeric forms. Thus, residues within the first part of the collagenous domain may be directly involved in both oligomer formation and MASP binding [23]. Evidence for the importance of serum MBP in the immune system has come from the identification of a deficiency syndrome in humans [47,48].This common genetic defect is characterized by an increased susceptibility to infections particularly during the first few years of life, before the adaptive immune system is fully established. Low titers of MBP are isolated from patients with this disorder [45]. Furthermore, the proteins consist predominantly of low molecular weight covalent structures that are inefficient at fixing complement, suggesting that both assembly and complement fixation are defective [40]. Three independent mutations within the gene encoding human MBP are associated with MBP deficiency [48-501. Each mutation leads to a single amino acid substitution within the first part of the collagen-like domain: Gly34+ Asp, Gly37+ Glu and Arg32+ Cys (Figure 5). In two of these substitutions, the glycine residue of the Gly-Xaa-Yaa collagen repeat is replaced by an acidic residue. These changes are likely to cause significant local disruption of the collagen structure as a

PHENOTYPE

LOW SERUM

POSSIBLE MECHANISM

1) REDUCED RATE OF SECRETION 2) INCREASED TURNOVER

1) LOWER PROPORTION OF HIGHER ORDER INABILITY TO FIX OLlGOMERS COMPLEMENT

2 ) DISRUPTION OF MASP BINDING SITE

Figure 5. MBP deficiency in humans.

36.3 Mannose-Binding Protein and Collectins

607

result of incorporation of the relatively bulky acidic side chains. This disruption in turn, probably prevents correct assembly of MBP into higher oligomers, resulting in the structurally aberrant protein isolated from patients with MBP deficiency. The structural and functional consequences of the Arg32 Cys substitution are less clear. This mutation results in both defective assembly and complement fixation by MBP. Mutagenesis studies indicate that the presence of the cysteine residue is responsible for both the altered oligomeric composition and the decreased complement fixation activity [51].One possibility is that the cysteine residue forms disulfide bonds or is modified during synthesis, preventing correct assembly of the oligomeric forms. The decreased ability to fix complement in protein isolated from patients with MBP deficiency may be a direct consequence of disruption of the binding site for the MASP proteases (Figure 4). All three substitutions occur within the N-terminal half of the collagen-like domain of MBP, regions identified as being critical for complement fixation [23]. Alternatively, low complement-fixing activity may result from altered oligomeric composition consisting of a decreased proportion of high molecular weight oligomers. The cause of the decreased titres of MBP in the serum is also unclear. The mutations may disrupt MBP secretion or lead to increased turnover from the serum. Serum MBP can function directly as an opsonin by binding to specific receptors on the surface of phagocytic cells [35].Several candidate receptors have been identified including a 60 kDa protein that is homologous to calreticulin [52]and a 126 kDa protein found on monocytes and macrophages [53].Although these and other proteins have been shown to bind to MBP and other members of the collectin family, the physiological significance of these interactions with respect to opsinization remain to be established. ---f

36.3.5 Liver MBP While humans appear to produce only serum MBP, genes encoding two distinct mannose-binding proteins have been cloned and sequenced in rats, mice and rhesus monkeys [20, 54, 551. The encoded proteins, serum and liver MBP, are both synthesized in the liver, have a similar domain organization and share a high degree of sequence identity but adopt different higher order structures [20, 561. Liver MBP consists of a single trimeric subunit [22] whereas serum MBP assembles into higher oligomers [29, 461. While the role of serum MBP in innate immunity is well established, the function of liver MBP is not known. Isolated CRDs from rat liver MBP have similar binding properties for monosaccharide ligands to those of serum MBP but have a much higher affinity for multivalent ligands containing clusters of mannose moieties [ 571. In addition, liver MBP is able to interact with certain mammalian glycoproteins [32]. These activities suggest that its retention within the liver may be a consequence of high affinity interactions with endogenous ligands. Alternatively, the liver protein may play a role within host defense. Liver MBP is able to activate complement in

608

36 C-Type Lectins and Collectins

an in vitro assay [23]. Thus, it may function like its serum counterpart to activate complement within the innate immune system. 36.3.6 Pulmonary Surfactant Proteins

SP-A and SP-D are synthesized mainly by type I1 alveolar cells and Clara cells in the lung and are secreted into the alveolar space where they form part of the surfactant lining [I]. Like MBP, SP-A and SP-D provide one of the first lines of host defense by neutralizing invading micro-organisms. Although neither protein is able to fix complement, both are highly polyvalent and probably act by binding to and aggregating pathogens through interactions involving the CRDs. Binding leads to elimination of the invading particles either directly due to physical clearance within the lung surfactant or by interacting with specific receptors on macrophages and other phagocytic cells. Several putative receptors have been identified including a 340 kDa glycoprotein that is present on macrophages and interacts with SP-D [58]. Consistent with their roles within the innate immune system, SP-A and SP-D recognize sugars found on the surfaces of microorganisms. SP-A interacts with a number of monosaccharides including mannose, glucose, fucose and galactose while SP-D binds to glucose-containing structures [59-611. SP-A and SP-D have been shown to bind to a variety of bacteria that can potentially cause respiratory infections. SP-A binding to Huemophilus injhenzue, Streptococcus pneumoniue, Stuphylococcus uureus and Pneumocystis carinii has been demonstrated, while SP-D can bind to strains of Escherichiu coli and other gram-negative bacteria [62]. The pulmonary collectins appear to have functions in addition to their roles within the immune system. SP-A can interact with endogenous glycoproteins and glycolipids [6I], while SP-D can bind to phosphatidylinositol [63]. Through these interactions, SP-A and SP-D may play a role in the recycling of the elements that constitute lung surfactant [64]. In addition, SP-A may be an important component in determining the structure of the surfactant layer itself through the formation of large intermolecular complexes [2 11. Mice deficient in SP-A and SP-D have been generated independently by homologous recombination of the genes in embryonic stem cells [65, 661. Mice lacking SPA are more susceptible to infection by group B Streptococcus than wild-type mice [67]. In addition, these mice have decreased amounts of myelin in the lung and produce surfactant with altered physical properties [65]. Mice deficient in SP-D are defective in surfactant homeostasis and progressively accumulate both protein and lipid in the alveolar space [66]. These findings suggest that the pulmonary collectins are multifunctional with important structural and regulatory roles in addition to their function within the innate immune system. 36.3.7 Conglutinin and CL-43

Conglutinin and CL-43 are collectins that have only been identified in the Bovidae. Both proteins are synthesized in the liver and secreted into the serum. The primary

References

609

structure of conglutinin is very similar to that of SP-D and the conglutinin gene almost certainly arose as a result of a relatively recent duplication event [30]. Both conglutinin and CL-43 are thought to function by binding to and neutralizing micro-organisms through interactions with collectin receptors on phagocytic cells. Neither protein can activate the complement cascade directly, although conglutinin is able to bind to sugar structures on the activation product of C3 (iC3b), one of the major products of complement activation [68]. By this mechanism, conglutinin may assist in complement-mediated clearance of pathogens by agglutination of complement-coated particles and presentation to phagocytic cells. Conglutinin and CL-43 have similar sugar specificities to other members of the collectin family. Both proteins bind to monosaccharides such as mannose, fucose and N-acetylglucosamine [21]. In addition, conglutinin has been shown to bind to components of the cell walls of Sacchromyces cerevisiae as well as to influenza virus, consistent with a role in host defense (211.

36.4 Conclusions Serum MBP and other collectins are key components of the mammalian innate immune system that recognize their targets through clustering of CRDs. A combination of biochemical, biophysical and molecular biological methods have provided the basis for understanding how these lectins are able to bind with high affinity to the surfaces of pathogenic microorganisms. A major goal for the future lies in establishing how target recognition leads to biological functions such as complement fixation and opsinization. Elucidating the structural organization of the non-lectin domains that in turn define the spatial arrangement of multiple clusters of CRDs will provide a useful starting point for this aim. Acknowledgments Funding was provided by a grant from the Wellcome Trust. I thank Kurt Drickamer for comments on the manuscript.

References 1. K. Drickamer and M.E. Taylor, Annu. Rev. Cell Biol., 1993, 9, 237-264. 2. W.I. Weis and K. Drickamer, Annu. Rev. Biochem., 1996, 65, 441-473. 3. K. Drickamer, Curv. Opin. Struct. Biol., 1993, 3, 393-400. 4. W.I. Weis, R. Kahn, R. Fourme, K. Drickamer and W.A. Hendrickson, Science, 1991, 254, 1608-1615. 5. W.I. Weis, K. Drickamer and W.A. Hendrickson, Nature, 1992, 360, 127-134. 6. B.J. Graves, R.L. Crowther, C. Chandran, J.M. Rumberger, S. Li, K . 3 . Huang, D.H. Presky, P.C. Familletti, B.A. Wolitzky and D.K. Burns, Nature, 1994, 367, 532-538.

610

36 C-Type Lectins and Collectins

7. S. Sheriff, C.Y.Y. Chang and R.A.B. Ezekowitz, Nature Struct. Biol.,1994, I , 789-794. 8. K.K.-S. Ng, K. Drickamer and W.I. Weis, J. Biol. Chem., 1996, 271, 663-674. 9. R.T. Lee, Y. Ichikawa, M. Fay, K. Drickamer, M.-C. Shao and Y.C. Lee, J. Biol. Chem., 1991,266, 4810-4815. 10. K. Drickamer, Nature, 1992, 360, 183-186. 11. A. Kolatkar and W.I. Weis, J. Biol. Chem., 1996,271, 6679-6685. 12. A.R. Kolatkar, A.K. Leung, R. Isecke, R. Brossmer, K. Drickamer and W.I. Weis, J. Biol. Chem., 1998, 273, 19502-19508. 13. R.T. Lee, K.G. Rice, N.B.N. Rao, Y. Ichikawa, T. Barthel, V. Piskarev and Y.C. Lee, Biochemistry, 1989,28, 8351-8358. 14. L. Burrows, S.T. Iobst and K. Drickamer, Biochem. J., 1997, 324, 673-680. 15. 0. Blanck, S.T. Iobst, C. Gabel and K. Drickamer, J. Biol. Chem., 1996,271, 7289-7292. 16. D. Torgersen, N.P. Mullin and K. Drickamer, J. Biol. Chem., 1998,273, 6254-6261. 17. K.K.3. Ng and W.I. Weis, Biochemistry, 1997, 36, 979-988. IS. K.G. Rice, O.A. Weisz, T. Barthel, R.T. Lee and Y.C. Lee, J. Biol. Chem., 1990,265, 1842918434. 19. M.E. Taylor and K. Drickamer, J. Biol. Chem., 1993,268, 399-404. 20. K. Drickamer, M.S. Dordal and L. Reynolds, J. Biol. Chem., 1986, 261, 6878-6887. 21. H.-J. Hoppe and K.B.M. Reid, Protein Sci., 1994, 3, 1143-1 158. 22. R. Wallis and K. Drickamer, Biochem. J., 1997, 325, 391-400. 23. R. Wallis and K. Drickamer, J. Biol. Chem., 1999, 274, 3580-3589. 24. A.B. Rothmann, H.D. Mortensen, U. Holmskov and P. Hojrup, Eur. J. Biochem., 1997, 243, 630-635. 25. P. Brown-Augsburger, D. Chang, K. Rust and E. Crouch, J. Biol. Chem., 1996, 271, 1891218919. 26. R.T. White, D. Damm, J. Miller, K. Spratt, J. Schilling, S. Hawgood, B. Benson and B. Cordell, Nature, 1985, 317, 361-363. 27. B.M. Elhalwagi, M. Damodarasamy and F.X. McCormack, Biochemistry, 1997, 36, 70187025. 28. K.J. Colley and J.U. Baenziger, J. Biol. Chem., 1987,262, 10290-10295. 29. J. Lu, S. Thiel, H. Wiedemann, R. Timpl and K.B.M. Reid, J. Immunol., 1990, 144, 22872294. 30. Y.-M. Lee, K.R. Leiby, J. Allar, K. Paris, B. Lerch and T.B. Okarma, J. Biol. Chem., 1991, 266, 2715-2723. 31. D. Resnick, J.E. Chatterton, K. Schwartz, H. Slayter and M. Krieger, J. Biol. Chem., 1996, 271, 26924-26930. 32. R.A. Childs, T. Feizi, C.T. Yuen, K. Drickamer and M.S. Quesenberry, J. Biol. Chem., 1990, 265, 20770-20777. 33. W.I. Weis and K. Drickamer, Structure, 1994, 2, 1227-1240. 34. K. Ikeda, T. Sannoh, N. Kawasaki, T. Kawasaki and I. Yamashina, J. Biol. Chem., 1987,262, 745 1-7454. 35. M. Kuhlman, K. Joiner and R.A.B. Ezekowitz, J. Exp. Med., 1989, 169, 1733-1745. 36. M. Super, R.J. Levinsky and M.W. Turner, Clin. Exp. Immunol., 1990, 79, 144-150. 37. M.E. Taylor, P.M. Brickell, R.K. Craig and J.A. Summerfield, Biochem. J . , 1989, 262, 763-771. 38. R.A.B. Ezekowitz, L.E. Day and G.A. Herman, J. Exp. Med., 1988, 167, 1034-1046. 39. H.O. Madsen, P. Garred, S. Thiel, J.A. Kurtzhals, L.U. Lamm, L.P. Ryder and A. Svejgaard, J. Immunol., 1995, 155, 3013-3020. 40. M.W. Turner, Immunology Today, 1996, 17, 532-540. 41. M. Matsushita and T. Fujita, J. Exp. Med., 1992, 176, 1497-1502. 42. A. Takahashi, Y. Takayama, H. Hatsuse and M. Kawakami, Biochem. Biophys. Rex Commun., 1993,190, 681-687. 43. S. Thiel, T. Vorup-Jensen, C.M. Stover, W. Schwaeble, S.B. Laursen, K. Poulsen, A.C. Willis, P. Eggleton, S. Hansen, U. Holmskov, K.B.M. Reid and J.C. Jensenius, Nature, 1997, 386, 506-510. Trans., 1983, 11, 1-12. 44. K.B.M. Reid, Biochem. SOC.

References

61 1

45. R.J. Lipscombe, M. Sumiya, J.A. Summerfield and M.W. Turner, Inimunology, 1995, 85, 660667. 46. Y. Yokota, T. Arai and T. Kawasaki, J. Biochem. (Tokyo), 1995,117, 414-419. 47. M. Super, S. Thiel, J. Lu, R.J. Levinsky and M.W. Turner, Lancet, 1989, ii, 1236-1239. 48. M. Sumiya, M. Super, P. Tabona, R.J. Levinsky, A. Takayuki, M.W. Turner and J.A. Summerfield, Lancet, 1991,337, 1569-1 570. 49. R.J. Lipscombe, M. Sumiya, A.V.S. Hill, Y.L. Lau, R.J. Levinsky, J.A. Summerfield and M.W. Turner, Human Mol. Genet., 1992, I , 709-715. 50. H.O. Madsen, P. Garred, J.A. Kurtzhals, L.U. Lamm, L.P. Ryder, S. Thiel and A. Svejgaard. Immunogenetics, 1994, 40, 31-44. 5 I . J. Cheng and R. Wallis, unpublished results. 52. E.I.B. Peerschke, R. Malhotra, B. Ghebrehiwet, K.B.M. Reid and R.B. Sim, J. Leukocyte B i d , 1993, 53, 179-184. 53. R.R. Nepomuceno, A.H. Henschen-Edman, W.H. Burgess and A.J. Tenner, Immunity, 1997, 6, 119-129. 54. K. Sastry, K. Zahedi, J.-M. Lelias, A S . Whitehead and R.A.B. Ezekowitz, J. Immunol., 1991, 147, 692-697. 55. T. Mognes, T. Ota, A.I. Tauber and K.N. Sastry, Glycobiology, 1996, 6, 543-550. 56. S. Oka, K. Ikeda, T. Kawasaki and I. Yamashina, Arch. Biochem. Biophys., 1988, 260, 257266. 57. M.S. Quesenberry, R.T. Lee and Y.C. Lee, Biochemistry, 1997, 36, 2724-2732. 58. U. Holmskov, P. Lawson, B. Teisner, I. Tornoe, A. Willis, C. Morgan, C. Koch and K.B.M. Reid, J. Biol. Chem., 1997, 272, 13143-13749. 59. H.P. Haagsman, S. Hawgood, T. Sargeant, D. Buckley, R.T. White, K. Drickamer and B.J. Benson, J. Biol. Chem., 1987,262, 13877-13880. 60. A. Persson, D. Chang and E. Crouch, J. Biol. Chem., 1990,265, 5755-5760. 61. R.A. Childs, J.R. Wright, G.F. Ross, C.-T. Yuen, A.M. Lawson, W. Chai, K. Drickamer and T. Feizi, J. Biol. Chem., 1992, 267, 9972-9979. 62. J. Epstein, Q. Eichbaum, S. Sheriff and R.A.B. Ezekowitz, Curr. Opin. Immunol., 1996, 8, 2935. 63. A.V. Persson, B.J. Gibbons, J.D. Shoemaker, M.A. Moxley and W.J. Longmore, Biochemistry, 1992, 31, 12183-12189. 64. J.R. Wright, J.D. Borchelt and S. Hawgood, Proc. Nut1 Acad. Sci. U.S.A., 1989, 86, 54105414. 65. T.R. Korfhagen, M.D. Bruno, G.F. Ross, K.M. Huelsman, M. Ikegami, A.H. Jobe, S.E. Wert, B.R. Stripp, R.E. Morris, S.W. Glasser, C.J. Bachurski, H.S. Iwamoto and J.A. Whitsett, Proc. Natl Acad. Sri. U.S.A . , 1996, 93, 9594-9599. 66. C. Botas, F. Poulain, J. Akiyama, C. Brown, L. Allen, J. Goerke, J. Clements, E. Carlson, A.M. Gillespie, C. Epstein and S. Hawgood, Proc. Natl Acad. Sci. U.S.A., 1998, 95, 118691 1874. 67. A.M. LeVine, M.D. Bruno, K.M. Huelsman, C.F. Ross, J.A. Whitsett and T.R. Korfhagen, J. Immunol. 1997, 158,4336-4340. 68. S. Hirani, J.D. Lambris and H.J. Muller-Eberhard, J. Immunol., 1985, 134, 1105-1109.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

37 Selectins Rodger P. McEver

37.1 Introduction The regulated expression of adhesion and signaling molecules directs the recruitment of leukocytes into lymphatic tissues or sites of inflammation [ I]. The critical first event in this multistep process is the adherence of circulating leukocytes to the vascular wall under shear forces. Interactions of selectins with cell-surface carbohydrate ligands initiate the tethering and rolling of leukocytes on endothelial cells, platelets, or other leukocytes. These reversible multicellular interactions enable the leukocytes to encounter regionally expressed chemokines and lipid autoacoids. The activated leukocytes then use integrins to arrest on the vessel wall and to emigrate into the underlying tissues in response to chemotactic gradients. This chapter provides a brief overview of the structure and function of selectins, with emphasis on their interactions with specific glycoconjugates and on the mechanisms by which they mediate cell adhesion under shear forces. Earlier reviews provide additional information and references [2-51.

37.2 Structure of Selectins Each selectin has an NHz-terminal C-type lectin domain, followed by an EGF-like domain, a series of consensus repeats, a transmembrane domain, and a short cytoplasmic tail (Figure I). L-selectin, expressed on most leukocytes, binds to constitutively expressed ligands on high endothelial venules (HEV) of lymph nodes, to inducible ligands on endothelium at sites of inflammation, and to ligands on other leukocytes. E- and P-selectin bind to ligands on myeloid cells and subsets of lymphocytes, and P-selectin also binds to ligands on HEV or activated endothelial cells. E-selectin is transiently synthesized by cytokine-activated endothelium. P-selectin,

614

37 Selectins

stored in membranes of secretory granules of platelets and endothelial cells, is rapidly redistributed to the cell surface by thrombin and other secretagogues. Some cytokines also increase synthesis of P-selectin in endothelial cells. Like all C-type lectins, the selectins bind to carbohydrate ligands in a Ca2+dependent manner. All selectins bind to the tetrasaccharide sialyl Lewis x (sLex, NeuAca2,3Galpl,4[ Fucal,3]GlcNAc). These and related sialylated, fucosylated structures are terminal components of glycans attached to proteins and lipids on the surfaces of most leukocytes and some endothelial cells, the usual target cells for selectins. The three-dimensional structure of the lectin and EGF domains of E-selectin has been solved by X-ray crystallography [6]. Site-directed mutagenesis of selectins, interpreted in the context of the E-selectin structure, suggests that sLeXbinds to the lectin domain on a shallow region that overlaps a single Ca2+coordination site opposite where the EGF domain is attached [2, 31. Substitution of a limited number of residues in another C-type lectin, mannose-binding protein, changes its properties such that it binds sLeXsimilarly to the selectins. Cocrystallization of this protein with sLeXdemonstrates docking of the fucose moiety to the predicted binding region that includes the Ca2+-coordination site. Surprisingly, the sialic acid moiety does not bind directly to this mutant lectin [7]. It should be emphasized that cocrystallization of an authentic selectin with a carbohydrate ligand has not yet been achieved. Thus, the manner in which selectins bind to sLeXor other glycoconjugates remains unclear. Several studies with selectin constructs in which the EGF domains and/or consensus repeats have been deleted, switched, or mutated suggest that these domains contribute to ligand recognition [2, 31. In most of these studies, it was not determined whether the selectin constructs underwent differential oligomerization, which could influence binding assays. A soluble form of P-selectin containing only the lectin and EGF domains is monomeric in aqueous solution, and binds to ligands with the same kinetics and affinity as soluble monomeric P-selectin containing all of the extracellular domains [8]. Coexpression of the EGF domain appears to be required for proper folding of the lectin domain and, in some cases, may also modulate the binding of the lectin domain to specific ligands through an undefined mechanism.

37.3 Selectin Ligands The selectins bind selectively, but with low affinity, to sLex-related glycans. L- and P-selectin, but not E-selectin, also bind to particular sulfated carbohydrates, such as heparan sulfate, that lack sialic acid and fucose. Although the selectins bind to many sialylated, fucosylated, and/or sulfated glycans, they bind with higher affinity or avidity to only a few appropriately modified glycoproteins on blood or vascular cells (Figure 1). Two topics have been of particular interest to investigators studying these glycoproteins: the post-translational modifications required for high affinity/ avidity binding to selectins, and the relative contributions of these glycoproteins to selectin-mediated leukocyte adhesion under physiological shear stress.

37.3 Selectin Ligands Endothelial Cell

615

Leukocyte

f---------)

Platelet

Leukocyte

P-selectin

PSGL-1

Platelet

Endothelial Cell

P-selectln

Leukocyte

0 Lectin domain 0 c3

EGFdomain Short consensus repeats

Leukocyte

0

*

Ig domain 0-linked glycosylation N-linked glycosylation

Figure 1. Selectins and their best-characterized glycoprotein ligands. The estimated lengths of the selectins and of PSGL-I are based on hydrodynamic data and electron microscopy. The lengths of GlyCAM-I, CD34, and MAdCAM-I are modeled from the dimensions of another sialomucin, CD43. Abbreviations: ESL-1, E-selectin glycoprotein ligand-I; GlyCAM-1, glycosylation cell adhesion molecule-I ; MAdCAM-1, mucosal addressin cell adhesion molecule-1; PSGL-1, P-selectin glycoprotein ligand-1.

With one exception, the best characterized glycoprotein ligands are sialomucins, i.e., glycoproteins with repeating peptide motifs that have multiple Ser/Thr-linked oligosaccharides (0-glycans). The sialomucins bind particularly well to L- and Pselectin, but may also bind to E-selectin. They must be appropriately sialylated and

616

37 Selectins

fucosylated to bind any of the selectins, and they must be sulfated to bind L- and P-selectin. The post-translational modifications of glycosylated cell adhesion molecule-1 (GlyCAM-1) and P-selectin glycoprotein ligand- 1 (PSGL-1) provide two contrasting ways to create binding sites for selectins. GlyCAM-1 is a relatively small mucin that is synthesized by HEV of murine lymph nodes [9]. It lacks a transmembrane domain and is secreted into the plasma. Rather than mediate cell adhesion] it may bind to and activate lymphocytes that circulate through lymph nodes. The structures of the 0-glycans of GlyCAM-1 have been extensively characterized [ 10-121. Most have a branched core-2 structure, in which a GlcNAc is linked via a p1,6 branch to GalNAc. Some of these 0-glycans have a capping group that includes sLeXand a sulfate linked to the C-6 position of the Gal or GlcNAc. There may be larger versions of this structure that are multiply fucosylated or sulfated. However, there appears to be relatively little polylactosamine, i.e., repeating disaccharides of the type [-+3Galp1-+4GlcNAcp -+In, on the p1,6 branch. Sialylation, fucosylation, and sulfation of the glycans are all required for GlyCAM-1 to bind to L-selectin [9]. Recent studies suggest that 6-sulfation of GlcNAc, which creates the capping group 6-sulfo sLex, is required for HEV mucins such as GlyCAM-1 to bind L-selectin [ 131. It is not known whether additional fucose residues or 6-sulfation of Gal enhances binding. The isolated sialylated, fucosylated 0-glycans, even with attached sulfate, still bind with low affinity to L-selectin [ 141. Thus the organization of the 0-glycans on GlyCAM-1 must enhance binding in some fashion. In one model, many clustered 0-glycans increase the avidity of binding of GlyCAM-1 for L-selectin [9]. In another model, specific groupings of 0glycans juxtapose sulfates, hydroxyls, or other groups to create composite recognition sites with higher affinity for L-selectin [ 151. Multimerization of GlyCAM-1 may also allow it to bind to L-selectin on leukocytes with high avidity [ 161. PSGL- 1 is a disulfide-bonded, homodimeric mucin expressed on leukocytes that, when appropriately modified] binds to all three selectins [ 171. The extracellular domain is rich in serines, threonines, and prolines, many of which are present in a series of 15 or 16 decameric consensus repeats [18] (Figure 2A). Biochemical analyses of native and recombinant PSGL-1 confirm that it is modified with many 0glycans [ 19-22]. PSGL-1 also has two or three N-glycans, but enzymatic removal of these structures does not detectably alter binding to selectins [ 191. PSGL-1 requires both a2,3 sialylation and a1,3 fucosylation of branched core-2 0-glycans to bind to selectins [18-20, 221. The structures of the 0-glycans on PSGL-1 from human HL60 cells have been determined [23]. Like those from GlyCAM-1, almost all the 0glycans have a branched, core-2 structure. A major difference is that the 0-glycans of PSGL-1 are not detectably sulfated. The structures of the fucosylated 0-glycans of PSGL-1 also differ from those of GlyCAM-1 (Figure 2B). The major fucosylated species has an extended, trifucosylated polylactosamine on the pl,6 branch. Only ~ 1 4 % of the 0-glycans of PSGL-1 from HL-60 cells are fucosylated, which suggests an upper limit on the 0-glycans on any polypeptide chain that might bind to selectins [23]. This suggests that PSGL-1 need not present many fucosylated 0glycans that increase binding avidity to selectins, and that 0-glycans containing fucose and/or polylactosamine are preferentially attached to specific regions of the polypeptide backbone. Consistent with this notion, a mAb to the NHz-terminal

37.3 Selectin Lipnds

A Signal Peptide Propeptide

PSGL-1

617

Transmembrane Domain

16 Decamer Repeats

COOH

QATE" E" LD: DFLPETEPPEM I I I I lycai-1

B Fuc

Fuc

Fuc

Figure 2. Primary structure and post-translational modifications of PSGL-I. (A) Schematic diagram of the domains of human PSGL-l. Not shown is a disulfide bridge near the transmembrane domain that enables PSGL-1 to form homodimers. The residue number that begins each region of the extracellular domain and the transmembrane domain is indicated. The NH2 terminus of mature PSGL-1 begins at residue 42, immediately after the propeptide. The sequence of the first 21 amino acids of mature PSGL-I is listed. The epitope for PL1, a mAb that blocks binding of PSGL-I to P- and L-selectin, spans residues 49-62 of this sequence. Sulfate is attached to one or more of the tyrosines at residues 46, 48, and 51. Mutational analysis suggests that binding to P-selectin requires sulfation of only one of these residues, plus a specific 0-glycan linked to Thr-57. (B) Structures of the two fucosylated core-2 0-glycans identified on PSGL-1 from human HL-60 cells. One of these is presumably attached to Thr-57.

region of PSGL-1 blocks binding to P- and L-selectin, although not to E-selectin [24-261. A striking feature of the NH2-terminal region of PSGL-1 is a cluster of three tyrosines in an anionic sequence that favors tyrosine sulfation (Figure 2A). PSGL-1 is sulfated virtually exclusively on these tyrosines [27], and enzymatic removal of sulfate, metabolic inhibition of sulfation, or substitution of the tyrosines with phenylanines blocks binding of PSGL-1 to P- and L-selectin, but not to E-selectin (22,

618

37 Selectins

26-29]. Detailed mutational analysis of PSGL-1 indicates that only one of the three tyrosines is required for binding to P-selectin, and also suggests that a critical fucosylated 0-glycan must be attached to a nearby threonine for binding to P-selectin [30] (Figure 2A). Thus, PSGL-1, like GlyCAM-I, requires both sulfation and specific 0-glycosylation to bind P- and/or L-selectin. However, the structures of the fucosylated 0-glycans differ on the two molecules. Furthermore, sulfate is attached to 0-glycans on GlyCAM-1 but to tyrosines on PSGL-1. Perhaps the C-type lectin domain of each selectin contains a specific region that binds fucosylated glycans, and the lectin domains of P- and L-selectin contain an additional region that binds anionic structures such as sulfated saccharides or tyrosines. Elucidation of the molecular contacts between a selectin and a specific ligand will require solving the structures of cocrystallized complexes of these molecules. As mentioned, PSGL- 1 requires neither tyrosine sulfation nor its NH2-terminal region to bind E-selectin, although it does require at least one sialylated and fucosylated 0-glycan [ 171. Depending on their state of differentiation, human or murine T cells express forms of PSGL-1 that bind only P-selectin or bind both P- and Eselectin. The ability of PSGL-1 to bind E-selectin is correlated with its ability to bind a mAb that recognizes a subset of oligosaccharides that include sLeX[31, 321. This raises the possibility that E-selectin binds preferentially to a specific type of fucosylated 0-glycan(s) on one or more regions of PSGL-1. It has been argued that cell adhesion molecules must bind to their ligands with low affinity so that multivalent cell attachment does not become irreversible [33]. Low affinity binding is clearly operative for some adhesion molecules such as the binding of L-selectin to GlyCAM-1 [ 161. However, surface plasmon resonance measurements indicate that soluble monomeric P-selectin binds to immobilized PSGL-1 with a relatively high affinity (Kd % 300 nM) [8]. With this technique, binding was shown to involve both very rapid association and dissociation kinetics, which are highly relevant for cell-cell interactions under flow conditions (see below). Thus, selectins may interact with mucins through multiple low-affinity but high-avidity contacts, through single high-affinity contacts, or by some combination of the two. PSGL-1 is the selectin ligand with the most clearly defined function in leukocyte adhesion under shear stress [17]. The mAb to the NH2-terminal region of human PSGL-1 that abrogates binding to P-selectin in purified systems also blocks the tethering and rolling of leukocytes on P-selectin under laminar flow conditions, both in vitro and in vivo [24, 341. A mAb to the analogous region of murine PSGL1 also blocks leukocyte adhesion to P-selectin under flow [35]. PSGL-1 is expressed in lower amounts on the cell surface than many other adhesion receptors, and it contains only a very small fraction of the total sLeXon leukocytes [ 171. Yet it represents all of the high affinity binding sites for P-selectin on leukocytes, and it is essential for leukocytes to tether to and roll on P-selectin under shear stresses. As mentioned, GlyCAM-1 is a secreted molecule, so it probably does not mediate cell adhesion directly. The role of other known glycoprotein ligands for selectins in leukocyte adhesion is less certain. As discussed below, the cell-surface presentation of these molecules may determine whether they participate in cell adhesion under

37.4 Requirenients f o r Selectins to Mediute Tethering

619

flow. It is possible that leukocyte tethering or rolling through L- or E-selectin requires binding to more than one ligand.

37.4 Requirements for Selectins to Mediate Tethering and Rolling of Leukocytes under Hydrodynamic Flow The requirements for cell adhesion under laminar shear stress are particularly demanding. To tether to a surface, a free-flowing leukocyte must form adhesive bonds very rapidly. For the cell to roll, these bonds must dissociate quickly at the trailing edge of the cell as new bonds form at the leading edge of the cell. The adhesive bonds must also resist premature dissociation by the forces applied to the bond; otherwise the cell would immediately detach into the fluid stream. The biochemical and biophysical features of selectin-ligand interactions have evolved to meet these specialized requirements. Leukocytes perfused over very low densities of immobilized P-selectin, E-selectin, or an L-selectin ligand such as CD34 form transient tethers that do not convert to rolling adhesions [36, 371. The number of tethers is linearly related to the selectin or selectin ligand density, suggesting that the transient tethers represent quanta1 units, or single selectin-ligand bonds. The durations of the tethers are very short ( 1 2 sec), particularly those for L-selectin-ligand interactions. But physiological shear forces do not significantly decrease tether durations, suggesting that the bonds have tensile strength that resists dissociation by applied force. The transient nature of selectin-ligand bonds requires that leukocytes be perfused at or above a minimum shear stress if they are to roll [38, 391. Shear stress allows the cell to rotate so that new bonds form to replace those that dissociate, and increased shear may also enhance the rate at which new bonds form [40]. If shear is too low, the cell does not roll, but instead detaches from the surface. This shear threshold requirement is particularly marked for rolling through L-selectin, consistent with its faster rate of dissociation [37]. In addition to these intrinsic features of selectin-ligand interactions, the cellsurface organization of selectins or their ligands contributes significantly to the efficiency of leukocyte adhesion under flow. For example, P-selectin is a long protein, 0 above with its nine consensus repeats extending the C-type lectin domain ~ 4 nm the cell surface. Flowing neutrophils tether to and roll on wild-type P-selectin expressed on a monolayer of transfected CHO cells [41]. However, very few rolling neutrophils accumulate on CHO cells expressing shortened forms of P-selectin that have less than five of the nine consensus repeats. Furthermore, those cells that adhere to shortened P-selectin roll much faster and are detached more readily by increasing shear stress. These data suggest that P-selectin must project its lectin doman sufficiently far above the plasma membrane to mediate optimal attachment of leukocytes under flow. This may increase its effective radius of contact with PSGL-1 on a flowing cell. By extending above most of the glycocalyx, it may also interact

620

37 Selectins

with PSGL- 1 under conditions that reduce electrostatic repulsion between cells. The lengths of other selectins or selectin ligands may also be important for function. Leukocytes have an irregular surface that contains many microvilli. Both Lselectin and PSGL-1 are concentrated on microvillous tips, the sites of earliest contact between a flowing leukocyte and another cell [24, 421. When expressed in transfected murine pre-B cells, wild-type L-selectin is concentrated on microvilli, where it supports tethering and rolling of the cells on immobilized L-selectin ligands [43, 441. A chimera with the extracellular domain of L-selectin fused to the transmembrane and cytoplasmic domains of CD44 is distributed on the cell body rather than on the microvilli [44]. Relatively few cells expressing the chimera tether to Lselectin ligands under flow; however, the cells that do tether roll with velocities and adhesive strengths that are similar to those of cells expressing wild-type L-selectin. Thus, the microvillous localization of L-selectin is particularly important for mediating the initial tethering event, but not the subsequent rolling event. Clustering of selectins or their ligands may also regulate leukocyte adhesion under flow. A form of L-selectin lacking most of the cytoplasmic tail is still targeted to microvilli of transfected pre-B cells [43]. However, cells expressing this tail-less L-selectin roll poorly on L-selectin ligands [45]. Interactions of the cytoplasmic domain of L-selectin with the cytoskeleton may be essential for its adhesive function [43, 451. PSGL-1 may also interact with the cytoskeleton; it is redistributed from microvilli to the uropods of activated leukocytes, but pretreatment with cytochalasin D prevents this redistribution [46]. Cytoskeletal interactions may enhance the adhesive function of selectins or selectin ligands in two ways. First, they may cluster the molecules, enhancing binding avidity by delaying the dissociation of selectinligand bonds. Second, they may prevent forced extraction of the molecules from the lipid bilayer during cell-cell contact. The cytoplasmic domain of P-selectin contains signals that mediate its rapid endocytosis in clathrin-coated pits. Interactions of the cytoplasmic domain of Pselectin with these structures were found to enhance its adhesive function under flow [47]. Transfected CHO cells were prepared that express wild-type P-selectin or Pselectin constructs with substitutions or deletions in the cytoplasmic domain that either increase or decrease its internalization rate. Under flow, neutrophils tether equivalently to all constructs when expressed at matched densities. However, neutrophils roll on the internalization-competent constructs with greater adhesive strength, at slower velocities, and with more uniform motion. Confocal immunofluorescence microscopy demonstrates colocalization of a-adaptin, a component of clathrin-coated pits, with wild-type P-selectin but not with internalization-defective P-selectin lacking the cytoplasmic domain. Thus, interactions of P-selectin with clathrin-coated pits provide an alternative to cytoskeletal interactions to enhance adhesive function. The association of P-selectin with clathrin-coated pits may delay dissociation of P-selectin-PSGL-1 bonds and/or prevent forced extraction of Pselectin from the membrane. Selectins or their ligands may also self-associate in the plane of the membrane. PSGL-1 is a dimer, which may improve binding avidity by contributing recognition determinants from both subunits. Hydrodynamic analysis indicates that P-selectin isolated from platelet membranes is oligomeric even in nonionic detergent above

3 7.6 Conclusions

62 1

the critical micellar concentration. Self-association may be mediated by the transmembrane domain, because recombinant P-selectin lacking this domain is monomeric [8]. Artificial dimerization of L-selectin improves its ability to support rolling adhesion of leukocytes; cell activation-induced dimerization of L-selectin may also transiently enhance its function [48].

37.5 Functions of Selectins and their Ligands in vivo The importance of selectins in humans is underscored by the discovery of a congenital disorder of fucose metabolism, termed leukocyte adhesion deficiency 2 (LAD-2) [49]. Because patients with LAD-2 lack fucosylated glycoconjugates, they do not express functional selectin ligands. Leukocytes from these patients do not tether to and roll on P- or E-selectin surfaces. Clinically, the patients have more infections, supporting the concept that the selectins have an important function in initiating recruitment of leukocytes. Mice made genetically deficient in each of the three selectins have defects in leukocyte trafficking in response to specific challenges [ 501. Lymphocytes from Lselectin-deficient mice home less efficiently to peripheral lymph nodes. Mice lacking L-, P-, or E-selectin demonstrate impaired rolling of leukocytes in postcapillary venules. Mice lacking both E- and P-selectin have more infections and shortened survival. Human and murine leukocytes express two al,3 fucosyltransferases, termed FucTIV and Fuc-TVII. Mice rendered deficient in Fuc-TVII have leukocyte trafficking defects much like those observed in mice lacking P- and E-selectin, demonstrating the essential role of this enzyme in fucosylating selectin ligands [51]. Mice lacking core 2 p 1,6GlcNAc transferase have significant impairment in trafficking of myeloid leukocytes to sites of inflammation, and myeloid cells roll poorly on P-, E-, or Lselectin [52].This result confirms the importance of core 2 O-glycans in construction of selectin ligands on these cells. Surprisingly, L-selectin-dependent trafficking of lymphocytes to lymph nodes is normal. This observation suggest that core 2 0-glycans on molecules such as GlyCAM-1 and CD34 may not be required for L-selectin to bind to lymph node HEV. The expression of selectins is normally tightly regulated to ensure that leukocytes tether to and roll on the vascular surface only at appropriate locations [53].However, dysregulated expression of selectins has been implicated in several forms of leukocyte-mediated tissue injury [ 541.

37.6 Conclusions The selectins are the best characterized group of mammalian lectins known to mediate cell adhesion. Much remains to be learned about the mechanisms for reg-

622

37 Selectins

ulating expression of selectins and their ligands, the molecular details of selectin binding to specific glycoconjugates, the biophysical principles that selectins employ to mediate rolling leukocyte adhesion under hydrodynamic flow, and the in vivo functions of selectins.

References 1. Springer, T. A. 1995. Traffic signals on endothelium for lymphocyte recirculation and leukocyte emigration. Annu. Rev. Physiol. 57927-872. 2. McEver, R. P., K. L. Moore, and R. D. Cummings. 1995. Leukocyte trafficking mediated by selectin-carbohydrate interactions. J. Biol. Chem. 270: 1 1025-1 1028. 3. McEver, R. P. 1997. Selectin-carbohydrate interactions during inflammation and metastasis. Glycoconj. J. 14:585 -591. 4. Tedder, T. F., D. A. Steeber, A. Chen, and P. Engel. 1995. The selectins: Vascular adhesion molecules. FASEB J. 9966-873. 5. Kansas, G. S. 1996. Selectins and their ligands: current concepts and controversies. Blood 88:3259-3287. 6. Graves, B. J., R. L. Crowther, C. Chandran, J. M. Rumberger, S. Li, K.-S. Huang, D. H. Presky, P. C. Familletti, B. A. Wolitzky, and D. K. Burns. 1994. Insight into E-selectin/ligand interaction from the crystal structure and mutagenesis of the lec/EGF domains. Nature 367:532-538. 7. Ng, K. K. S. and W. I. Weis. 1997. Structure of a selectin-like mutant of mannose-binding protein complexed with sialylated and sulfated Lewis’ oligosaccharides. Biochemistry 36:979-988. 8. Mehta, P., R. D. Cummings, and R. P. McEver. 1998. Affinity and kinetic analysis of P-selectin binding to P-selectin glycoprotein ligand-1. J. Biol. Chem. 273:32506-32513. 9. Rosen, S. D., S. T. Hwang, P. A. Giblin, and M. S. Singer. 1997. High-endothelial-venule ligands for L-selectin: identification and functions. Biochem. SOC.Trans. 25428-433. 10. Hemmerich, S., C. R. Bertozzi, H. Leffler, and S. D. Rosen. 1994. Identification of the sulfated monosaccharides of GlyCAM- 1, an endothelial-derived ligdnd for L-selectin. Biochemistry 33:4820-4829. 11. Hemmerich, S. and S. D. Rosen. 1994. 6’-sulfated sialyl Lewis x is a major capping group of GlyCAM-1. Biochemistry 33:4830-4835. 12. Hemmerich, S., H. Leffler, and S. D. Rosen. 1995. Structure of the 0-glycans in GlyCAM-I, an endothelial-derived ligand for L-selectin. J. Biol. Chem. 270: 12035- 12047. 13. Mitsuoka, C., M. Sawada-Kasugai, K. Ando-Furui, M. Izawa, H. Nakanishi, S. Nakamura, H. Ishida, M. Kiso, and R. Kannagi. 1998. Identification of a major carbohydrate capping group of the L-selectin ligand on high endothelial venules in human lymph nodes as 6-sulfo sialyl Lewis X. J. Biol. Chem. 273:11225-11233. 14. Koenig, A,, R. Jain, R. Vig, K. E. Norgard-Sumnicht, K. L. Matta, and A. Varki. 1997. Selectin inhibition: Synthesis and evaluation of novel sialylated, sulfated and fucosylated oligosaccharides, including the major capping group of GlyCAM-1. Glycobioloyy 7:79-93. 15. Varki, A. 1997. Selectin ligands: will the real ones please stand up? J. Clin. Invest. 99:158-162. 16. Nicholson, M. W., A. N. Barclay, M. S. Singer, S. D. Rosen, and P. A. Van der Menve. 1998. Affinity and kinetic analysis of L-selectin (CD62L) binding to glycosylation-dependent celladhesion molecule-1. J. Biol. Chem. 273:763-770. 17. McEver, R. P. and R. D. Cummings. 1997. Role of PSGL-1 binding to selectins in leukocyte recruitment. J. Clin.Invest. 100:485-492. 18. Sako, D., X.-J. Chang, K. M. Barone, G. Vachino, H. M. White, G. Shaw, G. M. Veldman, K. M. Bean, T. J. Ahern, B. Furie, D. A. Cumming, and G. R. Larsen. 1993. Expression cloning of a functional glycoprotein ligaud for P-selectin. Cell 75:1179-1186. 19. Moore, K. L., N. L. Stults, S. Diaz, D. L. Smith, R. D. Cummings, A. Varki, and R. P. McEver. 1992. Identification of a specific glycoprotein ligand for P-selectin (CD62) on myeloid cells. J. Cell. Biol. 118:445-456.

References

623

20. Norgard, K . E., K. L. Moore, S. Diaz, N. L. Stults, S. Ushiyama, R. P. McEver, R. D. Cummings, and A. Varki. 1993. Characterization of a specific ligand for P-selectin on myeloid cells. A minor glycoprotein with sialylated 0-linked oligosaccharides. J. Biol. Chem. 268: 1276412774. 21. Moore, K. L., S. F. Eaton, D. E. Lyons, H. S. Lichenstein, R. D. Cummings, and R. P. McEver. 1994. The P-selectin glycoprotein ligand from human neutrophils displays sialylated, fucosylated, 0-linked poly-N-acetyllactosamine. J. Biol. Chem. 269:233 18-23327. 22. Li, F., P. P. Wilkins, S. Crawley, J. Weinstein, R. D. Cummings, and R. P. McEver. 1996. Post-translational modifications of recombinant P-selectin glycoprotein ligand- 1 required for binding to P- and E-selectin. J. Biol. Chem. 271:3255-3264. 23. Wilkins, P. P., R. P. McEver, and R. D. Cummings. 1996. Structures of the 0-glycans on Pselectin glycoprotein ligand-l from HL-60 cells. J. Bid. Chem. 271:18732-18742. 24. Moore, K. L., K. D. Patel, R. E. Bruehl, L. Fugang, D. A. Johnson, H. S. Lichenstein, R. D. Cummings, D. F. Bainton, and R. P. McEver. 1995. P-selectin glycoprotein ligand-1 mediates rolling of human neutrophils on P-selectin. J. Cell Biol. 128:661-671. 25. Walcheck, B., K. L. Moore, R. P. McEver, and T. K. Kishimoto. 1996. Neutrophil-neutrophil interactions under hydrodynamic shear stress involve L-selectin and PSGL-1: a mechanism that amplifies initial leukocyte accumulation on P-selectin in vitro. J. Clin. Invest. 98: 1081-1087. 26. Spertini, O., A,-S. Cordey, N. Monai, L. Giuffre, and M. Schapira. 1996. P-selectin glycoprotein ligand-l (PSGL-I) is a ligand for L-selectin on neutrophils, monocytes and CD34+ hematopoietic progenitor cells. J. Cell Bid. 135523-53 1. 27. Wilkins, P. P., K. L. Moore, R. P. McEver, and R. D. Cummings. 1995. Tyrosine sulfation of P-selectin glycoprotein ligand-1 is required for high affinity binding to P-selectin. J. Biol. Chem. 270:22677-22680. 28. Pouyani, T. and B. Seed. 1995. PSGL-1 recognition of P-selectin is controlled by a tyrosine sulfation consensus at the PSGL-1 amino terminus. Cell 83:333--343. 29. Sako, D., K. M. Comess, K. M. Barone, R. T. Camphausen, D. A. Cumming: and G. D. Shaw. 1995. A sulfated peptide segment at the amino terminus of PSGL-1 is critical for Pselectin binding. Cell 83:323-331. 30. Liu, W.-J.. V. Ramachandran, J. Kang, T. K. Kishimoto, R. D. Cummings, and R. P. McEver. 1998. Identification of N-terminal residues on P-selectin glycoprotein ligand-l required for binding to P-selectin. J. Biol. Chem. 273:7078-7087. 31. Borges, E., G. Pendl, R. Eytner, M. Steegmaier, 0. Zollner, and D. Vestweber. 1997. The binding of T cell-expressed P-selectin glycoprotein ligand-1 to E- and P-selectin is differentially regulated. J. Biol. Chem. 272:28786-28792. 32. Fuhlbrigge, R. C., J. D. Kieffer, D. Armerding, and T. S. Kupper. 1997. Cutaneous lymphocyte antigen is a specialized form of PSGL-I expressed on skin-homing T cells. Nature 389:978-981. 33. Van der Merwe, P. A. and A. N. Barclay. 1994. Transient intercellular adhesion: the importance of weak protein-protein interactions. Trends Biochem. Sci 19:354-358. 34. Norman, K. E., K. L. Moore, R. P. McEver, and K. Ley. 1995. Leukocyte rolling in vivo is mediated by P-selectin glycoprotein ligand-l . Blood 86:4417-4421. 35. Borges, E., R. Eytner, T. Moll, M. Steegmaier, M. A. Campbell, K. Ley, H. Mossman, and D. Vestweber. 1997. The P-selectin glycoprotein ligand-1 is important for recruitment of neutrophils into inflamed mouse peritoneum. Blood 90: 1934-1942. 36. Alon, R., D. A. Hammer, and T. A. Springer. 1995. Lifetime of the P-selectin: carbohydrate bond and its response to tensile force in hydrodynamic flow. Nuture 374539-542. 37. Alon, R., S. Q. Chen, K. D. h r i , E. B. Finger, and T. A. Springer. 1997. The kinetics of L-selectin tethers and the mechanics of selectin-mediated rolling. J. Cell Biol. 138:1169%1180. 38. Finger, E. B., K. D. Puri, R. Alon, M. B. Lawrence, U. H. Von Andrian, and T. A. Springer. 1996. Adhesion through L-selectin requires a threshold hydrodynamic shear. Nature 379:266269. 39. Lawrence, M. B., G. S. Kansas, E. J. Kunkel, and K. Ley. 1997. Threshold levels of fluid shear promote leukocyte adhesion through selectins (CD62L,P,E).J. Cell Biol. 136:717-727. 40. Chen, S. Q. and T. A. Springer. 1999. An automatic braking system that stabilizes leukocyte rolling by an increase in selectin bond number with shear. J. Cell Biol. 144:185-200.

624

37 Selectins

41. Patel, K. D., M. U. Nollert, and R. P. McEver. 1995. P-selectin must extend a sufficient length from the plasma membrane to mediate rolling of neutrophils. J. Cell Biol. 131:1893-1902. 42. Picker, L. J., R. A. Warnock, A. R. Burns, C. M. Doerschuk, E. L. Berg, and E. C. Butcher. 1991. The neutrophil selectin LECAM-1 presents carbohydrate ligands to the vascular selectins ELAM-1 and GMP-140. Cell 66:921--933. 43. Pavalko, F. M., D. M. Walker, L. Graham, M. Goheen, C. M. Doerschuk, and G. S. Kansas. 1995. The cytoplasmic domain of L-selectin interacts with cytoskeletal proteins via a-actinin: Receptor positioning in microvilli does not require interaction with a-actinin. J. Cell Biol. 12911155-1164. 44. Von Andrian, U. H., S. R. Hasslen, R. D. Nelson, S. L. Erlandsen, and E. C. Butcher. 1995. A central role for microvillous receptor presentation in leukocyte adhesion under flow. CeN 821989-999. 45. Kansas, G. S., K. Ley, J. M. Munro, and T. F. Tedder. 1993. Regulation of leukocyte rolling and adhesion to high endothelial venules through the cytoplasmic domain of L-selectin. J. Exp. Med. 177:833-838. 46. Lorant, D. E., R. P. McEver, T. M. McIntyre, K. L. Moore, S. M. Prescott, and G. A. Zimmerman. 1995. Activation of polymorphonuclear leukocytes reduces their adhesion to P-selectin and causes redistribution of ligands for P-selectin on their surfaces. J. Clin. Invest. %: 171-1 82. 47. Setiadi, H., G. Sedgewick, S. L. Erlandsen, and R. P. McEver. 1998. Interactions of the cytoplasmic domain of P-selectin with clathrin-coated pits enhance leukocyte adhesion under flow. J. Cell Biol. 142:859-871. 48. Li, X., D. A. Steeber, M. L. K. Tang, M. A. Farrar, R. M. Perlmutter, andT. F. Tedder. 1998. Regulation of L-selectin-mediated rolling through receptor dimerization. J. Exp. Med. 188:1385-1390. 49. Etzioni, A,, M. Frydman, S. Pollack, 1. Avidor, M. L. Phillips, J. C. Paulson, and R. GershoniBaruch. 1992. Brief report: recurrent severe infections caused by a novel leukocyte adhesion deficiency. N. En& J. Med. 327:1789-1792. 50. Frenette, P. S. and D. D. Wagner. 1997. Insights into selectin function from knockout mice. Thromh. Haemost. 78:60-64. 51. Maly, P., A. D. Thall, B. Petryniak, G. E. Rogers, P. L. Smith, R. M. Marks, R. J. Kelly, K. M. Gersten, G. Y. Cheng, T. L. Saunders, S. A. Camper, R . T. Camphausen, F. X. Sullivan, Y. Isogai, 0. Hindsgaul, U. H. Von Andrian, and J. B. Lowe. 1996. The a(l,3)Fucosyltransferase Fuc-TVII controls leukocyte trafficking through an essential role in L-, E-, and Pselectin ligand biosynthesis. CeII 86:643-653. 52. Ellies, L. G., S. Tsuboi, B. Petryniak, J. B. Lowe, M. Fukuda, and J. D. Marth. 1998. Core 2 oligosaccharide biosynthesis distinguishes between selectin ligands essential for leukocyte homing and inflammation. Immunity 9% 1-890. 53. McEver, R. P. 1997. Regulation of expression of E-selectin and P-selectin. In: The Selectins: Initiators of Leukocyte Endothelial Adhesion. D. Vestweber, editor. Hanvood Academic Publishers, Amsterdam. 3 1-47. 54. Sharar, S. R., R. K. Winn, and J. M. Harlan. 1995. The adhesion cascade and anti-adhesion therapy: An overview. Springer Semin. Immunopathol. 16:359-378.

Carbohydrates in Chemistry and Biology Beat Emst, Gerald W. Hart, Pierre Sinay copyright Q WILEY-VCH Verlag GmbH. 2000

38 Galectins Douglas N. W. Cooper and Samuel H. Barondes

38.1 Introduction

Galectins are a family of proteins that bind P-galactosides by means of a carbohydrate recognition domain (CRD) that has many conserved sequence elements. First identified in 1975 in soluble extracts of electric organ tissue of an electric eel, galectins have since been identified in a wide range of organisms (reviewed in [ l, 21). In addition to vertebrates (including fish, birds, amphibians and mammals), galectins have also been found in invertebrates (worms and insects) and even in protists (sponge and fungus). Whereas the first galectins to be identified were discovered as proteins that bound to P-galactoside affinity columns, many related proteins that share conserved sequence elements are now being identified in DNA data banks. In addition to ten mammalian galectins that have been shown to bind to @-galactosides,eight related mammalian genes have been identified on the basis of sequence similarity. In the completely sequenced genome of a worm, Caenorhahditis elegans, more than 20 of the 20,000 genes encode proteins with conserved sequence elements that are characteristic of galectins. By definition all galectins bind galactoside sugars [ 3 ] , but individual galectins show significant differences in specificity for more complex carbohydrate chains and for potential glycoconjugate ligands. Studies of the crystal structures of several galectins have revealed the roles of specific conserved amino acids in sugar binding. However, in contrast with the detailed knowledge of galectin biochemistry, there is little agreement on their physiological ligands and functions. Mystery even surrounds the means by which galectins gain access to extracellular glycoconjugates. There is good evidence that these proteins can be secreted, but they lack signal sequences and do not pass through the standard ER/Golgi pathway. Nevertheless, binding of galectins to glycoconjugates on and around cells has been shown to result in many diverse biological effects.

626

38 Gulectins

The aim of this chapter is to provide a concise overview of current knowledge about the structure and function of galectins. This will prepare the reader for the many new developments that we anticipate in this rapidly changing field.

38.2 Galectin Structure All galectins share a core sequence consisting of about 130 amino acids, many of which are highly conserved. Crystallography has been used to determine the structure of several mammalian galectins and a frog galectin [4-lo], and it has been found that the core galectin sequence folds as a globule of two antiparallel P-sheets formed by five and six P-strands each. This structure is remarkably similar to that of legume lectins even though the binding sites are distinct and there is no significant primary sequence similarity. Galectin residues directly involved in carbohydrate binding are contained in a pocket formed by four adjacent P-strands. These strands are contained in a protein segment, referred to as the carbohydrate recognition domain (CRD), of about 60 amino acids generally encoded by a single exon, and this segment is especially highly conserved (Figure 1). While sharing this basic core structure, the amino acid sequences of all galectins identified to date can be structurally classified into three basic types [2] (Figure 2 and Tables 1 and 2). So-called “prototype” galectins, exemplified by mammalian galectin-1, are composed solely of a single core galectin domain. Most prototype galectins self-associate to form homodimers, or higher multimers in the case of two sponge galectins [ 111. However, certain prototype galectins, such as mammalian galectin-5 [I21 or chicken C-14 [2], are found only as monomers. Another major class is composed of the “tandem-repeat’’ galectins, which are composed of two non-identical tandem core galectin domains linked by a proline and glycine rich sequence of variable length. Although the sequence of this linker segment is only very weakly conserved, there is evidence that its nature may be functionally important. For example, galectin-4 and galectin-6, while distinct genes, are virtually identical except for their linker sequences [ 13). Furthermore, an intestine-specific splice variant of galectin-9 differs from the major form only in the linker sequence [14]. One possibility is that the linker structure is important in deb

Figure 1. Amino acid sequences of galectin CRDs are aligned to maximize sequence similarity, allowing variation in the size of some loops between beta strands. Highly conserved residues (specific amino acids or certain limited sets of similar amino acids) are capitalized with the most highly conserved residues also shaded. Beta strand positions are indicated at the top, with residues known to directly interact with carbohydrate marked by asterisks. Candidate galectin genes (i.e. not yet shown to bind galactosides) are indicated by an asterisk preceding their Genbank accession number. The mammalian galectins and candidates are all human, except for rat galectin-5 and mouse galectin-6. Further documentation for each of these genes is available at URL: http://www.sacs.ucsf.edu/home/cooper/galectins. htm.

38.2 Gulectin Structure

627

CARBOHYDRATE RECOGNITION DOMAINS OF KNOWN GALECTINS AND NEW CANDIDATES 53

5.1

Mammalian Galectins

al-1 PO9382 al-> PO5162 ~ l - 3P1793i d l - 4 382953

* * *

55* * -

A

L3G862 Gal-6 AF026794 A Gal-5

Gal-7 U06643 Gal-8 L78132

A

G d l - 9 13BCOS894 A

Gal-l@ 1,0166

F i s h Galectins E l ectro. A28302

Fugii

*AL020875

* FRaO73apsB12 *FKaO75apsH3 B i r d Galectins Galliis 000311

Insect Galectins Anopheles 26998%

Drosoph.*LP06039 Worm Galectins

C.elegans 063575 AB000802 A B

Sponge Galectins (;codla X93925

M97692 Fungus Galectins

Virus Galectins

ka Plant Galectins Arabid. *T7N9.14 qFmMetqgl*vdgedpprI

t s-9 f s-n la-a 11-n lm-n

56*

-__^_

* *

628

38 Galectins

1 CRD Monomer A

1 CRD Dimer B

2 CRDs Linked C

1 CRD+ N-Term. Repeat D

Figure 2. Schematic representation of the structural classification of galectins according to Hirabayashi and Kasai [2]: monomeric (A) or dimeric (B) prototype galectins, tandem-repeat galectins (C), and chimeric galectins (D)

termining the proximity and orientation of the galectin domains, properties expected to critically influence ligand crosslinking. The third class of galectins are the “chimeric” ones, exemplified by mammalian galectin-3, which is composed of an N-terminal domain with several repeats of a peptide sequence rich in proline, glycine and tyrosine residues followed by a Cterminal galectin domain. The repetitive domain is similar to ones found in certain other proteins (such as synexin and synaptophysin) and proposed to function in selfaggregation [15]. Indeed, this seems to be the case for galectin-3, which can form multimers, especially on binding to immobilized ligands. The isolated C-terminal galectin domain retains its lectin activity, but loses the propensity to multimerize [15-211. This could prove to be a biologically regulated property, because the Nterminal repetitive domain is susceptible to selective proteolysis with collagenases, including a calcium-dependent enzyme which copurifies with galectin-3 in the absence of calcium [15-191. Multimers of galectin-3 can also be stabilized by transglutaminase mediated crosslinking [22]. Another distinctive property of galectin-3’s N-terminus is its regulated phosphorylation on serine 6 [23], a residue conserved in each of the six mammalian galectin-3s which have been sequenced. This property may relate to regulated nuclear targeting of galectin-3, because in cultured fibroblasts phosphorylated galectin-3 is found in both the cytoplasm and the nucleus, but unphosphorylated galectin-3 is found exclusively in the nucleus [24]. There is as yet no evidence for such phosphorylation of other galectins, and there is no evidence for any other post-translational modification of galectins aside from N-terminal acetylation, a feature shared by almost all studied galectins. The number and sequence of the proline/glycine rich repeat units in galectin-3 varies across species (Table 3), but the significance of this is not yet clear. The mammalian galectin-3 repeat units have been classified into 7 distinct types (ILVII) [17]. These repeat units always are found in numerical order in a given mammalian galectin-3, but specific units may occur as multiple tandem repeats or may not occur at all. It is interesting to note that this classification can still be applied to the repeat

38.2 Gulrctin Structure

629

Table 1. Mammalian galectin and candidate genes. Sequencing of cosmids indicated with an asterisk is still in progress, so position numbering for the deduced exon boundaries may change with each update. Galectin

Gene

Message

Structure

1

22q13.1: gene 505303 cosmid Z83844* 49659-49650 50934-51010 52430-52603 53552-53695 22q13.1: gene M87860 cosmid AL0223 15* 86175-86169 78200-781 20 77006-76846 76682-76536 14q21.3: gene AF031421-5 sts G22378

common Unigene HS. 129924

1 CRD, dimer

intestine Unigene HS. 1 13987 ESTs: gall bladder, kidney

1 CRD, dimer

common Unigene HS.621

19ql3.1-13.3

intestine, esp. colon Unigene HS.5302 rat erythrocytes

1 CRD +N-term. repet. 2 CRDs

2

3

human not yet identified, rat gene L36862 human not yet identified, mouse gene tandem to gene for galectin-4 19q13.1 sts G38734 1q41-44 sts G22174 human not yet mapped mouse chr. 11. sts 236627 10

GRIFIN

AA3 11108

19q13.1: gene U68398 cosmid AC005393 17049-17035 14171-14095 13591Ll3381 10598-10473 7 cosmid AC004840* 118481 -1 18470 118100-118018 117926-117768 117407-117241 116973-116985 1 lq23 cosmid U73641 28867-28853

intestine

1 CRD, monomer 2 CRDs

stratified epithelia Unigene HS.99923 common Unigene HS.4082 lymphocytes, intestine specific isoform Unigene HS.81337 granulocytes Unigene HS.889

1 CRD, monomer 2 CRDs

rat lens

1 CRD, dimer

ESTs: breast, infant adrenal, Jurkat T-cell

1 CRD

2 CRDs

1 CRD, dimer

630

38 Galectins

Table 1 (continued) ~~

Galectin

H50946

N90645

N30757

R31311

A1138230

AC0055 15-11

N40740

AC000052

Gene 28206-28120 24394-24242 23714-23568 1 lq23 cosmid U73641 33498-33427 31401-31315 31181-30963 30179-30060 29969-29972 2P sts G30627 19q13.1 cosmid AC006133 ????- 8278 10331-10398 10913-11123 12932-13303 19q13.1 cosmid AC005205 14526-14541 16559-16633 17136-17348 19180-19293 19q13.1 cosmid AC0055 15 3464-3468 3925-4001 4502-47 12 6437-6559 19q13.1 cosmid AC0055 15 26891-26896 27476-27551 28040-28172 30073-30186 mouse gene U67985 19q13.1 cosmid AC005 176 22584-22658 23164-23374 cosmid AC0055 15 1866-1898 22ql1.2 cosmid AC000052* 41727-42113

Message

Structure

ESTs: spleen, gastric carcinoma, Jurkat T-cell

1 CRD

Unigene HS. I 14771 ESTs: fetal heart, fetal liver/spleen Unigene HS.24236 ESTs: placenta, aorta, embryo, fetal liver/spleen

1 CRD?

1 CRD

Unigene HS.23671 ESTs: placenta, fetal liver/spleen

1 CRD

Unigene HS.143557 EST A1 138230, placenta EST A1148582, placenta

1 CRD

1 CRD

Unigene HS. 146477 EST N40740, placenta EST AI128445, placenta

1 interrupted CRD

1 interrupted CRD

38.3 Novel Cundidute Gulectins

63 1

units in a chicken galectin [25] and even to a candidate galectin gene in zebrafish (A1384777).The only repeat unit type shared by all of these proteins is type I, which might indicate that galectin-3 evolved independently several times from a progenitor that included at least one copy of a type I proline/glycine-rich unit. Galectin-10, better known as the Charcot-Leyden crystal protein, is another unusual case, distinguished from other galectins by its lysophospholipase activity [26]. This protein is expressed at very high levels (1% of the soluble protein) and forms intracellular crystals in eosinophils. Even though the functional significance remains a mystery, the presence of distinctive hexagonal Charcot-Leyden crystals in tissues and exudates was long ago recognized and is still used as a marker for eosinophilassociated inflammation associated with allergic or parasitic diseases. Its function may be to detoxify lysophospholipids generated as byproducts during the production of inflammatory mediators at such sites. Galectin-10 deviates in its CRD sequence at four of the eight sugar-binding amino acid residues that are highly conserved in other galectins. So it is not surprising that galectin-10 seems to be only weakly active as a lectin [6, 271. It is, however, surprising that it is active as a lysophospholipase, because galectin-10 shows no structural similarity to other known lipases. Nevertheless, both lectin and enzyme activities have been confirmed for the recombinant protein [28]. No enzymatic activity has been reported for other galectins, and neither galectin-I nor galectin-3 showed detectable activity when tested in standard assays for lysophospholipases [ 1801. Therefore it seems likely that galectin- 10 has recently evolved as a phospholipase. Whether its lectin activity contributes to its physiological function remains unclear.

38.3 Novel Candidate Galectins A number of galectin-like sequences have been identified that cannot yet be considered as bona fide galectins, because they have either not yet been shown to bind galactoside sugars, or they have been found only in nucleic acid databases and not yet even been shown to be expressed as proteins. For instance, GRIFIN (galectin-:elated interfiber proteb), a protein apparently specific to the lens of the eye, is clearly closely related to the galectins in sequence, but does not meet the galectin definition, because it appears to lack lectin activity in assays used for other galectins [29]. However, it should be remembered that an apparent lack of lectin activity resulted in the initial exclusion of Charcot-Leyden crystal protein from the galectin family, and it could only be named galectin-I0 after it was later shown to weakly bind to galactoside affinity columns [6, 271. A quite different example is the candidate galectin discovered in an adenovirus [30]; Genbank U25120. Here a galectin-like sequence of the tandem-repeat type is part of the capsid fiber protein. The structure of this particular fiber protein is like those of other adenoviruses, except for inclusion of the galectin-like sequence as part of the head region, the portion believed to mediate high affinity binding to host cells. Therefore, it seems likely that this candidate galectin contributes galactosidespecific targeting to the process of infection by this virus.

Gal-3 ?

Sponge Geodia cydonium I I1

Insect Anopheles gambiae Drosophila melanogaster Worm Caenorhabditis elegans 16 32

Bird Gallus gallus C-14 C-16 Gal-3 ?

Fugu rubripes

Amphibian Xenopus laevis Bufo arenarum Fish Electrophorus electricus Conger myriater Con I Con I1 Danio rerio Gal-3 ?

Galectin

1 CRD, multimer 1 CRD, multimer

1 CRD, dimer 2 CRDs

063575; cosmid Y55B1 on chr.111-28-26 AB000802; cosmid 282081, chr. I1 7753-7767, alt. splices = W09H1.6a,6b = yk148a7.5, yk469b7.5

X93925 X70849

1 CRD 1 CRD

269982 *LP06039.5’

D00308-11, M 11674 M57240 *U50339

1 CRD, monomer 1 CRD, dimer 1 CRD + N-terminal Pro/Gly rich repeat

1 CRD, dimer 1 CRD, dimer 1 CRD, dimer 1 CRD 1 CRD + N-terminal Pro/Gly rich repeat 1 CRD I CRD 1 CRD + N-terminal Pro/Gly rich repeat

A28302 AB010276 AB010277 *G47571 *A1384777

*FRa073apsB 12 *FRa075apsH3 *AL020875: 199D14, 11 1H07

1 CRD, dimer 1 CRD, dimer

Structure

M88105 P562 17

Gene or Message

Table 2. Non-mammalian galectin and candidate genes. Asterisks mark candidate galectins (not yet shown to bind galactosides). Not included are numerous candidate nematode galectins (C. elegans and others).

Q

LJ

h,

38.3 Novel Candidate Gulectins

633

Species

I

I1 111

IV V = Ib

VI = Ilb

VII = IIIb

Table 3. Organization of N-terminal proline/glycine rich repeat units in mammalian galectin-3s, a similar chicken galectin, and a candidate zebrafish galectin using the classification scheme of Herrmann, et al. [ 171.

a w

38.4 Unorthodox Subcellular Targeting

635

Using search algorithms based on the structure of known galectins to screen Genbank databases, we have identified seven novel mammalian candidates for membership in the galectin family (Figure 1: only the exon-I1 encoded CRD domains are shown, but there is also considerable sequence similarity in other parts of each protein). All but one of these sequences (ACOO5515-11) appear not only in human genomic DNA, but also in expressed messages (Table I), proving that they are not pseudogenes. We have established a web address (URL: http://www.sacs.ucsf.edu/ home/cooper/galectins.htm) giving further documentation for each of these candidate galectins, including complete deduced protein sequence. Based on sequence comparison it seems quite likely that most of these galectin-like sequences have galactoside-binding activity, but one (N90645) lacks the tryptophan residue otherwise conserved in all established galectins. Four of the candidate galectins (N30757, R3131 l, AI138230, and AC00.5515-11) are very similar in sequence to galectin-10, raising the possibility that they too have lysophospholipase activity. These genes are also located close to the galectin-I0 gene on chromosome 19q13.1, suggesting that all are derived from an ancestral gene by tandem duplication. Two additional sequences also resemble galectin- 10, but appear to have stop codons interrupting the CRD. One of these appears to be expressed (EST = N40740), but no cDNA sequence has yet been recorded for the other gene. Candidate galectins are also apparent in the genomes of important model organisms. In the worm, C. elegans, two galectins have been isolated and shown to bind galactosides [31, 321, but another 26 candidate galectin genes (most also recorded as ESTs, not shown) are evident in the C. elegans genome. Thus, it is clear in this case, where the entire genome has been sequenced and estimated to contain 20,000 genes, that galectins are very highly represented. Similar galectins have been described in a number of other nematode species [33-351. Candidate galectin genes are also apparent in the fractional sequences so far obtained for genomes of other model organisms (Table 2), including Drosophila (LP06039), zebrafish (A1384777 and G47571), and Arahidopsis (AC000348, T7N9.14). The galectin-like sequence in Arabidopsis represents the first evidence for galectins in plants, where the whole class of lectin proteins was first discovered. This indicates that galectin family is evolutionarily even older than previously thought.

38.4 Unorthodox Subcellular Targeting Although galectins are frequently found on cell surfaces or in extracelluJar matrix, none have a secretion signal peptide with the possible exception of one sponge galectin [ 361. Instead, galectins have characteristics typical of cytoplasmic proteins, such as an acetylation N-terminus, free sulfhydryls, and lack of glycosylation. Indeed, inside cells galectins appear to be free in the cytosol and not compartmentalized in classical secretory compartments. Nevertheless, there is good evidence for several galectins that they are indeed secreted, albeit by novel non-classical mecha-

636

38 Gulectins

nisms. In this regard galectins belong to a small category of proteins [37] which must be secreted by mechanisms distinct from classical vesicle-mediated exocytosis. One possible mechanism is that specific transport proteins, like those that export some bacterial toxins, transport galectins directly across the plasma membrane. This could be the basis for export of rat galectin-1 expressed as a recombinant protein in yeast cells [38]. A specific yeast transmembrane protein was found to be required for secretion of transfected galectin-1. Whether similar proteins are involved in secretion of endogenous galectin-1 from other cells is still unknown. However, the yeast protein appears to belong to a family of proteins, the transmembrane-4-superfamily, which includes many mammalian representatives. A second possibility is that animal cells secrete galectins by a novel apocrine mechanism, as suggested for galectin-1 export from developing muscle, nerve and erythrocytes [39-421 and galectin-3 export from macrophages [43]. Prior to secretion, the galectins appear to become specifically concentrated under the plasma membrane and in plasma membrane evaginations which appear to pinch off to form galectin-rich extracellular vesicles. These vesicles presumably break to release stored galectin extracellularly in a process similar to holocrine release of a galectin from frog skin [44]. While there is some evidence that such a process might occur in vivo (451, there is still concern that this pathway might reflect an artifact of cell culture, perhaps even an aspect of cell blebbing during apoptosis. In any case, the striking concentration of galectin under the plasma membrane indicates a specificity to the process that is likely to be biologically meaningful. Besides their unorthodox secretion, several other distinctive aspects of galectin subcellular targeting have been described. Several galectins have been immunohistochemically localized not just in the cytoplasm, but also in unexpected cellular locations, such as in nuclei or in discrete submembranous regions. For instance, both galectin-1 and galectin-3 have been localized in nuclear speckles (46-521 where they have been implicated in mRNA splicing [53, 541. In osteoblasts, preferential retention of galectin-1 in the nucleus has been linked to differentiation [ 5 5 ] . In fibroblasts, nuclear accumulation of galectin-3 has been linked to cell proliferation and, in particular, the S-phase of the cell cycle [24, 48, 501. In differentiated colon epithelial cells, galectin-3 appears concentrated in the nuclei, and loss of this nuclear targeting has been associated with progression of colon carcinomas [56, 571. There is an intriguing similarity here to the behavior described for certain growth factors, such as FGF and IL-1, which also show both unorthodox secretion and regulated nuclear targeting [58]. Galectin-4 provides another example of unusual subcellular distribution. It was first localized as an adherens junction component in pig oral epithelium (591. In intestinal epithelial cells, galectin-4 appears to be enriched apically at the brush border [60], whereas in cultured colon adenocarcinoma cells it appears to be concentrated basally in a pattern clearly distinct from apical galectin-3 [61]. On the other hand, galectin-3 shifts from an apical to basolateral distribution when MDCK epithelial cells are shifted from culture on plates to three-dimensional matrices [62]. These results indicate that the subcellular distributions of different galectins can be independently regulated in response to the cellular environment.

38.5 Regulation of Gulectin Expression

637

38.5 Regulation of Galectin Expression Another outstanding feature of galectins is the marked developmental and physiological regulation of their expression, and in a number of cases the very high levels of expression that are achieved (even 1% or more of the soluble protein). While some galectins, such as galectin-1 and galectin-3, are expressed in many tissues and cell types, others, such as galectin -2 and galectin-7, are restricted to specific tissues (Table 4). But even the more widely expressed galectins show quite distinct expression patterns, e.g. [63] with marked developmental and physiological regulation. For instance, expression of galectin-1 is normally quite low in T lymphocytes, but rises to high levels when these cells are activated [64]. Expression of galectin-3 is similarly induced upon activation of macrophages [65] and Schwann cells [66]. Marked changes in the expression of specific galectins have also been noted in many cancers. In fact, galectins have often been identified as differentiation or tumor antigens and, as discussed further below, there is evidence that changes in galectin expression might contribute directly to cancer progression. The molecular mechanisms regulating galectin gene expression are still unclear and have so far been primarily studied with regard to galectin-1 and galectin-3. Upstream elements in these genes include sequences that could account for observed regulation by glucocorticoids, retinoids, and other factors. For instance, phorbol esters promote expression of both genes, and both genes include consensus binding sites for the phorbol sensitive transcription regulators AP- 1 and/or AP-2 [67-691. Both genes also include binding sites for the more general transcription factor Spl, which, at least for galectin-1, seems to account for induction by butyrate [70]. The galectin-1 gene also includes a consensus glucocorticoid response element [71] that presumably accounts for its induction by steroids [72, 731. The galectin-3 gene includes several serum response elements that presumably account for its induction by serum [68]. Galectin gene sequences have also been identified that could account for tissue specific expression. The galectin-3 promoter includes a putative ApoE sequence known to act as an enhancer during monocyte differentiation [67]. The galectin-10 gene includes several sequences described in other myeloid-specific promoters [74]. The GRIFIN gene has several large direct repeats with multiple consensus binding sites for 6EF1 [29], a transcription factor that has been implicated in lens specific expression. In addition to regulation by such transcription factors, promoter methylation also seems to be an important mechanism regulating expression of galectin- 1 [75, 761. Transcription factor binding sites have also been identified that could contribute to cancer related changes in galectin expression. For instance, loss of function of the p53 transcription factor is a major factor in many cancers, and the galectin-3 gene includes a consensus binding site for p53 [67, 771. In fact, transfection with p53 drastically inhibits galectin-3 expression. In contrast, p53 greatly enhances transcription of galectin-7 [781. While the purpose of galectin gene regulation by p53 remains unclear, it could be an important feature of tumor biology.

most abundant in muscle (smooth, skeletal, and cardiac), peripheral nerve, motor, sensory and olfactory neurons

epithelial cells of intestine and stomach [ 1751 also ESTs from gall bladder, kidney most abundant in many epithelial cells, activated macrophages, and osteoblasts

stomach, intestine (esp. colon) [ 131 rat erythrocytes [ 121 stomach, intestine [ 131 stratified epithelia [ 1761 most abundant in liver, kidney, muscle, lung, and brain [ 1771 lymphocytes, intestine (specific isoform) [14, 119, 1781

eosinophils and basophils [26]

1

2

4 5 6 7 8 9

10

3

Known Sites of Expression

Galectin

Table 4. Major sites of expression and biological activities for mammalian galectins.

stimulation of T-cell apoptosis chemoattraction of eosinophils transport of urate [ 1791 lysophospholipase

stimulation of cell adhesion to laminin stimulation of neurite outgrowth stimulation of cell proliferation stimulation of leukocyte activation inhibition of apoptosis participant in mRNA splicing binding of advanced glycosylation end products knockout mouse: subtle change in neutrophil targeting and apoptosis of some other cells

stimulation or inhibition of cell adhesion to laminin or fibronectin stimulation of neurite outgrowth stimulation or inhibition of cell proliferation stimulation of cell activation stimulation of T-cell apoptosis participant in mRNA splicing knockout mouse: subtle change in targeting of a subfamily of olfactory neurons

Known Biological Activities

6 2 1 2'

00

bJ

00

w

(s\

38.6 Gulectin Binding SpeciJcity and IdentiJed Ligands

639

38.6 Galectin Binding Specificity and Identified Ligands By definition all galectins bind galactosides [3], but those studied to date all show much higher affinity for type I (GalPl-3GlcNAc) or type I1 (GalPl-4GlcNAc) N-acetyllactosamine structures, reviewed in 1791. Oligosaccharides with multiple N-acetyllactosamine units, either as linear polylactosamine chains or as branched complex glycans, are recognized with even higher affinity. In general, terminal sialylation on the 3’-OH (on Gal) has little effect, but terminal sialylation on the 6’OH greatly decreases galectin affinity. However, individual galectins show significant differences in specificity for further modifications of the core lactosamine structure. Even individual domains of the tandem type galectins can show considerable differences in fine specificity [SO, 811. A major distinction amongst galectins in binding affinity occurs for lactose or lactosamine substituted at the 2’ or 3’-OH (on Gal), reviewed in [79]. As a result, is recognized the blood group A core structure (GalNAcal-3[Fucal-2]Gal~1-4Glc) with much higher affinity than lactose by some galectins, such as galectin-3, but with much lower affinity than lactose by other galectins, such as galectin-1. A sponge galectin shows even higher affinity for the Forsmann structure (GalNAcal3GalNAc~1-3Galcll-4Gal~l-4Glc) [82]. These differences in fine specificity suggest that different galectins have evolved to interact with specific complex carbohydrate structures. The complex carbohydrate structures preferentially bound by galectins are found in extracellular or cell surface glycoconjugates. A number of specific glycoconjugate ligands that might interact functionally with galectins have been identified, primarily by analyzing cell or tissue extracts using standard affinity techniques. Galectin specificity is underlined by the fact that in each of these studies just one or a few ligands stood out. For instance, major ligands for galectin-1 have been identified in skeletal muscle as laminin 1831, in cultured CHO fibroblasts and ovarian carcinoma cells as two membrane glycoproteins normally associated with lysosomes (LAMP-1 and LAMP-2) [84, 851, in olfactory neurons as laminin and lactosamine-containing glycolipids [86], in neuroblastoma cells as the ganglioside GM, [87], in T lymphocytes as CD43 and CD45 [88], and in smooth muscle as integrins [89, 901. Major glycoconjugate ligands identified for galectin-3 also include laminin and LAMPS 191-931, as well as the high affinity IgE receptor [94], certain glycoforms of IgE (951, a 90 kDa protein that has been named Mac-2-binding protein (Mac-2-BP) [92, 96, 971, and possibly certain non-enzymatic glycosylation products [98]. A controversial suggestion has also been raised that some cytokeratins are modified with teminal a1-3GalNAc in a novel type of cytoplasmic glycosylation recognized by some galectins [99].Lectin binding to cytokeratins from a variety of cultured cell lines was saturable, sensitive to mild periodate oxidation, inhibitable by GalNAc, and abolished after treatment of the cytokeratins with a-N-acetylgalactosaminidase. Furthermore, GalNAc residues were detected in highly purified cytokeratin preparatins using gas chromatography/mass spectrometry. These results challenge the standard dogma and raise the novel possibility that galectins might interact functionally with specific cytoplasmic glycoproteins.

640

38 Galectins

38.7 Physiological Functions Although identification of the potential physiological substrates described above immediately suggests a number of physiological functions for galectins, in no case have such interactions been unequivocally demonstrated to occur in vivo. In fact, despite extensive genetic and biochemical understanding of galectins, their physiological functions have been difficult to define. This has not been due to a lack of study. Indeed, abundant evidence has been accumulated showing that members of this family can interact with glycoconjugates on or around cells and influence a wide range of cell activities (Table 4). When taken together the results suggest that different galectins can have very different functions, and that even individual galectins can perform a variety of functions depending on their environment and available ligands. Indeed, the presence of galectins in so many evolutionarily divergent species suggests that they play fundamental roles in cell biology, while the presence of multiple galectins within a single species suggests that they have evolved to participate in a variety of more specific functions. It is notable that most galectins are functionally at least divalent, either as dimers or multimers or by tandem repeat of galectin domains. This suggests that an important aspect of galectin function is to crosslink or bridge appropriate ligands. The uncommon univalent galectins might then function to competitively inhibit such interactions. The striking tissue protein abundance that is achieved by some galectins suggests that they are likely to play a structural role, for instance in mediating cell-cell or cell-matrix crosslinking. Indeed, many reports have described effects of added galectin on cell adhesion, spreading, and migration through specific interactions with extracellular matrix glycoproteins, such as laminin 183, 86, 89, 1001041, fibronectin [89, 100, 1051, and Mac-2-BP [106], and cell surface glycoproteins, such as integrins [89, 1071 and LAMPS [108, 1091. In some cases galectin effects are attributed to crosslinking these matrix and cell surface proteins, but in other cases they are attributed to interactions just with matrix or just with cell surface ligands. In the tandem-repeat galectins, distinct specificity of each carbohydrate binding domain may promote crosslinking of distinct glycoconjugates. Indeed, when applied to epithelial tissue sections, the individual domains of galectin-4 bind to distinct areas at intercellular borders, with the isolated N-terminal domain appearing to bind to material in the intercellular spaces and the isolated C-terminal domain appearing to bind to the lateral cell membranes [ 1 101. This suggests that galectin-4 might function in mediating lateral cell interactions between intestinal epithelial cells by bridging membrane glycoconjugates and intercellular glycoconjugates. On the other hand, the ability of galectins to crosslink cell surface receptors suggests that they could function in transducing cellular signals. Indeed, many reports have described effects of added galectins on apoptosis [88, 111-1201, proliferation [121-126], chemotaxis [127], and immune regulation [64, 95, 128-1321, though the relevant receptors remain largely undefined. In a few cases binding to these receptors even seems to be carbohydrate independent [ 114, 125, 1271, because the effects of added galectin were not inhibited by lactose. Many of these studies have focused attention on possible galectin functions in regulating immune responses, reviewed

38.7 Physiological Functions

641

in [ 1331. Such involvement could be evolutionarily old, because a galectin has also been implicated in the immune response of a mosquito, Anopheles gambiae [134, 1351. Even a single galectin can apparently affect cells in a variety of ways depending on the cell type and circumstances. For instance, galectin-1 can either stimulate or inhibit cell adhesion to laminin [83, 86, 1021 and can either stimulate or inhibit cell proliferation 164, 114, 124--1261. While such diversity of galectin effects on cells might at first seem surprising, it is actually quite typical for extracellular matrix proteins to have both structural and regulatory roles that can vary depending on cell type and circumstance. There is also evidence that galectins can simultaneously have distinct intracellular and extracellular functions. For instance, both galectin- 1 and galectin-3 have been implicated as nuclear components important for pre-mRNA splicing in cell-free assays [53, 541. Galectin-3 has also been suggested to exert an intracellular function as an inhibitor of apoptosis [ 113, 1181, perhaps attributable to its lactose-inhibitable interaction with Bcl-2, a well known suppressor of apoptosis. For such intracellular functions, it is possible that the galectin carbohydrate binding site interacts not with sugar but with some structurally similar feature of specific protein sequences, as suggested for galectin-3 binding to Bcl-2 [113, 1181. A powerful new approach to study galectin function is to knock-out expression of individual galectin genes. Such mice engineered to lack galectin-1 or galectin-3 are fully viable, fertile, and on gross examination appear to be normal. However, closer study of these mice has so far revealed that the mice lacking galectin-1 have intriguing deficits in olfactory axon pathfinding [ 1371, and the mice lacking galectin3 have abnormalities in neutrophil accumulation during inflammation [ 1381. Although this approach can fail to detect normal biological functions of the missing protein, apparently because many functions can be performed by alternative or redundant systems, these initial positive results are very encouraging for further analysis of these mice and others engineered to eliminate additional galectin family members. Such altered galectin expression might even play an important role in certain diseases. In particular, galectins have been implicated in tumorigenesis, a complex multi-step process involving changes in cell proliferation, apoptosis, adhesion, and immune surveillance. To metastasize, malignant cells migrate away from the initial tumor into the circulation and then adhere to endothelial cells at distant sites where they proliferate to form new tumors. There is evidence that altered galectin expression could significantly influence each of these steps. There is an extensive literature reporting altered expression of galectins in various tumors [56, 100, 103, 138-1661 and effects on tumor proliferation [ 1211 and metastasis [ 167-1701. Both galectin-1 and galectin-3 have been reported to be expressed on some endothelial cells [ 171, 1721 where they could participate in metastatic arrest of blood-born tumor cells expressing potential ligands. Furthermore, malignant transformation has been associated with increased polylactosamine addition and cell surface expression of LAMPS on tumor cells, which would increase surface binding sites for galectins, e.g. [173, 1741. Given the multiple normal functions postulated for galectins, it is even possible that altered subcellular targeting of a galectin might alter its function in

642

38 Gulectins

a given cell type. For example, nuclear targeting of galectin-3 seems to be lost with progression of colon carcinomas [56, 571. Therefore, regardless of the normal functions of galectins, their altered expression in many tumors could make them useful not only as diagnostic markers, but even as targets for cancer therapy.

38.8 Summary In the quarter century since galectins were first discovered, a great deal has been learned about these galactoside-binding proteins. Ten have been discovered in mammalian tissue extracts, while another eight candidate galectins are apparent in human gene databases. Many more have been found in other species. All galectins share certain features, including a similar basic structure and conserved amino acid residues in the portions of the protein that participate in carbohydrate binding. A remarkable feature of galectins is that all are synthesized as cytoplasmic proteins, yet studies of their distribution in vivo or in culture indicate that a fraction of cellular galectin is on the cell surface or in an extracellular location. Determining the mechanisms used to translocate these cytoplasmic proteins to the cell exterior, where their glycoconjugate ligands reside, is one of the major challenges confronting the field. Another major challenge is determining the physiological functions of this large and widely distributed family of proteins. Although added galectins have been shown to have many effects on assays of cell behavior in culture, it has been difficult to demonstrate functions in vivo. This problem may finally be overcome with the advent of mice engineered to delete expression of specific galectin genes. However, it is already clear that phenotypes in such mice can be very subtle, requiring careful and clever analysis to be revealed. A major aid to such study would come from better identification of the ligands or receptors which physiologically interact with galectins. The considerable knowledge already available about galectin biochemistry, genetics and regulation should greatly facilitate such investigations. References 1. D. N. W. Cooper, S. H. Barondes, Glycobioloyy, 1999, in press. 2. K. I. Kasai, J. Hirabayashi, J. Biochem., 1996, 119, 1-8. 3. S. H. Barondes, V. Castronovo, D. N. Cooper, R. D. Cummings, K. Drickamer, T. Feizi, M. A. Gitt, J. Hirabayashi, C. Hughes; K. Kasai, et al., Cell, 1994, 76, 597--598. 4. D. D. Leonidas, E. H. Vatzaki, H. Vorum, J. E. Celis, P. Madsen, K. R. Acharya, Biochemistry, 1998, 37, 13930-13940. 5. Y. D. Lobsanov, M. A. Gitt, H. Leffler, S. H. Barondes, J. M. Rini, J. Biol. Chem., 1993,268, 27034-27038. 6. D. D. Leonidas, B. L. Elbert, Z. Zhou, H. Leffler, S. J. Ackerman, K. R. Acharya, Structure, 1995,3, 1379-1393. 7. J. Seetharaman, A. Kanigsberg, R. Slaaby, H. Leffler, S. H. Barondes, J. M. Rini, J. Bid. Chem., 1998, 273, 13047--13052.

References

643

8. D. I. Liao, G. Kapadia, H. Ahmed, G. R. Vasta, 0. Herzberg, Proc. Nut1 Acad. Sci. U S A ; 1994, 91, 1428-1432. 9. Y. Bourne, B. Bolgiano, M. P. Nesa, P. Penfold, D. Johnson, T. Feizi, C. Cambillau, J. Mol. Biol., 1994, 235, 787-789. 10. Y. Bourne, B. Bolgiano, D. I. Liao, G. Strecker, P. Cantau, 0. Herzberg, T. Feizi, C. Cambillau, Nature Struct. Biol., 1994, I , 863-870. 11. 9. Diehl-Seifert, G. Uhlenbruck, M. Geisert, R. K. Zahn, W. E. Miiller, EMU.J. Biochem., 1985, 147, 517-523. 12. M. A. Gitt, M. F. Wiser, H. Leffler, J. Herrmann, Y. R. Xia, S. M. Massa, D. N. W. Cooper, A. J . Lusis, S. H. Barondes, J. Biol. Chem., 1995, 270, 5032-5038. 13. M. A. Gitt, C. Colnot, F. Poirier, K. J. Nani, S. H. Barondes, H. Leffler, J. Biol. Chrm., 1998, 273, 2954-2960. 14. J. Wada, Y. S. Kanwar, J. Biol. Chrm., 1997, 272, 6078-6086. 15. S. M. Massa, D. N. W. Cooper, H. Leffler, S. H. Barondes, Biochemistry, 1993, 32, 260267. 16. N. Agrwal, Q. Sun, S. Y . Wang, J. L. Wang, J. Bid. Cliem., 1993,268, 14932-14939. 17. J. Herrmann, C. W. Turck, R . E. Atchison, M. E. Huflejt, L. Poulter, M. A. Gitt, A. L. Burlingame, S. H. Barondes, H. Leffler. J. Bid. Chem., 1993,268, 26704-267 11. 18. B. Mehul, S. Bawumia, S. R. Martin, R. C. Hughes, J. Biol. Chem., 1994, 269, 18250-18258. 19. J. Ochieng, B. Green, S. Evans, 0. James, P. Warfield, Biochim. Biophys Acta, 1998, J379, 97-106. 20. S. Kuklinski, R. Probstmeier, J. Neurochem., 1998, 70, 814-823. 21. R. Y. Yang, P. N. Hill, D. K. Hsu, F. T. Liu, Biochemistry, 1998, 37, 4086-4092. 22. B. Mehul, S. Bawumia, R. C. Hughes, FEBS Lett., 1995, 360, 160-164. 23. M. E. Huflejt, C. W. Turck, R. Lindstedt, S. H. Barondes, H. Leffler, J. Bid. Chem., 1993, 268, 2671 2-267 18. 24. E. A. Cowles, N. Agrwal, R. L. Anderson, J. L. Wang, J. Biol. Cliem., 1990, 265, 1770617712. 25. M. Nurminskaya, T. F. Linsenmayer, Deael. Dqwam., 1996, 206, 260-271. 26. S. J. Ackerman, S. E. Corrette, H. F. Rosenberg, J. C. Bennett, D. M. Mastrianni, A. Nicholson-Weller, P. F. Weller, D. T. Chin, D. G. Tenen, J. Iinmunol., 1993, 150, 456-468. 27. K. D. Dyer, H. F. Rosenberg, Life Sci., 1996, 58, 2073-2082. 28. Z. Zhou, D. G. Tenen, A. M. Dvorak. S. J. Ackerman, J. Leuko. B i d , 1992, 52, 588-595. 29. A. T . Ogden, 1. Nunes, K. KO, S. Wu. C. S. Hines, A.-F. Wang, R. S. Hegde, R. A. Lang, J. B i d . Clirin., 1998, 273, 28889-28896. 30. S. 9. Kleiboeker, Virus Rex. 1995, 3Y, 299-309. 31. Y . Arata, J. Hirabayashi, K. I. Kasai, J. Biol. Chern., 1997, 272, 26669-26677. 32. J. Hirabayashi, T. Ubukata, K. I. Kasai, J. Biol. Clzem., 1996,271, 2497-2505. 33. S. E. Newton, J. R. Monti, C . J. Greenhalgh, K. Ashman, E. N. T. Meeusen, Mol. Bioclzem. Parusit., 1997, 86, 143-153. 34. C. J. Greenhalgh, S. E. Newton, Profein Espr. Purg. 1999, 16, 152-159. 35. C. J. Creenhalgh, S. A. Beckham, S. E. Newton, Mol. Bioclzem. Parusit., 1999, 98, 285-289. 36. C. Wagner-Huelsmann, N. Bachinski, 9. Diehl-Seifert, B. Blurnbach, R. Steffen, Z. Pancer, W. E. G. Mueller, Glycohiology, 1996, 6 . 785-793. 37. K. Kuchler, Trendy Cell B i d , 1993, 3, 421 -426. 38. A. E. Cleves, D. N. W. Cooper, S. H. Barondes, R. 9 . Kelly, J. Cell Biol., 1996, 133, 10171026. 39. D. N. Cooper, S. H. Barondes, J. Cell Biol., 1990, 110, 1681-1691. 40. F. L. Harrison, T. J. Wilson, J. CrllSci., 1992, 101, 635-646. 41. V. Avellana-Adalid, G. Rebel, M. Caron, J. D. Cornillot, D. Bladier, R. Joubert-Caron, Glycoconj. J., 1994, 11, 286&291. 42. D. Lutomski, M. Fouillit, P. Bourin, D. Mellottee, N. Denize, M. Pontet, D. Bladier, M. Caron, R. Joubert-Caron, Glycohioloqy, 1997. 7, 1193-1199. 43. 9 . Mehul, R. C. Hughes, J. Cell Sci., 1997, JIO, 1169-1178. 44. N. C. Bols, M. M. Roberson, P. L. Haywood-Reid, R. F. Cerra, S. H. Barondes, J. Cell Biol., 1986, 102,492-499.

644

38 Galectins

45 S. Kuchler, R. Joubert, V. Avellana-Adalid, M. Caron, D. Bladier, G. Vincendon, J. P. Zanetta, Dev. Neurosci., 1989, 11, 414-421. 46. A. P. Seve, M. Felin, M. A. Doyennette-Moyne, T. Sahraoui, M. Aubery, J. Hubert, Glycobiology, 1993, 3, 23-30. 47 A. Vyakarnam, A. J. Lenneman, K. M. Lakkides, R. J. Patterson, J. L. Wang, Exp. Cell Res., 1998,242,419-428. 48 J. L. Wang, E. A. Werner, J. G. Laing, R. J. Patterson, Biochem. Soc. Trans., 1992, 20, 269274. 49 L. Wang, H. Inohara, K. J. Pienta, A. Raz, Biochenz. Biophys. Res. Commun., 1995,217, 292303. 50 M. Hubert, S. Y. Wang, J. L. Wang, A. P. Seve, J. Hubert, Exp. Cell Rex, 1995, 220, 397406. 51 M. Cuperlovic, M. Jankovic, K. Pfeifer, W. E. G. Mueller, Cell. Physiol. Biochem., 1995, 5, 33-44. 52 S. S. Craig, P. Krishnaswamy, A. M. A. Irani, C. L. Kepley, F. T. Liu, L. B. Schwartz, Anat. Rec., 1995,242, 211-219. 53 S. F. Dagher, J. L. Wang, R. J. Patterson, Proc. Nut1 Acad. Sci. U S A , 1995, 92, 1213-1217. 54 A. Vyakarnam, S. F. Dagher, J. L. Wang, R. J. Patterson, Mol. Cell Biol., 1997, 17, 47304737. 55 J.-Y. Choi, A. J. Van Wijnen, F. Aslam, J. D. Leszyk, J. L. Stein, G. S. Stein, J. B. Lian, S. Penman, J. Cell Sci., 1998, 111, 3035-3043. 56 H. L. Schoeppner, A. Raz, S. B. Ho, R. S. Bresalier, Cuncer, 1995, 75, 2818-2826. 57 M. M. Lotz, C. W. Andrews, Jr., C. A. Korzelius, E. C. Lee, G. D. Steele, Jr., A. Clarke, A. M. Mercurio, Proc. Nritl Acad. Sci, U S A , 1993, 90, 3466-3470. 58 N. R. Smalheiser, Mol. Biol. Cell, 1996, 7, 1003-1014. 59 M. L. Chiu, D. A. Parry, S. R. Feldman, D. G. Mapper, E. J. O’Keefe, J. Biol. Chem., 1994: 269, 31770-31776. 60 E. M. Danielsen, B. Van Deurs, Mol. Biol. Cell, 1997, 8, 2241-2251. 61 M. E. Huflejt, E. T. Jordan, M. A. Gitt, S. H. Barondes, H. Leffler, J. Biol. Chem., 1997,272, 14294-14303. 62 Q. Bao, R. C. Hughes, J. CellSci., 1995, 108, 2791-2800. 63 C. Colnot, M. A. Ripoche, F. Scaerou, D. Fowlis, F. Poirier, Biochem. Soc. Trans., 1996, 24, 141- 146. 64 C. Blaser, M. Kaufmann, C. Mueller, C. Zimmermann, V. Wells, L. Mallucci, H. Pirche, Eur. J. Zmmunol., 1998, 28, 2311-2319. 65 F. T. Liu, D. K. Hsu, R. I. Zuberi, I. Kuwabara, E. Y. Chi, W. R. Henderson, Jr., Am. J. Path., 1995,147, 1016-1028. 66 H. Be’eri, F. Reichert, A. Saada, S. Rotshenker, Eur. J. Neurosci., 1998, 10, 2707-2713. 67 J. C. Gaudin, C. Arar, M. Monsigny, A. Legrand, Glycobiology, 1997, 7, 1089-1098. 68 M. M. Kadrofske, K. P. Openo, J. L. Wang, Arch. Biochem. Biophys., 1998,349, 7-20. 69 P. Salvatore, C. Contursi, G. Benvenuto, C. B. Bruni, L. Chiariotti, FEBS Lett., 1995. 373, 159-163. 70 Y. Lu, R. Lotan, Biochim. Biophys. Acta, 1999, 1444, 85--91. 71 M. A. Gitt, S. H. Barondes, Biochemistry, 1991, 30, 82-89. 72 Y. S. Choe, C. Shim, D. Choi, C. S. Lee, K. K. Lee, K. Kim, Mol. Repro. Deuel., 1997, 48, 261-266. 73. G. L. Sanford, M. A. Owens; B. M. Odusanya, Exp. Lung Rex, 1993, 19, 91-104. 74. H. I. Gomolin, Y. Yamaguchi, A. V. Paulpillai, L. A. Dvorak, S. J. Ackerman, D. G. Tenen, Blood, 1993,82, 1868-1874. 75. P. Salvatore, G. Benvenuto, M. Caporaso, C. B. Bruni, L. Chiariotti, FEBS Lett., 1998, 421, 152-1 58. 76. G . Benvenuto, M. L. Carpentieri, P. Salvatore, L. Cindolo, C. B. Bruni, L. Chiariotti, Mol. Cell Biol., 1996, 16, 2736-2743. 77. J. Raimond, F. Rouleux, M. Monsigny, A. Legrand, FEBS Lett., 1995, 363, 165-169. 78. K. Polyak, Y. Xia, J. L. Zweier, K. W. Kinzler, B. Vogelstein, Nature, 1997, 389, 300-305. 79. H. Ahmed, G. R. Vasta, Glycobiology, 1994, 4 , 545-548.

References

645

80. Y. Arata, J. Hirabayashi, K. 1. Kasdi, J. Biochem., 1997, 121, 1002-1009. 81. Y. Oda, J. Herrmann, M. A. Gitt, C . W. Turck, A. L. Burlingame, S. H. Barondes, H. Leffler, J. Bid. Chem., 1993,268, 5929-5939. 82. F. G. Hanisch, S. E. Baldus, T. A. Kuemmel, Glycobioloyy, 1996, 6, 321-336. 83. D. N. Cooper, S. M. Massa, S. H. Barondes, J. Cell B i d , 1991, I15, 1437-1448. 84. K. Y. Do, D. F. Smith, R . D . Cummings, Biochem. Biophys. Rex Commun., 1990, 173, 11231128. 85. D. M. Skrincosky, H. J. Allen, R. J. Bernacki, Cuncer Res., 1993,53, 2667-2675. 86. N. K. Mahanthappa, D. N. W. Cooper, S. H. Barondes, G. A. Schwarting, Development, 1994,120, 1373-1384. 87. J. Kopitz, C. Von Reitzenstein, M. Burchert, M. Cantz, H. J. Gabius, J. Bid. Chem., 1998, 273, 11205-1 1211. 88. N. L. Perillo, K. E. Pace, J. J. Seilhamer, L. G. Baum, Nature, 1995, 378, 736-739. 89. E. P. Moiseeva, E. L. Spring, J. H. Baron, D. P. de Bono, J. Vusc. Res., 1999, 36, 47758. 90. V. S. Cannon, D. N. W. Cooper, E. M. Farrell, R. Clyman, FASEB J., 1995, 9, A1502. 91. D . W . Ohannesian, D. Lotan, P. Thomas, 3. M. Jessup, M. Fukuda, H. J. Gabius, R. Lotan, Cuncer Rex, 1995, 55, 2191-2199. 92. H. Inohara, A. Raz, Biochem. Biophys. Res. Commun., 1994, 201, 1366-1375. 93. S. Dong, R. C. Hughes, Glycoconj. J., 1997, 14, 267-274. 94. L. G. Frigeri, R. I. Zuberi, F. T. Liu, Biochemistry, 1993, 32, 7644-7649. 95. A. Wollenberg, H. De la Salle, D. Hanau, F. T. Liu, T. Bieber, J. Exp. Med., 1993, 178, 777785. 96. I. Rosenberg, B. J . Cherayil, K. J. Isselbacher, S. Pillai, J. Biol. Chem., 1991, 266, 1873118736. 97. K. Koths, E. Taylor, R. Halenbeck, C. Casipit, A. Wang, J. Bid. Chem., 1993, 268, 1424514249. 98. H. Vlassara, Y. M. Li, F. Imani, D. Wojciechowicz, Z. Yang, F. T. Liu, A. Cerami, Mol. Med., 1995, I , 634-646. 99. S. Goletz, F. G. Hanisch, U. Karsten, J. Cell Sci., 1997, 110, 1585-1596. 100. J. Ellerhorst, T. Nguyen, D. N. W. Cooper, D. Lotan, R. Lotan, Intl J. Oncol., 1999,14,217224. 101. Q. Zhou, R. D. Cummings, Arch. Biochem. Biophys, 1993, 300, 6-17. 102. F. A. Van Den Brule, C. B. M. Baldet, M. E. Sobel, D. N. W. Cooper, P. Marschal, V. Castronovo, Biochem. Biophj)s. Res. Conitnun., 1995, 209, 760-767. 103. P. R. Warfield, P. N. Makker, A. Raz, J. Ochieng, Invasion & Metastasis, 1997, 17, 101-112. 104. 1. Kuwabara, F. T. Liu, J. Zmnzunol., 1996, /56, 3939-3944. 105. Y. Ozeki, T. Matsui, Y. Yamamoto, M. Funahashi, J. Hamako, K. Titani, Glycobioloyy, 1995, 5, 255-261. 106. H. Inohara. S. Akahani, K. Koths, A. Raz, Cancer Rex, 1996,56,4530-4534. 107. J. Ochieng, M . L. Leite-Browning, P. Warfield, Biochem. Biophys. Rex Commun., 1998, 246, 788-791. 108. V. Sarafian, M. Jadot, J. M. Foidart, J. J. Letesson, F. Van Den Brule, V. Castronovo, R. Wattiaux, S. Wattiaux-De Coninck, Intl J. Cuncer, 1998, 75, 105-1 11. 109. B. Woynarowska, D. M. Skrincosky, A. Haag, M. Sharma, K. Matta, R. J. Bernacki, J. Biol. Ciiem., 1994,269, 22797-22803. 110. K . Wasano, Y. Hirakawa, J. Histochem. Cytochem., 1999, 47, 75S82. 1 1 1. M. Fouillit, M. Levi-Strauss, V. Giudicelli, D. Lutomski, D. Bladier, M. Caron, R. JoubertCaron, J. Chromatog. B, 1998, 706, 167-171. 112. N. L. Perillo, C. H. U. Gaart, J. T. Nguyen, L. G. Baum, J. Exp. Med., 1997,185, 1851-1858. 113. S. Akdhani, P. Nangi-Makker, H. Inohara, H . R. C. Kim, A. Raz, Cuncer Rex., 1997, 57, 5272- 5276. 114. A. Allione, V. Wells, G. Forni, L. Mallucci, F. Novelli, J. Zmmunol., 1998, 161, 2114-2119. 115. J. Ellerhorst, T. Nguyen, D. N. W. Cooper, Y. Estrov, D. Lotan, R. Lotan, Intl J. Oncol., 1999, 14, 225--232. 116. G. A. Rabinovich, N. M. Modesti, L. F. Castagna, C. A. Landa, C. M. Riera, C. E. Sotomayor, J. Biochem., 1997, 122, 365-373.

646

38 Gulectins

117. G. A. Rabinovich, M. M. Iglesias, N. M. Modesti, L. F. Castagna, C. Wolfenstein-Todel, C. M. Riera, C. E. Sotomayor, J. Immunol., 1998, 160,4831-4840. 118. R. Y. Yang, D. K. Hsu, F. T. Liu, Proc. Natl Acad. Sci. U S A , 1996, 93, 6737-6142. 119. J. Wada, K. Ota, A. Kumar, E. I. Wallner, Y. S. Kanwar, J. Clin. Invest., 1997, 99, 2452246 1. 120. G. N. Vespa, L. A. Lewis, K. R. Kozak?M. Moran, J. T. Nguyen, L. G. Baum, M. C. Miceli, J. Immunol., 1999, 162, 799-806. 121. F. A. Van Den Brule, A. Bellahcene, P. Jackers, F. T. Liu, M . E. Sobel, V. Castronovo, lntl J. Oncol., 1997, 11, 261-264. 122. G. L. Sanford, S. Harris-Hooker, FASEB J . , 1990, 4, 2912-2918. 123. H. Inohara, S. Akahani, A. Raz, Exp. Cell Rex, 1998, 245, 294-302. 124. L. Adams, G. K. Scott, C. S. Weinberg. Biochim. Biophjs. Acta, 1996, 1312, 137-144. 125. V. Wells, L. Mallucci, Cell, 1991, 64, 91-97. 126. K. Yamaoka, A. Ingendoh, S. Tsubuki, Y. Nagai, Y. Sanai, J. Biochem., 1996,119, 878-886. 127. R. Matsumoto, H. Matsumoto, M. Seki, M. Hata, Y. Asano. S. Kanegasaki, R. L. Stevens, M. Hirashima, J. Biol. Chem., 1998,273, 16976-16984. 128. S. Dong, R. C. Hughes, FEBS Lett., 1996,395, 165-169. 129. K . C. G. Jeng, L. G. Frigeri, F. T. Liu, Immunol. Lett., 1994, 42, 113-116. 130. A. Karlsson, P. Follin, H. Leffler, C. Dahlgren, Blood, 1998, 91, 3430-3438. 131. V. V. R. Swarte, R. E. Mebius, D. H. Joziasse, D. H. Van Den Eijnden, G. Kraal, Eur. J. Immunol., 1998,28,2864-287 I. 132. A. V. Timoshenko, H. J. Gabius. Bid. Chem. Hoppe-Seyler, 1993, 374, 237-243. 133. N. L. Perillo, M. E. Marcus, L. G. Baum, J. Mol. Med., 1998, 76, 402-412. 134. G. Dimopoulos, A. Richman, A. della Torre, F. C. Kafatos, C. Louis, Proc. Nutl Acad. Sci. CJ S A , 1996, 93, 13066-13071. 135. G. Dimopoulos, A. Richman, H. M. Muller, F. C. Kafatos, Proc. Nutl Acad. Sci. U S A , 1997, 94, 11508-11513. 136. A. C. Puche, F. Poirier, M. Hair, P. F. Bartlett, B. Key, Devel. Biol., 1996, 179, 274-287. 137. C. Colnot, M. A. Ripoche, G. Milon, X. Montagutelli. P. R. Crocker, F. Poirier, Immunology, 1998, 94, 290-296. 138. D. Cvejic, S. Savin, I. Paunovic, S. Tatic, M. Havelka, J. Sinadinovic, Anticancer Rex, 1998, 18, 2637-2641. 139. A. Gillenwater, X. C. Xu, Y. Estrov, P. G. Sacks, D. Lotan, R. Lotan, Intl J. Cancer, 1998, 75, 217-224. 140. F. Orlandi, E. Saggiorato, G. Pivano, B. Puligheddu, A. Termine, S. Cappia, P. De Giuli, A. Angeli, Cancer Res., 1998, 58, 301 5-3020. 141. M. J. Scanlan, Y. T. Chen, B. Williamson, A. 0. Cure, E. Stockert, J. D. Gordan, 0. Tureci, U. Sahin, M. Pfreundschuh, L. J. Old, In21 J. Cuncer, 1998, 76, 652-658. 142. H. Idikio, Intl J. Oncol., 1998, 12, 1287-1290. 143. X. Sanjuan, P. L. Fernandez, A. Castells, V. Castronovo, F. Van Der Brule, F. T. Liu, A. C. A. Campo, Gastroenterology, 1997, 113, 1906-191 5. 144. H. Rechreche, G. V. Mallo, G. Montalto, J. C. Dagorn, J. L. Iovanna, European J. Biochem., 1997,248, 225-230. 145. R. S. Bresalier, P. S. Yan, J. C. Byrd, R. Lotan, A. Raz, Cancer, 1997, 80, 776-787. 146. M. Ostergaard, H. H. Rasmussen, H. V. Nielsen, H. Vorum, T. F. Orntoft, H. Wolf, J. E. Celis, Cancer Res., 1997, 57, 4111-4117. 147. F. A. Van Den Brule, C. Buicu, A. Berchuck, R. C. Bast, M. Deprez, F. T. Liu, D. N. W. Cooper, C. Pieters, M. E. Sobel, V. Castronovo, Human Path., 1996, 27, 1185-1191. 148. A. Gillenwater, X. C. Xu, A. K. El-Naggar, G. L. Clayman, R. Lotan, Head & Neck, 1996, 18,422-432. 149. V. Castronovo, F. A. Van Den Brule, P. Jackers, N . Clause, F. T. Liu, C. Gillet, M. E. Sobel, J. Path., 1996, 179, 43-48. 150. K. N. Konstantinov, B. A. Robbins, F. T. Liu, Am. J. Path., 1996, 148, 25-30. 151. X. C. Xu, A. K. El-Naggar, R. Lotan, Am. J. Path., 1995,147, 815-822. 152. L. Chiariotti, M. T. Berlingieri, C. Battaglia, G. Benvenuto, M. L. Martelli, P. Salvatore, G. Chiappetta, C. B. Bruni, A. Fusco, Int. J. Cancer, 1995, 64, 171-175.

References

641

153. D. W. Ohannesian, D. Lotan, R. Lotan, Cuncer R e x , 1994, 54, 5992-6000. 154. E. Hebert, M. Monsigny, B i d . Cdl., 1994, 81. 73-76. 155. L. Chiariotti, G. Benvenuto, R. Zarrilli, E. Rossi, P. Salvatore, V. Colantuoni, C. B. Bruni, Cell Groivtli & Dif/i.rentiation, 1994, 5 , 769-775. 156. F. A. Van Den Brule, J. Price, M. E. Sobel, R. Lambotte. V. Castronovo, Biochem. Biophys. Res. Commun.: 1994,201, 388-393. 157. S. Kannan, H. J. Gabius, G. J. Chandran, M. R. Pillai, K. R. Nalinakumari, M. K. Nair, Cancer Lett., 1994, 85, 1-7. 158. R. Lotan, Y. Matsushita, D. Ohannesian, D. Carralero, D. M. Ota, K. R. Cleary, G. L. Nicolson, T. Irimura, Carbohyd. R e x , 1991, 213, 47-57. 159. T. Irimura, Y. Matsushita, R. C. Sutton, D. Carralero, D. W. Ohannesian, K. R. Cleary, D. M. Ota, G. L. Nicolson. R. Lotan. Cancer Res.; 1991, 51, 387-393. 160. R. Lotan, D. Lotan, D. M. Carralero, Cuncer Lett., 1989, 48, 115-122. 161. A. Raz, L. Meromsky, R. Lotan, Cuncrr R e s , 1986, 46, 3667-3672. 162. H. Inohara, Y. Honjo, T. Yoshii, S. Akahani, J . Yoshida, K. Hattori, S. Okamoto, T. Sawada, A. Raz, T. Kubo, Cancer, 1999, 85, 2475S2484. 163. G. Schwarz, Jr., M. Remmelink, C. Decaestecker, I. Gielen, V. Budel, M. Burchert, F. Darro, A. Danguy, H. J. Gabius, I. Salmon, R. Kiss, Am. J. Clin. Puthol., 1999, 111, 623-631. 164. L. Cindolo, G. Benvennto, P. Salvatore, R. Pero, G. Salvatore, V. Mirone, D. Prezioso. V. Altieri, C. B. Bruni, L. Chiariotti, Int. J . Cancer, 1999, 84, 39-43. 165. D. K . Hsu, C. A. Dowling, K. C. Jeng, J. T. Chen, R . Y. Yang, F. T. Liu, Znt. J. Cancer. 1999, 81, 5 19-526. 166. P. Santa Lucia, B. D. Wilson, H. J. Allen, J. Derniatol. Sury. Oncol., 1991, 17, 653-655. 167. R. S. Bresalier. N. Mzurek, L. R. Sternberg, J. C. Bryd, C. K. Yunker, P. Nangia-Makker. A. Raz, Gastroenterology. 1998, 115, 287- 296. 168. J. Beuth, H. L. KO,V. Schirrmacher. G. Uhlenbruck, G. Pulverer, Clin. Exp. Metast., 1988,6, 115-120. 169. N. Le Marer, R. C. Hughes, J. Cell. Physiol., 1996, 168, 51-58. 170. K. J. Pienta, H. Naik. A. Akhtar, K. Yamazaki, T. S. Replogle, J. Lehr, T. L. Donat, L. Tait, V. Hogan, A. Raz, J. Natl Cancer lnst., 1995, 87, 348-353. 171. R. Lotan, P. N . Belloni, R. J. Tressler, D. Lotan, X. C. Xu, G. L. Nicolson, G/ycoconj. J . , 1994, 11,462-468. 172. L. G. Baum, J. J. Seilhamer, M. Pang, W. B. Levine, D. Beynon, J. A. Berliner, Glycoconj. J.. 1995, 12, 63-68. 173. 0. Saitoh, W. C. Wang, R. Lotan, M. Fukuda, J. Biol. Chem., 1992,267; 5700-5711. 174. J. Garrigues, J. Anderson, K. E. Hellstrom, I. Hellstrom, J. Cell Biol., 1994, 125, 129-142. 175. T. Oka, S. Murakami, Y. Arata, J. Hirabayashi, K.-I. Kasai, Y. Wada, M. Futai, Arch. Biochem. Biophys., 1999, 361, 195-201. 176. T. Magnaldo, D. Fowlis, M. Darmon, Differentiation, 1998, 6.3: 159-168. 177. Y. R. Hadari, K. Paz, R. Dekel, T. Mestrovic, D. Accili, Y. Zick, J. Biol. Chem.; 1995, 270, 3447-3453. 178. 0. Tureci, H. Schmitt, N. Fadle, M. Pfreundschuh, U. Sahin, J. B i d . Chem., 1997,272, 64166422. 179. E. Leal-Pinto, W. Tao, J. Rappaport, M. Richardson, B. A. Knorr, R. G. Abramson, J. Biol. Chenz., 1997, 272, 617-625. 180. D. N. W. Cooper, unpublished results.

Part I1 Volume 4

IV Saccharide Biology

Carbohydrates in Chemistry and Biology Beat Emst, Gerald W. Hart, Pierre Sinay copyright Q WILEY-VCH Verlag GmbH. 2000

39 Structures and Functions of Nuclear and Cytoplasmic Glycoproteins Robert S. Haltiwanger

39.1 Introduction Glycoproteins are classically thought of as components of the extracellular environment. This view is based on several key observations. For instance the sugar-rich carbohydrate coat, or glycocalyx, of cells is located exclusively on the extracellular leaflet of the plasma membrane [ 11. In addition the vast majority of extracellular proteins are glycosylated, and the conventional glycosylation machinery is topologically oriented in such a way that it will only add carbohydrates to the extracellular portions of proteins [2]. Nonetheless, numerous examples of glycoproteins in the nucleus or cytoplasm have been reported over the years [3,4].In this Chapter, several reported forms of nuclear and/or cytoplasmic glycosylation will be reviewed. The first half of the Chapter deals with 0-linked N-acetylglucosamine (0-GlcNAc), the most studied form of glycosylation in the nucleus and cytoplasm. Several other unconventional forms of glycosylation are reviewed in the remainder of the Chapter. After summarizing what is known about the biochemistry of the modifications (structure and biosynthesis), potential functions will be discussed. Since this Chapter is intended to be a brief introduction to the field of nuclear and cytoplasmic glycosylation, the large number of reports requires that some things be left out. The reader is referred elsewhere [3, 41 for a more comprehensive listing of various reports of these modifications. Two criteria were used here to determine which reports to focus on: the glycosylation must exist on known nuclear or cytoplasmic proteins, and definitive structural evidence for covalent modification of the protein(s) with carbohydrates must exist. Many of the reports of glycosylation in the nucleus and cytoplasm in the literature satisfy only one of these criteria. In addition, several recent studies suggesting that earlier reports of nuclear and cytoplasmic glycosylation were false have also appeared [5-9]. For instance, it was reported over a decade ago that the high mobility group proteins (HMG proteins, abundant chromosomally associated proteins) from mammalian sources are glycosylated [ 10, l l]. Recent studies, using more sophisticated and sensitive approaches,

652

39 Structures and Functions of Nuclear and Cytoplasmic Glycoproteins

have failed to find any carbohydrate modifications on the HMG proteins [5, 7-91, Such studies have thrown into doubt some of the earlier observations, making it essential to apply fairly strict criteria before conclusions about nuclear or cytoplasmic glycosylation pathways can be made.

39.2 @Linked N-Acetylglucosamine (0-GlcNAc) The best documented form of nuclear and cytoplasmic glycosylation is 0-linked Nacetylglucosamine (0-GlcNAc; for recent reviews see: [ 12-15]). It consists of the monosaccharide GlcNAc in p-linkage to the hydroxyl groups of serines or threonines. This modification is structurally distinct from 0-linked N-acetylgalactosamine (0-GalNAc) which forms the core for mucin-type 0-glycosylation found in the secretory pathway. 0-GlcNAc was discovered by application of a novel technique: the specific radiolabeling of non-reducing, terminal GlcNAc residues using bovine milk galactosyltransferase and UDP-[3H]galactose [ 161. Upon applying this technique to murine T cells, Hart and coworkers discovered that the major form of terminal GlcNAc in cells is 0-GlcNAc [ 161. Subsequent studies demonstrated that the modification is found exclusively on proteins from the nuclear and cytoplasmic portions of the cell [17]. Definitive evidence for this localization was initially demonstrated with antibodies directed against the 0-GlcNAc moieties of nuclear pore proteins [ 18, 191. Immunoelectron microscopy of nuclear membranes clearly showed the presence of the 0-GlcNAc moieties in the nucleoplasmic and cytoplasmic portions of the cell. Some reports have suggested that 0-GlcNAc may also occur on proteins on the extracytoplasmic side of membranes [16, 20, 211, although these reports are the subject of debate [22].Rough estimates indicate that two thirds of the 0-GlcNAc modified proteins are nuclear and the remainder are mainly cytoplasmic [17]. Hundreds of different proteins appear to be modified with 0-GlcNAc in both of these compartments [ 12, 151. The first proteins to be identified with the 0-GlcNAc modification were structural proteins such as nuclear pore proteins [18, 231 and erythrocyte band 4.1 [24]. Since then, dozens of other proteins have been identified that bear 0-GlcNAc (see [12, 141 for comprehensive lists). A wide variety of proteins are modified, including proteins involved in transcriptional regulation, oncogenes, kinases, enzymes, heat shock proteins, as well as a number of structural proteins. The stoichiometry of modification varies widely, from heavily glycosylated proteins (e.g. nuclear pore proteins [ 181 and the transcription factor Spl [25] contain more than ten 0-GlcNAc moieties per polypeptide chain) to proteins where only a few percent of the molecules are glycosylated (e.g. band 4.1 [24]). The modification has been found on proteins from all multicellular eukaryotic organisms examined, as well as numerous unicellular parasites (e.g. Schistosoma [26], Leishmania [27]).In contrast, no definitive evidence for the presence of 0-GlcNAc modified proteins has been obtained in either yeast (e.g. Sacchromyces cerevisae) or prokaryotes. Sites of 0-GlcNAc modification have been identified in many of the proteins known to be modified (see [12]

39.2 O-Linked N-Acetylglucosamine (O-GlcNAc)

653

for a comprehensive list). N o consensus sequence can be discerned by comparison of the sites, although several occur on serines or threonines in close proximity to a proline. Nonetheless, based on the information obtained thus far, it is extremely difficult to predict whether a protein will be modified with O-GlcNAc based on primary sequence data. This is reminiscent of mucin-type O-GalNAc modification sites [28]. At this point, it is not clear if the O-GlcNAc sites are so different because the enzyme(s) that add GlcNAc recognize features not apparent in the primary structure, or if there are multiple enzymes with different specificities (as appears to be the case for mucin-type glycosylation [29-3 11). Enzymes capable of the addition and removal of O-GlcNAc have been purified and characterized [ 32-34]. The enzyme responsible for addition of O-GlcNAc to proteins (UDP-GlcNAc: peptide P-O-N-acetylglucosaminyltransferase or 0GlcNAc transferase) was originally identified using synthetic peptides based on known sites of glycosylation as acceptor substrates [32]. The enzyme purified from rat liver is a heterotrimer with two u (110 kDa) and one p (78 kDa) subunits 1331. The apparent molecular weight of the holoenzyme is 340,000. Photoaffinity probe studies demonstrated that the a-subunit contains the active site. The gene for 110 kDa subunit has been cloned and characterized [35, 361. Active O-GlcNAc transferase activity has been obtained by expression of this gene in a heterologous system, demonstrating that the gene encodes the appropriate enzyme. Tmmunoblotting data suggests that the 78 kDa subunit is immunologically related to the 110 kDa subunit [35], suggesting that the P-subunit may be a proteolytic fragment or possibly an alternatively spliced variant of the u-subunit. Northern blotting data demonstrates that the enzyme is widely expressed in mammalian tissues [35, 361, as suggested by the distribution of O-GlcNAc modified proteins, and homologs have been identified in numerous multicellular species including Caenorhabditis elegans [35, 361. The O-GlcNAc transferase contains 1 1 tetratrico peptide repeats (TPR) [35, 361. These motifs are believed to be involved in a wide variety of protein-protein interactions [37]. They may be involved in either substrate recognition or localization of the enzyme to particular regions in the cell. In addition to the O-GlcNAc transferase, an O-GlcNAc specific P-N-acetylglucosaminidase (O-GlcNAcase) has been purified and characterized from rat spleen cytosol [34]. The purified enzyme appears to be a heterodimer with an a-subunit of 54 kDa and a P-subunit of 51 kDa. This enzyme is distinct from lysosomal hexosaminidases in that it has a neutral pH optimum (pH 6.4) and is extremely specific for GlcNAc, whereas lysosomal enzymes have acidic pH optima and hydrolyze both GalNAc and GlcNAc containing substrates. Like the O-GlcNAc transferase, the O-GlcNAcase appears to be present in the cytoplasm of numerous cells [34]. 39.2.1 O-GlcNAc Appears to be a Regulatory Modification much like Phosphorylation The presence of enzymes capable of both the addition and removal of O-GlcNAc in the same cellular compartments is reminiscent of kinases and phosphatases and has raised the possibility that O-GlcNAc is a dynamic regulatory modification much

654

39 Structures and Functions of Nucleur and Cytoplasmic Glycoproteins 0-GlcNAc'ase GlcNAc

GlcNAc

0 Protein

A

Phosphatase

?"

B

Protein

UDP UDP-GlCNAC O-GlcNAc transferase

ATP

pP3

9

Protein

ADP Kinase

Figure 1. O-GlcNAc is a regulated modification and may compete with phosphate for modification sites on proteins. Cycle A represents the O-GlcNAc cycle. The italicized components (O-GlcNAcase, O-GlcNAc transferase, UDP-GlcNAc) are all potential targets for regulation. Changes in these components would result in changes in O-GlcNAc levels on proteins. Cycle B represents a typical kinase/phosphatase cycle. Modification of a protein at a particular serine/threonine with 0GlcNAc may block addition of phosphate at the same or a nearby site [46, 471.

like phosphorylation. There is abundant evidence supporting this view. For instance, several workers have demonstrated via pulse-chase analyses that the 0GlcNAc moieties on proteins turn over more rapidly than the proteins themselves [38, 391. This indicates that unlike all other forms of glycosylation, O-GlcNAc is added to and removed from proteins numerous times during their lifetime (see Figure 1). In addition, we have recently demonstrated that a potent in vitro inhibitor of the O-GlcNAcase (PUGNAc, see [34, 401) causes an increase in O-GlcNAc levels on proteins when added to cells in culture. This demonstrates that an O-GlcNAc transferase/O-GlcNAcase cycle like that shown in Figure 1 (cycle A) operates within cells. Presumably, such a system will respond to various upstream signals, causing changes in the level of glycosylation of proteins, resulting in downstream effects. Alterations in the activity of either enzyme or in the level of UDP-GlcNAc could effect changes in glycosylation, and thus could be potential mechanisms of regulating the system within cells. Several examples of this type of regulation exist. For instance, mitogenic activation of T cells results in rapid and dramatic changes in O-GlcNAc levels on numerous proteins [41]. Although the mechanism of change is not known in this case, it is likely to be mediated by an alteration in the activity of either the O-GlcNAc transferase or the O-GlcN Acase. Other workers have demonstrated that O-GlcNAc levels change in response to fluctuations in glucose concentrations [42, 431. Since glucose is a precursor to UDPGlcNAc, the mechanism for altering O-GlcNAc levels is believed to be modulation

39.2 O-Liiikrd N-Acetylqlucosamine (0-GlcNAc)

655

of intracellular UDP-GlcNAc levels. The pathway responsible for UDP-GlcNAc synthesis (hexosamine biosynthetic pathway) has been suggested as a glucose-sensing pathway in cells, and may play a role in type 11 diabetes (see below, [44, 451). Thus, several cases in which physiologic stimuli resulting in alterations in O-GlcNAc levels on proteins have been reported. The fact that O-GlcNAc modifies the hydroxyl group of serines/threonines has led to the suggestion that O-GlcNAc may block the addition of phosphate to proteins (see Figure 1) [ 12-1 51. Since O-GlcNAc has significantly different properties than phosphate, it is likely that modification of a protein with O-GlcNAc would result in different effects than phosphorylation. This suggests that one function of O-GlcNAc may be to prevent phosphorylation of certain proteins even in the presence of active kinases, permitting an extra level of regulation in signal transduction cascades. Numerous observations support this competition hypothesis. First, essentially all O-GlcNAc modified proteins are also known to be phosphoproteins. In addition, a reciprocal relationship between glycosylation and phosphorylation has been observed on several proteins. For instance, RNA polymerase I1 is modified on its C-terminal domain (CTD) with either phosphate or O-GlcNAc but not both at the same time [46]. Analyses of the sites of modification suggest that O-GlcNAc and phosphate compete for the same residues in the CTD. The oncoprotein c-myc can be modified with either O-GlcNAc or phosphate at threonine-58 [47]. a known glycogen synthase kinase-3 (GSK-3) site, suggesting that competition exists between GSK-3 and O-GlcNAc transferase for this threonine within cells. Both the tau protein [48] and keratins 8 and 18 [49] are known to exist in either glycosylated or phosphorylated states in cells, suggesting a reciprocal relationship on these proteins. Many of the O-GlcNAc sites analyzed on other proteins [12] resemble predicted kinase sites, especially those of MAP kinases and GSK-3 (47, SO], implying that similar competition between phosphorylation and glycosylation may occur at these sites. To examine whether the phosphorylation state of a protein can be altered by inducing a change in glycosylation, we examined both the phosphorylation and glycosylation of the transcription factor Spl from cells treated with and without the O-GlcNAcase inhibitor PUGNAc [40]. PUGNAc caused an increase in the level of O-GlcNAc and a corresponding decrease in phosphorylation of Spl . Although we have not yet determined that the basis of the reciprocity on Spl is at the level of individual serines or threonines, these results demonstrate that a dynamic reciprocal relationship between glycosylation and phosphorylation on proteins exists in cells.

39.2.2 Modulation of Protein Stability and Function by O-GlcNAc In addition to playing a role in signal transduction by blocking sites of phosphorylation, modulation of O-GlcNAc levels on proteins is known to directly affect both protein stability and function. A number of studies have been performed on the effects of glycosylation on the transcription factor Spl. Jackson and Tjian demonstrated that unglycosylated, bacterially expressed Spl had lower transcriptional

656

39 Structures und Functions of Nuclear und Cytoplasmic Glycoproteins

activity in vitro than fully glycosylated Spl from HeLa cells [25], suggesting that the sugar could modulate the activity of the protein. In addition, Kudlow and coworkers have demonstrated that 0-GlcNAc plays several specific roles in the function of Spl. They showed that they can reduce 0-GlcNAc levels on Spl in NRK cells by a combination of glucose starvation (to reduce UDP-GlcNAc levels in the cells) and forskolin treatment (to increase cAMP levels) [43]. The link between cAMP levels and 0-GlcNAc is not clear. Nonetheless, the reduction in 0-GlcNAc levels on Spl caused a dramatic reduction in Spl protein levels that could be blocked by proteosome inhibitors. These results indicate that 0-GlcNAc plays a role in controlling the stability of Spl . In another report, Kudlow’s group demonstrated that a specific 0-GlcNAc site near the glutamine-rich transactivation domain of Spl plays a role in modulation of protein-protein interactions [ 511. This domain is known to interact with the TATA-binding-protein associated factor (TAF1lo), a component of TFIID, mainly through hydrophobic interactions. The presence of 0-GlcNAc on this domain significantly reduces the interaction, suggesting that 0-GlcNAc plays a role in modulating the interaction between these two proteins. Another example of how 0-GlcNAc can control protein stability has been described by Gupta and coworkers [52, 531. They have suggested that 0-GlcNAc functions in regulation of translation by controlling the level of a widely expressed regulatory protein called p67. Gupta’s group has demonstrated that p67 protects eIF-2a from phosphorylation by eIF-2a kinases, allowing translation to proceed [54, 551. Phosphorylation of the a-subunit of eukaryotic initiation factor 2 (eIF-2) is known to inhibit protein translation in animal cells [52].Interestingly, p67 is heavily modified with 0-GlcNAc [52], and removal of 0-GlcNAc from p67 results in its rapid degradation [53]. They have shown that the deglycosylation of p67 is activated under certain conditions, resulting in degradation of the protein and inhibition of protein synthesis. For instance, depletion of hemin from reticulocyte lysates causes deglycosylation of p67, allowing phosphorylation of eIF-2a by the heme regulated protein synthesis inhibitor (an eIF-2a kinase) [53, 561. They have also provided evidence that deglycosylation of p67 can be induced by viral infection [57, 581. In this case the loss of p67 inhibits protein synthesis by allowing phosphorylation of eIF-2a by the double stranded RNA-dependent protein kinase (also an eIF-2a kinase). The p67 deglycosylase has recently been identified [58], but its relationship to the 0-GlcNAcase described above is unknown. These results strongly suggest a role for 0-GlcNAc in regulation of protein synthesis. Other recent observations also demonstrate that the presence or absence of 0GlcNAc on proteins results in specific biological effects. The a-toxin from Clostridium novyi was recently shown to be an 0-GlcNAc transferase [59]. Upon entering cells it modifies the Rho subfamily of proteins (Rho, Rac, Cdc42, RhoG) with a single 0-GlcNAc. The modification occurs on threonine-37 of Rho in the effector domain of the protein. Rho subfamily proteins are members of the small GTPase family and are known to be involved in cytoskeletal rearrangements. Glycosylation of Rho inactivates the protein and results in a total redistribution of the actin cytoskeleton. These results suggest that the addition of a single 0-GlcNAc to a protein is sufficient to alter function. We have recently demonstrated that the levels of the 0-GlcNAc transferase are regulated by 0-GlcNAc levels in the cell (Li and

39.2 0-Linked N-Acet)~lglucosamine( 0 - G k N A c )

651

Haltiwanger, unpublished observation). As mentioned earlier, treatment of cells with PUGNAc, the 0-GlcNAcase inhibitor, results in an overall increase in 0GlcNAc levels on proteins within cells [40]. Levels of the 0-GlcNAc transferase decrease under these conditions, suggesting a feedback inhibitory effect. It is not yet clear whether the decrease in 0-GlcNAc transferase protein levels is caused by a decrease in its transcription rate or by an increase in its turnover rate, but these results suggest that cells are protected against hyper-glycosylation by this feedback inhibition mechanism. In addition to these clear examples of cases in which 0-GlcNAc modifies protein stability or function, a number of intriguing observations suggest additional roles for 0-GlcNAc in various biological phenomena. For instance, the reciprocity between glycosylation and phosphorylation of RNA polymerase 11’s CTD suggests that 0-GlcNAc may play a role in transcriptional regulation [46]. Since the hyperphosphorylation of the CTD is known to be the signal for initiation of mRNA synthesis [60], removal of 0-GlcNAc from the CTD may be a necessary precondition for mRNA elongation. In an intriguing preliminary report supporting this idea, Hart’s group showed that inhibitors of 0-GlcNAc removal prevent mRNA transcription mediated by the major late promoter of adenovirus in vitro [61]. These results strongly suggest a role for 0-GlcNAc in the regulation of the basal transcriptional machinery. 0-GlcNAc has also been implicated in glucose homeostasis and type I1 diabetes. In the last few years a clear link has emerged between the hexosamine biosynthetic pathway (HBP) and the development of insulin resistance [44, 451, an identifying characteristic of type I1 diabetes. The HBP is believed to function as a “glucose sensor” in the cell [44]. One to three percent of glucose entering the cell enters the HBP, the major product of which is UDP-GlcNAc. Thus, changes in flux through the HBP result in changes in UDP-GlcNAc levels. The flux of glucose through the HBP is known to regulate the expression of several genes including transforming growth factors a [62, 631 and p [64], leptin [65], pyruvate kinase [66], glycogen synthase [67] and glutamine: fructose 6-phosphate amidotransferase [65, 681. Several groups have suggested that the observed changes in gene expression are likely to be the result of changes in the level of 0-GlcNAc on transcription factors [45, 65, 691. Although a direct link between increased flux through the HBP and alterations in transcription due to glycosylation of transcription factors has not yet been made, changes in 0-GlcNAc levels on proteins in skeletal muscle have been shown to occur in response to increases in cellular UDP-GlcNAc levels [69]. Since one of the problems in type I1 diabetes is a lack of responsiveness to insulin signaling (insulin resistance), these results suggest that the defect may be hyper-glycosylation of transcription factors. This hyper-glycosylation may prevent insulin signaling in cells, possibly by blocking phosphorylation sites modified by kinases downstream of insulin activation (e.g. glycogen synthase kinase-3). If so, potential therapies for type I1 diabetes could be based on reducing the level of 0-GlcNAc addition to transcription factors. Current work in this extremely interesting area is aimed at providing a causal link between increased flux through the HBP, increases in 0-GlcNAc levels on specific transcription factors, and changes in transcriptional regulation of specific genes.

658

39 Structures and Functions of Nuclear and Cytoplasmic Glycoproteins

39.3 Other Forms of Nuclear and Cytoplasmic Glycosylation Although 0-GlcNAc has been studied most extensively, several other carbohydrate modifications on proteins in nuclear or cytoplasmic compartments have been reported [3, 41. The examples presented here fall into two major classes. The first class consists of structurally unconventional forms of glycosylation which are added to proteins in the cytoplasm. Interestingly, in contrast to 0-GlcNAc, each of these modifications appears to occur on a single protein (SKPI, phosphoglucomutase/ parafusin, and glycogenin). The second class consists of conventional forms of glycosylation (e.g. classical N-glycans or glycosaminoglycans) found on proteins in subcellular compartments not normally associated with the secretory pathway (e.g. cytoplasm, nucleus, mitochondria). The presence of glycoproteins in these compartments suggests a unique transport mechanisms which allows glycoproteins synthesized in the lumen of the endoplasmic reticulum and Golgi apparatus to move out of the normal secretory pathway and into other portions of the cell. Although some of the reports describing nuclear and cytoplasmic glycosylation resulting from these unique transport mechanisms lack definitive data localizing the modified proteins to the nucleus or cytoplasm, much of the data are compelling and are described below. At the end of this section, lectins in the nucleus and cytoplasm are briefly discussed, especially as they relate to the potential ligands with which they interact in these compartments. 39.3.1 Unique Cytoplasmic Forms of Glycosylation

The most thoroughly characterized example in this class is the glycosylation of SKPl, a cytoplasmic F-box binding protein from Dictyostelium [70-721. West and coworkers have shown that proline-143 of SKPl becomes hydroxylated and glycosylated with a linear pentasaccharide with the following structure: galp pal ,6D-Galpal, ~ - F u c p ,2-~-Galpp d 1,3GlcNAc-O-HyPro [73]. Several of the enzymes involved in addition of the saccharide have been identified [70, 71, 741, and the fucosyltransferase has recently been purified [75]. These data demonstrate the presence of a unique glycosylation pathway in the cytoplasm of Dictyostelium. Homologues of SKPl are widely expressed, from yeast to humans [73]. The modified proline residue is present in a highly conserved C-terminal region, and a proline at the equivalent position of proline-I43 is present in SKPl homologues from yeast, fungal, plant, and Caenorhabditis elegans. Although the known mammalian SKPl homologues do not contain an equivalent proline, several uncharacterized SKPl loci exist [73]. Thus, it is possible that this modification will be found in mammalian systems, too. SKPl is known to play a role in post-translational modification (ubiquitination, phosphorylation) of target proteins. It interacts with cullin and an F-box containing protein to form the SCF complex [73], and this complex targets modifying enzymes (kinases, ubiquitination machinery) to particular proteins. West and coworkers have suggested that the glycosylation of SKPl may play a role in the regulation of the SCF complex.

39.3 Other Forms qf Nuclear and Cytoplasmic Glycosylution

659

Another example of a cytoplasmic glycoprotein is phosphoglucomutase. Marchase and coworkers had originally identified a novel UDP-glucose: glycoprotein glucose l-phosphotransferase (Glc-phosphotransferase) in rat liver that transfers glucose 1-phosphate from UDP-glucose selectively to a 62 kDa cytoplasmic protein [ 761. The 62 kDa protein was subsequently identified as phosphoglucomutase (PGM) [77] and has been demonstrated to be modified by glucose l-phosphate in a number of species, including mammals and Saccharomyces cerevisiae [77--791. A closely related protein from the Purumecium tetraureliu called parafusin is also modified by glucose I-phosphate [78, SO]. Although parafusin shares over 50% sequence identity with rabbit muscle PGM [81], they appear to be distinct gene products since mammalian homologues of parafusin with a higher sequence identity have been identified [82]. These highly related proteins appear to be the predominant acceptors for the Glc-phosphotransferase, and these results suggest that this unique form of glycosylation is conserved across species. The glucose 1 -phosphate is believed to be part of a larger oligosaccharide containing mannose [78, 831. Although structural details are still lacking, the oligosaccharide appears to contain glucose, phosphate and mannose. The mannose is believed to be in O-linkage to serine with a terminal glucose l-phosphate [83]. A preliminary report suggested that there may be more than one mannose in the saccharide [84], although this remains unconfirmed. A cytoplasmic phosphodiesterase has been described which selectively removes glucose 1-phosphate from PGM [ 8 5 ] ,suggesting that a regulated cycle for the addition and removal of glucose 1-phosphate is present in cells, much like that for O-GlcNAc (see Figure 1). In this case it is thought that the mannose remains associated with the protein while the glucose l-phosphate cycles on and off. Changes in the glycosylation state of PGM and parafusin have been demonstrated to occur in several biological systems [79, 80, 861, demonstrating that such a cycle is indeed functional in cells. Changes in cytoplasmic calcium levels appear to alter the glucosylation state of both PGM (in the pheochromocytoma PC-12 cell line) [79] and parafusin (in Parumeciunz) [SO]. The change in glucosylation state correlates with an association with membranes [80, 861 and has been suggested to be involved in the regulation of secretion. Interestingly, other stresses have also been correlated with changes in the glucosylation of PGM. In the yeast S. cereviJiae, both heat shock and carbon sources other than glucose induce a decrease in glucose 1-phosphate on PGM [86], suggesting that the glucosylation of PGM may be important in responding to nutrient deprivation in these cells. These results suggest that glucose I-phosphate modification of PGM is a conserved, regulated modification in cells. The third example of unique glycosylation in the cytoplasm is that of glycogenin [87]. Glycogenin is the primer for glycogen biosynthesis [88]. It is modified with O-linked glucose on tyrosine- 194 [ 891. Glycogenin has an endogenous glucosyltransferase activity with which it modifies itself and adds the first few a1,4 linkedglucose residues of the glycogen chain, until a sufficiently large polymer is made for glycogen synthase to use as a primer. Two forms of glycogenin (glycogenin-1 and glycogenin-2) have been identified in mammalian systems [90]. It is widely expressed, being especially abundant in liver and skeletal muscle. Homologs exist in numerous other species, including C. eleyuns and Succhromyces cerevisue [90].

660

39 Structures and Functions of Nucleur and Cytoplasmic Glycoproteins

39.3.2 Conventional Forms of Glycosylation in the Nucleus and Cytoplasm Glycosaminoglycans have been reported to exist in the nucleus for over 20 years 191, 921. Most of these reports are based on identification of various glycosaminoglycans in purified nuclear fractions. Although many of these studies are highly suggestive of the presence of glycosaminoglycans in the nucleus, none of the data is definitive in terms of localization. It is very difficult to completely rule out possible contamination from other subcellular compartments in the nuclear preparations [93]. One set of studies performed using rat hepatoma cells by Conrad and coworkers [94, 951 suggested that heparan sulfate proteoglycans are secreted from cells and taken back up by an unknown pathway to the nucleus. These studies were especially compelling because the heparan sulfate isolated from the nuclei of these cells was structurally different than the heparan sulfate that was secreted. The nuclear heparan sulfate was unique in that it was in the form of free chains (not protein bound) and had a high content of 2-sulfated glucuronic acid. This suggested that the heparan sulfate chains were processed as they were taken up by the cell and transported to the nucleus. Interestingly, subsequent studies in the same cells demonstrated that exogenously added heparan sulfate proteoglycans could be taken up and processed in the same manner [94] and that these nuclear heparan sulfates appeared to slow cell growth in the G1 phase of the cell cycle 1961. Similar studies by Hascall’s group suggested that processed forms of dermatan sulfate are transported from the cell surface to the nucleus of rat ovarian granulosa cells 1931. Importantly, this same report clearly demonstrated the difficulty in isolating “pure” nuclear fractions, and thus drawing firm conclusions on localization from this type of analysis. In addition to the work from Conrad’s and Hascall’s groups, Margolis and Margolis’ group has provided strong localization data for the presence of proteoglycans in the nucleus. Instead of using a purely biochemical approach like those described above, they have utilized antibodies directed towards specific proteoglycan molecules in conjunction with immunofluorescence to demonstrate nuclear localization. In a recent report they demonstrated that two neuronal proteoglycans, glypican and biglycan, appear in the nucleus in a cell cycle dependent manner [97]. Using antibodies to these proteoglycans coupled with confocal microscopy, they showed that a significant portion of both of the proteoglycans are localized in the nucleus in rat central nervous tissue as well as C6 glioma cells. Transfection of epitope-tagged versions of glypican into COS-1 cells resulted in transport of the proteoglycan to the nucleus. Interestingly, both glypican and biglycan have functional nuclear localization sequences which can be used to drive a fusion protein (pgalactosidase) into the nucleus. The mechanism for the transport of these proteoglycans into the nucleus remains unknown. Since glycosaminoglycans are known to mimic nucleic acids in their interactions with proteins, it is possible that the nuclear glycosaminoglycans play a role in modulating the interactions between DNA binding proteins and DNA. There is also an extensive literature dealing with proteins bearing N-glycans in the nuclear and cytoplasmic compartments 13, 4, 981. Many of these reports consist of metabolic radiolabeling of cells in culture with sugars (usually [2-3H]mannose)

39.3 Other Forms qf Nuclear and Cytoplasmic Glycosylution

66 1

followed by examination of radiolabeled proteins isolated from either “cytoplasmic” or “nuclear” fractions. The results provided by these studies are very intriguing, and the presence of N-glycans on the proteins from the fractions isolated is often unequivocal. As with the studies described above dealing with nuclear glycosaminoglycans, completely ruling out contamination of a “cytoplasmic” or a “nuclear” fraction with proteins derived from other compartments is virtually impossible. Thus, unequivocal localization of these glycoproteins to cytoplasmic, nuclear, or mitochondrial compartments remains an open question. Nonetheless, several of these reports are highly suggestive. Recently Spiro’s group [99] provided evidence for N-glycosylation of a 45 kDa protein putatively from the inner mitochondrial membrane. There is good evidence that the protein is modified by high-mannose type N-glycans, and that the sugars are added to the protein in the ER. The protein is present in preparations of inner mitochondrial membrane proteins, and it coprecipitates with known inner membrane protein complexes (NADH-ubiquinone oxidoreductase or FIFo-ATPase (complex V)). These data strongly suggest that 45 kDa protein is in fact an inner mitochondrial membrane protein and make it an excellent candidate for further analysis. In future studies, antibodies to this protein could be generated and used in conjunction with immunofluorescence and immunoelectron microscopy to unequivocally localize the protein to the mitochondria. Once the localization is confirmed, the mechanisms for the unusual transport need to be elucidated. 39.3.3 Nuclear and Cytoplasmic Lectins Several other indirect observations have contributed to the proposal that nuclear and cytoplasmic glycoproteins are more widespread than generally accepted. In particular, a number of carbohydrate binding proteins (e.g. galectins [ 1001, CBP70 [ 1011, [ 1021) have been definitively localized to nuclear and/or cytoplasmic compartments. This localization raises the question of the ligands with which these lectins interact. Some of the galectins are known to be secreted by an unusual pathway directly through the plasma membrane and to bind to extracellular ligands [ 1001. Thus, one possibility is that there are no intracellular ligands for these lectins, but that their presence in the cytoplasm is only a result of this unusual secretory pathway. Since the galectins are galactose specific [ 1001, they are believed to be secreted in this unusual fashion to prevent interaction with the many galactose-terminated proteins moving through the Golgi apparatus. Although identification of intracellular ligands for these lectins has been difficult, several good candidates have recently been identified. An intriguing study [ 1031 suggests that galectins interact with carbohydrate modifications on keratins. They provide carbohydrate compositional analysis data suggesting the presence of GalNAc-containing saccharides on the keratins to which the galectins are believed to bind. CBP70 is a nuclear and cytoplasmic lectin with affinity for glucose and GlcNAc [ 1041. A recent report [ 105) suggests that CBP70 interacts with proteins bearing 0-GlcNAc. Thus, evidence for an intracellular role for these lectins involving specific carbohydrate interactions with nuclear and cytoplasmic glycoproteins is mounting. The finding that both

662

39 Structures and Functions of Nuclear und Cytoplusmic Glycoproteins

galectin-1 and galectin-3 are necessary components of a pre-mRNA splicing complex [ 106, 1071 demonstrates that the galectins are involved in intracellular biological processes. As more of the intracellular ligands for these carbohydrate binding proteins are identified, the biological roles of these uniquely placed lectins will become clearer.

39.4 Conclusions The presence of carbohydrate modifications on proteins in the nuclear and cytoplasmic portions of the cell is now well established. The 0-GlcNAc modification is already known to play roles in modulating the stability and function of a wide variety of proteins, and there is increasing evidence for a direct role in transcriptional regulation. Several of the other modifications discussed above are not as well understood, but now that there is definitive evidence for their existence in nuclear and cytoplasmic compartments, more effort can be directed towards determining the functional implications of these modifications. In the ten years since the most extensive review of this literature [3], a tremendous amount has been learned about several of the forms of glycosylation described in this chapter. Hart’s review ended with the statement: “It is likely that we will continue to discover unique saccharide structures and glycosyltransferases in these compartments that may play critical roles in many important intracellular processes” [3]. This continues to be true today.

Acknowledgments

The author would like to thank Dr. Deborah Brown, Dr. Daniel Moloney, Mr. Scott Busby, Mr. Sean Li, Ms. Li Shao and Ms. Kathleen Grove for helpful discussions and comments in the preparation of this manuscript. All original work was supported by a grant from the National Institutes of Health (NIH G M 46888) and a GRANT Award from Neose Technologies, Inc.

References 1, G.M.W. Cook, Glycohiol., 1995, 5, 449-458, Glycobiology of the cell surface: The emergence

of sugars as an important feature of the cell periphery. 2. C.B. Hirschberg and M.D. Snider, Annu. Rev. Biochem., 1987, 56, 63-87, Topography of glycosylation in the rough endoplasmic reticulum and Golgi apparatus. 3. G.W. Hart, R.S. Haltiwanger. G.D. Holt and W.G. Kelly, Annu. Rev. Biochem., 1989, 58, 841-874, Glycosylation in the nucleus and cytoplasm. 4. J. Hubert, A.P. Seve, P. Facy and M. Monsigny, Cell Dzze., 1989, 27, 69-81, Are nuclear lectins and nuclear glycoproteins involved in the modulation of nuclear functions? 5 . L. Medina and R.S. Haltiwanger, GEycohiology, 1998, 8, 191-198, Calf thymus high moblility group proteins are non-enzymatically glycated but not significantly glycosylated.

Rejerences

663

6. L. Medina, K. Grove and R.S. Haltiwanger, Glycohiology, 1998, 8, 383-391, 9 4 0 Large T antigen is modified with 0-linked N-acetylglucosamine but not with other forms of glycosylation. 7. P. Ferranti, A. Malorni, G. Marino, P. Pucci, G.H. Goodwin, G. Manfioletti and V. Giancotti, J. Bid. Chem., 1992, 267, 22486-22489, Mass spectrometric analysis of the HMGY protein from Lewis lung carcinoma. Identification of phosphorylation sites. 8. Y.B. Chao, W.M. Scovell and S.B. Yan, Protein Sci., 1994, 3. 2452-2454, High mobility group protein, HMG-1, contains insignificant glycosyl modification. 9. V.A. Boumba, 0. Tsolas, D. Choli-Papadopoulou and K. Seferiadis, Arch. Biochem. Biophys., 1993,303,436-442, Isolation by a new method and sequence analysis ofchromosomal HMG17 protein from porcine thymus. 10. R. Reeves and D. Chang, J. Biol. Chem., 1983, 258, 679-687, Investigations of the possible functions for glycosylation in the high mobility group proteins. Evidence for a role in nuclear matrix association. 11. R. Reeves, D. Chang and S.C. Chung, Pror.. Nut/ Acud. Sci. U . S . A . , 1981, 78, 6704-6708, Carbohydrate modifications of the high mobility group proteins. 12. R.S. Haltiwanger, S. Busby, K. Grove, S. Li, D. Mason, L. Medina. D.J. Moloney, G.A. Philipsberg and R. Scartozzi, Biochem. Bioph-vs. Res. Commun., 1997, 231, 237-242, O-Glycosylation of Nuclear and Cytoplasmic Proteins: Regulation Analogous to Phosphorylation? 13. G.W. Hart, Annu. Rev. Biochem., 1997, 66, 315-335, Dynamic 0-linked glycosylation of nuclear and cytoskeletal proteins. 14. G.W. Hart, L.K. Kreppel, F.I. Comer, C.S. Arnold, D.M. Snow, Z. Ye, X. Cheng, D. DellaManna, D.S. Caine. B.J. Earles, Y. Akimoto, R.N. Cole and B.K. Hayes, Glycobioloyy, 1996, 6, 71 1-716, 0-GlcNAcylation of key nuclear and cytoskeletal proteins: reciprocity with 0-phosphorylation and putative roles in protein multimerization. 15. D.M. Snow and G.W. Hart, h t . Rev. Cy/o/., 1998, 181, 43-74, Nuclear and cytoplasmic glycosylation. 16. C.R. Torres and G.W. Hart, J. Biol. Cl7em., 1984, 259, 3308-3317, Topography and polypeptide distrubution of terminal N-Acetylglucosamine residues on the surfaces of intact lymphocytes. Evidence for 0-GlcNAc. 17. G.D. Holt and G.W. Hart, J. Bid. Ch~wi.,1986, 261, 8049-8057, The subcellular distribution of terminal N-acetylglucosamine moieties. Localization of a novel protein-saccharide linkage, 0-linked GlcNAc. 18. G.D. Holt, C.M. Snow, A. Senior, R.S. Haltiwanger, L. Gerace and G.W. Hart, J. Ce//Biol.. 1987, 104, 1 157-1 164, Nuclear pore complex glycoproteins contain cytoplasmically disposed 0-linked N-acetylglucosamine. 19. C.M. Snow, A. Senior and L. Gerace, J. Cell Bid., 1987, 104, 1143-1 156, Monoclonal antibodies identify a group of nuclear pore complex glycoproteins. 20. C. Abeijon and C.B. Hirschberg, Proc. Nu/l Acud. Sci. USA., 1988, 85, 1010-1014, Intrinsic membrane glycoproteins with cytosol-oriented sugars in the endoplasmic reticulum. 21. J.M. Capasso, C. Abeijon and C.B. Hirschberg, J. Bid. Chem., 1988, 263, 19778-19782, An intrinsic membrane glycoprotein of the Golgi apparatus with 0-linked N-acetylglucosamine facing the cytosol. 22. K.P. Kearse and G.W. Hart, Arch. Biochem. Biophys., 1991, 290, 543-548, Topology of 0linked N-acetylglucosamine in murine lymphocytes. 23. J.A. Hanover, C.K. Cohen, M.C. Willingham and M.K. Park, J. Biol. Chem., 1987, 262; 9887-9894, 0-linked N-acetylglucosamine is attached to proteins of the nuclear pore. Evidence for cytoplasmic and nucleoplasmic glycoproteins. 24. G.D. Holt, R.S. Haltiwanger, C.R. Torres and G.W. Hart, J. Bid. Chenz., 1987, 262, 14847-14850, Erythrocytes contain cytoplasmic glycoproteins. 0-linked GlcNAc on Band 4.1. 25. S.P. Jackson and R. Tjian, Cell, 1988, 55, 125-133, 0-Glycosylation of eukaryotic transcription factors: Implications for mechanisms of transcriptional regulation. 26. K. Nyame, R.D. Cummings and R.T. Damian, J. Purusitol., 1988, 74, 562-572, Characterization of the N- and 0-linked oligosaccharides in glycoproteins synthesized by Schistosomn mansoni schistosomula.

664

39 Structures and Functions of Nuclear and Cytoplasmic Glycoproteins

27. E. Handman, L.D. Barnett, A.H. Osborn, J.W. Goding and P.J. Murray, Mol. Biochem. Parasitol., 1993, 62, 61-72, Identification, characterisation and genomic cloning of a 0-linked N-acetylglucosamine-containing cytoplasmic Leishmania glycoprotein. 28. J.E. Hansen, 0. Lund, K. Rapacki and S. Brunak, Nucleic Acids Res., 1997, 25, 278-282, 0GLYCBASE version 2.0: A revised database of 0-glycosylated proteins. 29. H.H. Wandall, H. Hassan, E. Mirgorodskaya, A.K. Kristensen, P. Roepstorff, E.P. Bennett, P.A. Nielsen, M.A. Hollingsworth, J. Burchell, J. Taylor-Papadimitriou and H. Clausen, J. Biol. Chem., 1997, 272, 23503-23514, Substrate specificities of three members of the human UDP-N-acetyl-a-D-galactosamine: Polypeptide N-acetylgalactosaminyltransferase family, GalNAc-TI, -T2, and -T3. 30. F.K. Hagen and K. Nehrke, J. Biol. Chem., 1998, 273, 8268-8277, cDNA cloning and expression of a family of UDP-N-acetyl-D-galactosamine: polypeptide N-acetylgalactosaminyltransferase sequence homologs from Caenorhabditis elegans. 31. H. Clausen and E.P. Bennett, Glycohiology, 1996, 6 , 635-646, A family of UDP-GalNAc: Polypeptide N-acetylgalactosaminyl-transferasescontrol the initiation of mucin-type 0-linked glycosylation. 32. R.S. Haltiwanger, G.D. Holt and G.W. Hart, J. Biol. Chem., 1990, 265, 2563-2568, Enzymatic addition of 0-GlcNAc to nuclear and cytoplasmic proteins. Identification of a uridine diphospho-N-acetylglucosamine:peptide 0-N-acetylglucosaminyltransferase. 33. R.S. Haltiwanger, M.A. Blomberg and G.W. Hart, J. Biol. Chem., 1992, 267, 9005-9013, Glycosylation of nuclear and cytoplasmic proteins. Purification and characterization of a uridine diphospho-N-acetylglucosamine:polypeptide P-N-acetylglucosaminyltransferase. 34. L.-Y.D. Dong and G.W. Hart, J. Biol. Chem., 1994, 269, 19321-19330, Purification and Characterization of an 0-GlcNAc Selective N-Acetyl-P-D-glucosaminidase from Rat Spleen cytosol. 35. L.K. Kreppel, M.A. Blomberg and G.W. Hart, J. Biol. Chem., 1997, 272, 9308-9315, Dynamic glycosylation of nuclear and cytosolic proteins-Cloning and characterization of a unique 0-GlcNAc transferase with multiple tetratricopeptide repeats. 36. W.A. Lubas, D.W. Frank, M. Krause and J.A. Hanover, J. Biol. Chem., 1997, 272, 93169324, 0-linked GlcNAc transferase is a conserved nucleocytoplasmic protein containing tetratricopeptide repeats. 37. J.R. Lamb, S. Tugendreich and P. Hieter, TZBS, 1995, 20, 257-259, Tetratrico peptide repeat interactions: to TPR or not to TPR? 38. C.F. Chou, A.J. Smith and M.B. Omary, J. Bid. Chem., 1992, 267, 3901-3906, Characterization and Dynamics of 0-linked Glycosylation of Human Cytokeratin 8 and 18. 39. E.P. Roquemore, M.R. Chevrier, R.J. Cotter and G.W. Hart, Biochemistry, 1996, 35, 35783586, Dynamic 0-GlcNAcylation of the small heat shock protein aB-crystallin. 40. R.S. Haltiwanger, K. Grove and G.A. Philipsberg, J. Biol. Chem., 1998, 273, 3611-3617, Modulation of 0-Linked N-acetylglucosamine Levels on Nuclear and Cytoplasmic Proteins In Vivo Using the Peptide 0-GlcNAc-P-N-acetylglucosaminidase Inhibitor 0-(2-Acetamido2-deoxy-D-glucopyranosylidene) amino-N-Phenylcarbamate. 41. K.P. Kearse and G.W. Hart, Proc. Natl Acad. Sci. USA, 1991, 88, 1701-1705, Lymphocyte activation induces rapid changes in nuclear and cytoplasmic glycoproteins. 42. M.D. Roos, 1.0. Han, A.J. Paterson and J.E. Kudlow, Am. J. Plzysiol. Cell Physiol., 1996, 270, C803-C811, Role of glucosamine synthesis in the stimulation of TGF-a gene transcription by glucose and EGF. 43. I. Han and J.E. Kudlow, Mol. Cell. B i d , 1997, 17, 2550-2558, Reduced 0 glycosylation of Spl is associated with increased proteasome susceptibility. 44. S. Marshall, W.T. Garvey and R.R. Traxinger, FASEB J., 1991, 5, 3031-3036, New insights into the metabolic regulation of insulin action and insulin resistance: role of glucose and amino acids. 45. D.A. McClain and E.D. Crook, Diabetes, 1996, 45, 1003-1009, Hexosamines and insulin resistance. 46. W.G. Kelly, M.E. Dahmus and G.W. Hart, J. Biol. Chem., 1993, 268, 10416-10424, RNA polymerase I1 is a glycoprotein. Modification of the COOH-terminal domain by 0GlcNAc.

Rejerences

665

47. T.-Y. Chou, G.W. Hart and C.V. Dang, J. Biol. Chern.; 1995, 270, 18961-18965, c-Myc is glycosylated at threonine 58, a known phosphorylation site and a mutational hot spot in lymphomas. 48. C.S. Arnold, G.V.W. Johnson, R.N. Cole, D.L.-Y. Dong, M. Lee and G.W. Hart, J. Bid. Chem., 1996, 271, 28741-28744, The Microtubule-associated Protein Tau is Extensively Modified with 0-linked N-acetylglucosaniine. 49. N.-0. Ku and M.B. Omary, J. Bid. Clzem., 1995, 270, 11820-11827, Identification and mutational analysis of the glycosylation sites of human keratin 18. 50. B.E. Kemp and R.B. Pearson, Trends Biochem. Sci., 1990, 15, 342-346, Protein kinasc recognition sequence motifs. 51. M.D. Roos, K.H. Su, J.R. Baker and J.E. Kudlow, Mol. Cell. Biol., 1997, 17, 6472--6480,0glycosylation of an Spl -derived peptide blocks known Spl protein interactions. 52. B. Datta, M.K. Ray, D. Chakrabarti, D.E. Wylie and N.K. Gupta, J. Biol. Chem., 1989,264, 20620-20624, Glycosylation of eukaryotic peptide chain initiation factor 2 (eIF-2)-associated 67-kDa polypeptide (p6’) and its possible role in the inhibition of eIF-2 kinase-catalyzed phosphorylation of the eIF-2 u-subunit. 53. M.K. Ray, B. Datta, A. Chakraborty, A. Chattopadhyay, S. Meza-Keuthen and N.K. Gupta. Proc. Nut1 Acad. Sci. USA, 1992, 8Y, 539-543, The eukaryotic initiation factor 2-associated 67-kDa polypeptide (p6’) plays a critical role in regulation of protein synthesis initiation in animal cells. 54. B. Datta, D. Chakrabarti, A.L. Roy and N.K. Gupta, Proc. Nail Acad. Sci. U.S.A . , 1988, 85, 3324-3328, Roles of a 67-kDa polypeptide in reversal of protein synthesis inhibition in hemedeficient reticulocyte lysate. 55. S. Wu, S. Gupta, N. Chatterjee, R.E. Hileman, T.G. Kinzy, N.D. Denslow, W.C. Merrick, D. Chakrabarti, J.C. Osterman and N.K. Gupta, J. Biol. Chem., 1993,268, 10796-10801, Cloning and characterization of complementary DNA encoding the eukaryotic initiation factor 2associated 67-kDa protein (p6’). 56. A. Chakraborty, D. Saha, A. Bose; M. Chatterjee and N.K. Gupta, Biochemistry, 1994, 33. 6700-6706, Regulation of elF-2 a-Subunit Phosphorylation in Reticulocyte Lysate. 57. A. Bose, D. Saha and N.K. Gupta, Arch. Biochem. Biophys., 1997, 342, 362-372, Viral infection. 1. Regulation of protein synthesis during vaccinia viral infection of animal cells. 58. D. Saha, S.Y. Wu, A. Bose. N. Chatterjee, A. Chakraborty, M. Chatterjee and N.K. Gupta, Arch. Biochem. Biophys., 1997, 342, 373-382, Viral infection. 2. Hemin induces overexpression of p67 as it partially prevents appearance of an active p67-deglycosylase in baculovirus-infected insect cells. 59. J. Selzer, F. Hofmann, G. Rex, M. Wilm, M. Mann, I. Just and K. Aktories, J. Biol. Chem., 1996, 271; 25173 -25 177: Clostridium nouyi a-Toxin-catalyzed Incorporation of GlcNAc into Rho Subfamily Proteins. 60. A.J. Koleske and R.A. Young, Trend.?Biochem. Sci., 1995, 20, 113-1 16, The RNA polymerase I1 holoenzymc and its implications for gene regulation. 61. F.I. Comer and G.W. Hart, Mol. Biol. Cell, 1995, 6, 415a, Dynamic Glycosylation of RNA Polymerase 11. (Abstract) 62. D.A. McClain, A.J. Paterson, M.D. Roos, X. Wei and J.E. Kudlow, Proc. Nut/ Acud. Sci. U.S. A . , 1992, 89, 8 150-8 154, Glucose and glucosamine regulate growth factor gene expression in vascular smooth muscle cells. 63. M.C. Daniels, P. Kansal, T.M. Smith, A.J. Paterson, J.E. Kudlow and D.A. McClain, Mol. Endocrinol., 1993, 7, 1041-1048, Glucosc Regulation of Transforming Growth Factor-u Expression is Mediated by Products of the Hexosamine Biosyntheis Pathway. 64. V. Kolm-Litty, U. Sauer, A. Nerlich, R. Lehmann and E.D. Schleicher, J. Clin. Znoest., 1998, 101, 160-169, High glucose-induced transforming growth factor D l production is mediated by the hexosamine pathway in porcine glomeruiar mesangial cells. 65. J. Wang, R. Liu, M. Hawkins. N. Barzilai and L. Rossetti, Nature, 1998, 393, 684-688, A nutrient-sensing pathway regulates leptin gen expression in muscle and fat. 66. R.R. Traxinger and S. Marshall, J. Bid. Chem., 1992,267, 9718-9723, Insulin Regulation of Pyruvate Kinase Activity in Isolated Adipocytes: Crucial Role of Glucose and the Hexosamine Biosynthesis Pathway in the Expression of Insulin Action.

666

39 Structures and Functions o j Nuclear und Cytoplasmic Glycoproteins

67. E.D. Crook, J. Zhou, M. Daniels, J.L. Neidigh and D.A. McClain, Diabetes, 1995, 44, 314320, Regulation of glycogen synthase by glucose, glucosamine, and glutamine: fructose-6phosphate amidotransferase. 68. R.R. Traxinger and S. Marshall, J. Bid. Chem., 1991, 266, 10148-10154, Coordinated Regulation of Glutamine: Fructose-6-Phosphate Amidotransferase Activity by Insulin, Glucose, and Glutamine: Role of Hexosamine Biosynthesis in Enzyme Regulation. 69. H. Yki-Jarvinen, A. Virkamaki, M.C. Daniels, D. McClain and W.K. Gottschalk, Metabolism, 1998, 47, 449-455, Insulin and glucosamine infusions increase 0-linked N-acetylglucosamine in skeletal muscle proteins in vivo. 70. B. Gonzalez-Yanes, J.M. Cicero, R.D. Brown, Jr. and C.M. West, J. Biol. Chem., 1992, 267, 9595-9605, Characterization of a cytosolic fucosylation pathway in Dictyostelium. 71. E. Kozarov, H. Van der Wel, M. Field, M. Gritzali, R.D. Brown, Jr. and C.M. West, J. Biol. Chem., 1995, 270, 3022-3030, Characterization of FP21, a cytosolic glycoprotein from Dictyostelium. 72. C.M. West, E. Kozarov and P. Teng-umnuay, Gene, 1997, 200, 1-10, The cytosolic glycoprotein FP21 of Dictyostelium discoideum is encoded by two genes resulting in a polymorphism at a single amino acid position. 73. P. Teng-umnuay, H.R. Morris, A. Dell, M. Panico, T. Paxton and C.M. West, J. Biol. Chem., 1998,273, 18242-18249, The cytoplasmic F-box binding protein SKPl contains a novel pentasaccharide linked to hydroxyproline in Dictyostelium. 74. P. Teng-umnuay and C.M. West, Glycobioloyy, 1998,8, 1105, Evidence that a protein involved in ubiquitination is modified by a novel cytoplasmic glycosylation pathway. (Abstract) 75. C.M. West, T. Scott-Ward, P. Teng-umnuay, H. Van der Wel, E. Kozarov and A. Huynh, J. Biol. Chem., 1996,271, 12024-12035, Purification and characterization of an ul,2-L-fucosyltransferase, which modifies the cytosolic protein FP21, from the cytosol of Dictyostelium. 76. C. Srisomsap, K.L. Richardson, J.C. Jay and R.B. Marchase, J. Biol. Chem., 1988, 263, 17792-1 7797, Localization of the Glucose Phosphotransferase to a cytoplasmically Accessible Site on Intracellular Membranes. 77. R.B. Marchase, P. Bounelis, L.M. Brumley, N. Dey, B. Browne, D. Auger, T.A. Fritz, P. Kulesza and D.M. Bedwell, J. Bid. Chem., 1993, 268, 8341-8349, Phosphoglucomutase in Saccharomyces cerevisiaeis a cytoplasmic glycoprotein and the acceptor for a Glc-phosphotransferase. 78. B.H. Satir, C. Srisomsap, M. Reichman and R.B. Marchase, J. Cell Biol., 1990,111, 901-907, Parafusin, an Exocytic-sensitive Phosphoprotein, Is the Primary Aceeptor for the Glucosylphosphotransferase in Paramecium tetraurelia and Rat Liver. 79. N.A. Veyna, J.C. Jay, C. Srisomsap, P. Bounelis and R.B. Marchase, J. Neurochem., 1994,62. 456-464, The addition of glucose-1-phosphate to the cytoplasmic glycoprotein phosphoglucomutase is modulated by intracellular calcium in PC 12 cells and rat cortical synaptosomes. 80. S.V. Subramanian and B.H. Satir, Proc. Nut1 Arud. Sci. USA, 1992, 89, 11297-11301, Carbohydrate cycling in signal transduction: Parafusin, a phosphoglycoprotein and possible Ca2+-dependent transducer molecule in exocytosis in Paramecium. 81. S.V. Subramanian, E. Wyroba, A.P. Andersen and B.H. Satir, Proc. Nut1 Acad. Sci. U S A , 1994, 91, 9832-9836, Cloning and sequencing of parafusin, a calcium-dependent exocytosisrelated phosphoglycoprotein. 82. E. Wyroba, A.W. Hoyer, P. Storgaard and B.H. Satir, Eur. J. Cdl Biol., 1995, 68, 419-426, Mammalian homologuc of the calcium-sensitive phosphoglycoprotein, parafusin. 83. R.B. Marchase, K.L. Richardson, C. Srisomsap, R.R. Drake, Jr. and B.E. Haley, Arch. Biochem. Biophys., 1990, 280, 122-129, Resolution of phosphoglucomutase and the 62-kDA acceptor for the glucosylphosphotransferase. 84. C. Srisomsap, S.A. Mattox, R.D. Cummings and R.B. Marchase, J. Cell B i d , 1988, 107,406, The Glucosylphophotransferase Recognizes 0-linked Mannose Residues on its 62 kDa Acceptor. (Abstract) 85. C. Srisomsap, K.L. Richardson, J.C. Jay and R.B. Marchase, J. Biol. Chem., 1989, 264, 20540-20546, An a-glucose-1-phosphate phosphodiesterase is present in rat liver cytosol. 86. N.B. Dey, P. Bounelis, T.A. Fritz, D.M. Bedwell and R.B. Marchase, J. Biol. Chem., 1994, 269, 27143-27148, The glycosylation of phosphoglucomutase is modulated by carbon source and heat shock in Saccharomyces cerevisiae.

References

661

87. J. Pitcher, C. Smythe, D.G. Campbell and P. Cohen, Eur. J. Biochem., 1987, 169, 497-502, Identification of the 38-kDa subunit of rabbit skeletal muscle glycogen synthase as glycogenin. 88. J. Pitcher, C. Smythe and P. Cohen, Eur. J. Biochem., 1988, 176, 391-395, Glycogenin is the priming glucosyltransferase required for the initiation of glycogen biogenesis in rabbit skeletal muscle. 89. C. Smythe, F.B. Caudwell, M. Ferguson and P. Cohen, EMBO J., 1988, 7, 2681-2686, Isolation and structural analysis of a peptide containing the novel tyrosyl-glucose linkage in glycogenin. 90. J. Mu, A.V. Skurat and P.J. Roach, J. Biol. Chrm., 1997,272, 27589-27597, Glycogenin-2, a novel self-glucosylating protein involved in liver glycogen biosynthesis. 91. V.P. Bhavanandan and E.A. Davidson, Proc. Nut/ Acad. Sci. USA., 1975, 72, 2032-2036, Mucopolysaccharides associated with nuclei of cultured mammalian cells. 92. G.S. Stein, R.M. Roberts, J.L. Davis, W.J. Head, J.L. Stein, C.L. Thrall, J. Van Veen and D.W. Welch, Nature, 1975, 258, 639-641 Are glycoproteins and glycosaminoglycans components of the eukaryotic genome? 93. D.R.R. Hiscock. M. Yanagishita and V.C. Hascall, J. Bid. Chem., 1994, 269, 4539-4546, Nuclear localization of glycosaminoglycans in rat ovarian granulosa cells. 94. M. Ishihara, N.S. Fedarko and H.E. Conrad, J. Biol. Chern., 1986, 261, 13575-13580, Transport of heparan sulfate into the nuclei of hepatocytes. 95. N.S. Fedarko and H.E. Conrad. J. Cell Biol., 1986, 102, 587-599, A unique heparan sulfate in the nuclei of hepatocytes: structural changes with the growth state of the cells. 96. N.S. Fedarko, M. Ishihara and H.E. Conrad, J. Cell Physiol., 1989, 139, 287-294, Control of cell division in hepatoma cells by exogenous heparan sulfate proteoglycan. 97. Y. Liang, M. Haring, P.J. Roughley, R.K. Margolis and R.U. Margolis, J. Cell Bid., 1997, 139, 851-864, Glypican and biglycan in the nuclei of neurons and glioma cells: Presence of functional nuclear localization signals and dynamic changes in glypican during the cell cycle. 98. P. Codogno, C. Bauvy, A.-P. Seve, M. Hubert, E. Ogier-Denis, M. Aubery and J. Hubert, J. Cell. Biochem., 1992, 50, 93-102, Evidence for the Presence of Complex High-Molecular Mass N-Linked Ohgosaccharides in Intranuclear Glycoproteins From HeLa Cells. 99. N.C. Chandra, M.J. Spiro and R.G. Spiro, J. Biol. Chem., 1998, 273, 19715-19721, Identification of a glycoprotein from rat liver mitochondria1 inner membrane and demonstration of its origin in the endoplasmic reticulum. 100. S.H. Barondes, D.N.W. Cooper, M.A. Gitt and H. Leffler, J. Bid. Chem., 1994,269, 2080720810, Galectins: Structure and Function of a Large Family of Animal Lectins. 101. Y.H. Sahraoui, A.P. Seve, M.A. Doyennette-Moyne, L. Saffar. M. Felin, M. Aubery, L. Gattegno and J. Hubert, J. Cell. Biochem., 1996, 62, 529-542, Nuclear and cytoplasmic expressions of the carbohydrate-binding protein CBP70 in tumoral or healthy cells of the macrophagic lineage. 102. J.L. Wang, J.G. Laing and R.L. Anderson, GIycobiology, 1991, I , 243-252, Lectins in the Cell Nucleus. 103. S. Goletz, F.G. Hanisch and U. Karsten, J. Cell Sci., 1997, 110, 1585-1596, Novel alpha GalNAc containing glycans on cytokeratins are recognized in vitro by galactins with type I1 carbohydrate recognition domains. 104. M. Felin, M.-A. Doyennette-Moyne, Y. Hadj-Sahraoui, M. Aubery, J. Hubert and A.-P. Seve, J. Cell. Biochem., 1994,56, 527--535,Identification of two nuclear N-acetylglucosaminebinding proteins. 105. M. Felin, M.A. Doyennette-Moyen, C. Rousseau, H.C. Schroder and A.P. Seve, Glycobiology, 1997, 7, 23-29, Characterization of a putative 82 kDa nuclear ligand for the N-acetylglucosamine-binding protein CBP70. 106. S.F. Dagher, J.L. Wang and R.J. Patterson, Proc. Nut1 Acad. Sci. USA, 1995, 92, 1213-1217, Identification of galectin-3 as a factor in pre-mRNA splicing. 107. A. Vyakarnam, S.F. Dagher, J.L. Wang and R.J. Patterson, Mol. Cell Biol., 1997, 17, 47304737, Evidence for a role for galectin-1 in pre-mRNA splicing.

Carbohydrates in Chemistry and Biology Beat Emst, Gerald W. Hart, Pierre Sinay copyright Q WILEY-VCH Verlag GmbH. 2000

40 Structure and Functions of Mucins Joyce Taylor-Papadimitriou and Joy M. Burchrll

40.1 Classification of Mucins Mucins have been defined as glycoproteins with more than 50% of the molecule being made up of oligosaccharides attached in O-linkage to serines and threonines. Traditionally, epithelial mucins have been isolated from mucus, focusing on the GI tract, tracheobronchial products, cervical and submaxillary mucins. These are the extracellular gel forming mucins and the biophysical properties as well as the carbohydrate structures were of interest. Studies with reducing agents and proteases suggested that the molecules were made up of regions which were heavily glycosylated (protease insensitive) and domains which were not heavily glycosylated (protease sensitive domains). Effects of reducing agents suggested that multimers were made with disulfide bridges. More recent work has confirmed this basic structure. Several developments influenced the progress of mucin research in the 1980s and early 1990s. The first was perhaps serendipitous, and came out of studies in oncology where many investigators were developing antibodies to tumor-associated antigens. Because of their repetitive structure, mucins are highly immunogenic and many antibodies raised to epithelial cells, and the carcinomas which develop from them, reacted with these molecules, bringing in a new focus of interest, and a new group of investigators. The second factor which has influenced mucin research dramatically, relates to the cloning of the genes coding for the core proteins or apomucins, which has allowed both elucidation of mucin structure and manipulation of expression. Furthermore) cloning of genes coding for the glycosyltransferases involved in mucin-type O-glycosylation has made it possible to relate O-glycan structure to the activity of these enzymes. The use of antibodies to the core proteins of mucins allowed the cloning of the apomucin, and the first human mucin gene to be cloned was named MUCl [l]. MUCl is a membrane mucin, normally expressed at the apical surface of most glandular epithelial cells and is simpler in structure than the complex extracellular mucins, but may have complex functions. Other mucin genes cloned subsequently by similar methods are referred to as MUCZ, 3

670

40 Structure and Functions of Mucins MEMBRANE

LIGANDS FOR SELECTINS on LEUKOCYTES and ENDOTHELIAL CELLS (E.G. PSGL-1)

EXTRACELLULAR

( e.g. MUC2 )

EPITHELIAL

MUCINS

Figure 1. Types of mucin classified according to structure and type of cell expressing the mucin.

etc., and there are to date eight human mucins identified, although the full sequence of the core protein is known for only some (e.g. MUC1, MUC2, MUC4). Homologous rodent mucins are also being cloned as well as the bovine and porcine submaxillary mucins, and frog integumentary mucins. In addition to the epithelial mucins, membrane proteins which are heavily 0glycosylated, and are considered mucins, are expressed in leukocytes and endothelial cells; these are the selectin ligands such as PSGL-1. Although several carbohydrate structures, (based on sialylated and/or sulfated LeXand Lea) have been identified as being involved in the selectin-ligand interaction, the core protein carrying the oligosaccharides is also important. Figure 1 summarizes the types of mucin, and this Chapter will focus on the epithelial mucins since the structure and function of the selectin ligands is reviewed in detail in Chapter 37.

40.2 The Epithelial Mucins All mucins share a common structural feature-they contain one or more domains made up of tandem repeats rich in proline serine and threonine which serve as a scaffold for the addition of 0-glycans. Because of the conservation of sequence in the tandem repeats of an epithelial mucin, recombination is common and the number of tandem repeats in a specific mucin is variable, making the gene polymorphic [ l , 21. The actual sequence and length of the tandem repeat domain varies with the different mucins and in homologs of the same mucin in different species, conservation being limited to maintaining serines and threonines as sites of attachment. The final mucin product is a highly glycosylated molecule, and since the pattern of glycosylation may vary in different tissues and in disease, the final mucin may have different properties and functions in different tissues.

40.3 Mucin Type 0-Glycosylation Pathways 0-Glycosylation pathways have been better defined in recent years, partly due to the cloning of the enzymes involved. It is clear that several enzymes may catalyze

40.3 Mucin Type O-Glycosylution Puthwuys

671

the same reaction and that enzymes yielding different products can act on the same substrate. This means that competition can occur for the substrate, providing the positions of the different enzymes overlap in the Golgi pathway. Thus the position of the enzymes involved in mucin-type O-glycosylation is as important as the level of activity in determining the final composition of O-glycans added.

40.3.1 Initiation of O-Glycosylation This begins in the Golgi apparatus, and a family of enzymes catalyze the transfer of UDP GalNAc to serine or threonine (UDP-Ga1NAc:polypeptide glycosyltransferases, described in detail in Chapter 16) The different ppGalNAc Ts have distinct but overlapping specificities, defined by the peptide sequence, and the same enzyme can add GalNAc to both serine and threonine [ 3 ] . Thus the sequences flanking the serine or threonine do influence whether it is glycosylated, but cannot be inferred merely from a database analysis of in vivo substrates [4]. The three GalNAcTs which have been localized have been found to be present in the cis, medial and trans Golgi apparatus, indicating that chain initiation can proceed throughout the Golgi pathway [ 5 ] .

40.3.2 Chain Extension After the addition of GalNAc, various core structures are formed by the addition of different sugars (see Chapter 18 for details). Chains are extended from these cores by the addition of N-acetylglucosamine (GlcNAc) and galactose alternately, to give polylactosamine side chains which may be straight or branched. The formation of the different cores varies with the tissue, however, a pathway which is commonly used is through core 1 and core 2. Where biosynthesis is through core 2 the addition of the GlcNAc is crucial for chain extension (see Figure 2).

40.3.3 Chain Termination The terminal epitopes of the O-glycans on mucins are probably the most important in determining whether the molecule plays a role in cell adhesion phenomena. The epitopes recognised by antibodies related to the ABO, and Lewis blood group antigens are also found in this terminal region. Terminal sugars added in a-linkage, include sialic acid, fucose, galactose, GalNAc and GlcNAc and some sulfation of the sugars in terminal structures may also occur. The number of enzymes which catalyze the addition of sialic acid is increasing daily and at present is greater than ten [6]. Those acting on the early stages of O-glycosylation can influence dramatically the final O-glycan structure as they inhibit chain extension (see Figure 2).

672

40 Structure and Functions of Mucins GalNAc- (serine or threonine)

I

(core 1)Gal f~1,3 GalNAc

SA

c(

2,6 GalNAc (Sidlyl Tn)

GlcNAc

B 1,6 Gal p 1,3 GalNAc (core 2 )

SA a 2,3 Gal p 1,3 GalNAc) (Sialyl T)

Figure 2. Chain extension or termination of 0-glycans on MUCl expressed by normal and malignant mammary epithelial cells.

40.4 Expression of Epithelial Mucins Identification of specific mucins expressed in individual tissues can be done by in situ hybridization studies on tissue sections [7]. Antibody staining can be equivocal because epitopes on the core protein which define the mucin may be masked by the carbohydrate side chains. Studies to date indicate that MUCl shows the widest range of expression, and is upregulated in breast and other carcinomas [8, 91. Many of the other mucins are more restricted in their expression, for example MUCSAC is more or less restricted to the stomach and respiratory tract [lo]. However, rat sialomucin (the rat homologue of MUC4) is expressed by a number of tissues and is found in the lactating mammary gland, large and small intestines, trachea and uterus. Clearly more than one mucin can be expressed by the same tissue but it is not yet clear whether an individual cell can synthesize more than one mucin.

40.5 The Complex Gel-Forming Mucins: Processing and Function Secreted mucins are the glycoproteins found in mucus giving this material its elasticity and viscosity. Mucus represents one of the main interfaces between the organism and its environment and clearly the major function of the large mucins such as the intestinal mucin MUC2 is to protect the underlying epithelial cells from injury or from infection by invading micro-organisms. The large number of 0-glycans carried on the tandem repeat domains are largely responsible for the molecular properties which lead to this protective function. Other cysteine-rich sequences in the protein core which are conserved in several of the complex mucins are respon-

40.5 The Complex Gel-Forming Mucins: Processing and Function

673

MUC2

I

I

I pro-peptide

mature vWF

Figure 3. Domain structure of MUC2 and vWF. The four D domains of MUC2 and the domain structure of vWF are illustrated.

sible for the oligomerization and packaging of the molecules in the cell. A comparison of the sequences of several complex mucins with that of von Willebrand factor (vWF) shows conservation of the cysteine-rich domains which, in this molecule have been shown to be involved in its oligomerization (see Figure 3). The half cystine residues at the carboxyl end of vWF are also found in porcine submaxillary mucin, frog integumentary mucin FIMB. 1, rat intestinal mucin, bovine submaxillary mucin and the human complex mucins MUC2, MUCSB and MUCSAC. The amino terminal region of vWF contains three disulfide rich domains (D domains) which share homology with similar domains in porcine submaxillary mucin, with the frog mucin and with human MUC2. In vWF, the carboxyl cysteine-rich domains are involved in dimerisation, while the D domains at the amino end are involved in the formation of multimers (see Figure 3). Recently, the availability of clones for MUC2 and for porcine submaxillary mucin has allowed the transfection of cells with specific domains to test for their role in oligomerization and trafficking [ I 1, 121. These experiments have shown that the D1, D2 and D3 domains of the porcine mucin are involved in the formation of trimers in the Golgi, after Nglycosylation occurs in the ER. To obtain the large multimeric mucin, both the amino and carboxyl domains were required. Figure 4 illustrates the processing pathway as described by Perez-War et al. [ 111. Dimerization of MUC2 is initiated in the ER, preceeding N-glycosylation, which is apparently required for correct folding. Thus, intracellular dimerization and trimerization means that the size of the molecule being 0-glycosylated in the Golgi is very large indeed. How these molecules interact in an ordered fashion with the glycosylating enzymes is difficult to vizualize, particularly since in the submaxillary mucin most of the available sites have been shown to be glycosylated [4]. (The reader is referred to [ 11, 121 for references and a comprehensive description of mucin assembly.)

674

0

Y

1

40 Structure and Functions of Mucins

Amino terminal D1, D2, D3 domains Tandem repeat and unique domains Disulfide-rich,carboxyl domains N-linked glycosylation 0-linked glycorylation

Figure 4. Diagramatic representation of the processing of porcine submaxillary mucin (based on 1111). Step 1, the much is synthesizes in the ER-bound ribosomes and then translocated and folded in the ER. Step 2, immediately after synthesis the apomucin much dimerizes in the ER through inter-chain disulfide bonds at the carboxy-terminal domains. Step 3, the dimers are N-glycosylated and transported to the Golgi apparatus. Step 4, the dimeric molecules are 0-glycosylated and the biosynthesis of N- and 0-linked glycans proceeds throughout the medial and cis Golgi. Step 5 , in the trans Golgi the multimeric species are assembled by interchain disulfide bonds involving the amino terminal D domains.

40.6 Epithelial Membrane Mucins Until recently the only human epithelial membrane mucin was the MUC1 mucin although a rat sialomucin (sialomucin complex, SMC) with a similar structure has also identified and characterized [ 131. With the determination of the full length sequence of the MUC4 apomucin, it has become apparent that the rat sialomucin is

40.7 Studies Related to the MUCl Muciri

675

the homolog of MUC4 [ 141. All three of these mucins are processed in a similar fashion in that immediately post translation the transmembrane and cytoplasmic domain of the apomucin is cleaved from the major part of the extracellular domain but remains attached to form a heterodimer at the cell membrane [13, 151. The bonds involved in maintaining the heterodimer are not defined, and only seem to be disrupted by SDS, but not by urea or reducing agents. Rat SMC heterodimer consists of an 0-glycosylated mucin subunit (ASGP-1) bound to an integral membrane protein, ASGP-2. ASGP-2 contains two EGF-like domains [ 161 which have all the consensus residues present in active members of the E G F growth factor family. Interestingly, SMC has been shown to bind to the erbB2 receptor and it has been speculated that this interaction may play a role in the constitutive phosphorylation of erbB-2 observed in the 13762 ascites cells. SMC was originally isolated from 13762 which is a highly metastatic rat mammary adenocarcinoma [ 171 and the complex has been implicated in metastasis and in the protection of tumor cells from natural killer cells. Over-expresssion of SMC accomplished by transfection into human melanoma cells, induces cell detachment and cell-cell dissociation [ 181, similar to that observed with MUCl (see below). In the normal situation, the expression of soluble SMC and membrane-associated SMC on the apical surface of cells in the rat airways suggest a role for SMC in the formation of the protective barrier in airway epithelium. Furthermore, the expression of SMC by the adult rat intestine, lactating mammary gland and uterus, where it is hormonally regulated, suggests that it may have complex functions in the normal, adult rat. In the case of MUC1, the sequence of the transmembrane and cytoplasmic tail in various mammalian species is highly conserved, suggesting an important function [ 191. This has not however been defined, but interactions with signalling molecules have been reported [20]. An alternatively spliced variant of MUCl missing the tandem repeat domain has been described as being expressed in breast cancers, and the signal transduction properties of MUCl have been attributed to this variant P11. MUCl was the first mucin to be cloned and compared to the large complex mucins has a relatively simple structure. Consequently this mucin has been extensively studied. It has been the focus of attention in a variety of studies related to cell adhesion and various formulations based on MUCl are being tested for immunotherapy of cancer patients. To put these studies in context, it is relevant to consider the changes in the pattern of glycosylation of MUCl which can occur in cancer.

40.7 Studies Related to the MUCl Much 40.7.1 Changes in the Patterns of 0-Glycosylation of MUCl in Breast Cancer Differences in sites of glycosylation

There are five potential 0-glycosylation sites in each tandem repeat of MUCl and the use of these sites in the normal and malignant mammary gland has recently

676

40 Structure and Functions of Mucins

been analyzed. Analysis of the mucin produced by the normal lactating mammary gland has shown that while all of these sites can be glycosylated, the average number of 0-glycans added to each tandem repeat is around 2.5 [22]. Analysis of the sites of glycosylation of the MUCl mucin produced by the breast cancer cell line T47D however suggests that all five sites are occupied in all the tandem repeats (FG Hanisch, personal communication). One possible explanation for this difference is that the 0-glycans on the T47D mucin which are shorter (see below) allow initiation of glycosylation to proceed throughout the Golgi apparatus, while the longer side chains which are built on to the normal mucin block access of the GalNAcTs at a certain point.

Changes in the composition of 0-glycans added to MUCl in Breast Cancer In the normal human mammary gland, the addition of GalNAc to serines or threonines is followed by the addition of galactose to form core 1 which then acts as a substrate for the core 2P1,6GlcNAc T (C2GnT) enzyme, leading to the formation of core 2 (see Figure 2). Type I1 polylactosamine side chains are then formed and terminated by sialic acid or fucose [23, 241. Thus the 0-glycans on the normally glycosylated mucin are core 2-based. Analysis of the structure of the 0-glycans attached to MUCl produced by breast cancer cell lines however has shown the structure of the 0-glycans is core 1 and not core 2 based [25, 261. The differences in glycosylation seen in cancer results in the exposure of core protein epitopes which are normally masked: The SM3 antibody recognises such an epitope [27] and specifically binds to malignant but not normal breast epithelial cells. The direct analyses with MUCl confirm the earlier work of the late George Springer, who demonstrated the appearance of the T epitope in breast cancer [28] and used a preparation of erythrocytes expressing this antigen in the treatment of breast cancer patients. The sialyl Tn epitope (NeuAca2,6 GalNAc) has been reported to be specifically expressed in several cancers, including gastric and breast, and has been used, coupled to a carrier protein as an immunogen for cancer patients (see below).

Correlation in changes of Glycosyltransferase activities with changes in 0-glycan structure in Breast Cancer An examination of Figure 2 shows that core 1 can be a substrate for an enzyme leading to chain extension via core 2 or for an enzyme which, by adding sialic acid to core 1 to form sialyl T, terminates chain extension. Analysis of the activity of the enzymes catalyzing these reactions in extracts of normal and malignant breast epithelial cell lines showed that the a2,3 sialyltransferase activity was increased 8-10 fold in three breast cancer cells lines relative to the normal cell line (MTSVl-7), while the C2GnT enzyme activity was absent in two of the breast cancer cell lines (T47D and BT20) and reduced by 50%)in a third [29]. Observations at the mRNA level revealed that C2GnT mRNA was essentially absent in the T47D and BT20 lines. Clearly in measuring the activity of a cell extract several enzymes may be involved in the same reaction. In the case of the activity catalyzing the formation of core 2 from core 1, it is likely that the L enzyme [30] is being measured as the M

40.7 Studies Related to the M U C l Mucin

617

enzyme (also catalyzing the formation of core 4 [31]), is catalytically absent in these cells [29]. There are at least three sialyltransferases which could theoretically catalyze the addition of sialic acid in a2,3 linkage to GalP1,3GalNAc [6]. However, in situ analysis of tumor sections with a specific ST3Gal I probe [32] shows a marked increase in expression of this specific sialyltransferase in primary breast cancers. Significantly, the increased expression appeared to be correlated with the invasiveness of the tumor, and with the expression of a tumor-associated epitope recognized by the antibody. It should be noted of course that the changes in expression of glycosyltransferases seen in breast cancer will affect the glycosylation not only of MUCl but of any other molecule undergoing mucin-type O-glycosylation.

40.7.2 Changes in Glycosylation of MUCl in other Cancers Changes in the carbohydrate epitopes carried on MUCl have been detected in colon cancer by purifying components from cancer cell lines which express the sialyl LeXepitope which is not well expressed in the normal colon [33]. The full structures have been analyzed, but the direct comparison with the complete structures carried on the much produced by normal colon has not been done. However, it is clear that the SM3 epitope, which is absent from normal colon can be exposed in some colon cancers [34]. This epitope also appears in ovarian cancers [34] and gastric carcinomas [35], suggesting some modification of the side chains in these diseases. The sialyl Le" epitope is expressed on the glycoform of the selectin ligand PSGLl which interacts with P and E selectin (see Chapter 37). Data from Hansson and colleagues [33] show that MUCl carrying sialyl Lewis' purified from a colon carcinoma cell line, inhibited the adhesion of a leukocyte cell line to cells transfected with E-selectin (see below).

40.7.3 Effects of MUCl Expression on the Behavioral Properties of Cancer Cells Effects on cell interactions and tumourogenicity The changes in expression and post-translational modification of MUCl has stimulated investigations into whether this affects the behavior of cancer cells, particularly relating to interactions with other cells and with the extracellular matrix. In the normal glandular epithelial cell, MUCl expression is limited to the apical surface bordering a lumen. In cancer cells however, which have lost polarity, the mucin is expressed all over the surface. Because of its rod-like structure, the molecule extends more than 100-200 nm above the surface, which is 5-10-fold the length of most membrane molecules. By virtue of the high level of sialic acid, MUCl is also negatively charged and cells expressing high levels may repel each other. Such repulsive effects have been demonstrated by showing that MUCl transfectants show reduced aggregation as compared to the non-expressing parental cells [36] and interactions with the extracellular matrix are also inhibited (371. Such effects have also been demonstrated to be caused by the over-expression of rat SMC [18]. With E-

678

40 Structure and Functions of Mucins

cadherin mediated cell interactions, MUCl has been reported to be inhibitory (in L cells transfected with E-cadherin and MUC1) [38], or to enhance adhesion by interacting with p-catenin [39]. In considering the effects of MUCl on cell-cell interactions, it is clear that without specific interactions, for example with a lectin molecule, the long highly charged molecule can result in repulsion between cells. These inhibitory effects on cell interactions appear to depend on both the large size of the molecule and the negative charge [37, 381. However, where a specific interaction is possible-for example a particular carbohydrate epitope binding to a lectin, then cell interactions may be enhanced. Thus, MUCl has been reported to be a ligand for ICAM1 expressed by endothelial cells [40]. Moreover, MUCl can enhance antigen presentation to Tcells, possibly by operating through a lectin interaction [41]. Furthermore, MUCl has been shown to be a ligand for sialoadhesin, a macrophage restricted adhesion molecule, which specifically binds NeuSAca2,3Gal and so may be involved in recruiting macrophages into the tumor site [67]. When the glycosylation pattern of the much is changed in carcinomas, resulting in the production of different glycoforms, such carbohydrate-dependent interactions will be affected. The 0-glycans on the cancer mucin will also vary with the specific carcinoma so that, while selectin ligands may appear on MUCl produced by colon carcinomas, these will not be present in breast cancers. How the surface MUCl on cancer cells may influence metastatic progression is not clear, although in Muc-1 null mice, mammary tumor progression has been reported to be delayed [42]. 40.7.4 MUCl Expression and Immune Responses Cell-cell interactions in the context of cellular immune responses are of particular interest, since MUCl is under intensive study as a possible immunogen for immunotherapy of some cancer patients. Before considering these studies, it is appropriate to consider briefly the framework of the different compartments of the immune response to see how these might be effectively recruited by MUCl-based immunogens to kill cancer cells. The immune system was developed to reject invading pathogens, be these extracellular as with most bacteria and fungi, or intracellular parasites such as viruses. In vertebrates, there are two major compartments namely the “Innate” Immune Response and the “Adaptive” Immune Response. Lower organisms use only Innate Immunity where molecules with a specific structural pattern, characteristic of a class of pathogens (pathogen-associated molecular patterns or PAMPs) are recognised [43]. Examples of these are the mannans of yeasts, lipopolysacharides of Gram negative bacteria or bacterial DNA which contains unmethylated CpGs. Many of these PAMPs are rich in carbohydrates and can interact with lectins. Their interaction with lectins on antigen-presenting cells (APCs) activates the cells to increase expression of molecules (MHC, co-stimulatory molecules and cytokines) required to stimulate the adaptive immune response mediated by T cells and B cells. These cells recognise small domains or “epitopes” which, in the case of B cells are present in a large molecule or antigen and recognised by surface immunoglobulin. T cells

40.7 Studies Reluted to the MUCl Mucin

679

however, recognise peptides derived from the larger molecule which are presented either by MHC Class I molecules, which activate the T cell receptor (TCR) of CD8+ cytotoxic T cells (CTL), or by MHC Class I1 molecules which activate the TCR of CD4+ helper T cells. Depending on the cytokines produced, the adaptive immune response may be geared towards a Thl response where CTL develop, or towards a Th2 response with a bias towards the induction of antibodies. It is generally thought that a Thl response is required for effective rejection of cancer cells, although a role for antibodies is not excluded. In considering MUCl as an antigen, it is clear that where the response is to the whole molecule, the glycoform is of paramount importance. This will apply to the antibody response, and to lectin mediated interactions with the effector cells of the immune system. MUCl is a highly repetitive molecule, carrying multiple 0glycans which are different in the cancer mucin. It is therefore not unlikely that this particular glycoform may be capable of acting as a PAMP as well as enhancing cell-cell interactions. Whether the presentation of peptides by MHC molecules is occurring and whether this is affected by glycosylation remains to be clarified. The first indication that the molecule may be immunogenic in humans came from the observations of Finn and colleagues who isolated CTL from breast and ovarian cancer patients which killed MUC 1 expressing cells in a non-HLA-restricted fashion [43]. These studies indicated that only the carcinoma associated glycoform was recognised by the CTL [45] and it was suggested that the intact molecule made multiple interactions with the T cell receptors (TCR), the important epitope being found within the tandem repeat [46]. Only later was it reported that antibodies recognizing sequences in the tandem repeat are also found in cancer patients [47, 481. These original observations have been followed by studies from several groups, working with mouse models and human PBL (peripheral blood leukocytes), attempting to define the cellular responses to MU C l. Classical MHC restricted responses to epitopes within the tandem repeat have been reported for most of the common mouse strains [49], while HLA restricted CTL have also been isolated from HLA-A2 transgenic mice (501. Proliferative responses to the MUCl tandem repeat peptide possibly representing a helper T-cell response have been reported 1511. In contrast to the studies directed towards understanding and enhancing the immune response to MUC1, inhibitory effects of the mucin on T-cell responses have been reported [52, 531. Clearly, the possibility of the dual function of repelling cells by the highly charged extended mucin or enhancing adhesion via lectin interactions bring a complexity to the function of the molecule in the context of cell interactions which is not easily resolved.

40.7.5 Active Specific Immunotherapy Based on MUCl Animal models Before clinical studies can be initiated, some form of preclinical testing in animal models is necessary. To this end, syngenic and transgenic mouse models have been

680

40 Structure and Functions of Mucins

developed [54-561. Syngenic models have been used to investigate the efficacy of immunogens based on MUC1, including naked DNA [54], viral vectors [57],peptides [%] and liposome encapsulation of peptides [59]. Using MUCl transgenic mice it has been shown that tolerance to the mucin can be overcome without inducing autoimmunity [60, 611. However, even when using MUCl transgenic mice the immune response observed is representative of the murine repertoire. Studies using HLA A2 transgenic mice have shown that there are functional CTL A2 epitopes within MUCl [50] and probably the next generation of preclinical testing should involve the use of MUC1, HLA A2 double transgenic mice. However, even in this model other compartments of the immune response, eg receptors on APCs are still murine and so no animal model can truly predict the response that will be observed in humans.

Clinical studies Some of the humoral and cytotoxic immune responses to MUCl in cancer patients are known to be directed to specific sequences in the tandem repeat. As a result, and because there are many pragmatic advantages in using relatively short defined amino acid sequences, the clinical studies involving the use of peptides as the immunogen have concentrated on the tandem repeat of MUCl . Phase 1 studies have been reported that use tandem repeat peptides coupled to diphtheria toxid, BCG or KLH (for review see [62]). In the latter trial, class I restricted CTL were demonstrated after immunization with 16 amino acids from the tandem repeat coupled to KLH, and increase in CTL precursor frequency has been reported when patients have been immunized with five tandem repeats mixed with BCG. One formulation that looked very promising in the animal model was the use of a MUC 1 tandem repeat fusion protein coupled to oxidized mannan, which should target mannose receptors on APCs. However, although a predominantly CTL response was demonstrated in mice [63],early clinical trials conducted in patients with advanced breast and colorectal cancer show that in the clinical setting the predominant response was humoral, with few CTL being observed [64]. This apparent discrepancy highlights the limitations of translating studies in animal models into the clinical setting. The use of tandem repeat peptides as immunogens has the disadvantage that the response is being restricted to one domain of the mucin. A recombinant vacinnia virus expressing the full length cDNA for MUCl and IL-2 (VV-MUCl/IL2) has been shown to induce CTL in mouse models and used in a phase I study of patients with advanced breast cancer. Immunization was not associated with significant toxicity and immune responses were detected in some patients. A phase I1 multicentre trial using the VV-MUC 1/IL2 construct in patients with metastatic breast cancer is now in progress [62]. A direct way to exploit the aberrant glycosylation of MUCl for immunotherapy is the use of tumor-associated carbohydrate antigens which are found on MUCl, but may also be carried on other glycoproteins. A prospective, randomized clinical trial using STn in patients with breast cancer has recently been reported [65]. The treatment had minimal toxicity, and the highest antibody titres and survival were in patients pre-treated with low dose intravenous cyclophosphamide which appears to

References

68 1

reduce immunosuppression [66]. The results suggest a therapeutic effect for pretreatment with IV cyclophosphamide followed by immunisation with STn-KLH. A large multicentre trial comparing IV cyclophosphamide and STn-KLH/DETOX-B with IV cyclophosphamide and KLH/DETOX-B in the treatment of patients with breast cancer is about to commence.

40.8 Comments The field of mucin research has expanded dramatically in the last 10-15 years, bringing oncologists and immunologists alongside glycobiologists. The cloning of the apomucin genes has allowed the identification and characterization of individual mucins and provided tools for the experimental analysis of expression and function. It is to be expected that the continued study of these complex molecules will bring further insights into their biological function. References I . S.J. Gendler, C.A. Lancaster, J. Taylor-Papadimitriou, T. Duhig, N. Peat, J. Burchell, L. Pemberton, E.N. Lalani, D. Wilson. J. Biol. Chem., 1990, 265, 15286-15293. 2. V. Debailleul, A. Laine, G. Huet, P. Mathon, M.C. d’Hooghe, J.P. Aubert, N. Porchet. J. Biol. Chem., 1998,273, 881-890. 3. H.H. Wanda]], H. Hassan, K. Mirgorodskaya, A.K. Kristensen, P. Roepstorff, E.P. Bennett, P.A. Nielsen, M.A. Hollingsworth, J. Burchell, J. Taylor-Papadimitriou, H. Clausen. J. Biol. Chem., 1997,272, 23503-23514. 4. T.A. Gerken, C.L. Owens, M. Pasumarthy. J. Biol. Chem., 1997,272, 9709-9719. 5. S. Rottger, J. White, H.H. Wandall, J.C. Olivo, A. Stark, E.P. Bennett, C. Whitehouse, E.G. Berger, H. Clausen, T. Nilsson. J. Cell. Sci., 1998, 111, 45-60. 6. S. Tsuji, A.K. Datta, J.C. Paulson. Glycohiology, 1996, 6, v-vii. 7. J.P. Audie, A. Janin, N. Porchet, M.C. Copin, B. Gosselin, J.P. Aubert. J. Histochem. & Cytochem., 1993, 41, 1479-1485. 8. S. Zotter, P.C. Hageman, A. Lossnitzer, W.J. Mooi, J. Hilgers. Cuncer Rev., 1988, 11-12, 55101. 9. S.J. Gendler, A.P. Spicer. Annu. Rev. Physiol., 1995, 57, 607-634. 10. N . Porchet, P. Pigny, M.P. Buisine, V. Debailleul, A. Laine, J.P. Aubert. Biochem. Soc. Trans., 1995,23, 800-805. 11. J. Perez-Vilar, A.E. Eckhardt, A. DeLuca, R.L. Hill. J. Biol. Chern., 1998,273, 14442-14449. 12. N. Asker, M.A.B. Axelsson, S.-0. Olofsson, G.C. Hansson. J. Biol. Chem., 1998, 273, 1885718863. 13. Z. Sheng, S.R. Hull, K.L. Carraway. J. B i d Chem., 1990,265, 8505-8510. 14. S. Nollet, N. Moniaux, J. Maury, D. Petitprezk, P. Degand, A. Laine, N. Porchet, J.P. Aubert. Biochem. J. 1998, 332, 739-748. 15. M.J.L. Ligtenberg, L. Kruijshaar, F. Buijs, M. van Meijer, S.V. Litvinov, J. Hilkens. J. Biol. Chem., 1992,267, 6171-6177. 16. Z. Sheng, K. Wu, K.L. Carraway, N . Fregien. J. Biol. Chem., 1992,267, 16341-16346. 17. A.P. Sherblom, R.L. Buck, K.L. Carraway. J. Biol. Chem., 1980,255, 783-790. 18. M. Komatsu, C.A. C. Carraway, N.L. Fregien, K.L. Carraway. J. Bid. Chem., 1997, 272, 33245 -33254.

682

40 Structure and Functions of Mucins

19. A.P. Spicer, G. Parry, S. Patton, S.J. Gendler. J. Biol. Chem., 1991, 266, 15099-15109. 20. P. Pandey, S. Kharbanda, D. Kufe. Cancer R e x , 1995,55,4000-4003. 21. S. Zrihan-Licht, H.L. Vos, A. Baruch, 0. Elroy-Stein, D. Sagiv, I. Keydar, J. Hilkens, D.H. Wreschner. Eur. J. Biochem., 1994, 224, 787-795. 22. S. Muller, S. Goletz, N. Packer, A. Gooley, A.M. Lawson, F.G. Hanisch. J. Biol. Chem.. 1997; 272, 24780-24793. 23. F.G. Hanisch, G. Uhlenbruck, J. Peter-Katalinic, H. Egge, J. Dabrowski, U. Dabrowski. J. Biol. Chem., 1989,264, 872-873. 24. F.G. Hanisch, J. Peter-Katalinic, H. Egge, U. Dabrowski, G. Uhlenbruck. Glycoconj. J . , 1990, 7, 525-543. 25. K.O. Lloyd, J. Burchell, V. Kudryashov, B.W.T. Yin, J. Taylor-Papadimitriou. J. Biol. Chem., 1996.271. 33325-33334. 26. S.R. 'Hull: A. Bright, K.L. Carraway, M. Abe, D.F. Hayes, D.W. Kufe. Cancer Comm., 1989, 1,261-267. 27. J.M. Burchell, S. Gendler, J. Taylor-Papadimiriou, A. Girling, A. Lewis, R. Millis, D. Lamport. Cancer Rex, 1987, 47, 5476-5482. 28. G.F. Springer. J. Mol. Med., 1997, 75, 594-602. 29. I. Brockhausen, J.M. Yang, J. Burchell, C. Whitehouse, J. Taylor-Papadimitriou. Eur. J. Cancer, 1995,233, 607-617. 30. M.F.A. Bierhuizen, M. Fukuda. Proc. Nutl Acad. Sci. USA., 1992,89, 9326-9330. 31. M.F.A. Bierhuizen, M.G. Mattei, M. Fukuda. Genes & Devel., 1993, 7, 468-478. 32. J. Burchell, R. Poulsom, A. Hanby, C. Whitehouse, L. Cooper, D. Miles, J. TaylorPapadimitriou. Glycobiology-Submitted. 33. K. Zhang, D. Baeckstrom, H. Brevinge, G.C. Hansson. J. Cell. Biochem., 1996, 60, 538-549. 34. A. Girling, J. Bartkova, J. Burchell, S.J. Gendler, C. Gillett, J. Taylor-Papadimitriou. Int. J. Cancer, 1989, 43, 1072-1076. 35. C.A. Reis, L. David, M. Sixas, J. Burchell, M. Sobrinho-Simoes. Int. J. Cancer, 1998, 79, 1-9. 36. M.J.L. Litenberg, F. Buijs, H.L. Vos, J. Hilkens. Cancer R e x , 1992, 52, 2318 -2324. 37. J. Wesseling, S.W. van der Valk, H.L. Vos, A. Sonnenberg, J. Hilkens. J. Cell. Biol., 1995, 129, 255-265. 38. J. Wesseling, S.W. van der Valk, J. Hilkens. Mol. Biol. Cell, 1996, 7, 565-577. 39. M. Yamamoto, A. Bharti, Y. Li, D. Kufe. J. Biol. Chem., 1997,272, 12492-12494. 40. L.H. Regimbald, L.M, Pilarski, B.M. Longenecker, M.A. Reddish, G. Zimmermann, J.C. Hugh. Cuncer Rex, 1996, 56, 4244-4249. 41. C.M. Bohm, M.C. Mulder, R. Zennadi, M. Notter, A. Schmitt-Graff, O.J. Finn, J. TaylorPapadimitriou, H. Stein, H. Clausen, E.O. Riecken, C. Hanski. Scand. J. Immunol., 1997, 46, 27-34. 42 A.P. Spicer, G.J. Rowse, T.K. Lidner, S.J. Gendler. J. Biol. Chem., 1995, 270, 30093-30101. 43 R. Medzhitov, C.A. Janeway. Cell, 1997, 91, 295-298. 44 D.L. Barnd, M.S. Lan, R.S. Metzgar, O.J. Finn. Proc. Nutl Acad. Sci. USA., 1989, 86, 7159 7163. 45 K.R. Jerome, D.L. Barnd, K.M. Bendt, C.M. Boyer, J. Taylor-Papadimitriou, I.F.C. McKenzie, R.C.J. Basr, O.J. Finn. Cancer Rex, 1991, 51, 2908-2916. 46 O.J. Finn, K.R. Jerome, R.A. Henderson, G. Pecher, N. Domenech, J. Magarian-Blander, S.M. Barratt-Boyes. Immunol. Rev., 1995, 145, 61-89. 47 M.M. Gourevitch, S. von Mensdorff-Pouilly, S.V. Litvinov, P. Kenemans, G.J. van Kamp, A.A. Verstraeten, J. Hilgers. Br. J. Cancer, 1995, 72, 934-938. 48 S. von Mensdorff-Pouilly, M.M. Gourevitch, P. Kenemdns, A.A. Verstraeten, G.J. van Kamp, A. Kok, K. van Uffelen, F.G. Snijdewint, M.A. Paul, S. Meijer, J. Hilgers. Tum. Biol., 1998, 19, 186-195. 49 V. Apostolopoulos, B.E. Loveland, G.A. Pietersz, I.F.C. McKenzie. J. Immunol., 1995, 155, 5089-5094. 50 V. Apostolopoulos, V. Karanikas, J.S. Haurum, I.F.C. McKenzie. J. Immunol., 1997, 159, 5211-5218. 51. B. Agrawal, M.A. Reddish, B.M. Longenecker. J. Immunol., 1996, 157, 2089-2095. 52. B. Agrawal, M.J. Krantz, M.A. Reddish, B.M. Longenecker. Nature Med., 1998, 4, 43-49. ~

References

683

53. E. van de Wiel-van Kemenade, M.J. Ligtenberg, A.J. de Boer, F. Buijs, H.L. Vos, C.J. Melief, J. Hilkens, C.G. Figdor. J. Immunol. 1993, 151, 767-716. 54. R.A. Graham, J.M. Burchell, P. Beverley, J. Taylor-Papadimitriou. Int. J. Cancer, 1996, 65, 664-610. 55. N.Peat, S.J. Gendler, E.N. Lalani, T. Duhig, J. Taylor-Papadimitriou. Cancer Res., 1992, 52, 1954-1960. 56. G.J. Rowse, R.M. Tempero, M.L. van Lith, M.A. Hollingsworth, S.J. Gendler. Cancer Rex, 1998, 58, 315-321. 57. R. Acres, M. Hareuveni, J. Balloul, M. Kieny. J. Immunother., 1993, 14, 136-143. 58. L. Ding, E.N. Lalani, M. Reddish, R. Koganty, T. Wong, J. Samuel, M.B. Yacyshyn, A. Meikle, P.Y.S. Fung, J. Taylor-Papadimitriou, B.M. Longenecker. Cancer Immunol. & Immunother., 1993, 36, 9-17. 59. J. Samuel, W.A. Budzynski, M.A.Reddish, L. Ding, G.L. Zimmermann, M.J. Krantz, R.R. Koganty, B.M. Longenecker. Int. J. Cancer, 1998, 75, 295-302. 60. M. Smith, J.M. Burchell, R. Graham, E.P. Cohen, J. Taylor-Papadimitriou. Immunology1999, 97, 648-655. 61. J. Gong, D. Chen, M. Kashiwaba, Y. Li, L. Chen, H. Takeuchi, H. Qu, G.J. Rowse, S.J. Gendler, D. Kufe. Proc. Nut1 Acud. Sci. USA. 1998, 95, 6279-6283. 62. D.W. Miles and J. Taylor-Papadimitriou. J. Exp. Opin. Invest. Drugs, 1998, 7, 1865-1877. 63. V. Apostolopoulos, G. Pietersz, B. Loveland, M. Sandrin, I. McKenzie. Proc. Nut1 Acud. Sci. USA, 1995, 92, 10128-10132. 64. V. Karanikas, L.A. Hwang, J. Pearson, C.S. Ong, V. Apostolopoulos, H. Vaughan, P.X. Xing, G. Jamieson, G. Pietersz, B. Tait, R. Broadbent, G. Thynne, I.F. McKenzie. J. Clin. Invest., 1997, 100, 2138-2192. 65. G.D. MacLean, M.A. Reddish, R.R. Koganty, B.M. Longenecker. J. Immunother., 1996, 19, 59-68. 66. G. MacLean, D. Miles, R. Rubens, M. Reddish, B. Longenecker. J. Immunother., 1996, 19, 309-3 16. 67. D. Math, A. Hartnell, L. Happerfield, D.W. Miles, J. Burchell, J. Taylor-Papadimitriou, P.R. Crocke. Immunology, 1999, 98, 213-219.

Carbohydrates in Chemistry and Biology Beat Emst, Gerald W. Hart, Pierre Sinay copyright Q WILEY-VCH Verlag GmbH. 2000

41 Biological Roles of Hyaluronan Brym P. Toole

41.1 Introduction Hyaluronan is a uniformly repetitive, linear glycosaminoglycan (GAG) composed of disaccharides of glucuronic acid and N-acetylglucosamine: [-p(1,4)-GlcUAp( 1,3)-GlcNAc-],. Depending on the tissue source, the polymer usually consists of 2,000-25,000 disaccharides, giving rise to molecular weights ranging from 1O6 to lo7 Da and extended lengths of 2-25 pm. Despite these relatively simple characteristics, hyaluronan has unusual physical and biochemical properties. As a consequence of these unusual attributes, hyaluronan fulfils several distinct physiological functions that contribute both to structural properties of tissues and to cell behavior during formation or remodeling of tissues. In this chapter, I will summarize current knowledge concerning these functions and the mechanisms whereby hyaluronan exerts its wide biological influence. First among these functions is the direct contribution of hyaluronan itself to tissue homeostasis and biomechanics due to its unique charge characteristics and biophysical properties. Second, interactions of hyaluronan with link proteins and proteoglycans are of fundamental importance to the structural integrity of extracellular and pericellular matrices. Third, interactions with cell surface hyaluronan receptors mediate significant influences on cell behavior during morphogenesis, tissue remodeling, inflammation, and diseases such as cancer and atherosclerosis.

41.2 Hyaluronan is a Biopolymer with Unusual Physical Properties The conformation of hyaluronan in aqueous solution is a gently undulating, tape or ribbon-like, two-fold helix which forms as a result of 180" rotations between alternating disaccharides [ 1 , 21 (Figure 1A,B). Although, overall, hyaluronan is highly

686

A

41 Biological Roles of Hyaluronan

B

C

Figure 1. Models depicting the polymeric structure and interaction of hyaluronan molecules. A-B. Projections, at right angles to each other, showing the two-fold helix that a hyaluronan molecule adopts in aqueous solution. The dotted lines in B represent hydrogen bonds; the circle-square pairs joined by dotted lines represent water bridges between acetamido groups and carboxyls. C. Proposed mode of interaction between two anti-parallel hyaluronan molecules in which hydrophobic patches are apposed and acetamido and carboxyl groups are within hydrogen bonding distance. Shading, hydrophobic patches; circles, acetamido groups; squares, carboxyl groups. From [2].

hydrophilic, it also has repeating hydrophobic patches of approximately three sugars in length. These are arrayed along the two flat sides of the tape-like polymer, with sequential patches alternating between the two sides. The patches most likely form hydrophobic bonds between corresponding regions of neighboring hyaluronan chains aligned in an antiparallel fashion. In addition, hydrogen bonds would occur between the opposing acetamido and carboxyl groups within these antiparallel regions [ l , 21 (Figure 1C). Since these hydrophobic and hydrogen bonds would form on both sides of the hyaluronan polymer, higher order aggregates can assemble, and accordingly, high molecular weight hyaluronan has been shown to form virtually infinite networks, even at concentrations as low as 1 pg/ml. As the concentration of hyaluronan increases, the branches within the network become thicker [ 1-31. In addition, interactions with other tissue components in vivo would add further

41.3 Hyuluronan Binds to Scverul Types o j Proteins (Hvaladherins)

681

ordered complexity to the meshwork-forming properties of hyaluronan. On the other hand, some of these interactions, e.g. with proteoglycans such as aggrecan and versican, would cause hyaluronan molecules to become extended and separated (see Figure 3B), and would block network-forming self-interactions. High molecular weight hyaluronan in very dilute saline solution occupies an enormous domain wherein the mass of hyaluronan itself is 0.1% or less and solvent occupies the remaining volume of this domain [4].As the hyaluronan concentration is increased, the interactions described in the previous paragraph would occur at greater and greater frequency. These properties give rise to high viscosities at concentrations of 0.5 mg/ml or more, as found in many tissues such as synovial fluid, umbilical cord and skin. However the viscosity of hyaluronan solutions is highly dependent on flow rates; i.e. high shear causes a reversible decrease in viscosity. This visco-elastic property of hyaluronan may be important in soft tissue lubrication. The rheological and network-forming properties of hyaluronan also contribute to water homeostasis, tissue hydration and transport of macromolecules within tissues [4], and form the basis of current clinical applications of hyaluronan, e.g. stabilizing eye tissues during surgery, reducing soft tissue adhesions, and alleviating joint problems [5-71.

41.3 Hyaluronan Binds to Several Types of Proteins (Hyaladherins) 41.3.1 General Properties of Hyaladherins Hyaluronan binds to a wide variety of proteins, termed hyaladherins. Two well characterized groups of hyaladherins are: i) structural hyaluronan-binding proteins of the extracellular matrix, such as link protein and the aggregating proteoglycans; and ii) cell surface hyaluronan receptors, such as CD44 and the receptor for hyaluronic acid-mediated motility (RHAMM). Most of these well characterized hyaLdherins have structurallysimilar hyaluronan-binding domains with sequence homologies of 30-40'1/0. These domains, sometimes termed link modules or proteoglycan tandem repeats, form disulfide-bonded loops and, in many hyaladherins, two modules are arranged in tandem array. Two link modules form the hyaluronanbinding region of link proteins and the aggregating proteoglycans [8, 91 whereas only a single link module is found in the hyaluronan-binding domains of CD44 [ 10121 and TSG-6 [13]. Recent structural work demonstrates that link modules have a very similar conformation to that of the carbohydrate-binding region in C-type selectins despite the absence of sequence homology [ 141. Some hyaladherins, notably RHAMM, do not have link modules. However, mutation and sequence-swapping studies of RHAMM have revealed a hyaluronanbinding motif which is present, not only within RHAMM, but also within or adjacent to the link modules of several of the hyaladherins mentioned in the previous paragraph, e.g. link protein and CD44 [15]. The proposed motif is B(X7)B, where B is arginine or lysine and X is any non-acidic amino acid. Variations of this motif,

688

41 Biological Roles of Hyaluronan

e.g. B(Xg)B, also bind to hyaluronan with significant affinity, and clearly clustering of basic amino acids within and around the motif is the key feature that determines binding [ 151. Several other hyaluronan-binding proteins, especially intracellular hyaluronan-binding proteins (see section 41.3.4) lack domains with homology to link modules but contain B(X7)B and related sequences. Although clusters of basic amino acids clearly contribute to hyaluronan binding in several hyaladherins, other structural features, e.g. glycosylation and conformational effects, are also involved 111, 121.

41.3.2 Structural Hyaluronan-Binding Proteins The integrity and distinctive biomechanical properties of the extracellular matrix of cartilages is in great part dependent on interactions between hyaluronan, link protein and the proteoglycan, aggrecan, that give rise to an enormous ternary complex. Both link protein and aggrecan contain tandem link modules and B(X7)B motifs that mediate formation of this complex with hyaluronan. The biophysical properties of the complex underly the unique capability of cartilage to resist compression and to recover its shape after deformation. The chemistry and functional properties of the ternary complex have been extensively studied and are reviewed in detail elsewhere (e.g. 18, 91 and Chapter 44 in this volume). Aggrecan is now known to be just one member of a family of so-called aggregating proteoglycans (also known as the hyalectan or lectican family); other members of this family include versican, brevican and neurocan 191. All members contain tandem link modules and are believed to contribute importantly to the unique properties of extracellular matrices within various connective tissues (versican and aggrecan) and the brain (neurocan and brevican). Hyaluronan-aggrecan interactions are also vital in pericellular matrix assembly (see section 41.4.2). A similar role for the other aggregating proteoglycans is probable.

41.3.3 Hyaluronan Receptors Early work on embryonic development and tissue repair suggested strongly that hyaluronan influences cell behavior. Consequently, evidence was sought and obtained for the presence of hyaluronan-binding proteins on the surface of cells [ 16, 171. Subsequent investigations led to the full molecular characterization of two classes of cell surface hyaluronan receptors, namely CD44 and RHAMM. Numerous reviews (e.g. [lo-12, 18, 191) have been published that amply describe work on the structures and functions of CD44 and RHAMM and this work will only be summarized here. CD44 is a widely distributed cell surface glycoprotein that is encoded by a single gene but expressed as numerous isoforms as a result of alternative splicing. The extracellular region of all CD44 isoforms includes an amino-terminal domain which contains a link module and two B(X7)B motifs, and is the binding site for hyalur-

41.3 Hyaluronan Binds to Seurrul Types of Proteins (Hya1udherin.s)

689

onan. However, hyaluronan binding to CD44 is subject to numerous additional positive and negative influences. Among these are: glycosylation, alternative splicing, dimerization, clustering in the plasma membrane, and integrity of the cytoplasmic domain [ 12, 191. Since cell surface hyaluronan-CD44 interactions mediate many cellular effects of hyaluronan, the biochemical mechanisms by which these interactions are transduced into intracellular signals that bring about these effects are now being intensely studied by several groups. It is probable that, in at least some cell types, multivalent interaction of polymeric hyaluronan with CD44 [20] causes clustering of CD44 in the plasma membrane and that this event is associated with phosphorylation of CD44 and interactions with the cytoskeleton [ 1 1, 2 1, 221. However, there is little consensus with respect to the specific signaling pathways initiated by hyaluronan-CD44 interaction. Less work has been published on RHAMM than CD44 but what has appeared is provocative and potentially interesting. Alternative splicing generates several isoforms of RHAMM, including a major intracellular and a transient cell surface isoform [ 181. Hyaluronan-RHAMM interactions induce transient phosphorylation of ~ 1 2 . in5 concert ~ ~ ~with turnover of focal adhesions in ras-transformed cells, thus leading to initiation of locomotion. Suppression of normal HA-RHAMM interactions inhibits both locomotion and growth in vitro, and leads to inhibition of tumor growth in vivo; over-expression of RHAMM leads to enhanced tumor growth and metastasis [23, 241.

41.3.4 Intracellular Hyaluronan-Binding Proteins A third and, as yet, puzzling group of proteins appears to be emerging, i.e. intracellular hyaluronan-binding proteins. These include Cdc37 [25], P-32 [26], LH21 [27], and an intracellular form of RHAMM [18, 281. Since hyaluronan is synthesized on the cytoplasmic side of the plasma membrane ([29] also see Chapter 21 in this volume) and since hyaluronan has been localized to intracellular compartments [28], it is possible that these proteins may play some role in chaperoning or transporting newly synthesized hyaluronan to these compartments. Also, since RHAMM, Cdc37 and LH21 have all been shown to associate with kinases, they may serve to coordinate hyaluronan synthesis and various cellular activities.

41.3.5 Inter-a-Trypsin Inhibitor Another unusual hyaluronan-binding protein is inter-a-trypsin inhibitor [ 301. This serum protein is composed of a light chain, also known as bikunin, and two heavy chains which are covalently cross-bridged by chondroitin sulfate. Hyaluronan can replace chondroitin sulfate by trans-esterification or bind non-covalently to the heavy chains. This interaction has been shown to bind hyaluronan to cell surfaces [30] and to participate in forming a matrix around the oocyte during ovulation [31, 321.

690

41 Biological Roles of Hyaluronan

41.4 Hyaluronan-Dependent Pericellular Matrices Assemble Around Several Cell Types 41.4.1 Hyaluronan-Dependent Cellular “Coats” Several cell types exhibit highly hydrated, hyaluronan-dependent, pericellular matrices or “coats” that are destroyed by treatment with hyaluronan-specific hyaluronidase [16, 171. In culture, these matrices cannot be analyzed readily by conventional light microscopy. However the coats can be visualized indirectly by exclusion of particles and are usually 5-10 pm in thickness (Figure 2). These pericellular matrices provide the milieu in which numerous cellular activities take place and they influence the behavior of cells in many circumstances. For example, during tissue formation or remodeling, such matrices would provide a hydrated, fluid pericellular environment in which assembly of other matrix components and presenta-

Figure 2. Hyaluronan-dependent pericellular matrices. A-D. Large pericellular matrices surrounding rat fibrosarcoma cells (A) and chick embryo chondrocytes (C) are visualized by their ability to exclude particles (fixed red blood cells are used here). The matrices are removed by treatment with hyaluronidase (B, D). E-F. Myoblasts (E ) also exhibit large pericellular matrices that are destroyed by hyaluronidasc treatment (not shown); these pericellular matrices are lost during myoblast fusion (F).

41.4 Hyuluronun- Dependent Pericellular Matrices

69 1

tion of growth and differentiation factors could readily occur without interference from the highly structured fibrous matrix usually found in fully differentiated tissues. In some cases: such as in cartilage, the pericellular matrix is a unique structural component that protects the cells and contributes to the characteristic properties of the differentiated tissue. 41.4.2 Assembly of Chondrocyte Pericellular Matrix Chondrocytes exhibit particularly prominent pericellular matrices (Figure 2C). Their function and assembly have been studied extensively and shown to be dependent on three features [ l l , 17, 33, 341. First, their integrity is dependent on hyaluronan (Figure 2C vs D). Second, the assembly and density of chondrocyte pericellular matrix depend on specific interaction of hyaluronan with the proteoglycan, aggrecan. Third, hyaluronan must be tethered to the cell surface. In the case of chondrocytes, tethering is mediated mainly by interaction of hyaluronan with CD44. Pericellular matrix can be reconstituted on living or fixed, matrix-free cells that lack a pericellular matrix but possess hyaluronan receptors, by adding hyaluronan and aggrecan [ 11, 33, 341. Analysis of the motion of gold particles attached to hyaluronan or aggrecan in the pericellular matrix of chondrocytes has clearly illustrated several features of these matrices [35](Figure 3 ) . First, individual tethered hyaluronan molecules can extend as far as -10 pm from the cell surface. Second, extension of hyaluronan molecules from the cell surface depends on interaction with aggrecan. Third, hyaluronanaggrecan complexes move in a restricted, cone-shaped area consistent with tethering of hyaluronan to the cell surface. Fourth, such motion would exclude large, but not small, particles from the pericellular zone. This behavior of pericellular hyaluronanaggrecan complexes fits well with the predicted behavior of polymers tethered to a surface at high density [35].The high proteoglycan concentration in cartilage would increase the density of the pericellular matrix and thus increase resistance to compression, a fundamental characteristic of cartilage.

41.4.3 Tethering of Cell Surface Hyaluronan to Hyaluronan Synthase Tethering of hyaluronan to different cell types can occur by at least two independent mechanisms. One of these is by binding to the hyaluronan receptor, CD44, as described above for chondrocytes. A second likely mechanism is transmembrane interaction of “nascent” hyaluronan with hyaluronan synthase. Thus, in several cell types, hyaluronan in the process of extrusion across the plasma membrane appears to be tethered by sustained attachment to hyaluronan synthase or associated proteins on the cytoplasmic face of the plasma membrane ([29] also Chapter 21 in this volume). Some pathogenic bacteria have hyaluronan capsules that may also be tethered in this way. These capsules would represent a primitive form of pericellular matrix which acts to facilitate bacterial invasion by inhibiting phagocytosis [36] and/or by promoting adhesion to host tissues [37].

692

41 Biological Roles of Hyaluronan

A HA

Figure 3. Model of chondrocyte pericellular matrix. The de Gennes model for polymers grafted at an interface is adapted to the structure of chondrocyte pericellular matrix [35].A. When hyaluronan (HA) alone is bound to the cell surface it collapses in a “mushroom” configuration, as predicted for grafted polymers at low density. In this configuration pericellular matrix is not evident. B. When aggrecan interacts with hyaluronan, the resulting negative charge repulsion between GAG sidechains simulates the effect of increased polymer density in the de Gennes model. This causes the hyaluronan-aggrecan complexes to extend out from the cell surface, forming a “brush” which can be visualized as pericellular matrix. Theoretically, the same effect could be achieved at very high hyaluronan concentration at the cell surface. C. Tethered motion (arrows) in the latter configuration excludes large particles, such as red blood cells. The size of particle excluded will depend on polymer density. From [35].

41.5 H.yaluronan Injuencrs Cell Behavior During Morphogenesis

693

41.5 Hyaluronan Influences Cell Behavior During Morphogenesis and Tissue Remodeling 41.51 Migratory and Proliferating Cells are Surrounded by Hyaluronan-enriched Matrices

The pericellular matrices surrounding migrating and proliferating cells, during morphogenesis of embryonic organs and during regeneration and healing, are enriched in hyaluronan and resemble the pericellular matrices described in the previous section. Striking examples of such hyaluronan-rich matrices occur: around mesenchymal cells invading the primary corneal stroma; around neural crest cells traveling from the neural tube to form peripheral ganglia; around somite cells encompassing the notochord to form vertebrae; around cushion cells migrating from endocardium to myocardium during formation of heart valves; around proliferating and migrating cells during brain development; and, around proliferating mesenchymal cells during embryonic limb development, salamander limb regeneration, tendon regeneration, and fetal wound repair [ 16, 331. 41.5.2 Hydrated Pericellular Milieux Provide Cellular Pathways

One important way in which hyaluronan-rich matrix would promote cell proliferation is by provision of a hydrated pericellular zone that facilitates cell rounding during mitosis. Hyaluronan synthase activity has been shown to fluctuate with the cell cycle and to peak at mitosis [16]. Thus extrusion of hyaluronan onto the cell surface at mitosis would create a hydrated micro-environment that promotes partial detachment and rounding of the dividing cells. In support of this idea, inhibition of hyaluronan synthesis has been shown to lead to cell cycle arrest at mitosis. just before cell rounding and detachment [38]. It is also possible that intracellular hyaluronan-binding proteins, e.g. the cell cycle regulatory factor, Cdc37 [25], and an intracellular form of RHAMM [18, 281, may be involved in regulation of these events. With respect to cell migration and invasion, hyaluronan-enriched matrices create hydrated pathways that separate cellular or fibrous barriers to penetration by the invading cells [16]. In addition, increases in density of these matrices due to increased binding of proteoglycans to pericellular hyaluronan may transform the pericellular matrix from conducive to inhibitory for cell migration, e.g. during neural crest cell migration and ncurite outgrowth [33].

41.53 Receptors Mediate Effects of Hyaluronan In addition to providing a suitably hydrated pericellular milieu, a second manner in which hyaluronan influences cell behavior is via interaction with its cell surface

694

41 Biological Roles of Hyuluronun

receptors. Several investigators have demonstrated that cell movement in vitro is promoted in the presence of hyaluronan, that invasion into three dimensional collagen gels is dependent on hyaluronan synthesis, and that these phenomena are inhibited as a result of degrading hyaluronan or blocking binding of hyaluronan to its receptors, CD44 or RHAMM. Similar studies have shown that these interactions also influence cell growth [ 16-18, 22-24, 39-42]. Such interactions clearly initiate signaling pathways that promote cell movement or proliferation but, as mentioned in section 41.3.3, the details are still unclear. 41.5.4 Hyaluronan-Cell Interactions in Limb Development

Studies carried out on embryonic limb development provide an illustrative example of the way in which modulation of pericellular hyaluronan concentration and organization influences several of the events leading to differentiation in vivo. First, the volume of hyaluronan-rich matrix separating cells at different stages of limb development closely parallels their differentiation [43]. Early limb mesodermal cells are surrounded and separated by extensive, hyaluronan-enriched matrix in vivo and express voluminous, hyaluronan-enriched, hyaluronan-dependent, pericellular matrices in culture [43].At this stage pericellular hyaluronan appears to be tethered to the cell via transmembrane interaction with hyaluronan synthase since it is retained at the cell surface in a non-receptor mediated manner [44]. The pericellular matrix maintains separation of early mesodermal cells [45], consistent with the predicted behavior of surface-associated polymers, i.e. polymer molecules on apposing surfaces do not interdigitate due to their constant motion [35] (see Figure 3C). This hydrated pericellular matrix would facilitate proliferation and migration of early mesenchymal precursors of limb tissues in the manner described in the preceding paragraphs. This has been demonstrated directly in the case of muscle differentiation, where it has been shown that exposure of myoblasts to hyaluronan supports proliferation and migration but inhibits differentiation [ 17, 461. Subsequent to the early stage of limb development described in the previous paragraph, the mesoderm condenses, i.e. the intercellular matrix decreases in volume, at sites of future cartilage and muscle differentiation. This is paralleled by loss of ability of the mesodermal cells to form hydrated pericellular matrices in culture. During this condensation, much of the hyaluronan is removed from the intercellular matrix, thus accounting for the decreased intercellular volume [43]. However, cell surface hyaluronan is now retained via interaction with receptors that appear on the mesodermal cells at this stage [44]. This cell surface hyaluronan interacts multivalently with receptors on adjacent cells thus crossbridging them within the condensate [45].Crossbridging occurs in an analogous fashion to that previously shown for aggregation of various cell lines in culture [ 16, 171. Similar events comprise an early step in mesodermal condensation in other developing tissues also, e.g. skin (47, 481. Further differentiation of condensed limb mesoderm to cartilage is accompanied by extensive matrix formation in vivo and recovery of the ability to form extensive hyaluronan-dependent pericellular matrices in culture [43] (Figure 2C). In these

41.5 Hyaluronan Injuences Cell Behavior During Morphoyenesis

695

matrices, however, hyaluronan is tethered to the cell surface by interaction with CD44 and the proteoglycan concentration is much higher than in the matrices surrounding early mesodermal cells [43-45]. Thus, the pericellular matrix of differentiated chondrocytes would be denser than that of mesodermal cells [35], reflecting its structural rather than morphogenetic role. With respect to muscle differentiation in the limb, mononucleated myoblasts are also initially separated from each other by a hyaluronan-rich matrix in vivo and express pericellular matrices in culture [49] (Figure 2E). However during the processes of condensation and fusion to give myotubes these coats are lost [49] (Figure 2F), thus allowing the interactions required for these cells to differentiate. During long bone differentiation, lacunae surrounding hypertrophic chondrocytes are highly enriched in hyaluronan, and the swelling pressure exerted by this hyaluronan causes expansion of lacunae as bone growth occurs [50].Subsequently, in the zone of erosion, the hyaluronan within these lacunae is removed via CD44mediated endocytosis. CD44-mediated internalization of hyaluronan is an important step in differentiation of several other tissues also, e.g. skin [47, 481 and lung 1511. 41 S.5 Hyaluronan-Cell Interactions in Other Physiological and Developmental Systems

Another interesting system in which hyaluronan plays an important role is expansion of the cumulus oophorus during ovulation. In response to follicle stimulating hormone and a factor produced by the oocyte, hyaluronan synthesis by the cumulus cells surrounding the oocyte increases dramatically [32]. A stable, gel-like matrix is formed between the cumulus cells due to crosslinking of hyaluronan by inter-atrypsin inhibitor and TSG-6 [31, 321. This matrix is responsible for the integrity of the cumulus cell-oocyte complex which is required for protection and transport of the oocyte during ovulation, entry into the oviduct and fertilization. Spermassociated hyaluronidases allow penetration of this matrix at fertilization. Recent studies indicate that hyaluronan is important in branching morphogenesis of epithelial organs [52] and in formation and remodeling of blood vessels [53-551. In these cases it appears again that hyaluronan provides a hydrated milieu that favors proliferation and movement of epithelial, endothelial and smooth muscle cells. The influence of hyaluronan on vascular smooth muscle cell movement and proliferation may be an important factor in progression of atherosclerotic lesions [54, 551. A hyaluronan-enriched pericellular milieu is also important in control of cellular behavior in the mature organism. For example, hyaluronan surrounds individual cells within many complex epithelia, e.g. epidermis and esophagus, and may facilitate movement of cells between epithelia strata as they differentiate [56, 571. Extravasation of lymphocytes at sites of inflammation involves adhesive interactions between CD44 on the lymphocyte surface and hyaluronan on the endothelial cell surface; expression of hyaluronan on the latter is induced by proinflammatory cytokines such as TNFa and IL-lp, and by bacterial lipopolysaccharide [ 5 8 ] .

696

41 Biological Roles of Hyaluronan

41.6 Hyaluronan Plays a Crucial Role in Cancer Most malignant solid tumors contain elevated levels of hyaluronan [ 17, 591. High levels of hyaluronan expression correlate with poor differentiation in human ductal breast carcinomas [60] and with poor survival rates in human colorectal adenocarcinomas [61]. Enrichment of hyaluronan in tumors can be due to increased production by tumor cells themselves or to interactions between tumor cells and surrounding stromal cells that induce increased production by the latter. In accordance with the latter, hyaluronan accumulation occurs at the interface of tumor invasion into host tissues in various tumor types [17, 57, 59, 621. Direct experimental evidence has been obtained implicating hyaluronan and hyaluronan receptors in progression of a variety of tumor types [22, 41, 42, 63-67]. However, many apparently contradictory studies have also been published [ 171, especially with respect to CD44, although a recent publication has thrown some light on these contradictions. In this study 1221, it was shown that hyaluronan-induced clusters of CD44 in the plasma membrane bind gelatinase B, which in turn promotes tumor cell invasion. Over-expression of soluble or intact CD44 in the tumor cells causes disruption of the CD44 clusters and loss of tumorigenicity [22]. Despite the caveats above, it is abundantly clear that different interventions aimed at perturbing events involving hyaluronan and its receptors lead to impressive inhibition of tumor progression in vivo. For example, treatment with hyaluronan oligomers or with soluble CD44, both of which compete for hyaluronan polymer-CD44 interaction, inhibits melanoma, lymphoma and mammary carcinoma growth in vivo [63, 66, 671, and transfection of tumor cells with soluble CD44 inhibits growth and metastasis of murine mammary carcinoma [22, 41, 421. It is possible that some of these approaches may form the basis of new therapies.

References 1. Scott JE. Chemical morphology of hyaluronan. In: Laurent TC, ed. The Chemistry, Biology and Medical Applications of Hyaluronan and Its Derivatives. London: Portland Press, 1998:715. 2. Scott JE. Secondary and tertiary structures of hyaluronan in aqueous solution. Some biological consequences. Glycoforum: Science of hyaluronan . http: {jwww.glycoforum.gr .jp 3. Mikelsaar R-H, Scott JE. Molecular modelling of secondary and tertiary structures of hyaluronan, compared with electron microscopy and NMR data. Possible sheets and tubular structures in aqueous solution. Glycoconj J 1994; 11:65-71. 4. Laurent TC, Fraser JRE. Hyaluronan. FASEB J 1992; 6:2397-2404. 5. Balazs EA, Denlinger JL. Clinical uses of hyaluronan. Ciba Found Symp 1989; 143:265-275. 6. Bertolami CN, Gay T, Clark GT, Rendell J, Shetty V, Liu C, Swann DA. Use of sodium hyaluronate in treating temporomandibular joint disorders: a randomized, double-blind, placebocontrolled clinical trial. J Oral Maxillofuc Surg 1993; 51:232-242. 7. Burns JW, Skinner K, Colt MJ, Burgess L, Rose R, Diamond MP. A hyaluronate based gel for the prevention of postsurgical adhesions: evaluation in two animal species. Fertil Steril 1996; 661814-821.

References

697

8. Hardingham TE. Cartilage: Aggrecan-link protein-hyaluronan aggregates. G1)~~ofbrutn: Science of hyaluronan. http://www.glycoforum.gr.jp 9. Iozzo, RV. Matrix proteoglycans: from molecular design to cellular function. Annu Rev Biochem 1998; 67:609-652. 10. Sherman L, Sleeman J, Herrlich P, Ponta H. Hyaluronate receptors: key players in growth, differentiation, migration and tumor progression. Curr Opin Cell B i d 1994; 6:726-733. 1 1. Knudson W, Knudson CB. The hyaluronan receptor, CD44. Glycoforum: Science of hjJaluronan. http://www.glycoforum.gr.jp 12. Kincade PW, Zheng 2, Katoh S . Hanson L. The importance of cellular environment to function of the CD44 matrix receptor. Curr Opin Cell Biol 1997; 9:635-642. 13. Lee TH, Wisniewski H-G, Viltek J. A novel secretory tumor necrosis factor-inducible protein (TSG-6) is a member of the family of hyaluronate binding proteins, closely related to the adhesion receptor CD44. J Cell Biol 1992; 116:545-557. 14. Kohda D, Morton CJ, Parkar AA, Hatanaka H, Inagaki FM, Campbell ID, Day AJ. Solution structure of the link module: a hyaluronan-binding domain involved in extracellular matrix stability and cell migration. Cell 1996; 86:767- 775. 15. Yang B, Yang BL, Savani RC, Turley EA. Identification of a common hyaluronan binding motif in the hyaluronan binding proteins RHAMM, CD44 and link protein. EMBO J 1994; 131286-296. 16. Toole BP. Proteoglycans and hyaluronan in morphogenesis and differentiation. In: Hay E, ed. Cell Biology of Extracellular Matrix, 2nd ed. New York: Plenum 1991:305-341. 17. Toole BP. Hyaluronan in morphogenesis and tissue remodeling. Glycqjbrum: Science of’lzyaluronan. http://www.glycoforum.gr.jp 18. Entwistle J, Hall CL, Turley EA. HA receptors: regulators of signalling to the cytoskeleton. J Cell Biochem 1996; 61:569-577. 19. Lesley J. Hyaluronan binding function of CD44. In: Laurent TC, ed. The Chemistry, Biology and Medical Applications of Hyaluronan and Its Derivatives. London: Portland Press, 1998:123134. 20. Underhill CB, Toole BP. Physical characteristics of hyaluronate binding to the surface of simian virus 40-transformed 3T3 cells. J Biol Chem 1980; 255:4544-~4549. 21. Lokeshwar VB, Fregien N, Bourguignon LYW. Ankyrin-binding domain of CD44 (gp85) is required for the expression of hyaluronic acid-mediated adhesion function. J Cell Biol 1994; 126:1099-1109. 22. Yu Q, Stamenkovic I. Localization of matrix metalloproteinase 9 to the cell surface provides a mechanism for CD44-mediated tumor invasion. Genes Deli 1999; 13:35-48. 23. Hall CL, Yang B, Yang X, Zhang S , Turley M, Samuel S , Lange LA, Wang C, Curpen GD, Savani RC, Greenberg AH, Turley EA. Overexpression of the hyaluronan receptor RHAMM is transforming and is also required for H-ras transformation. Cell 1995; 82:19-28. 24. Mohapatra S , Yang X, Wright JA, Turley EA, Greenberg AH. Soluble hyaluronan receptor RHAMM induces mitotic arrest by suppressing Cdc2 and cyclin BI expression. J Exp Med 1996; 183:1663-1668. 25. Grammatikakis, N, Grammatikakis A, Yoneda M, Yu Q, Banerjee SD, Toole BP. A novel glycosaminoglycan-binding protein is the vertebrate homologue of the cell cycle control protein, Cdc37. J Biol Chem 1995; 270:16198-16205. 26. Deb TB, Datta K. Molecular cloning of human fibroblast hyaluronic acid-binding protein confirms its identity with P-32, a protein co-purified with splicing factor SF2. J Biol Clwn 1996; 271:2206-22 12. 27. Huang L, Grammatikakis N, Banerjee SD, Yoneda M. Toole BP. Molecular characterization of a novel hyaluronan-binding protein related to the intracellular Ki-l antigen. Submittedfbr publication. 28. Collis L, Hall C, Lange L, Ziebell M, Prestwich R, Turley EA. Rapid hyaluronan uptake is associated with enhanced motility: implications for an intracellular mode of action. FEBS Lett 1998; 440:444-449. 29. Weigel PH, Hascall VC, Tammi M. Hyaluronan synthases. J Biol Chem 1997; 272:1399714000.

698

41 Biological Roles of Hyuluronun

30. Huang L, Yoneda M, Kimata K. A serum-derived hyaluronan-associated protein (SHAP) is the heavy chain of the inter a-trypsin inhibitor. J Biol Chem 1993; 268:26725-26730. 31. Chen L, Mao SJT, McLean LR, Powers RW, Larsep WJ. Proteins of the inter-a-trypsin inhibitor family stabilize the cumulus extracellular matrix through their direct binding with hyaluronic acid. J Biol Chem 1994; 269:28282-28287. 32. Salustri A, Fulop C. Role of hyaluronan during ovulation and fertilization. Glycoforum: Science of hyaluronan. http://www.glycoforum.gr.jp 33. Knudson CB, Knudson W. Hyaluronan-binding proteins in development, tissue homeostasis, and disease. FASEB J 1993; 7:1233-1241. 34. Knudson CB. Hyaluronan receptor-directed assembly of chondrocyte pericellular matrix. J Gel[ Biol 1993; 120:825-834. 35. Lee GM, Johnstone B, Jacobson K , Caterson B. The dynamic structure of the pericellular matrix on living cells. J Cell Biol 1993; 123:1899-1907. 36. Moses AE, Wessels MR, Zalcman K, Alberti S, Natanson-Yaron S, Meiies T, Hanski E. Relative contributions of hyaluronic acid capsule and M protein to virulence in a mucoid strain of the group A Streptococcus. Infect Immun 1997; 65:64-71. 37. Schrager HM, Alberti S, Cywes C, Dougherty GJ, Wessels MR. Hyaluronic acid capsule modulates M protein-mediated adherence and acts as a ligand for attachment of group A Streptococcus to CD44 on human keratinocytes. J Clin Invest 1998; 101:1708-1716. 38. Brecht M, Mayer U, Schlosser E, Prehm P. Increased hyahronate synthesis is required for fibroblast detachment and mitosis. Biochem J 1986; 239:445-450. 39. Koochekpour S, Pilkington GJ, Merzak A. Hyaluronic acid/CD44H interaction induces cell detachment and stimulates migration and invasion of human glioma cells in vitro. Int J Cunccrr 1995; 63:450-454. 40. Ellis I, Banyard J, Schor SL. Differential response of fetal and adult fibroblasts to cytokines: cell migration and hyaluronan synthesis. Development 1997; 124:1593- 1600. 41. Yu Q , Toole BP, Stamenkovic I. Induction of apoptosis of metastatic mammary carcinoma cells in vivo by disruption of tumor cell surface CD44 function. J Exp Med 1997; 186:19851996. 42. Peterson RM, Yu Q, Stamenkovic I, Toole BP. Perturbation of hyaluronan interactions prevents murine mammary ascites tumorigenesis in vivo. Submitted. 43. Knudson CB, Toole BP. Changes in the pericellular matrix during differentiation of limb bud mesoderm. Deo Biol 1985; 112:308-318. 44. Knudson CB, Toole BP. Hyaluronate-cell interactions during diRerentiation of chick embryo limb mesoderm. Deo Biol 1987; 124:82-90. 45. Maleski MP, Knudson CB. Hyaluronan-mediated aggregation of limb bud mesenchyme and mesenchymal condensation during chondrogenesis. Exp Cell Res 1996; 225:55-66. 46. Kujawa MJ, Pechak DG, Fiszman MY, Caplan AI. Hyaluronic acid bonded to cell culture surfaces inhibits the program of myogenesis. Deo Biol 1986; 11310-16. 47. Underhill CB. Hyaluronan i s inversely correlated with expression of CD44 in the dermal condensation of the embryonic hair follicle. J Invest Dermatol 1993; 101:820-826. 48. Kaya G, Rodriguez 1; Jorcano JL, Vassalli P, Stamenkovic I. Selective suppression of CD44 in keratinocytes of mice bearing an antisense CD44 transgene driven by a tissue-specific promoter disrupts hyaluronate metabolism in the skin and impairs keratinocyte proliferation. Genes Dev 1997; 11:996-1007. 49. Orkin RW, Knudson W, Toole BP. Loss of hyaluronate-dependent coat during myoblast fusion. Deo Biol 1985; 107527-530. 50. Pavasant P, Shizari T, Underhill CB. Hyaluronan contributes to the enlargement of hypertrophic lacunae in the growth plate. J CeN Sci 1996; 109:327-334. 51. Underhill CB, Nguyen HA, Shizari M, Culty M. CD44 positive macrophages take up hyaluronan during lung development. Dev Bid 1993; 155:324-336. 52. Gakunga P, Frost G, Shuster S, Cunha G, Formby 9 , Stern R. Hyaluronan is a prerequisite for ductal branching morphogenesis. Development 1997; 124:3987-3997. 53. Banerjee SD, Toole BP. Hyaluronan-binding protein in endothelial cell morphogenesis. J Cell Biol 1992; 119:643-652.

References

699

54. Evanko SP, Angello JC, Wight TN. Formation of hyaluronan- and versican-rich pericellular matrix is required for proliferation and migration of vascular smooth muscle cells. Avterioscler Thromh Vasc Biol 1999; 19:1004--1013. 55. Wight T. Hyaluronan in atherosclerosis and restenosis. Glycoforum: Science of hyaluronun. http://www.glycoforum.gr.jp 56. Tammi R, Tammi M. Hyaluronan in the epidermis. Glycoforum: Science ofhyaluronan. http:// www.glycoforum.gr.jp 57. Wang C, Tammi M, Guo H, Tammi R. Hyaluronan distribution in the normal epithelium of esophagus, stomach, and colon and their cancers. Am J Puthol 1996; 148:1861-1869. 58. Mohamadzadeh M, DeGrendele H, Arizpe H, Estess P, Siegelman M. Proinflammatory stimuli regulate endothelial hyaluronan expression and CD44/HA-dependent primary adhesion. J Clin Invest 1998; 101 :97- 108. 59. Knudson W, Biswas C, Li XQ, Nemec RE, Toole BP. The role and regulation of tumourassociated hyaluronan. Cibu Found Synzp 1989; 143:150- 159. 60. Auvinen PK, Parkkinen JJ, Johansson RT, Agren UM, Tammi RH, Eskelinen MJ, Kosma V-M. Expression of hyaluronan in benign and malignant breast lesions. fnt J Cancer 1997; 74:477-48 1. 61. Ropponen K, Tammi M, Parkkinen J, Eskelinen M, Tammi R. Lipponen P, Agren U: Alhava E, Kosma V-M. Tumor cell-associated hyaluronan as an unfavorable prognostic factor in colorectal cancer. Cancer Res 1998; 58:342--347. 62. Yeo T-K, Nagy JA, Yeo K-T, Dvorak HF, Toole BP. Increased hyaluronan at sites of attachment to mesentery by CD44-positive mouse ovarian and breast tumor cells. Am J Pathol

64.

65. 66. 67.

, Aruffo A, Stamenkovic I. Interaction between CD44 and hyaluronate is directly implicated in the regulation of tumor development. J E.xp Med 1994; 180:53-66. Schmits R, Filmus J, Genvin N, Senaldi G , Kiefer F, Kundig T, Wakeham A, Shahinian A, Catzavelos C, Rak J, Furlonger C, Zakarian A, Simard JJL, Ohashi PS, Paige CJ, GutierrezRamos JC, Mak TW. CD44 regulates hematopoietic progenitor distribution, granuloma formation, and tumorigenicity. Blood 1997; 90:22 17-2233. Zahalka MA, Okon E, Gosslar U, Holzmann B, Naor D. Lymph node (but not spleen) invasion by murine lymphoma is both CD44- and hyaluronate-dependent. J fmnzunol 1995; 1545345-5355. Zeng C, Toole BP, Kinney SD, Kuo J-W, Stamenkovic I. Inhibition of tumor growth in vivo by hyaluronan oligomers. Int J Cancer 1998; 77:396-401. Sy MS, Guo Y-J, Stamenkovic I. Inhibition of tumor growth in vivo with a soluble CD44immunoglobulin fusion protein. J Esp Med 1992; 176:623-627.

Carbohydrates in Chemistry and Biology Beat Emst, Gerald W. Hart, Pierre Sinay copyright Q WILEY-VCH Verlag GmbH. 2000

42 Biological Roles of Heparan Sulfate Proteoglycans Ofer Reizes, Pyong Woo Park, and Merton Bernjield

42.1 Introduction Heparan sulfate (HS) was distinguished from the pharmaceutical product heparin by its lower extent of sulfation about 50 years ago, but its importance to the cellular economy has only recently become apparent [ I ] . Since the discovery about 30 years ago that it is on the surface of nearly all animal cells, it has been found to be nearly ubiquitous in distribution: within secretory vesicles, at cell surfaces and within the extracellular matrix (ECM) [2, 31. Indeed, HS is covalently associated with distinct proteins at each of these sites. HS is an ancient molecule, appearing in organisms as early as Cnidaria, which likely accounts for its large number of ligands [4]. HS is the most structurally variable glycosaminoglycan (GAG), distinguished from all others by the presence of N-sulfated glucosamine residues, and the most acidic molecule made by animal cells [ 5 ] .HS is a linear polysaccharide, containing a repeating disaccharide unit of hexuronic acid, usually iduronic but also glucuronic acid (GlcA), alternating with an N-substituted glucosamine on which the substituents are either acetate (GlcNAc) or sulfate (GlcNS04) [ 5 ] .Additionally, hydroxyl groups of the polysaccharide backbone are often esterified with sulfate.

42.2 Heparan Sulfate Biosynthesis

HS biosynthesis is initiated by the enzymatic formation of a covalent bond between the reducing end of a xylosyl residue and the hydroxyl of certain serine residues on specific proteoglycan core proteins [6]. Sequential enzymatic addition of two galactosyl and a single glucuronosyl residue to the non-reducing end of the xylose forms a tetrasaccharide linkage region which serves an identical role in chondroitin

702

42 Biological Roles of Heparan Sulfate Proteoglycans

sulfate biosynthesis. The HS chain is initiated by the addition of a GlcNAc residue. The HS chain is then extended by a single HS copolymerase which adds GlcA alternating with GlcNAc residues to form the repeating 1,4-linked disaccharide chain [7, 81. This polymer, heparosan, undergoes a series of sequential modification reactions in which the product of one reaction is a substrate for the next. The initial modification is a concerted reaction catalyzed by a single enzyme in which N deacetylation of GlcNAc is coupled to N-sulfation. This reaction is followed by epimerization of GlcA residues adjacent to N-sulfates, forming L-iduronic acid. The next modifications are a series of sulfotransferase reactions, yielding 2-O-sulfated uronic acids and 3 - 0 and 6-O-sulfated hexosamines [9, 101. Because these modification reactions do not go to completion, the resulting HS chains are highly diverse and no two chains are likely identical. The modification reactions occur to different extents along the length of the polymer, resulting in domains that are highly sulfated (highly sulfated domains, HSDs) [ 111 and domains that are nearly unmodified (unmodified domains, UMDs) [ 121 separated by mixed domains in which sulfated residues alternate with unsulfated residues. Although HS biosynthesis appears to proceed by the same steps in all cell types and organisms, the macroscopic structure of the HS chain, including chain length (ca. 20-150 disaccharides) and number and size of HSDs and UMDs, can vary reproducibly between cell types [13, 141. For example, the HS chains produced by mast cells, the source of heparin, are highly sulfated and contain mostly HSDs. In part, these structural differences arise from multiple isoforms of each modification enzyme [15-17]. Different cell types may have distinct complements of these isoforms. These differences can be functionally significant because the binding avidity of HS chains for proteins may depend on the abundance and spacing of the HSDs.

42.3 Functions of Heparan Sulfate HS chains bind under physiological conditions to a large number of ECM components, coagulation factors, growth factors, growth factor binding proteins, proteinases, proteinase inhibitors, cytokines, chemokines, cell adhesion molecules as well as other ligands (reviewed in [IS]). Included among the HS-binding proteins are surface proteins of several families of viruses, notably the herpes viruses, and a variety of proteins associated with the surface of microbial pathogens, e.g. Neisseria yonorrhoeae [ 19, 201. Many of these proteins bind most avidly to discrete sequences in HS chains [21]. This binding can modify the interactions of these proteins with their receptors, affect the stability of protein-protein complexes or act to catalyze interactions between proteins. The proteins bind to the HS chains via basic residues on their surfaces. However, despite the very large number of protein ligands for HS chains, there are few consensus binding sequences on the proteins. The reason for this is not clear. One possibility is that proteins whose function depends on interaction with

42.5 lntrucellulur Proteoglycans

703

HS have co-evolved with the HS chains, and because HS is an ancient molecule, the sequence similarity may have eroded over evolutionary time [22]. A useful explanation for the function of HS chains is that they catalyze molecular interactions between proteins, as seen most commonly in the inhibition of thrombin by the serpin antithrombin 111 [23, 241. The idea is that the HS chain provides a surface to which both proteins bind and thus will encounter each other at a much more rapid rate than they would in solution. This effect of reduced dimensionality can explain both the acceleration of molecular interactions and the hyperbolic effect of the HS chain on the reaction [24].

42.4 Heparan Sulfate Proteoglycans Except for their intracellular degradation products, HS chains are found solely linked covalently to a variety of distinct core proteins, and these complexes are known as HS proteoglycans (HSPGs) [18, 251. The core proteins direct the intracellular and extracellular trafficking of the HSPGs and thus determine where the proteoglycan will be located. Consequently, we have classified the HSPGs by their location: i) intracellularly; ii) on the cell surface; and iii) within the ECM (Figure 1). HSPG core proteins contain evolutionarily conserved domains with unique functions in addition to conserved sites for HS attachment. The attachment sequence is not invariant, and includes a Ser-Gly dipeptide sequence often repeated two or more times. The attachment site is flanked, often on the N- and C-terminal side, by clusters of acidic residues as well as hydrophobic residues, usually C-terminal of the attachment site. For certain core proteins, the HS attachment site can serve to attach chondroitin sulfate chains [26, 271. Although heparin, a highly sulfated GAG derived from an intracellular HSPG has been known and its anticoagulant function exploited as a pharmaceutical product for many years, recent evidence has provided information on the relationship of heparin to HS and on the biological roles of the major HSPGs.

42.5 Intracellular Heparan Sulfate Proteoglycans HSPGs found within the cell are within secretory granules or, much less well characterized, within the nucleus. These PGs must be distinguished from HSPGs either undergoing biosynthesis, on ER membranes and in the Golgi, or degradation in the lysosomes. HSPGs in the nucleus have been described in several cell types 128, 291, but the core proteins of these PGs have not been defined. Moreover, in some cases, the possibility of adventitious association of highly anionic soluble HSPGs with basic nuclear proteins has not been rigorously excluded.

704

42 Biological Roles of Heparan Sulfate Proteoglycans

Basement Membrane

Figure 1. Schematic representation of major HSPGs. Serglycin is an intracellular HSPG that bears 7-15 HS chains attached to a continuous sequence of 9-24 Ser-Gly repeats. Serglycins of connective tissue mast cells within bovine lung or porcine intestinal mucosa serve as the common source of commercially available heparin. The syndecans and glypicans represent the major cell surface HSPGs. Syndecans are associated with the cell surface as a single pass transmembrane protein whereas glypicans are covalently associated via GPI linkage. The HS chains of syndecans are attached to the linear core protein distal to the plasma membrane and that of glypicans are attached to the extensively disulfide-linked core protein proximal to the plasma membrane. Perlecan and agrin are the major extracellular HSPGs and they exist mostly in BMs. Both perlecan and agrin are expressed in a polarized manner where their N-terminus is oriented towards the cell surface (motor neuron for agrin). The five and four distinct domains of perlecan and agrin, respectively, are shown as globular domains, and the HS attachment sites in specific domains are decorated with HS chains in the figure. The HS chains are shown as a chain of closed diamonds.

42.5.1 Serglycin and Heparin The HSPG of secretory granules in connective tissue mast cells, basophils and NK cells contains the distinctive serglycin core protein attached to highly sulfated HS chains also referred to as macromolecular heparin [30, 311. These cells are prominent in the connective tissue adjacent to blood vessels. In contrast, serglycin in mucosal mast cells and a variety of hematopoietic cells, including lymphoid cells, myeloid cells, macrophages and megakaryocytes, contains predominantly highly sulfated chondroitin sulfate chains; thus the identical core protein can contain distinct GAG chains depending on the cell type. The serglycin core protein is derived from proteolytic processing of the translation product, and bears 7-15 GAG chains attached to a continuous sequence of 9-24 serine-glycine repeats [32, 331. This amino acid sequence and the closely juxtaposed GAG chains impart substantial proteinase resistance to serglycin.

42.6 Cell Surfme Heparan Sulfate Proteoglycans

705

Within the mast cell secretory granule, serglycin binds a large number of basic proteases (tryptases and chymases) and vasoactive amines, especially histamine, which are released upon degranulation of mast cells from a variety of stimuli, including injury, cytokines and immune reactions, especially those involved in IgEdependent allergic responses (reviewed in [ 341). The HS chains of serglycin are highly sulfated and rich in iduronic acid and, when processed by commercial means, yield the pharmaceutical product heparin. There is no qualitative difference between the HS chains of serglycin, sometimes referred to as heparin chains, and the HS chains on other HSPGs; rather the structures represent a continuum, with the serglycin HS chain likely reflecting the lack of unmodified domains 1351. These differences presumably result from distinct isoforms of the biosynthetic enzymes. The protein-free heparin pharmaceutical product is commonly derived from the serglycin of connective tissue mast cells within bovine lung or porcine mucosa [25]. The fractions with highest anticoagulant activity (1 50-1 80 IU/mg) are considered to be heparin. This fraction contains 5- 15 kDa GAG chains. Heparin is a clinically useful anticoagulant that activates antithrombin 111, which then inhibits thrombin, Factor Xa and other coagulation enzymes [36]. At high concentrations, heparin interacts with heparin cofactor I1 to form another thrombin inhibitor and can influence platelet activity. The anticoagulant activity of heparin depends on a specific pentasaccharide sequence containing a 3-0-sulfated GlcNS04 residue within a larger highly sulfated domain. Thus, only a fraction of heparin molecules have anticoagulation activity. Recently, the use of low molecular weight heparin (3-5 kDa) has expanded because of its greater reliability as an anticoagulant in clinical situations [37]. The extensive literature on heparin and its clinical utility is beyond the scope of this review.

42.6 Cell Surface Heparan Sulfate Proteoglycans The core proteins determine the localization, expression, turnover and placement of HS chains a t the cell surface. The predominant source of cell surface HS are the syndecans and glypicans though several other proteins (betaglycan, CD44, and a581 integrin) have also been shown to carry HS chains. All these core proteoglycans (PGs) except glypicans are type I transmembrane protein, whereas glypicans are associated with the plasma membrane via a glycosylphosphoinositol linkage. The only conserved domain among these PGs are their GAG attachment sites. This section will focus on the cell surface HSPGs and their functional roles. 42.6.1 Syndecans Mammalian syndecans are a family of four widely expressed genes that show sequence conservation between their transmembrane and cytoplasmic domains [ 381.

106

42 Biological Roles of Heparan SuEfute Proteoglycuns

They are named syndecans, from the Greek syndein meaning to bind together, because they were originally thought to link cells to the ECM [39]. In Drosophila, there is a single syndecan gene [40] suggesting that syndecans arose early in development and have subsequently undergone two gene duplications. Their extracellular domains are variable in length and probably extended in conformation, show anomalous migration in polyacrylamide gels and contain HS chains distal from the plasma membrane. The syndecans are proposed to contain eight distinct domains [41].Within their extracellular region (ectodomain) in addition to a signal sequence leader peptide, they contain sites for GAG attachment, putative cell interaction, proteolytic cleavage, and oligomerization domains [ 181. The syndecan-4 ectodomain appears to contain a unique site for interaction with a cell surface ligand(s) [42, 431. Proteolytic cleavage (shedding) from the cell surface is a regulated process and is accelerated by various cellular effectors [44, 451. In syndecan-1, the proteolytic cleavage site likely resides within ten amino acids of the plasma membrane (unpublished observation). The shed ectodomains have similar binding and signaling activities as their cellular counterparts suggesting that they may act either as paracrine effectors or dominant negative inhibitors of ligand function. Following the ectodomain is a highly conserved transmembrane domain thought to localize the cores to distinct membrane compartments. It may also contain an additional oligomerization domain. Finally, there appear to be three domains within the short (34-38 amino acid) cytoplasmic region, membrane proximal and carboxyl terminal common regions, that are split by a variable domain [41].The cytoplasmic domains contain sites for phosphorylation within the membrane proximal domain, and sites for interaction with cell scaffolding proteins, the actin cytoskeleton, and cell signaling pathways [46-501. The cytoplasmic domains share sequence homology with the neurexins including a binding site for PDZ (PSD-95, Discs-large, and Zonula occludens-1] domain-containing proteins. This region has been shown to interact with several PDZ containing proteins including syntenin [51] and CASK, a homolog of the membrane-associated guanylate kinase family (MAGUK) [52, 531. CASK and its Cuenorhubditis elegans ortholog LIN2A, a gene involved in cell scaffolding contain binding sites for the ERM (ezrin, moesin, radixin) family of proteins and talin, all known to associate with the actin cytoskeleton and thought to organize and assemble protein complexes at the inner face of the plasma membrane PI. 42.6.2 Glypicans

Glypicans are a family of six genes in mammals of molecular weight 50-60 kDa containing a cysteine-rich extracellular domain and a site for HS attachment near the plasma membrane [54, 551. They associate with the plasma membrane via a linkage to a glycosylphosphatidyinositol (GPI) group and are localized within ordered microdomains in the plasma membrane called “rafts” [56]. GPI anchored proteins turnover rapidly, so the glypicans may mediate the internalization and recycling of HS binding proteins (e.g. superoxide dismutase, antithrombin 111, folli-

42.7 Purt-time Cell Surfuce Hepuran Suljute Proteoglycans

707

statin, and lipoprotein lipase). Release of glypicans from the cell surface may occur either via proteolysis or by hydrolysis via a phosphatidylinositol specific phospholipase C . Soluble glypicans are detected in the conditioned media from cultured cells, though neither the presence of soluble glypicans in uivo nor regulated shedding in culture have been described. A D. melunogaster mutant called dully (division abnormally delayed) codes for the fly glypican homolog, and shares all the characteristics of its mammalian counterparts [57]. Dully loss of function mutants resemble mutants in the wingless signaling pathway and recent epistatic analysis revealed that dully modulates signaling via the wingless as well as the decapentaplegic pathways [58, 591. In humans, glypican-3 mutants give rise to the Simpson-GolabiBehmel Syndrome, an X-linked overgrowth disorder [60].The defect is likely due to either growth inhibition or apoptosis during development. Interestingly, in vitro, glypican-3 is able to induce apoptosis in a cell-specific manner though HS chains are not required [61]. Thus, unlike the syndecans, glypicans may have HS-independent functions similar to the part-time proteoglycans described below.

42.7 Part-time Cell Surface Heparan Sulfate Proteoglycans In contrast to the abundance of HS chains presented on the cell surface by the syndecans and glypicans, several other HSPGs have been identified that express HS chains, though their expression is less abundant and is dependent on alternative splicing. Betaglycan and CD44 are thus considered part-time proteoglycans, which present both HS and chondroitin sulfate, though both have GAG-independent functions. They are discussed in this Chapter because despite their low abundance their HS chains function analogously to full-time HSPGs.

42.7.1 Betaglycan Betaglycan is the TGFD type 111 receptor and binds all three TGFB isoforms in an HS independent manner, though, TGFPl and p2 will bind to the highly sulfated HS chains [62 -641. Betaglycan is expressed in multiple embryonic and adult tissues, including mesenchyme, epithelia, neurons, but not endothelia. It is composed of a large extracellular domain (853 amino acids) with the HS chains attached within the juxtamembrane region, a transmembrane domain, and a short [41 amino acids) cytoplasmic domain [65].Shed (or secreted) betaglycan has been detected in conditioned media of cultured cells that overexpress betaglycan, though this release is a slow process and apparently not regulated, in contrast to syndecan shedding [66]. While the HS chains are not necessary for TGFP binding to betaglycan, FGF-2 does bind to these GAG chains [67]. The function of betaglycan in TGFP biology and the relevance of its HS chains remain unclear.

708

42 Biological Roles of Heparan Sulfate Proteoglycans

42.7.2 CD44 CD44 is the receptor for hyaluronan, a nonsulfated GAG of the interstitial space, and has been implicated in diverse functions including lymphocyte homing, inflammation, hematopoiesis, tumor progression and metastasis [68]. The CD44 gene is composed of 21 exons leading to expression of multiple splice variants. The most common CD44 core protein is composed of a 250 amino acid extracellular domain, a 22 amino acid transmembrane region, and a 72 amino acid cytoplasmic region. Splicing of the 11 variable exons occurs between amino acids 222 and 223 of the extracellular domain. The CD44E splice variant, originally identified in cultured keratinocytes as a novel HSPG, contains exon v3 that codes for the HS and chondroitin sulfate attachment sites. These HS chains are able to specifically bind to many HS binding proteins such as FGF-2, hepatocyte growth factor/scatter factor (HGF/SF) and the chemokine RANTES (yegulated upon activation, gormally 1 cell expressed and secreted) [69-711. Binding of HGF/SF to the CD44 HS chains leads to activation of c-Met, the HGF/SF receptor [72]. The cytoplasmic domain is specifically phosphorylated on a serine residue, a requirement for cell migration in melanoma cells, and interacts with ankyrin and the actin cytoskeleton via ERM proteins [73, 741. Additionally, the extracellular domain may be cleaved from the cell surface by a regulated process that likely involves a tissue inhibitor of metalloproteases-1 (TIMP-1) sensitive metalloprotease [75]. Inhibition of shedding suppresses lung cancer cell migration on hyaluronate surfaces. Splice variants bearing the HS containing exons have been identified in colorectal tumors, normal and neoplastic epithelia, as well as activated lymphocytes and malignant lymphomas [76, 771. A specific function for CD44 and the HS bearing splice variants remains unclear.

42.8 Functions of Cell Surface Heparan Sulfate Proteoglycans Cell surface HSPGs may act as receptors or coreceptors for a variety of extracellular ligands, both soluble and insoluble, either at the cell surface or as shed soluble effectors. Multiple ligand-receptor encounters have been shown to be modulated by HS based on studies using heparin, HS, modified oligosaccharides, GAG degrading enzymes, and the intact PGs. Where studied the soluble oligosaccharide functions in an analogous manner to the intact cell surface HSPGs. 42.8.1 Ligand Receptors Cell surface HS can bind a variety of extracellular effectors, acting as clearance/ internalization receptors. This is an important process in the uptake of FGF-2, follistatin/activin complex, vitronectin and thrombospondin [78-80]. Additionally, HSPGs are important in the uptake and activity of antithrombin 111, and lipo-

42.8 Functions of Cell Surface Heparan Sulfate Proteoglycans

709

protein lipase (LPL) [81]. HS may also function in localization of their activity at the cell surface. In hepatocytes and endothelia, the active dimeric LPL binds to cell surface HSPGs, concentrating it at the surface where it degrades the lipids from chylomicrons and very low density lipoproteins [82].

42.8.2 Ligand Coreceptors Ligand receptor encounter and subsequent signal transduction is often regulated by cell surface HS. Both soluble (growth factor, cytokine) and insoluble (microbe, matrix component, cell) ligands bind to HSPGs and ligand activity is modulated at their cognate receptors (reviewed in [IS]). The mechanism of action of HS on soluble ligands remains controversial and no unifying model exists. A simple paradigm has been proposed for the single pass transmembrane receptor tyrosine kinases (e.g. FGF receptors), in which HS facilitates the formation of ligand dimers or higher order oligomers. Subsequently, the ligand interacts with its high affinity receptor and facilitates the formation of a dimeric receptor state that is capable of activating the receptor’s cytoplasmic kinase [83, 841. Despite its simplicity there is no consensus on the validity of this model. In contrast, the chemokine receptors (seven pass transmembrane receptors) which are also modulated by HS, formation of dimeric receptor complexes has not been established as a mechanism for facilitating signal transduction, though several seven pass transmembrane receptors may form dimers [85, 861. It is quite likely that a different model will be required to model the action of HS on the seven pass transmembrane receptors. HSPG binding to insoluble ligands leads to the formation of an immobile complex between the ligand and the actin cytoskeleton [38, 871. The complexes strengthen the formation of cell-cell adhesion by binding to secondary sites on homophilic and heterophilic cell adhesion molecules [88-901. HSPGs are also integral in cell-ECM adhesion as coreceptors for the integrins [91]. The recent demonstration that a5pl is a part time proteoglycan emphasizes the role of the HSPG partner in integrin ECM interactions [ 921. Finally, microbial pathogenesis is facilitated by interactions with cell surface HSPGs by serving as initial attachment sites for microbes which subsequently fuse with the host cell membrane via specific interactions between microbial-host internalization determinants [ 181.

42.8.3 Shed Effectors Shedding of cell surface HSPGs instantly converts these tethered PG modulators to soluble effectors. While all cell surface HSPGs are shed in vitro, regulated shedding and soluble HSPGs in tissue fluids have been documented only for the syndecans and CD44. Constitutive shedding likely is part of the process of normal HS turnover, while regulated shedding which is stimulated by a diverse set of non-HS binding ligands likely serves a distinct function. Because the HS chains of the soluble extracellular domains are equivalent to those of the cell surface proteoglycan, they can function either as agonists or antagonists of ligand binding [93]. In an

710

42 Biological Roles o j Heparan Sulj'ute Proteoglycans

analogous manner to heparin or HS oligosaccharides, the soluble ectodomain may provide a template for ligand receptor encounters that would depend on the ectodomain concentration. Thus, at low concentrations they may participate in complex formation, while at higher concentrations they may disrupt ligand-receptor interaction and behave as antagonists. Finally, the soluble HSPG may serve to protect ligands from proteolysis in the extracellular environment, as such they may act as carriers or transporters of ligands from one cell or tissue to another [25, 941.

42.9 Extracellular Matrix Heparan Sulfate Proteoglycans and Their Functions The ECM refers to the connective tissue that surrounds mesenchymal cells and to the basement membrane (BM) that encompasses parenchymal cells. The ECM contains a large variety of PGs, including members of the aggrecan family, which interact with hyaluronan, and of the decorin family, which have leucine-rich repeats. However, the former are predominantly chondroitin sulfate or keratan sulfate PGs. These PGs can be major tissue constituents, such as aggrecan in cartilage or decorin in tendon. Interestingly, the HSPGs in the ECM are mostly restricted to BMs. BMs are highly specialized ECMs generally distinguished by two major properties [95, 961. First, they are found as thin sheets at the basolateral side of polarized cells (epithelial, mesothelial, endothelial) or surrounding muscle fibers, peripheral nerves, and adipocytes. Second, BMs are comprised of distinct ECM components, of which the ubiquitous components are laminins, collagen IV, entactin/nidogen, and HSPGs. Several lines of evidence indicate that BMs are formed by self-assembly of laminins and collagen IV, which is followed by linkage of these two distinct networks by entactin and subsequent binding of HSPGs and other ECM components to this primary scaffold. Perlecan and agrin are the best characterized extracellular HSPGs, and both HSPGs have been found to accumulate within BMs at various tissue sites. These HSPGs play critical roles in essential functions of the BM, such as compartmentalization of tissues, sequestration and presentation of growth factors, induction of synaptic differentiation, and regulation of macromolecule permeability. This section reviews the functional roles of extracellular HSPGs, primarily in the context of BM function. 42.9.1 Perlecan

Perlecan, initially isolated from the Englebreth-Holm-Swarm murine tumor, is one of the largest HSPGs with a core protein size of approximately 400 kDa [97, 981. The name perlecan was derived from rotary shadowing electron microscopy images showing that the HSPG contains globular domains resembling a string of pearls. Subsequent molecular characterization of the murine and human cDNA clones showed that perlecan is comprised of five distinct domains. Importantly, each

42.9 Extracellular Matrix Heparan Sulfute Protroglycans and Their Functions

7 11

domain has specific functions in mediating the multiple and diverse physiological activities of perlecan. Starting at the N-terminus, domain I consists of primary sequences unique to perlecan, is oriented towards the cell surface, and is the only domain with HS attachment sites. The three Ser-Gly-Asp GAG attachments sites in domain I are generally decorated with HS chains, but in some cases, chondroitin sulfate chains have also been detected [99]. HS attachment in domain I is regulated by an SEA region, named after the three proteins with this motif (Sperm protein, enterokinase, agrin), in the distal portion of the three Ser-Gly-Asp HS attachment sites. The HS chains of perlecan are partially responsible for the charge-selective permeability properties of BMs. For example. removal of the highly charged HS component from the glomerular BM has been found to increase protein permeability and lead to proteinuria [ 1001. The HS chains of domain I can also bind to various ligands including growth factors (e.g. FGF-2, TGFP-1 and 21 and other major BM components (e.g. laminin), and it is thought that these interactions mediate specific aspects of cell growth and differentiation, tumor cell metastasis, and assembly and maintenance of BMs. Binding of perlecan HS to growth factors is physiologically significant in several aspects. For instance, the interaction between FGF-2 and the HS chains of perlecan can: i) protect the growth factor from degradation [ 1011; ii) provide a reservoir of FGF-2 which can be released by the action of specific proteases [102]; and iii) present the growth factor to its high affinity receptor on the cell surface [ 1031. These perlecan HS-growth factor interactions influence the growth, differentiation and invasive properties of normal and tumor cells. Perlecan HS has also been shown recently to function in a non-BM compartment. Perlecan was found to be a marker of chondrogenesis and potentiated chondrogenic differentiation in vitro [ 1041. Whether perlecan is required for normal chondrogenesis awaits development of perlecan mutant animal models. Domain I1 contains four repeats homologous to the low density lipoprotein (LDL) binding region of the LDL receptor and it has also been shown that this domain binds to entactin. Although these properties suggest that domain I1 participates in LDL binding and incorporation of perlecan into BMs, these hypotheses have not been tested. Domain I11 is homologous to the N-terminal third of the laminin A chain, and consists of laminin-like globular domains separated by cysteinerich EGF-like regions. Interestingly, domain 111 of murine perlecan has been demonstrated to promote adhesion of various cells through an Arg-Gly-Asp motif present in this domain [105]. However, since the R G D sequence is not present in human perlecan, it is likely that other domains are involved in the cell adhesive activities of perlecan. Domain IV is the largest and consists of variably spliced immunoglobulin-like repeats similar to those found in NCAM. By inference, this domain is thought to be important in homophilic polymerization of perlecans. Domain V, at the C-terminus, consists of three distinct globular domains that alternate with EGF-like repeats. The globular domains in domain V are homologous with the G-domain of the long arm of the A chain of laminin-1. Domain V can self-assemble and also contains a Leu-Arg-Glu (LRE) cell adhesive motif found in laminin-3 (S-laminin). Thus, this domain is implicated in BM assembly and cell adhesion. There are several other core protein-mediated functions of perlecan that have not been ascribed to specific domains. For example, perlecan core protein can

112

42 Biological Roles of Heparan Sulfate Proteoglycans

bind to fibronectin [ 1061 and FGF-7 [ 1071, and these interactions are thought to play important roles in BM assembly and perlecan-mediated angiogenesis, respectively.

42.9.2 Agrin Agrin, first isolated from the BM fraction of the synapse-rich electric organ of Torpedo californica (marine ray), is a major 200 kDa HSPG core protein of neuromuscular junction and renal tubular BMs [108, 1091. Localized secretion of agrin into BMs of the neuromuscular junction is of paramount importance since agrin induces post-synaptic differentiation [ 1081. Mice lacking agrin lack normal synapses [ 1 101. Agrin is required for clustering nicotinic acetylcholine receptors (nAChRs), acetylcholinesterase, rapsyn, utrophin, neuregulin, and neuregulin receptors at postsynaptic sites. Accordingly, agrin associates with neuromuscular BMs in a polarized manner. The N-terminal domain of agrin, containing its BM targeting function, is located near the motor neuron cell surface, whereas its C-terminal domain, containing its nAChR clustering activity, is at the muscle cell surface [ 1111. Similar to perlecan, agrin is a multidomain HSPG with four distinct functional domains. Domain I, named NtA for N-terminal in agrin, encompasses the first 130 amino acids and mediates binding to laminins. This binding interaction with laminin is critical in that it localizes secreted agrin to synaptic BM [ 1121. The signature characteristic of domain I1 is the nine follistatin-like repeats that are homologous to the Kazal-type of protease inhibitors, suggesting that this domain may function to inhibit proteolysis and mediate growth factor binding. Domain I1 also contains one of the three HS attachment sites located in the amino-end of the eighth follistatin-like repeat. Domain I11 contains the other two conserved HS attachment sites, and is characterized by a central SEA HS attachment regulatory motif flanked by two Ser/ Thr-rich regions. Clustering of neuromuscular synaptic components is mediated by domain IV. This domain, comprised of four EGF-like and three laminin G-like domains, is homologous to domain V of perlecan, and by analogy, may also self assemble. Furthermore, alternatively spliced variants of domain IV with insertion of 4-19 amino acids have been described. This alternative splicing has been found to regulate the ability of agrin to interact with heparin, a-dystroglycan, and the cell surface [ 1131. Although agrin clearly exists as an HSPG in viuo, the physiological role of agrin HS chains is not fully understood. Available evidence indicates that HS chains of agrin are not involved in agrin-induced post-synaptic differentiation. However, since agrin is not only found in neuromuscular junctions, but also in the BMs of kidney, brain, skin, gastrointestinal tract and heart, agrin HS may have functions similar to perlecan HS, such as in regulating BM assembly and macromolecule permeability, at these tissue sites. Furthermore, several studies have demonstrated that heparin inhibits the ability of agrin to cluster nAChRs, agrin binds heparin, and muscle cell lines deficient in HS biosynthesis are unable to form nAChR clusters in response to agrin [114-1161. These findings suggest that agrin itself is an HS binding protein and the ability of agrin to bind muscle cell surface HS may be required for agrin-induced post-synaptic differentiation.

42.10 Conclusions

7 13

42.9.3 Other Extracellular HSPGs In addition to perlecan and agrin, collagen XVIII has been recently identified as a third HSPG component of BMs [117].In its intact form, collagen XVIIl migrates as a smear of approximately 300 kDa in SDS-PAGE. Although the exact physiological role is unknown, collagen XVIII is abundant in BMs of various tissues including retina, epidermis, cardiac and striated muscles, blood vessels, lung and kidney, suggesting that collagen XVIII may have functions similar to perlecan in BMs. Recently, proteolytic cleavage of the C-terminal end of collagen XVIII (NC1) has been found to produce an 18 kDa peptide, named endostatin, which has an antiangiogenic activity [118]. Because NC1 and endostatin can bind to heparin [119], the HS chains of collagen XVIII may also act to regulate the generation and activity of endostatin. In addition to perlecan, agrin and collagen XVIII, other HSPGs of unknown identity have been found in BMs [120]. Available data indicate that cell surface HSPGs, such as syndecans, can be secreted by a proteolytic mechanism known as shedding [44]. Whether shed cell surface HSPGs define yet another component of BMs (and other ECMs) remains to be investigated.

42.10 Conclusions The function of HS and HSPGs continues to be defined and refined. The HS chains serve as ligand protectors, transporters, internalizers, and coreceptors in receptor encounters. It is evident that as new HS binding ligands are discovered additional functions will be uncovered. For example recent studies suggest that HSPGs are involved in pathophysiological processes, such as angiogenesis, apoptosis, microbial pathogenesis, tumorigenesis, and energy balance. Clinically, heparin continues to be used as an anticoagulant, though, new HS oligosaccharides with greater specificity may be developed with improved clinical efficacy. Finally, mutants in the various PGs and GAG biosynthetic enzymes will undoubtedly provide new insights in our understanding of these complex molecules. References 1. 2. 3. 4. 5. 6. 7. 8.

J. E. Jorpes, S. Gardeli, J. Biol. Chem. 1948 176, 267-75. P. M. Kraemer, Biochemistry 1971 10, 1437-45. P. M. Kraemer, Biochemisfry 1971 10, 1445-51. H. B. Nader, T. M. P. C. Ferreira, L. Toma, S. F. Chavante, C. P. Dietrich, B. Basu, G. Torri, Curhohydr. Rex 1988 184, 292-300. M. Salmivirta, K. Lidholt, U. Lindahl, FASEB J. 1996 10, 1270-79. J. D . Esko, L. Zhang, Curr. Opin. Struct. Bid. 1996 6 , 663-70. T. Lind, U. Lindahl, K. Lidholt, J. Bid. Chern. 1993 268, 20705-8. T. Lind, F. Tufaro, C. McMormick, U. Linddhl, K. Lidholt, J . Bid. Chem. 1998 273, 2626568.

714

42 Biological Roles of Hepurun Sulfute Proteoglycuns

9. H. Habuchi, 0. Habuchi, K. Kimata, Trends Glycosci. Glycotechnol. 1998 10, 65-80. 10. R. D. Rosenberg, N. W. Shworak, J. Liu, J. J. Schwartz, L. Zhang, J. Clin. Invest. 1997 99, 2062-70. 11. M. Maccarana, Y. Sakura, A. Tawada, K. Yoshida, U. Lindahl, J. Biol. Chem. 1996 271, 17804-1 0. 12. T. Toida, R. J. Linhardt, Trends Glycosci. Glycotechnol. 1998 10, 125-36. 13. R. D. Sanderson, J. E. Turnbull, J. T. Gallagher, A. D. Lander, J. Biol. Chem. 1994 269, 13100-6. 14. M. Kato, H. Wang, M. Bernfield, J. T. Gallagher, J. E. Turnbull, J. Bid. Chem. 1994 269, 18881-90. 15. J. Aikawa, J. D. Esko, J. Biol. Chem. 1999274, 2690-5. 16. J. Liu, N. W. Shworak, P. Sinay, J. J. Schwartz, L. Zhang, L. M. Fritze, R. D. Rosenberg, J. Biol. Chem. 1999 274, 5185-92. 17. N. W. Shworak, J. Liu, L. M. Petros, L. Zhang, M. Kobayashi, N. G. Copeland, N. A. Jenkins, R. D. Rosenberg, J. Biol. Chem. 1999 274, 5170-84. 18. M. Bernfield, M. Gotte, P. W. Park, 0. Reizes, M. L. Fitzgerald, J. Lincecum, M. Zako, Annu. Rev. Biochem. 1999 68, 729-777. 19. J. M. Patti, B. L. Allen, M. J. McGavin, M. Hook, Annu. Rev. Microbiol. 1994 48, 585617. 20. K. S. Rostand, J. D. Esko, Infect. Zmmun. 1997 65, 1-8. 21. M. Lyon, J. T. Gallagher, Matrix Biol. 1998 17, 485 -93. 22. R. F. Doolittle, Annu. Rev. Biochem. 1995 64, 287-314. 23. R. D. Rosenberg, Fed. Proc. 1985 44, 404-9. 24. A. D. Lander, Matrix Biol. 1998 17, 465-72. 25. H. E. Conrad, Heparin-Binding Proteins, Academic Press, San Diego 1998. 26. L. Zhang, G. David, J. D. Esko, J. Biol. Chem. 1995270, 27127-35. 27. L. Zhang, J. D. Esko, J. Bid. Chem. 1994269, 19295-9. 28. N. S. Fedarko, H. E. Conrad, J. Cell Bid. 1986 102, 587-99. 29. M. Ishihara, N. S. Fedarko, H. E. Conrad, J. Biol. Chem. 1986261, 13575-80. 30. U. Gullberg, E. Andersson, D. Garwicz, A. Lindmark, 1. Olsson, Eur. J. Haematol. 1997 58, 137-53. 31. H. Kresse, H. Hausser, E. Schonherr, Experientia 1993 49, 403-16. 32. M. A. Bourdon, M. Shiga, E. Ruoslahti, J. Biol. Chem. 1986261, 12534-7. 33. M. A. Bourdon, A. Oldberg, M. Pierschbacher, E. Ruoslahti, Proc. Natl Acad. Sci. USA 1985 82, 1321-5. 34. R. L. Stevens, K. F. Austen, Immunol. Today 1989 10, 381-6. 35. S. R. Delaney, H. E. Conrad, Biochem. J. 1983209, 315-22. 36. J. A. Marcum, Perspect. Biol. Med. 1996 39, 610-25. 37. N. S. Gunay, R. J. Linhardt, Planta Med. 1999 65, 301-6. 38. M. Bernfield, R. Kokenyesi, M. Kato, M. T. Hinkes, J. Spring, R. L. Gallo, E. J. Lose, Ann. Rev. Cell Biol. 1992 8, 365-93. 39. S. Saunders, M. Jalkanen, S. O’Farrell, M. Bernfield, J. Cell Biol. 1989 108, 1547-56. 40. J. Spring, S. E. Paine-Saunders, R. 0. Hynes, M. Bernfield, Proc. Natl Acad. Sci. USA 1994 91, 3334-38. 41. A. C. Rapraeger, V. L. Ott, Curr. Opin. Cell. Bid. 1998 10, 620-8. 42. A. J. McFall, A. C. Rapraeger, J. Biol. Chem. 1998273, 28270-6. 43. A. J. McFall, A. C. Rapraeger, J Bid. Chem. 1997272, 12901-4. 44. S. V. Subramanian, M. L. Fitzgerald, M. Bernfield, J. Biol. Chem. 1997 272, 14713-20. 45. M. L. Fitzgerald, P. W. Park, Z. Wang, J.-S. Chun, G. Murphy, M. Bernfield, J. Cell Biol. 1999 Submitted. 46. E. S. Oh, J. R. Couchman, A. Woods, Arch. Biochem. Biophys. 1997 344, 61-74. 47. J. Reiland, V. L. Ott, C. S. Lebakken, C. Yeaman, J. McCarthy, A. L. Rapraeger, Biochem. J. 1996 319, 39-41. 48. T. Prasthofer, B. Ek, P. Ekman, R. Owens, M. Hook, S. Johansson, Biochem. Mol. Bid. Znt. 1995 36, 793-802. 49. A. Horowitz, M. Simons, J. Biol. Chem. 1998 273, 10914-18.

Rejerences

115

SO. T. Kinnunen, M. Kaksonen, J. Saarinen, N. Kalkkinen, H. B. Peng, H. Rauvala, J. Bid. Chem. 1998 273, 10702-8. 51. J. J. Grootjans, P. Zimmermann, G. Reekmans, A . Smets, G. Degeest, J. Durr, G. David, Proc. Natl Acad. Sci. USA 1997 94, 13683-88. 52. A. R. Cohen, D. F. Woods, S. M. Marfatia, Z. Walther, A. H. Chishti, J. M. Anderson, D. F. Woods, J. Cell Biol. 1998 142, 129-38. 53. Y. P. Hsueh, F. C. Yang, V. Kharazia, S. Naisbitt, A. R. Cohen, R. J. Weinberg, M. Sheng, J. Cell Biol. 1998 142, 139-51. 54. G. David, V. Lories, B. Decock, P. Marynen, J.-J. Cassiman, H. Van den Berghe, J. Cell Biol. 1990 111, 3165-76. 5 5 . A. D. Lander, C. S. Stipp, J. K. Ivins, Perspect. Dev. Neurobiol. 1996 3, 347-58. 56. R. Varma, S. Mayor, Nature 1998 394, 798-801. 57. H. Nakato, T. A. Futch, S. B. Selleck, Deuelopment 1995 121, 3687-702. 58. M. Tsuba, K. Kamimura, H. Nakato, M. Archer, W. Staatz, B. Fox, M. Humphrey, S. Olson, V. Kaluza, E. Siegfried, L. Stam. S. B. Selleck, Nature 1999 400, 276-280. 59. X. Lin, N. Perrimon, Nature 1999 400, 281-284. 60. G. Pilia, R. M. Hughes-Benzie, A. MacKenzie, P. Baybayan, E. Y. Chen, R. Huber, G. Neri, A. Cao, A. Forabosco, D. Schlessinger, Nut. Genet. 1996 12, 241-7. 61. A. D. Gonzalez, M. Kaya, W. Shi, H. Song, J. R. Testa, L. Z. Penn, J. Filmus, J. Cell Biol. 1998 141, 1407-14. 62. F. Lopez-Casillas, S. Cheifetz, J. Doody, J. L. Andres, W. S. Lane, J. Massague, Cell 1991 67, 785-95. 63. X. F. Wang, H. Y. Lin, E. Ng-Eaton, J. Downward, H. F. Lodish, R. A. Weinberg, Cell 1991 67, 797-805. 64. J. L. Andres, L. Ronnstrand, S. Cheifetz, J. Massague, J. Biol. Clzem. 1991 266, 23282-7. 65. J. Massague, J. Andres, L. Attisano, S. Cheifetz, F. Lopez-Casillas, M. Ohtsuki, J. L. Wrana, Mol. Reprod. Dev. 1992 32, 99-104. 66. F. Lopez-Casillas, H. M. Payne, J. L. Andres, J. Massague, J. Cell Biol. 1994 124, 557-68. 67. J. L. Andres, D. DeFalcis, M. Noda, J. Massague, J. Biol. Chem. 1992 267, 5927-30. 68. D. Naot, R. V. Sionov, D. Ish-Shalom, Adu. Cancer. Res. 1997 71, 241-319. 69. K. L. Bennett, D. G. Jackson, J. C. Simon, E. Tanczos, R. Peach, B. Modrell, I. Stamenkovic, G. Plowman, A. Aruffo, J. Cell Biol. 1995 128, 687798. 70. D. G. Jackson, J. I. Bell, R. Dickinson, J. Timans, J. Shields, N. Whittle, J. Cell Biol. 1995 128, 673-85. 71. B. Greenfield, W. C. Wang, H. Marquardt, M. Piepkorn, E. A. Wolff, A. Aruffo, K. L. Bennett, J. Biol. Chem. 1999 274, 251 1-7. 72. R. van der Voort, T. E. Taher, V. J. Wielenga, M. Spaargaren, R. Prevo, L. Smit, G. David, G. Hartmann, E. Gherardi, S. T. Pals, J. Bid. Chem. 1999 274, 6499 -506. 73. S. Yonemura, S. Tsukita, J. Cell Biol. 1999 145, 1497-509. 74. S. Yonemura, M. Hirao, Y . Doi, N. Takahashi, T. Kondo, S. Tsukita, J. Cell Biol. 1998 140, 885-95. 75. 1. Okamoto, Y. Kawano, H. Tsuiki, J. Sasaki, M. Ndkao, M. Matsumoto, M. Suga, M. Ando, M. Nakajima, H. Saya, Oncogene 1999 18, 1435-46. 76. K. Nackaerts, E. Verbeken, G. Deneffe, B. Vanderschueren, M. Demedts, G. David, Intl J. Cancer 1997 74, 335-45. 77. R. J. Sneath, D. C. Mangham, Mol. Pathol. 199851, 191-200. 78. 0. Hashimoto, T. Nakamura, H. Shoji, S. Shimasaki, Y. Hayashi, H . Sugino, J. Biol. Chenz. 1997 272, 13835-42. 79. P. E. Gleizes, J. Noaillac-Depeyre, M. A. Dupont, N. Gas, Eur. J. Cell. Biol. 1996 71, 144-53. 80. M. J. Wells, M. A. Blajchman, J. Riol. Chem. 1998 273, 23440-7. 81. F. Rinninger, T. Kaiser, W. A. Mann, N. Meyer, H. Greten, U. Beisiegel, J. Lipid Res. 1998 39, 1335-48. 82. A. Lookene, R. Savonen, G. Olivecrona, Biochemistry 1997 36, 5267-75. 83. J. Schlessinger, I. Lax, M. Lemmon, Cell 1995 83, 357-60. 84. M. S. Filla, P. Dam, A. C. Rapraeger, J. Cell. Physiol. 1998 174, 310-21. 85. B. A. Jordan, L. A. Devi, Nature 1999 399, 697-700.

716

42 Biological Roles of Hepuran Surfate Proteoglycans

86. M. Bai, S. Trivedi, 0. Kifor, S. J. Quinn, E. M. Brown, Proc. Natl Acad. Sci. U S A 1999 96, 2834-9. 87. K. Hayashi, M. Hayashi, M. Jalkanen, J. Firestone, R. L. Trelstad, M. Bernfield, J. H i m chem. Cytochem. 1987 35, 1079-88. 88. S. M. Albelda, C. W. Smith, P. A. Ward, FASEB J. 1994 8, 504-12. 89. A. A. Reyes, R. Akeson, L. Brezina, G. J. Cole, Cell Reg. 1990 I , 567-76. 90. K. L. Bennett, J. Bradshaw, T. Youngman, J. Rodgers, B. Greenfield, A. Aruffo, P. S. Linsley, J. Bid. Chem. 1997 272, 26940-6. 91. A. Woods, J. R. Couchman, S. Johansson, M. Hook, EMBO Jourrtal1986 5, 665-70. 92. S. S. Veiga, M. Elias, W. Gremski, M. A. Porcionatto, R. da Silva, H. B. Nader. R. R. Brentani, J. Bid. Chem. 1997 272, 12529-35. 93. M. Kato, H. Wang, V. Kainulainen, M. F. Fitzgerald, S. Ledbetter, D. M. Ornitz, M. Bernfield, Nature Med. 1998 4 , 691-97. 94. P. E. Bickel, 0. Reizes, 0. Goldberger, P. E. Scherer, H. Lodish, M. Bernfield, J. Biol. Chem. 1999 submitted. 95. R. Timpl, J. C. Brown, Bioessuys 1996 18, 123-131. 96. P. D. Yurchenco, J. J. O’Rear, Curr. Opin. Cell Biol. 1994 6, 674-681. 97. D. M. Noonan, A. Fulle, P. Valente, S. Cai, E. Horigan, M. Sasaki, Y. Yamada, J. R. Hassell, J. Biol. Chem. 1991 266, 22939-22947. 98. R. V. Iozzo, Annu. Rev. Biochem. 199867, 609-652. 99. M. Dolan, T. Horchar, B. Rigatti, J. R. Hassell, J. Bid. Chem. 1997272, 4316-4322. 100. Y. S. Kanwar, A. Linker, M. G. Farquhar, J. Cell Biol. 1980 86, 688-693. 101. 0. Saksela, D. Moscatelli, A. Sommer, D. B. RiRcin, J. Cell Biol. 1988 197, 743-751. 102. J. M. Whitelock, A. D. Murdoch, R. V. Iozzo, P. A. Underwood, J. Bid. Chem. 1996 271, 10079-86. 103. D. Aviezer, D. Hecht, M. Safran, M. Eisinger, G. David, A. Yayon, Cell 1994 79, 1005-1013. 104. M. M. French, S. E. Smith, K. Akanbi, T. Sanford, J. Hecht, M. C. Farach-Carson, D. D. Carson, J. Cell. Biol. 1999 145, 1103 -15. 105. S. Chakravarti, T. Horchar, B. Jefferson, G. W. Laurie, J. R. Hassell, J. Bid. Chem. 1995 270, 404-409. 106. A. Heremans, B. De Cock, J. J. Cassiman, H. Van den Berghe, G . David, J. Bid. Chein. 1990 265, 8716-8724. 107. B. Sharma, M. Handler, I. Eichstetter, J. M. Whitelock, M. A. Nugent, R. I. Iozzo, J. Clin. Invest. 1998 102, 1599-1608. 108. S. J. Burden, Genes Den 1998 12, 133-148. 109. G. Tsen, W. Halfter, S. Kroger, G. J. Cole, J. Bid. Chem. 1995 270, 3392-3399. 110. M. Gautam, P. G. Noakes, L. Moscoso, F. Rupp, R. H. Scheller, J. P. Merlie, J. R. Sanes, Cell 1996 85, 525-535. 111. A. J. Denzer, T. Schulthess, C. Fauser, B. Schumacher, R. A. Kammerer, J. Engel, M. A. Ruegg, EMBO J. 1998 17, 335-343. 112. A. J. Denzer, R. Brandenberger, M. Gesemann, M. Chiquet, M. A. Ruegg, J. Cell Biol. 1997 137,671-683. 113. J. J. O’Toole, K. A. Deyst, M. A. Bowe, M. A. Nastuk, B. A. McKechnie, J. R. Fallon, Proc. Nut1 Acad. Sci. U S A 1996 93, 7369-7374. 114. B. G. Wallace, J. Neurosci. 1990 10, 3576-3582. 115. M. J. Ferns, J. T. Campanelli, W. Hoch, R. H. Scheller, Z. Hall, Neuron 1993 11, 491-502. 116. H. Gordon, M. Lupa, D. Bowen, Z. Hall, J. Neurosci. 1993 13, 586-595. 117. W. Halfter, S. Dong, B. Schurer, G. J. Cole, J. Biol. Chem. 1998 273, 25404-25412. 118. M. S. O’Reilly, T. Boehm, Y. Shing, N. Fukai, G. Vasios, W. S. Lane, E. Flynn, J. R. Birkhead, B. R. Olsen, J. Folkman, Cell 1997 88, 277-285. 119. T. Sasaki, N. Fukai, K. Mann, W. Gohring, B. R. Olsen, R. Timpl, EMBO J. 1998 17,42494256. 120. J. R. Couchman, A. Woods, in D. D. Roberts, R. P. Mecham (Eds.): Cell surface extracellular glycoconjugutes: structure and function, Academic Press, Inc., San Diego 1993, p. 33-82.

Carbohydrates in Chemistry and Biology Beat Emst, Gerald W. Hart, Pierre Sinay copyright Q WILEY-VCH Verlag GmbH. 2000

43 Biological Roles of Keratan Sulfate Proteoglycans Gary W. Conrad

43.1 Introduction Historically, structural analysis of keratan sulfate proteoglycans began with pioneering studies of Meyer, Mathews, and Cifonelli in characterizing the structure of the sulfated polysaccharide chains in detail [l-91. Their decision, to study the polysaccharide component first, reflected the inadequacy of techniques at that time for purifying and sequencing core proteins, compared with techniques available for analyzing polysaccharides and sugars. Such structural studies of the keratan sulfate polysaccharide component have been continued most recently using NMR spectroscopy [ 10-201. Such detailed information on keratan sulfate polysaccharide chain structures is required for elucidating their biological roles, information as important as the sequences of the core proteins and their genes. Keratan sulfate (KS) exists in vivo as sulfated polysaccharide (glycosaminoglycan, GAG) chains covalently linked to various core proteins, mostly in the extracellular matrix: molecules known as keratan sulfate proteoglycans (KSPGs). These KS polysaccharide chains, composed of repeating disaccharides of galactose and Nacetylglucosamine (sulfated poly N-acetyllactosamine), often are sulfated in the C-6 positions of both the galactose and N-acetylglucosamine residues; chain lengths may be long (40 highly-sulfated N-acetyllactosamine units, cornea) or short (8-9 poorly-, or non-sulfated N-acetyllactosamine units, aorta) [211. Chains of KS are assembled on core proteins in the Golgi apparatus during biosynthesis (see accompanying chapter 24, Part 11, Volume 3: J. Funderburgh, Biosynthesis of Proteoglycans with Keratan Sulfates) and therefore represent tissue-specific posttranslational modifications of core proteins, modifications not encoded in DNA. Other such modifications also include addition of sulfate to tyrosines in the core proteins, specification of GAG chain length, and regulated addition of sulfate and sialic acid to those GAG chains. Genes for at least five KSPG core proteins have been identified recently: lumican, keratocan, mimecan (also known as osteoglycin), fibromodulin, and osteoadherin. Based on cDNA sequences, the fibromodulin gene

718

43 Biological Roles of Kevatan Sulfate Pvoteoglycuns

is most closely related to lumican, and the osteoadherin gene is most closely related to kerdtocan, whereas the mimecan/osteoglycin gene is very divergent in sequence from that of the other four KSPG family members. These five KSPG proteins are members of a larger family of leucine-rich repeat (LRR) proteins, other members of which carry GAG chains of chondroitin sulfate and/or dermatan sulfate. Patterns of tissue expression of KSPG mRNAs and core proteins suggest that keratocan is the most tissue-specific, with expression exclusively in the eye (corneal stroma or cornea and sclera, depending on the species), with mimecan expressed in a wider assortment of tissues, and lumican expressed in virtually all mesenchymal tissues [22-25] and some epithelial cells [26]. Multiple mRNA transcripts for mimecan are generated by alternative splicing and polyadenylation [27], processes that also occur in a tissue-specific manner and reflect biological control of KSPG biosynthesis. At least one KSPG, mimecan, is synthesized and secreted as a 37 kDa core protein that then undergoes posttranslational extracellular proteolytic processing, yielding in some tissues a 25 kDa core protein (corneal stroma), but in others generating an even smaller protein, osteoglycin (bone). The biological roles of the KSPGs therefore can be expected to derive not only from the properties of each specific core protein, predictable from their cDNA sequences, but also from the exact, tissuespecific, final post-translational modifications of each such core protein in vivo (e.g. full-length vs proteolytically processed; +/- tyrosine sulfation; +/- KS chains on -3-4 potential sites on core proteins; length, pattern of sulfation, and pattern of sialylation of KS chains; and +/- other glycosylation), structural fingerprints that are not encoded in DNA, but which are required for the normal: tissue-specific, biological roles of each type of KSPG.

43.2 Corneal Transparency Chakravarti et al. [28] prepared knock-out mice to inactivate the lumican gene and observed that collagen fibril assembly was disrupted in the corneal stroma and in the skin dermis. The biological consequences were that the cornea eventually (over the first 6 -7 months, post-birth) became cloudy and the dermis became fragile. This experiment in vivo suggests that this technique will be useful for collecting preliminary evidence on biological roles of lumican, and will be equally useful in elucidating roles of the other major KSPGs in the corneal stroma, keratocan and mimecan. However, these experiments do not reveal whether biologically critical features of the targeted KSPG occur on the core protein per se or on posttranslationally-added modifications that may not be added normally, or at all, on a transgenic core protein. Thus, for example, the knock-out mice of Chakravarti et al. [28] appear to be full null mutants, with neither mRNA nor core protein for lumican produced. The rationale for this experiment relied on results of cell-free experiments [29] suggesting that lumican core protein, with or without KS chains attached, was equally effective in regulating collagen fibril polymerization, producing uniformly small diameter fibrils characteristic of normal corneal stroma in vivo. However, even though that conclusion appears to be true, under physiological conditions in vivo, biological functions of other domains of the normal lumican core protein molecule bearing KS

43.3 Nerve Growth Cone Guidunce

I19

chains (i.e. the normal, complete KSPG molecule) would be compromised because of the absence of KS chains, for example, not because those functions derived directly from properties of the core protein. Duplication of collagen fibril polymerization patterns under cell-free conditions does not necessarily predict a duplication of all other biological roles of the normal molecule. Thus, as discussed by Chakravarti et al. [28] for example, the compromised transparency of the cornea in vivo, may have arisen from the absence of the KS chains normally carried by the lumican core protein (and thought to be necessary for normal, regular, fibril-to-fibril spacing), rather than from the absence of the core protein per se (the physical property of transparency has not been demonstrated directly to be duplicated under cell-free conditions with KS-free lumican core protein molecules). This reservation therefore leads directly to two questions, both of which are equally applicable to all other studies of KSPG biological roles discussed in this review: What chemical or physical properties of KSPG molecules are altered upon removal or alteration of the KS chains, i.e. what are the demonstrated biological roles of the KSPG core proteins per se? and: What are the demonstrated biological roles of the KS chains per se? Human corneas progressively become opaque in patients afflicted with macular corneal dystrophy (MCD), which is inherited as an autosomal recessive. The result is that normal KSPG molecules are not synthesized in the cornea, apparently because the sugar backbone of the KS chain does not become sulfated during biosynthesis, or is not sulfated to a normal degree or in a normal pattern of “hot-spots’’ and domains [30].In MCD type I, no KS is detected in the cornea (using an antibody that reacts specifically with normally highly-sulfated KS chains), whereas in MCD type 11, a low but detectable level of KS is present in the corneal stroma. The protein affected by MCD may be the enzyme that normally transfers sulfate from the biologically universal sulfate PAPS (3’-phosphoadenosine-5’-phosphosulfate, donor) to the growing KS chain (i.e. a KS sulfotransferase) [31]. The gene(s) for both MCD types I and I1 has been localized to chromosome 16q22 [24]. If, in fact, MCD does arise from an absence of sulfation of KS chains on the corneal KSPG core proteins during biosynthesis, then the clinical symptoms of progressive corneal opacification may represent a demonstration of the biological role of KS chains per se in the cornea. However, that conclusion would assume that the physical interactions of the KSPG core proteins, bearing only or predominately non-sulfated KS chains (keratan chains) would be normal. This would be a questionable assumption, because it is known that if all KS chains are removed from KSPG core proteins, the proteins become nearly insoluble in physiological solutions (the high leucine content of the core proteins determines that the proteins, by themselves, will be very hydrophobic); whether such proteins can be maintained in soluble form if synthesized with only (non-sulfated) keratan chains attached is not known.

43.3 Nerve Growth Cone Guidance Several studies have determined that KSPGs may act as barriers to migrating neuronal growth cones during nomal development of the central and peripheral ner-

720

43 Biological Roles of Kerutun Sulfute Proteoglycans

vous systems [32-361.The hypothesized inhibition of neurite outgrowth has been seen in vitro when neurites extend from rat cerebellar and dorsal root ganglia neurons onto laminin-coated substrates in the presence of free KS chains (although not These conditions when the substratum was coated with Ll, instead of laminin) [37]. do not duplicate those in vivo, for cells normally only encounter KS chains that are bound to the core proteins of generally immobilized KSPGs, whereas free KS chains are freely water-soluble. Nevertheless, these data emphasize that neuronal growth cones may perceive KS chains as inhibitory or non-inhibitory depending on the context of other extracellular matrix molecules present in the same environment (the extent to which laminin and Ll bind free KS chains differentially, e.g. electrostatically, was not determined but might be relevant to the interpretation of these experimental results). Inhibition of neurite outgrowth also has been seen in vitro as embryonic chick brain optic lobe cells extend neurites onto a substratum containing a normal brain KSPG (claustrin) [38].If the claustrin is first digested with keratanase, in the presence of protease inhibitors, to remove KS chains from the core protein, presumably without modifying the latter, then neurites extend on the substratum without inhibition, equivalently to controls; other controls demonstrated that it is the removal of the KS chains, rather than the presence of keratanase or the protease inhibitors, that cancels the inhibition of neurite extension by this KSPG. The microtubule-associated protein, MAP1 B, carries KS chains and appears to be related to, or identical with, claustrin [39]. In addition to serving as barriers and guidance cues for neuronal growth cone migration, it is likely that KSPGs play several other roles in the developing nervous system. For example, some KS expression is seen in discrete patches in areas not obviously serving as barriers in the developing rat striatum [40].Several studies have noted an association of KS with microglia [41-431,including an up-regulation of KSPG expression in wounds to the cerebral cortex [44].In addition, a synaptic vesicle-associated protein, SV2, is a keratan sulfate proteoglycan [45];the SV1 epitope is on the KS chains, whereas the SV2 epitope is on the KSPG core protein. Morover, neurites from retinal ganglion cells in the chick normally migrate along a pathway that contains both chondroitin sulfate proteoglycans (CSPGs) and KSPGs [46], suggesting that KSPGs may not be inhibitory to outgrowth of neurites from all types of nerves. The relationship between KSPGs and neuronal growth cone movement seems most important to elucidate in the developing cornea, a tissue that eventually contains both the highest concentration of KSPGs of any body tissue [47],and also the highest density of peripheral nerves [48].Because KSPGs are made by stromal keratocytes and deposited in the corneal stroma of developing chicks [49]and humans [ 501 during the period in which primary stromal innervation occurs [ 5 11, the hypothesis of KSPG as a molecule generally inhibitory to growth cone movement would appear to be contradicted dramatically. However, the explanation may be in the detailed biochemistry of the post-translational modifications of the KSPG core proteins. For example, neurite outgrowth from hippocampal neurons is promoted by a CSPG synthesized by glial cells and localized in the central nervous system; the neurite-promoting activity requires a specific pattern of sulfation on the CS chains [52].Thus, corneal KS chains are very long and highly sulfated, compared to those

43.4 Cell Adhesion

721

synthesized on KSPG core proteins in other tissues, and they carry multiple residues of sialic acid (“polysialic acid”), a highly negatively charged sugar at the terminii of most KS chains [ 19, 201. It therefore is of special interest to note that polysialic acid has been demonstrated to regulate the movement of motor nerve growth cones, at least [53, 541, and thus may play a corresponding role for the sensory nerves that innervate the cornea, principally from the trigeminal ganglia. Finally, it should be noted that not only are KSPGs present in the normal brain, associated with most neurons of the cerebral cortex, for example, but also that the epitope for highly sulfated KS chains disappears from these neurons in Alzheimer’s disease [ 551. However, the periphery of Alzheimer’s disease neuritic plaques is positive for KS [56].

43.4 Cell Adhesion Cell adhesion to extracellular matrices of hyaluronate is mediated by a cell surface receptor, CD44, that can undergo a variety of types of alternative splicing. The most commonly expressed form in many normal tissues is CD44H, which lacks several exons from a central domain. However, variance in CD44 arises not only from such alternative splicing, but also from significant post-translational modifications, such as 0-,and N-glycosylation and substitution with GAG chains, including those of KS. The degree to which CD44H is substituted with KS chains determines the biological role of this receptor: CD44H, when expressed in a colon carcinoma cell line that adds KS chains to CD44H, causes that population of cells to be highly metastatic and not to bind significantly to hyaluronate substrata. Conversely, if CD44H is expressed in a clonal variant of the cell line above that does not add KS chains to CD44H, such cells are poorly metastatic and bind significantly to hyaluronate substrata [57]. Removal of the KS chains from CD44H, with KSase or via site-directed mutagenesis, greatly enhanced the adhesion of the formerly metastatic cells to hyaluronate substrata. These data suggest that the biological role of CD44H, a cell-surface receptor important in cell adhesion, can be regulated by the presence or absence of KS chains on the CD44 core protein. Whether the functions of the receptor can be regulated by the chain length or degree or pattern of sulfation or degree of polysialyation of the KS chains has not been studied. Cell adhesion also can be affected significantly by the presence of KSPGs if cells have cell surface receptors for KSPGs and if those KSPGs occur in a variety of forms, e.g., from alternative splicing, and/or from post-translational modifications of the KSPG core proteins. For example, the first and only receptor yet characterized that recognizes a KSPG has been described in mouse peritoneal macrophages [%]. Macrophages do not adhere to the type of lumican KSPG synthesized in the corneal stroma (core protein carrying long, highly-sulfated KS chains), but do adhere to corneal lumican after removal of the KS chains. In addition, macrophages adhere to the type of lumican found in the walls of major blood vessels (core

722

43 Biological Roles of Keratan Suljute Pvoteoglycans

proteins carrying short, poorly-sulfated KS chains). Thus, in this case, as in the experimental system above [57], the biological role of the KSPG derives from the effect of the KS chains on modulating the properties of the core protein, and the effect, in both cases, is to inhibit cell adhesion.

43.5 Other Possible Roles of KSPGs Although first described as a sulfated polysaccharide component of connective tissue [7], and subsequently found as a covalently bound prosethetic group of not only the KSPGs of cornea (lumican, mimecan, and keratocan), but also of aggrecan, the fundamental extracellular building block molecule of hyaline cartilage [ 591, KS chains covalently-bound to core proteins subsequently have been detected in several non-connective tissues. These KSPGs undoubtedly serve many functions distinct from those described above, so their descriptions below serve only as introductions to the functions that they may serve. They have been detected because of the availability of monoclonal antibodies to the highly sulfated epitope that characterizes KS polysaccharide chains (5D4 and others [60]; and I22 [61]). KS chains of lower degrees of sulfation, or entirely non-sulfated (poly-N-acetyllactosaminoglycans), together with their core proteins, may serve equally important roles; antibodies specifically recognizing such epitopes have been reported [ 621. Meyer et al. 18, 63, 641 detected a KSPG in the allantoic fluid of chick embryos, using chemical methods. Its biological role remains unelucidated. Epithelial cells covering the surfaces of animals, keratinocytes, express abundant amounts of one of the classes of intermediate filaments, i.e., keratins. Four molecular weight classes of keratin filaments react with antibody 5D4 and therefore evidently bear KS chains [65]. Of seven monoclonal antibodies to KS epitopes, five displayed positive reactions in keratinocytes, depending on the state of differentiation of these cells. All, or a portion, of this reactivity was sensitive to digestion with the degradative enzymes, keratanase and endo-P-galactosidase, indicating structurally-specificKS-like carbohydrate expression during epithelial cell differentiation 1661. In addition to keratinocytes, several other epithelial cell-types express KS epitopes on their cell surfaces or intracellularly, in a broad perinuclear zone [66, 671, or transiently express lumican mRNA [26]. Cervix of non-pregnant and pregnant human uterus contains a very large molecular weight KSPG, with a core protein estimated to be -220 kDa; such tissue also contains lumican core protein not bearing KS chains 1681. In addition, the luminal surface of the epithelial cells that line the uterus (endometrium) bear a KSPG on their microvilli. Reactivity for KS (with antibody 5D4) is virtually negative in the uterus of control and progesterone-injected rats, but strong staining for KS appears on the luminal epithelium following injection of 17-P-estradiol[69]; that KS expression almost disappears if progesterone is injected simultaneously with the 17P-estradiol, suggesting that, in this tissue, KSylation is cell-type specific and can be up-regulated and down-regulated hormonally. Studies of the human endometrium

References

723

have confirmed these observations, and demonstrated the presence of not only a KSPG associated with the luminal surface of endometrial epithelial cells, but also of a KSPG secreted from the glandular epithelium of the endometrium [70]. Secretion of the glandular KSPG occurs in synchrony with the menstrual cycle, with significantly increased expression seen during the secretory phase of the cycle; intracellular levels of KS in glandular epithelial cells reach maximum levels 3 days after the peak in luteinizing hormone (LH) in the normal menstrual cycle, with an increase in secreted KS seen 1 day later [71]. In addition to this secreted form, it has been demonstrated that this molecule, termed MUCl, also is expressed in a different form, complete with KS chains terminated with sialic acid residues, localized in association with the apical surface of the glandular endometrial epithelium [72]. The exact molecular features of the KS chains on this KSPG therefore may determine the success with which an embryonic blastocyst, after hatching from the zona pellucida, can attach to the lumenal surface of the maternal endometrial epithelium as its first step toward invading and implanting in the endometrium. Such a biological role for a mammalian KSPG will be important to study at the level of the molecular features of the KS chains, as well as at the level of core protein expression. Preliminary data also suggest biological roles for KS chains or core proteins of KSPGs in the inner ear [73], as potential oncofetal antigens, as in papillary carcinomas [74], and as inhibitors of their own secretion [75], yet as strong stimulators of proteoglycan biosynthesis [ 761. The apparently contrasting effects seen in the latter two studies may reflect differences in animal species used as sources for chondrocytes (adult rabbits vs chick embryos) and for KS (shark and whale vs mammalian), or mode of cell culture (monolayer vs suspension). In addition, the major CSPG of hyaline cartilage, aggrecan, normally bears several 0-linked chains of KS near the hyaluronic acid-binding region which are thought to add to the charge density in that domain; thus, as their biological role, these KS chains may add to the inherent resistance of aggrecan to mechanical deformation [77]. However, in hyaline cartilage of rat Swarm chondrosarcoma and of normal mice and rats, aggrecan molecules lack KS chains [78, 791, therefore raising uncertainty about the biological role of KS chains in this tissue.

Acknowledgment The author gratefully acknowledges the support of NIH grant EY00952.

References 1. Bhavanandan, V.P., and K. Meyer. 1966. Mucopolysaccharides: N-Acetylglucosamine- and galactose-6-sulfates from keratosulfate. Science 151: 1404-1405. 2. Bhavanandan, V.P., and K. Meyer. 1967. Studies on keratosulfates. J. Biol. Chem. 242: 43524359. 3. Bhavanandan, V.P., and K. Meyer. 1968. Studies on keratosulfates. J. Biol. Cliem. 243: 10521059.

124

43 Biological Roles of Keratan Sulfate Proteoglycans

4. Cifonelli, J.A., A. Saundcrs, and J.I. Gross. 1967. Keratan sulfate fractions from bovine and human tissues. Carbohyd. Rex 3: 478-485. 5. Mathews, M.B. 1975. Connective Tissue. Marromoleculur Structure and Evolution. SpringerVerlag, New York.. 6. Mathews, M.B., and J.A. Cifonelli. 1965. Comparative biochemistry of the keratosulfates. J. Biol. Chem. 240: 4140-4145. 7. Meyer, K., A. Linker, E.A. Davidson, and B. Weissmann. 1953. The mucopolysaccharides of bovine cornea. J. Biol. Chem. 205: 611-616. 8. Meyer, K., V.P. Bhavanandan, D. Yung, L.T. Lee, and C. Howe. 1967. The keratosulfate-like mucopolysaccharide of chick allantoic fluid. Proc. Nut1 Acad. Sci. USA 58: 1655-1659. 9. Seno, N., K. Meyer, B. Anderson, and P. Hoffman. 1965. Variations in keratosulfates. J. Biol. Chem. 240: 1005-1010. 10. Hounsell, E.F., J. Feeney, P. Scudder, P.W. Tang, and T. Feizi. 1986. ‘H-NMR studies at 500 MHz of a neutral disaccharide and sulphated di-, tetra-, hexa-, and larger oligosaccharides obtained by endo-P-galactosidase treatment of keratan sulphate. Eur. J. Biochem. 157: 375384. 11. Stuhlsatz, H.W., R. Keller, G. Becker, M. Oeben, L. Lennartz, D.C. Fischer, and H. Greiling. 1989. Structure of keratan sulphate proteoglycans: core proteins, linkage regions, carbohydrate chains. In: “Keratan sulphate. Chemistry, Biology, Chemical Puthology,” H. Greiling and J.E. Scott, eds., The Biochemical Society, London, pp.: 1-1 1. 12. Brown, G.M., T.N. Huckerby, H.G. Morris, B.L. Abram, and I.A. Nieduszynski. 1994. Oligosaccharides derived from bovine articular cartilage keratan sulfates after keratanase I1 digestion: Implications for keratan sulfate structural fingerprinting. Biochemistry 33: 48364846. 13. Brown, G.M., T.N. Huckerby, M.T. Bayliss, and I.M. Nieduszynski. 1998. Human aggrecan keratan sulfate undergoes structural changes during adolescent development. J. Biol. Chem. 273: 26408-26414. 14. Huckerby, T.N., G.M. Brown, J.M. Dickenson, and I.A. Nieduszynski. 1995. Spectroscopic characterisation of disaccharides derived from keratan sulfates. Eur. J. Biochem. 229: 119-131. 15. Huckerby, T.N., G.M. Brown, and I.A. Nieduszynski. 1998. ”C-NMR spectroscopy of keratan sulphates. Eur. J. Biochem. 251: 991-997. 16. Lauder, R.M.: T.N. Huckerby, and LA. Niedusznyski. 1994. Structure of the keratan sulphate chains attached to fibromodulin isolated from bovine tracheal cartilage. Biochem. J. 302: 417423. 17. Lauder, R.M., T.N. Huckerby, I.A. Nieduszynski, and A.H.K. Plaas. 1998. Age-related changes in the structure of the keratan sulphate chains attached to fibromodulin isolated from articular cartilage. Biochem. J. 330: 753-757. 18. Tai, G.-H., G.M. Brown, H.G. Morris, T.N. Huckerby, and I.A. Nieduszynski. 1991. Fucose content of keratan sulphates from bovine articular cartilage. Biochem. J. 273: 307-310. 19. Tai, G.-H., T.N. Huckerby, and I.A. Nieduszynski. 1996. Multiple non-reducing chain termini isolated from bovine corneal keratan sulfates. J. Biol. Chem. 271: 23535-23546. 20. Tai, G.-H., I.A. Nieduszynski, N.J. Fullwood, and T.N. Huckerby. 1997. Human corneal keratan sulfates. J. Biol. Chem. 272: 28227-2833 I . 21. Funderburgh, J.L., M.L. Funderburgh, M.M. Mann, and G.W. Conrad. 1991. Arterial lumican. Properties of a corneal-type keratan sulfate proteoglycan from bovine aorta. J. Biol. Chem. 266: 24773-24771. 22. Corpuz, L.M., J.L. Fundcrburgh, M.L. Funderburgh, G.S. Bottomley, S. Prakash, and G.W. Conrad. 1996. Molecular cloning and tissue distribution of keratocdn. Bovine corneal keratan sulfate proteoglycan 37A. J. Biol. Chem. 271: 9759-9763. 23. Funderburgh, J.L., L.M. Corpuz, M.R. Roth, M.L. Funderburgh, E.S. Tasheva, and G.W. Conrad. 1997. Mimecan, the 25-kDa corneal keratan sulfate proteoglycan, is a product of the gene producing osteoglycin. J. Biol. Chem. 272: 28089-28095. 24. Liu, C.-Y., A. Shiraishi, C.W.-C. Kao, R.L. Converse, J.L. Funderburgh, L.M. Corpuz, G.W. Conrad, and W.W.-Y. Kao. 1998. The cloning of mouse keratocan cDNA and genomic DNA and the characterization of its expression during eye development. J. Biol. Chem. 273: 2258422588.

References

725

25. Tasheva, E.S., J.L. Funderburgh, L.M. Corpuz, and G.W. Conrad. 1998. Cloning, characterization and tissue-specific expression of the gene encoding bovine keratocan, a corneal keratan sulfate proteoglycan. Gene 218: 63-68. 26. Ying, S., A. Shiraishi, C.W.-C. Kao, R.L. Converse, J.L. Funderburgh, J. Swiergiel, M.R. Roth, G.W. Conrad, and W.W.-Y. Kao. 1997. Characterization and expression of the mouse lumican gene. J. Biol. Chem. 272: 30306-30313. 27. Tasheva, E.S., L.M. Corpuz, J.L, Funderburgh, and G.W. Conrad. 1997. Differential splicing and alternative polyadenylation generate multiple mimecan mRNA transcripts. J. Biol. Chem. 272: 32551-32556. 28. Chakravarti, S., T. Magnuson, J.H. Lass, K.J. Jepsen, C. LaMantia, and H. Carroll. 1998. Lumican regulates collagen fibril assembly: Skin fragility and corneal opacity in the absence of lumican. J. Cell Biol. 141: 1277-1286. 29. Rada, J.A., P.K. Cornuet, and J.R. Hassell. 1993. Regulation of corneal collagen fibrillogenesis in vitro by corneal proteoglycan (lumican and decorin) core proteins. Exp. Eye Rex 56: 635-648. 30. Hassell, J.R., D.A. Newsome, J.H. Krachmer, and M. Rodriques. 1980. Macular corneal dystrophy. Failure to synthesize a mature keratan sulfate proteoglycan. Proc. Nut1 Acad. Sci USA 77: 3705-3709. 31. Hassell, J.R., and G.K. Klintworth. 1997. Serum sulfotransferase levels in patients with macular corneal dystrophy type I. Arch. Ophthulmol. 115: 1419-1421. 32. Geisert, E.E., Jr., and D.J. Bidanset. 1993. A central nervous system keratan sulfate proteoglycan: localization to boundaries in the neonatal rat brain. Deu. Brain Rex 75: 163-173. 33. Hamanaka, H.. N. Maeda, and M. Noda. 1997. Spatially and temporally regulated modification of the receptor-like protein tyrosine phosphatase {/p isoforms with keratan sulphate in the developing chick brain. Eur. J. Neurosci. 9: 2297-2308. 34. Hemming, F.J., and R. Saxod. 1997. Keratan sulphate is present in developing chick skin in vivo where it could constitute a barrier to advancing neurites as observed in vitro. J. Neurosci. Res. 48: 133-145. 35. Robson, J.A., and E.E. Geisert, Jr. 1994. Expression of a keratan sulfate proteoglycan during development of the dorsal lateral geniculate nucleus in the ferret. J. Comp. Neurol. 340: 349-360. 36. Seo, H., and E.E. Geisert, Jr. 1995. A keratan sulfate proteoglycan marks the boundaries in the cortical barrel fields of the adult rat. Neurosci. Lett. 197: 13-16. 37. Dou, C.L., and J.M. Levine. 1995. Differential effects of glycosaminoglycans on neurite growth on laminin and L1 substrates. J. Neurosci. 15: 8053-8066, 38. Cole, G.J., and C.F. McCabe. 1991. Identification of a developmentally regulated keratan sulfate proteoglycan that inhibits cell adhesion and neurite outgrowth. Neuron 7: 1007-1018. 39. Burg, M.A., and G.J. Cole. 1994. Claustrin, an antiadhesive neural keratan sulfate proteoglycan, is structurally related to MAPIB. J. Neurobiol. 25: 1-22. 40. Charvet, I., F.J. Hemming, C. Feuerstein, and R. Saxod. 1998. Mosaic distribution of chondroitin and keratan sulphate in the developing rat striatum: possible involvement of proteoglycans in the organization of the nigrostriatal system. Deu. Bruin Rex 109: 229-244. 41. Bertolotto, A,, B. Caterson, G. Canavese, A. Migheli, and D. Schiffer. 1993. Monoclonal antibodies to keratan sulfate immunolocalize ramified microglia in paraffin and cryostat sections of rat brain. J. Histochem. Cytochem. 41: 481-487. 42. Bertolotto, A,, E. Manzardo, M. Iudicello, R. Guglielmone, and A. Riccio. 1995. Keratan sulphate is a marker of differentiation of ramified microglia. Deu. Bruin Res. 86: 233-241, 43. Jander, S., and G. Stoll. 1996. Strain-specific expression of microglial keratan sulfate proteoglycans in the normal rat central nervous system: Inverse correlation with constitutive expression of major histocompatibility complex class I1 antigens. Gliu 18: 255-260. 44. Geisert, E.E., Jr., D.J. Bidanset, N. Del Mar, and J.A. Robson. 1996. Up-regulation of a keratan sulfate proteoglycan following cortical injury in neonatal rats. Zntl J. Deu. Neuro.sci. 14: 257-267. 45. Scranton, T.W., M. Iwata, and S.S. Carlson. 1993. The SV2 protein of synaptic vesicles is a keratan sulfate proteoglycan. J. Neurochem. 61: 29-44. 46. McAdams, B.D., and S.C. McLoon. 1995. Expression of chondroitin sulfate and keratan sulfate proteoglycans in the path of growing retinal axons in the developing chick. J. Comp. Neurol. 352: 594-606.

726

43 Biological Roles of Keratan Suljute Proteoglycans

47. Funderburgh, J.L., B. Caterson, and G.W. Conrad. 1987. Distribution of proteoglycans antigenically related to corneal keratan sulfate proteoglycan. J. Biol. Chem. 262: 11634-1 1640. 48. Rozsa, A.J., and R.W. Beuerman. 1982. Density and organization of free nerve endings in the corneal epithelium of the rabbit. Pain 14: 105-120. 49. Funderburgh, J.L., B. Caterson, and G.W. Conrad. 1986. Keratan sulfate proteoglycan during embryonic development of the chicken cornea. Dev. Biol. 116: 267-277. 50. Azuma, N., A. Hirakata, T. Hida, and S. Kohsaka. 1994. Histochemical and immunohistochemical studies on keratan sulfate in the anterior segment of the developing human eye. Exp. Eye Rex 58: 277-286. 51. Bee, J.A. 1982. The development and pattern of innervation of the avian cornea. Deu. Biol. 92: 5-15. 52. Clement, A.M., S. Nadanaka, K. Masayama, C. Mandl, K. Sugahara, and A. Faissner. 1998. Thc DSD-I carbohydrate epitope depends on sulfation, correlates with chondroitin sulfate D motifs, and is sufficient to promote neurite outgrowth. 1998. J. Biol. Chem. 273: 2844428453. 53. Tang, J., L. Landmesser, and U. Rutishauser. 1992. Polysialic acid influences specific pathfinding by avian motoneurons. Neuron 8: 1031-1044. 54. Tang, J., U. Rutishauser, and L. Landmesser. 1994. Polysialic acid regulates growth cone behavior during sorting of motor axons in the plexus region. Neuron 13: 405-414. 55. Lindahl, B., L. Eriksson, D. Spillmann, B. Caterson, and U. Lindahl. 1996. Selective loss of cerebral keratan sulfate in Alzheimer’s Disease. J. Biol. Chem. 271: 16991-16994. 56. Snow, A D . , D. Nochlin, R. Sekiguchi, and S.S. Carlson. 1996. Identification and immunolocalization of a new class of proteoglycan (keratan sulfate) to the neuritic plaques of Alzheimer’s disease. Exp. Neurol. 138: 305 -317. 57. Takahashi, K., I. Stamenkovic, M. Cutler, A. Dasgupta, and K.K. Tanabe. 1996. Keratan sulfate modification of CD44 modulates adhesion to hyaluronate. J. Biol. Chem. 271: 94909496. 58. Funderburgh, J.L., R.R. Mitschler, M.L. Funderburgh, M.R. Roth, S.K. Chapes, and G.W. Conrad. 1997. Macrophage receptors for lumican. A corneal keratan sulfate proteoglycan. Invest. Ophthalmol. Vis. Sci. 38: 1159-1 167. 59. Hascall, V.C.: and S.W. Sajdera. 1970. Physical properties and polydispersity of proteoglycan from bovine nasal cartilage. J. Biol. Chen?. 245: 4920-4930. 60. Caterson, B.. J.E. Christner, and J.R. Baker. 1983. Identification of a monoclonal antibody that specifically recognizes corneal and skeletal kerdtan sulfate. J. Biol. Chem. 258: 8848-8854. 61. Funderburgh, J.L., P.R. Stenzel-Johnson, and J.W. Chandler. 1983. Monoclonal antibodies to rabbit corneal keratan sulfate proteoglycan. Curr. Eye Res. 2: 769-775. 62. Feizi, T., E.F. Hounsell, J. Alais, A. Veyrieres, and S. David. 1992. Further definition of the size of the blood group-I antigenic determinant using a chemically synthesised octasaccharide of poly-N-acetyllactosamine type. Carbohyd. Res. 10: 289-297. 63. Choi, H.U., and K. Meyer. 1974. The structure of a sulfated glycoprotein of chick allantoic fluid. J. Biol. Chenz. 249: 932-939. 64. Choi, H.U., and K. Meyer. 1975. The structure of a sulfated glycoprotein of chick allantoic fluid: Methylation and periodate oxidation. Carbohyd. Res. 40: 77-88. 65. Schafer, LA., and J.M. Sorrell. 1993. Human keratinocytes contain keratin filaments that are glycosylated with keratan sulfate. Exp. Cell Res. 207: 213-219. 66. Sorrell, J.M., B. Caterson, A.1. Caplan, B. Davis, and LA. Schafer. 1990. Human keratinocytes contain carbohydrates that are recognized by keratan sulfate-specific monoclonal antibodies. J. Invest. Dermatol. 95: 347-352. 67. Sorrell, J.M., and B. Caterson. 1990. Monoclonal antibodies specific for keratan sulfate detect epithelial-associated carbohydrates. Histochemistry 94: 269-275. 68. Fisher, D.-C., A. Henning, M. Winkler, W. Rath, H.-D. Haubeck, and H. Greiling. 1996. Evidence for the presence of a large keratan sulphate proteoglycan in the human uterine cervix. Biochem. J. 320: 393-399. 69. CidadBo, A.J., S. Thorsteinsdottir, and J.F. David-Ferreira. 1990. Immunocytochemical study of tissue distribution and hormonal control of chondroitin-, dermatan-, and keratan sulfates from rodent uterus. Eur. J. Cell Bid. 52: 105-1 T6.

References

121

70. Hoadley, M.E., M.W. Seif, and J.D. Aplin. 1990. Menstrual-cycle-dependent expression of keratan sulphate in human endometrium. Biocl?enz. J. 266: 757-763. 71. Graham, R.A., T.C. Li, I.D. Cooke. and J.D. Aplin. 1994. Keratan sulphate as a secretory product of human endometrium: cyclic expression in normal women. Humun Reprod. 9: 926930. 72. Aplin, J.D., N.A. Hey, and R.A. Graham. 1998. Human endometrial MUCl carries keratan sulfate: characteristic glycoforms in the lumenal epithelium at receptivity. Glycobiohgy 8: 269276. 73. Thalmann, I., K. Machiki, A. Calabro, V.C. Hascall, and R. Thalmann. 1993. Uronic acidcontaining glycosaminoglycans and keratan sulfate are present in the tectorial membrane of the inner ear: functional implications. Arch. Biochenz. Biophys. 307 391-396. 74. Ito, N., M. Yokota, C. Nagaike, Y. Morimura, K. Hatake, 0 . Tanaka, and T. Matsunaga. 1996. Simultaneous expression of keratan sulphate epitope (a sulphated poly-N-acetyllactosamine) and blood group ABH antigens in papillary carcinomas of the human thyroid gland. Histoclzem. J. 28: 613-623. 75. Fukuda, K., K. Ohtani, F. Matsumurd, and S. Tanaka. 1993. Keratan sulfate inhibits its release in rabbit chondrocyte. Conn. Tim Rex SO: 75-83. 76. Nevo, Z., and A. Dorfman. 1972. Stimulation of chondromucoprotein synthesis in chondrocytes by extracellular chondromucoprotein. Proc. Nut1 Acad. Sci. USA 69: 2069-2072. 77. Hascall, V.C., D.K. Heinegird, and T.N. Wight. 1991. Proteoglycans. Metabolism and pathology. In: “Cell Biologv of Extracellulur Matrix,” E.D. Hay, ed., 2”* ed., Plenum Press, New York, pp. 149-175. 78. Oegema, T.R., Jr., V.C. Hascall, and D.D. Dziewiatkowski. 1975. Isolation and characterization of proteoglycans from the Swarm rat chondrosarcoma. J. Biol. Chem. 250: 6151-6159. 79. Watanabe, H., Y. Yamada, and K . Kimata. 1998. Roles of aggrecan, a large chondroitin sulfate proteoglycan, in cartilage structure and function. J. Biochenz. 124: 687-693.

Carbohydrates in Chemistry and Biology Beat Emst, Gerald W. Hart, Pierre Sinay copyright Q WILEY-VCH Verlag GmbH. 2000

44 Developmental and Aging Changes of Chondroitin/ Dermatan Sulfate Proteoglycans J. Michael Sorrell, David A . Carrino, and Arnold I. Caplan

Development and aging are complex, multidimensional processes in which the structure and function of tissues alter in response to encoded cellular and genomic transformations. Extracellular matrices play critical structural and functional roles in tissues, and, thus, are integral to these processes. Chondroitin/dermatan sulfate proteoglycans, which are an important component of these extracellular matrices, display developmental and age-related changes in a wide variety of tissues and organs.

44.1 Proteoglycans A proteoglycan is a protein which is substituted with one or more glycosaminoglycan chains [26]. These linear carbohydrate chains attach covalently to their respective core protein, mostly through 0-glycosidic linkages, although in some noncartilaginous tissues, the keratan sulfate chains are attached through N-glycosidic linkages [8, 24, 281. One type of glycosaminoglycan, hyaluronan, is not covalently bound to protein; however, because of its important interactions with proteoglycans, it is commonly included in the discussions of proteoglycans [8, 241.

44.2 Glycosaminoglycans A glycosaminoglycan is a heteropolysaccharide consisting of repeating disaccharide sequences, where each disaccharide contains a hexosamine and a hexuronic acid (chondroitin sulfate, derrnatan suljbte, heparin, and heparan sulfate). However, in keratan sulfate, the hexuronic acid moiety is replaced by a hexose [8, 281. These

730

44 Developmental and Aging Changes Chondroitin Sulfate

c-4-s

C-6-S

cs D

cs

c-0-s

E

Dermatan Sulfate

6SOa ~~~

CS B

e

Di-di B

N-acetylgalactosarnine

~

Chondroitin Sulfate

B

Q

0 glucuronic

cs

acid

iduronic

acid

Figure 1. Chondroitin sulfate and dermatan sulfate are diagrammatically compared with respect to their principal disaccharide compositions. Chondroitin sulfate contains disaccharides which consist of glucuronic acid and N-acetylgalatactosamine units, which can be variously sulfated or remain unsulfated. Most chondroitin sulfate disaccharides exist in the form of chondroitin 4-sulfate (C-4-S) or chondroitin 6-sulfate ( C - 6 3 ) .Unsulfated chondroitin (C-0-S)occurs as a minor constituent of most chondroitin sulfate chains. The oversulfated disaccharides, chondroitin sulfate D (CS D)and chondroitin sulfate E (CS E ) , are rare entities for most mammalian chondroitin sulfate chains, but appear more commonly in chondroitin sulfate chains from lower vertebrates or invertebrates. Dermatan sulfate chains are heteropolysaccharides in which the principal disaccharides are chondroitin sulfate B (CS B) units; they consist of an iduronic acid and a 4-sulfated N-acetylgalactosamine. An oversulfated version of this disaccharide ( D i d B) appears as a minor constituent in most dermatan sulfate chains. All dermatan sulfate chains contain chondroitin sulfate domains; however, the size and frequency of these domains are highly regulated by cells.

chains are characterized by their high negative charge which results from the presence of carboxyl groups and/or from sulfate ester groups attached to various locations within the disaccharide unit. Figure 1 compares chondroitin sulfate and dermatan sulfate and illustrates the major disaccharides found in each of these glycosaminoglycan chains, which typically consist of 50- 100 disaccharide units. A given chondroitin sulfate chain contains 4-, and 6-sulfated7 and to a lessor extent 0-sulfated, disaccharide units in varying proportions; the oversulfated chondroitin sulfate D and E units are less common and do not appear within all chondroitin sulfate chains [ 13, 14, 4.51. Dermatan sulfate is synthesized from chondroitin sulfate by epimerization of some proportion of the uronic acid residues to form iduronic acid residues. Consequently, varying proportions of these chains contain chondroitin sulfate domains in which epimerization did not occur. The iduronate-rich domains tend to be 4-sulfated, although oversulfated disaccharides in which there is a 2-sulfated iduronic acid-4-sulfated N-acetylgalactosamine also occur [ 13, 14, 451.

44.3 Cow Proteins

731

Thus, both chondroitin sulfate and dermatan sulfate are heterogeneous structures based upon their varied disaccharide compositions. Interestingly, this heterogeneity is regulated by cellular activity [13]. As discussed below, the distribution of the different disaccharides along the length of the chain may also be characteristic of its cellular source. The degree of sulfation and the distribution of variously sulfated disaccharides within the linear structure of glycosaminoglycan chains confers important properties upon the proteoglycan.

44.3 Core Proteins The protein portion of a proteoglycan is termed the core protein [28]. There is no overall unifying feature which defines a core protein; however, it is now known that many of the major proteoglycan species belong to discrete families based upon their genomic structures and amino acid sequences [28, 29, 331. For the purposes of this discussion only two major proteoglycan families will be discussed in detail. They are the hyalectans, interstitial chondroitin sulfate proteoglycans which bind hyaluronan, and the small, leucine-rich proteoglycans, or SLRPs. Members of both of these two families appear in interstitial spaces of tissues and organs, and, together, comprise the majority of interstitial proteoglycans [29, 331. Consequently, they participate in tissue modifications related to development and aging. Table 1 provides a list of chondroitin/dermatan sulfate matrix and cell surface proteoglycans that includes molecules in addition to those in these two families. Although this list is not comprehensive, it demonstrates the large variety of chondroitinldermatan sulfate proteoglycans which have been identified. 44.3.1 Hyalectans The principal members of this family are agyrecan, versican, neurocan and brevican [33]. However, only aggrecan and versican will be discussed in detail here. The reader is directed to other reviews for a discussion of neurocan and brevican [22, 421. The matrix proteoglycans of this family bind hyaluronan and thereby form large multimolecular complexes in interstitial spaces [8, 24, 281, All members of this family contain three basic core protein domains: an hyaluronan-binding domain, a glycosaminoglycan-attachment domain, and a lectin-binding domain [24, 28, 331. Agyrecan, so termed because of its ability to form supermolecular aggregates in the presence of hyaluronan, is the principal proteoglycan species in hyaline cartilage, and it has also been identified in neural tissues [55]. Domain I of the core protein contains an immunoglobulin-like repeat and four link protein-like modules which form two tandem globular domains, called G1 and G2 [24, 28, 331. An interglobular region has a rod-like structure and contains cleavage sites for proteases involved in the degradation of this proteoglycan. The G1 domain serves as the functional hyaluronan-binding region. In con-

732

44 Developmental and Aging Clzanges

Table 1. Chondroitin/Dermatan Sulfate Proteoglycans.

I. Matrix Proteoglycans Basement membrane proteoglycans Proteoglycan Bamacan

Glycosaminoglycan chains Chondroitin sulfate

Anatomic distribution Skin

Chondroitin/keratan sulfate Chondroitin sulfate Chondroitin sulfate Chondroitin sulfate

Cartilage, Neural Skin, Neural Neural Neural

H yalectans

A ggrecan Versican Neurocan Brevican

Small, leucine-rich proteoglycans (SLRPs) Decorin Biglycan Epiphycan

Dermatan/chondroitin sulfate Dermatan/chondroitin sulfate Dermatan/chondroitin sulfate

Cartilage, Skin, Neural Cartilage, Skin, Neural Cartilage

Fibril-associated collagens with interrupted triple-helices Type IX collagen Type XI1 collagen Type XIV collagen

Chondroitin sulfate Chondroitin sulfate Chondroitin sulfate

Cartilage Cartilage, Skin Cartilage, Skin

Heparan/chondroitin sulfate Heparan/chondroitin sulfate

Cartilage, Skin, Neural Cartilage, Skin, Neural

11. Cell surface proteoglycans Syndecan Family CD44

GIycosylphosphatidylinositol-linked Phosphacan

Chondroitin sulfate

Neural

trast, the G2 domain, which shares structural homology to the G1 domain, appears not to bind hyaluronan, and its function is not yet clearly understood [33]. Immediately following the G2 domain appears a small region where keratan sulfate glycosaminoglycan chains are attached to the core protein via O-glycosidic linkages. The presence and length of this region is species-dependent 141. In chick and human, it is present, but it is absent in mouse. Domain I1 is the largest structural unit of aggrecan and it contains the bulk of the glycosaminoglycan chains, primarily chondroitin sulfate, but also some keratan sulfate [8, 24, 281. Up to 100 chondroitin sulfate chains may be covalently bound to serine-glycine repeats that characterize this domain and which serve as attachment sites for these chains 18, 24, 26, 281. This domain is divided into two sub-regions. The CS1 sub-region, adjacent to the keratan sulfate-attachment region, contains a zone of densely packed chondroitin sulfate chains, and distal to this is another sub-region, CS2, which contains tufts of more widely spaced chondroitin sulfate chains. The presence of these large numbers of closely apposed chondroitin sulfate chains produces the so-called bottle brush appearance of aggrecan that is seen in rotary shadowed images viewed in the electron microscope 1281. Domain I11 contains structural motifs which are similar to those found in versican (see below). These include epidermal growth factor (EGF)-repeats

44.3 Core Proteins

733

which can be alternatively spliced and a lectin-like region which forms a third globular domain, G3. The lectin-like domain binds simple sugars such as fucose and galactose in a Ca++-dependent manner (331. Aggrecan proteoglycans, or more specifically multimolecular aggregates of aggrecan and hyaluronan, in combination with type I1 collagen, are primarily responsible for the visco-elastic properties of hyaline cartilage [ 81. Proteoglycan aggregates, which are held in place by their interactions with collagenous fiber meshworks, bind and structure water molecules by virtue of their high negative charge. Compressive forces result in the redistribution of these water molecules. Upon release of the compressive force, water molecules reassemble around proteoglycan aggregates. Aggrecan, itself, may act as a linking molecule. The hyaluronanbinding region near the N-terminus may bind to cell surface-associated hyaluronan, while the lectin-like domain near the C-terminus may link with other extracellular matrix molecules [33]. This feature of aggrecan may, in part, account for the presence of an unique pericellular matrix which surrounds chondrocytes in cartilage 131, 491. Versican is the mammalian counterpart to the avian proteoglycan, PG-M [28, 33, 58, 721. Like aggrecan, its core protein is subdivided into three domains. Domain I contains one immunoglobulin repeat followed by two link protein modules. Versican binds to hyaluronan with a Kd of about 4 nM, which is similar to that for aggrecan [33]. Domain I1 contains two glycosaminoglycan-attachment regions, GAG-a and GAG-P. Within these two subdomains are 12-15 attachment sites for chondroitin sulfate chains [28, 331. Domain I11 contains two EGF-like repeats, a C-type lectin region, and a complement regulatory protein-like module [28. 331. The recombinant human lectin region can bind fucose and N-acetylglucosamine as well as tenascin-R [2, 28, 331. There are at least four possible splice variants of mammalian versican [27, 32, 571. The largest, termed VO, contains both GAG-a and GAG-P chondroitin sulfate-attachment sub-regions. The V I variant contains only the GAG-P sub-region, the V2 variant contains only the GAG-a sub-region, and the V? variant lacks both of these chondroitin sulfate-attachment regions and, thus, exists simply as a glycoprotein. Versican, as its name implies, appears to be an extremely versatile molecule. It is highly expressed in embryonic and fetal, non-cartilaginous tissues and is also a major component of tumor stromas [34, 611. In these situations it is thought that versican plays roles in cellular proliferation and cellular migration [33, 721. Recent studies have provided evidence that the EGF-like motifs in the C-terminal domain of versican enhance cellular proliferation of fibroblasts through interaction with the EGF-receptor on cell surfaces [71]. Versican, like aggrecan, interacts with hyaluronan; consequently, versican may participate in determining the visco-elastic properties of non-cartilaginous tissues and in structuring pericellular matrices. 44.3.2 Small Leucine-rich Proteoglycans All of the members of this family contain a central domain containing leucine-rich repeats [28, 29, 331. This hydrophobic central domain is essential for many of the molecular interactions engaged in by these molecules. The proteoglycans of this

134

44 Developmental and Aging Changes

family can be substituted with chondroitin sulfate, dermatan sulfate, or keratan sulfate chains; some of these proteoglycans can exist as simple glycoproteins [33]. The glycosylation of these proteoglycans is highly regulated and is dependent upon the tissue and species of origin, as well as on development and aging. The three dermatan/chondroitin sulfate members of this family will be briefly described. Decorin is a proteoglycan which is principally produced by mesenchymal cells [ 6, 28, 29, 331. It is a major proteoglycan species in tissues which contain types I and I1 collagen fibers, and it acquired its name because of its ability to bind to, or decorate, these collagens at definitive locations [56].Decorin mRNA codes for four core protein domains. The first domain contains the signal peptide and a propeptide; it is not expressed in the final translational product [33]. The three domains which are expressed are the cysteine-rich domain 11, which contains the single glycosaminoglycan attachment site, the leucine-rich domain 111, which contains ten tandem leucine-rich repeats, and domain IV, which contains a relatively large loop with two cysteine residues. Many of the functional characteristics of decorin depend upon its leucine-rich domain. Its ability to bind to collagens and to inhibit type I collagen fibrillogenesis depends upon this hydrophobic domain [ 54, 681. The physiologic importance of the interactions between decorin and type 1 collagen fibers is demonstrated in decorin-null mice [ 151. These mice grow normally to adulthood, but exhibit increased skin fragility. The bundles of type I collagen fibers in the mutant skin appear as unusually coarse and irregular structures, an indication that decorin is essential for regulating their formation. In addition, decorin binds, through its core protein, the growth factor TGF-p and in doing so, temporarily inactivates this molecule [53]. This interaction may also provide a storage compartment in the extracellular matrix. Decorin, through its binding to specific cell surface receptors, can modify cellular physiology and, in this way has been shown to inhibit cellular proliferation [33].This proteoglycan, but not biglycan, activates fibroblasts that are adhereing to vitronectin to produce matrix metalloproteinase- 1, a tissue collagenase [32]. Thus, this proteoglycan has multiple gene regulatory effects which facilitates both structuring and remodeling of the extracellular matrix, events that have important implications in morphogenesis. Biglycun shares substantial sequence homology with decorin, and it has four core protein domains as described above for decorin [29, 331. However, it differs from decorin in that it is substituted with two dermatan sulfate or chondroitin sulfate chains near its N-terminus. Despite the overall similarity between decorin and biglycan, these two molecules have, in many instances, different tissue distributions. This is especially apparent in human skin where decorin is associated with extracellular matrix and fibrillar collagen bundles, while biglycan is principally found in association with epithelial structures, the epidermis, and vascular endothelial cells [61. Epiphycun, the mammalian counterpart for the chick proteoglycan PG-Lb, appears in cartilage, particularly in the flattened zone of chondrocytes of chick embryonic limb cartilage [29, 351. Thus, this proteoglycan has a significantly more restricted tissue distribution than does either decorin or biglycan. This proteoglycan is usually substituted with a single dermatan/chondroitin sulfate glycosaminoglycan chain near its N-terminus [35].

44.4 ChondrotinlDermatan Sulfate Proteoglycans in Development and Aging

735

44.4 Chondrotin/Dermatan Sulfate Proteoglycans in Development and Aging Because of their complexity, developmental, age-related, and pathologic changes in proteoglycans may occur through three basic mechanisms, two of which will be briefly discussed here. First, changes in the production and distribution of specific core proteins occur during development and in tumor growth [lo, 23, 34, 611. These changes in core protein expression also include the controlled expression of specific splice variants for a given core protein. Second, many developmental and pathologic situations are accompanied by post-translational changes in the glycosaminoglycan chains attached to a particular proteoglycan species. Third, the extracellular degradation of proteoglycans is a highly regulated phenomenon that is now known to be an important factor in age-related changes of some tissues, especially cartilage. This latter mechanism, although important, will not be discussed further.

44.4.1 Core Proteins in Development, Aging, and Pathologies

Cartilage development provides a paradigm for the study of proteoglycans in development and aging. In limb morphogenesis, an undifferentiated mesenchyme is replaced by chondrogenic and myogenic tissue [ 161. Condensation of the central core of mesenchymal cells denotes the start of chondrogenesis. Undifferentiated mesenchymal cells produce a different complement of proteoglycans than do chondrogenic cells, which appear in the condensation region. Versican, a product of most mesenchymal cells, is replaced by aggrecan, a cartilage-specific proteoglycan [57]. However, versican continues to be produced in the surrounding myogenic regions. Aggrecan remains the principal proteoglycan species in cartilage; however, as discussed below, this proteoglycan undergoes post-translational modifications that are related to development, maturation, and aging. There are several other tissues in which versican is the major proteoglycan synthesized early in development, and also where its synthesis declines as development proceeds. These include skeletal muscle, liver, and cornea [9, 10, 21, 27, 391. In contrast to cartilage, where aggrecan remains the principal proteoglycan species, versican is replaced in these other tissues by small proteoglycans, often decorin [9, 10, 23, 611. As discussed below, skin follows the same pattern as for other noncartilaginous tissues. The appearance of epiphycan in the flattened zone of chondrocytes of chick embryonic limb cartilage is developmentally regulated. Its presence is postulated to delay the onset of calcification or to arrange the matrix in preparation for the extensive remodeling that occurs during calcification 1351. Other small proteoglycans, decorin and biglycan, are present throughout articular cartilage; however, the relative amounts of these proteoglycans differ in the various regions of cartilage; the functional significance of this regional distribution of decorin and biglycan has not been elucidated [SO].

136

44 Developmental and Aging Changes

Skin is another tissue in which changes in proteoglycans occur. Human fetal skin differs from adult skin in that, on a relative basis, the principal proteoglycan species is versican [61]. Hyaluronan is also a significant component of fetal skin. Presumably much of the versican interacts with hyaluronan to form multi-molecular aggregates, much like the aggregates of aggrecan and hyaluronan in cartilage matrix. In adult human skin, decorin becomes, on a relative basis, the principal proteoglycan species [61]. A similar increase in the ratio of decorin to versican has been reported for developing rat skin [23]. In some skin pathologies, such as hypertrophic scars and keloids, substantial changes occur in the proportions of different proteoglycans. Biglycan, which under normal circumstances is a relatively minor proteoglycan species in dermis, becomes a major proteoglycan species [30]. This switch between decorin and biglycan production may be related to the action of specific growth factors. Growth factors and other tissue regulatory factors differentially regulate the production of decorin and biglycan [29, 33, 36, 371. Thus, changes in growth factor/cytokine profiles which might occur during development or in pathologies might have an important impact on proteoglycans.

44.4.2 Chondroitin/DermatanSulfate Glycosaminoglycan Chains in Development, Aging, and Pathologies Cartilage serves as a model for the study of changes in glycosaminoglycan structure. Much of the early work on glycosaminoglycan structure was performed on aggrecan extracted from chick limb cartilage [ 161. Chondroitin sulfate chains attached to aggrecan undergo significant post-translational modifications related to development and maturation. First, chondroitin sulfate chains attached to embryonic aggrecan are significantly longer than are those chains attached to adult aggrecan. Second, the ratio of chondroitin 6-sulfate disaccharides to chondroitin 4-sulfate disaccharide declines with age. Chick aggrecan is also substituted with keratan sulfate chains which become longer as the donor matures and ages. Analogous changes have been reported for aggrecan in human, bovine, and ovine cartilage [ 5 , 18, 51, 65, 661. Recently, more refined analysis of the chondroitin sulfate of aggrecan from human cartilage has shown age-related changes in the sulfation pattern of the non-reducing terminal residues [48]. The basis for these changes in glycosaminoglycan composition appears to be determined at the level of enzymes located within the Golgi apparatus. Recent evidence for this has been found in neural tissue, another tissue in which there are significant, physiologically relevant changes in chondroitin sulfate proteoglycans [20, 42, 44, 561. These studies demonstrated developmentally regulated changes in chondroitin 4- and chondroitin 6-sulfotransferases in extracts of chick brain tissue [ 381. The ratio of activity of the 6-sulfotransferase to 4-sulfotransferase declined with development in concert with a decline in the ratio of chondroitin 6-sulfate to chondroitin 4-sulfate in the tissue. Furthermore, the total amount of enzyme activity and glycosaminoglycan content in tissues was highest in embryos. These results mirror those found in cartilage and highlight the point that changes in the types and

44.4 ChondrotinjDermatan Sulfute Profeoglyc’ans in Development and Aging

737

amount of sulfation in a given tissue or organ are a common occurrence in normal development. Changes in sulfation are not limited to developmental situations. In malignancies, similar changes in sulfation have also been observed [ 11. A physiologic role for chondroitin sulfate chains in neural development is illustrated by the work of Faissner and co-workers [20].They identified developmentally related changes in rat neural tissue associated with chondroitin sulfate proteoglycans and produced monoclonal antibodies that enabled them to identify a specific, developmentally related proteoglycan. This DSD-1 proteoglycan promoted neurite outgrowth in an in vitro assay. Removal of the glycosaminoglycan chain eliminated this in vitro activity. Addition of monoclonal antibody 473HD, which recognizes an epitope on this glycosaminoglycan chain, to the in vitro system also inhibited neurite outgrowth. These data strongly suggest that a specific carbohydrate epitope has physiologic relevance. Recent studies from this group have partially identified the epitope as containing at least one oversulfated chondroitin sulfate D disaccharide, with the remaining disaccharides being chondroitin 6-sulfate [ 14, 441. In contrast to these studies, it has been demonstrated that other chondroitin sulfate proteoglycans, such as aggrecan extracted from cartilage, inhibit neurite outgrowth [42, 591. Thus, these apparently contradictory data emphasize the point that proteoglycan core proteins and their glycosaminoglycan chains, in combination, play an important role in neural development. Chondroitin sulfate D is an unusual oversulfated moiety that is found in low abundance in most mammalian chondroitin sulfate and dermatan sulfate chains (Figure 1). Chondroitin sulfate proteoglycans appear to be important in the developmental patterning of the central nervous system. This is exemplified in a recent study where embryonic thalamic neurons were plated onto living slices of mouse forebrain [ 191. Cell attachment and neurite outgrowth on different layers of the developing cerebral cortex varied in ways that correlated with timing and pattern of thalamocortical innervation. The cortical plate possessed inhibitory activity, while the intermediate zone and subplate regions possessed stimulatory activity. Both the inhibitory and stimulatory layer-specific differences were abolished by pretreatment of tissue slices with chondroitinase. The cell-cell and cellLmatrix interactions that are responsible for the phenomena described above have not been identified. One potential mechanism is the interaction of chondroitin sulfate chains with cell adhesion molecules. Chondroitin sulfate chains appear to play a major role in the binding of phosphacan and the neural cell adhesion molecule TAG- l/axonin- I [44]. Interestingly, chondroitin sulfate chains have little, if any, role in the binding of neurocan to the same adhesion molecule. These results emphasize the importance of integrated protein/glycosaminoglycan chain interactions in development. These studies of neural development not only emphasize the importance of chondroitin/dermatan sulfate glycosaminoglycan chains, but also illustrate the utility of carbohydrate-specific monoclonal antibodies in studies of development and aging. In recent years a large number of monoclonal antibodies have been developed which recognize carbohydrate epitopes associated with chondroitin and dermatan sulfate glycosaminoglycan chains [3, 7, 20, 43, 62, 64, 70). Many of these

738

44 Developmental and Aging Changes

antibodies have been employed in studies which demonstrate that there are changes in carbohydrate structure which are apparently related to tissue development and pathologies. There are two groups of these antibodies. One group of antibodies only recognizes epitopes in glycosaminoglycan chains that have been previously modified by chondroitinase treatment [ 111. Bacterial chondroitinases are eliminases which leave an unsaturated uronic acid residue at the non-reducing termini of chondroitin sulfate and dermatan sulfate chains. The unsaturated uronic acid residue is an obligatory component of the epitope for this first group of antibodies. A second set of monoclonal antibodies recognizes epitopes that are located within or at the nonreducing termini of intact glycosaminoglycan chains [ 12,25,48,62].No chemical or enzymatic treatment is required to generate these epitopes. These latter epitopes have been termed native epitopes to differentiate them from those that require prior enzymatic treatment. A number of other native anti-chondroitin sulfate antibodies have been developed. The epitopes for some of these antibodies are located within the interior of chondroitin sulfate chains. This was demonstrated for embryonic chick aggrecan by performing controlled chondroitinase ABC treatments [25, 621. Some epitopes were removed from the aggrecan by very brief treatments, while other epitopes required extensive treatments for their removal. These data suggest that specific epitopes exist within defined domains that are located within chondroitin sulfate chains (Figure 2). Modification of chondroitin or dermatan sulfate chain chemistries could presumably result in changes in immunological expression in tissues. However, another consideration must also be taken into account. The immunologic expression of a particular epitope depends not only on its presence, but also on its presentation. The location of the epitope within the three-dimensional structure of a

Linkage Region

Chondroitin Sulfate Chain

I

+

4C3RD4

4D316C3iCS-56

Native 383

Core Protein

Figure 2. Chondroitin sulfate chains covalently attach to a core protein via a carbohydrate linkage region. The native chondroitin sulfate epitopes for various anti-chondroitin sulfate monoclonal antibodies have been mapped to different regions of the chondroitin sulfate chains of embryonic chick aggrecan. The native 3B3 epitope [12, 481 appears at the non-reducing terminus. Interior domains contain epitopes for antibodies 4D3/6C3/CS-56 [62]. Epitopes for antibodies 4C3/704 are located adjacent to the linkage region [25, 621. Figure modified from [62].

44.4 ChondrotinlDermatan Suljiate Proteoylycans in Development und Aging

739

given proteoglycan affects immunoreactivity, as does interaction of a glycosaminoglycan chain with other matrix molecules [48, 621. Immunohistochemical studies employing various anti-chondroitin sulfate monoclonal antibodies have demonstrated spatio-temporal changes in chondroitin sulfate chains in a variety of tissues and developmental situations [7, 12, 20, 43, 46, 61, 631. These changes presumably result from the controlled modification of the sulfation in these glycosaminoglycan chains, although changes in the presentation or masking of epitopes is a factor that must be considered in these studies [48, 621. Similarly, changes in the sulfation of chondroitin sulfate chains related to tissue pathologies have also been demonstrated [ 1, 121. Human skin provides an example where spatiotemporal changes in chondroitin sulfate structure have been shown to occur during development [ 6 11. These changes in sulfation were demonstrated with monoclonal antibodies which specifically recognize chondroitin sulfate glycosaminoglycans, which, for the most part, are attached to the interstitial proteoglycan versican [61]. Such spatio-temporal distributions of proteoglycans, based upon the expression of chondroitin sulfate epitopes, have been shown in other developing systems. A recent example is demonstrated in the developing chick heart where the appearance of a specific chondroitin sulfate epitope is associated with subdomains of the endocardial cushions as well as with trabecular and atrial septa1 formation [7]. The physiologic relevance of changes in the spatio-temporal distribution of chondroitin sulfate epitopes is not yet fully understood. However, clues as to the relevance of such tissue modifications come from the study of other glycosaminoglycans such as heparin and heparan sulfate. These two glycosaminoglycan chains contain various domains situated within their linear sequence that have been identified by combined chemical and enzymatic depolymerization studies [40, 52, 671. One such domain in heparin/heparan sulfate, enriched for a 2-sulfated iduronic acid and N-sulfated glucosamine, binds basic fibroblast growth factor FGF-2, an interaction which potentiates the cellular proliferation activity of this growth factor [67]. Dermatan sulfate, like heparin/heparan sulfate, also contains domains that are enriched in disaccharides that are sulfated in the 2-position of iduronic acid, and, consequently, bind FGF-2 [47]. Dermatan sulfate proteoglycans are an important constituent of skin and in wounds, where the content of dermatan sulfate proteoglycans is significantly increased over that in normal skin [47]. The dermatan sulfate from wound fluids binds FGF-2, and this interaction is sufficient to support the ability of FGF-2 to modulate cellular proliferation [47]. Thus, one potential mechanism for dermatan sulfate glycosaminoglycan function in development is to regulate cellular proliferation, migration, and differentiation through its growth factor interactions. This may be significant in skin development where it has been shown that FGF-2 is one of the growth factors that appears at sites of incipient hair and feather formation [ 17, 691. Experimental studies with embryonic chick skin show that insertion under the epidermis of microbeads soaked in FGF-2 initiates feather formation [60]. The role of dermatan sulfate in binding and regulating growth factors in developing tissues such as skin needs to be more carefully investigated. The studies described above indicate that chondroitin sulfate and dermatan sulfate glycosaminoglycan chains, like their heparin/heparan sulfate counterparts, participate in highly defined molecular interactions which depend upon the existence of specific

740

44 Developmental and Aging Changes

structural domains. However, our current knowledge of the chemical structures of these molecules remains limited. Therefore, to fully appreciate the roles that these glycosaminoglycans play in development, it will be necessary to develop the technology to chemically sequence these chains and to determine the physiologic relevance of these chemical sequences.

44.5 Summary Chondroitin/dermatan sulfate proteoglycans are critical components of extracellular matrices and these molecules play important, if as yet poorly understood, roles in tissue development, maturation, and pathology. The recent development of an extensive array of core protein- and glycosaminoglycan-specific monoclonal antibodies has revealed tantalizing information regarding the developmental importance of this diverse and complex group of molecules. However, a detailed appreciation of the physiologic roles that these molecules play awaits a more thorough understanding of their interactive nature. Such interactions include:

1) the ability of these molecules to bind, store, and regulate the activity of growth factors; 2) the ability of these molecules to interact with cell surface receptors and cell adhesion molecules in order to direct cellular migration; 3) the ability of these molecules to interact with cell surface and matrix molecules to structure the pericellular and interstitial matrices; and 4) the ability of these molecules to bind to cell surface receptors and thereby to directly influence cellular metabolism. These interactions involve both core proteins and glycosaminoglycan chains, sometimes both in combination. All of these interactive events contribute to tissue differentiation and morphogenetic movements that are critical for tissue development and pathologies.

References I. Addny, R., Heimer, R., Caterson, B., Sorrell, J.M., and Iozzo, R.V., J. B i d Chem., 1990, 265, 11389-11396. 2. Aspberg, A,, Binkert, C., and Ruoslahti, E., Proc. Nut1 Acud. Sci. USA, 1995, 92, 1059010594. 3. Avnur, Z., and Geiger; B., Cell, 1984, 38: 811-822. 4. Barry, F.P., P.J. Neame, J. Sasse, and D. Pearson, Matrix Biol., 1994, 14, 323-328. 5. Bayliss, M.T. and Ah, S.Y.. Biochem. J., 1978, 176, 683-693. 6. Bianco, P.: Fisher., L.W., Young, M.F., Termine, J.D., and Robey, P.G., J. Histochem. Qtochem., 1990, 38, 1549-1563.

References

741

7. Capehart, A.A., Mjaatvedt, C.H., Hoffman, S., and Kmg, E.L., Anut. Rec., 1999,254, 181-195. 8. Carney, S.L., Muir, H., Physiol. Rev., 1988, 68, 858-910. 9. Carrino, D.A., Dennis, J.E., Drushel, R.F., Haynesworth, S.E., and Caplan, A.I., Biochem. J . , 1994,298, 5 1-60. 10. Carrino, D.A. and Caplan, A.I., J. Bid. C'hem., 1982, 257, 14145-14154. 1 I. Caterson, B., Christner, J.E., Baker, J.R., and Couchman, J.R., Fed. Proc., 1985, 44, 386-393. 12. Caterson, B., Mahmoodian, F., Sorrell, J.M., Hardingham, T.E., Bayliss, M.T., Carney, S.L., Ratcliffe, A,, and Muir, H., J. Cell Sci., 1990, 97, 411-417. 13. Cheng, F., Heinegdrd, D., Malmstrom, A., Schmidtchen, A,, Yoshida, K., and Fransson, L-A,, Glycobioloyy, 1994,4, 685-696. 14. Clement, A.M., Nadanaka, S., Masdyama, K., Mandl, C., Sugahara, K., and Faissner, A., J. Biol. Chem., 1998,273, 28444-28453. 15. Danielson, K.G., Baribault, H., Holmes, D.F., Graham, H., Kadler, K.E., and Iozzo, R.V., J , Cell Biol., 1997, 136, 729-73. 16. DeLuca, S., Heinegdrd, D., Hascall, V.C., Kimura, J.H. Caplan, A.I., J. Biol. Chem., 1977, 252, 6600-6609. 17. du Cros, D.L., Isaacs, K., and Moore, G.P., J. Cell Sci., 1993, 105, 667-674. 18. Dziewiatkowski, D.D., Lavalley, J., and Beaudoin, A.G., Connect. Tis.Res., 1989, 19, 103120. 19. Emerling, D.E. and Lander, A.D., Neuron, 1996, 17, 1089-1100. 20. Faissner, A,, Clement, A., Lochter, A., Streit, A,, Mandl, C. and Schachner, M., J. Cell Biol., 1994, 126, 783-799. 21. Gressner, A.M. and Vasel, A,, Mech. Ageing Devel., 1985, 31, 307-327. 22. Grumet, M., Friedlander, D.R., and Sakurai, T., Persp. Devel. Neurobiol., 1996, 3, 319-330. 23. Habuchi, H., Kimata, K., and Suzuki, S., J. Biol. Chem., 1986, 261, 1031-1040. 24. Hardingham, T.E. and Fosang, A.J., FASEB J., 1992, 6, 861-870. 25. Hardingham, T.E., Fosang, A.J., Hey, N.J., Hazell, P.K., Kee, W.J., and Ewins, R.J.F., Carbohydr. Res., 1994, 255, 241-254. 26. Hascall, V.C. and Kimura, J.H., Meth. Eniymol., 1982, 82, 769-800. 27. Hassell, J.R., Cintron, C., Kublin, C., and Newsome, D.A.. Arch. Biochem. Biophys., 1983, 222, 362-369. 28. Heineg;rd, D. and Oldberg, In: Royce PM. Steinmann B (eds) Connective Tissue and its Heritable Disorders: Molecular, Genetic, and Medical Aspects, Wiley-Liss, New York, 1993, pp., 189-209. 29. Hocking, A.M., Shinomura, T., and McQuillan, D.J., Matrix Biology, 1998, 17, 1Ll9. 30. Hunzelmann, N., Anders, S., Sollberg, S., Schonherr, E., and Krieg, T., Br. J. Dermatol., 1996, 135, 394-399. 31. Hunziker, E.B. and Herrmann, W., J. Histochem. Cytochem., 1987,35, 647-655. 32. Huttenlocher, A., Werb, Z., Tremble, P., Huhtala, P., Rosenberg, L., Damsky, C.H., Matrix Biology, 1996, 15, 239-250. 33. Iozzo, R. Ann. Rev. Biochem., 1998, 67, 609-652. 34. Isogai, Z., Shinomura, T., Yamakawa, N., Takeuchi, J., Tsuji, T., Heinegird, D., Kimata, K., Cancer Res., 1996, 56, 3902-3908. 35. Johnson, H.J., Rosenberg, L., Choi, H.U., Garza, S., Hook, M., and Neame, P.J., J. Biol. Chem., 1997,272, 18709-18717. 36. Kahari, V.-M., Hakkinen, L., Westermarck, J. and Larjava, H., J. Invest. Dermatol., 1995, 104, 503-508. 37. Kahari, V.-M., Larjava, H., and Uitto, J., J. Biol. Chem., 1991, 266, 10608-10615. 38. Kitagawa, H., Tsutsumi, K., Tone, Y., and Sugahara, K., J. Biol. Chem., 1997, 272, 3137731381. 39. Lennon, D.P., Carrino, D.A., Baber, M.A., and Caplan, A.I., Matrix, 1991, 11, 412-427. 40. Lindahl, U., Kusche-Gullberg, M., and Kjellen. L., J. Bid. Chem., 1998,273, 24979-24982. 41. Lyon, M., Deakin, J.A., Rahmoune, H., Fernig, D.G., Nakamura, T., and Gallagher, J.T., J. Biol. Chent., 1998, 273, 271-273. 42. Margolis, R.K.. Rauch, U., Maurel, P., and Margolis, R.M., Persp. Devel. Neurobiol., 1996, 3, 273-290.

A.,

142

44 Developmental and Aging Changes

43. Mark, M.P., Butler, W.T., and Ruch, J-V., Develop. Biol., 1989, 133, 475-488. 44. Milev, P., Maurel, P., Haring, M., Margolis, R.K., and Margolis, R.U., J. Bid. Chem., 1996, 271, 15716-15723. 45. Nadanaka, S., Clement, A., Masayama, K., Faissner, A,, and Sugahara, K., J. Bid. Chem.. 1998,273, 3296-3307. 46. Nieminen, P., Vainio, S.. Jernvall, J., Lukinmaa, P.-L.. and Thesleff, I.. J. Dent. Res., 1993, 72. 460- 1472. 47. Penc, S.F., Pomahac, B., Winkler, T., Dorschner, R.A., Eriksson, E., Herndon, M. and Gallo, R.L., J. Biol. Chem., 1998,273, 28116-28121. 48. Plaas, A.H.K., Wong-Palms, S., Roughley, P.J., Midura, R.J., and Hascdlt, V.C., J. Biol. Chem., 1997,272, 20603-20610. 49. Poole, C.A., J. Anat., 1997, 191, 1-13. 50. Roughley, P.J., Melching, L.I., and Recklies, A.D., Matrix Biol.. 1994, 14, 51-59. 51. Roughley, P.J., and White, R.J., J. Biol. Chern., 1980, 255, 217-224. 52. Salmivirta, M., Lidholt, K., and Lindahl, U., FASEB J., 1996, 10, 1270-1279. 53. Schonherr, E., Broszat, M., Brandan, E., Bruckner, P., and Kresse, H., Arch. Biochem. Biophys., 1998,355, 241-248. 54. Schonherr, E., Hausser, H. Beavan, L. and Kresse H., J. Biol. Chem., 1995, 270, 8877-8883. 55. Schwartz, N.B., Domowicz, M., Krueger, R.C., Li, H., and Mangoura, D., Persp. Devel. Neurobiol., 1996, 3, 291-306. 56. Scott, J.E., FASEB J., 1992, 6, 2639-2645. 57. Shinomura, T. Jensen, K.L., Yamagata, M. Kimata, K., and Solursh, M., Anat. Embryol., 1990,181,227-233. 58. Shinomura, T., Zako, M., Ito, K., Ujita, M., and Kimata, K., J. Bid. Chenz., 1995, 270, 10326-10333. 59. Snow, D.M., Lemmon, V., Carrino, D.A., Caplan, A.I., and Silver, J., Exp. Neurol., 1990, 109, 111- 130. 60. Song, H.K., Wang, Y., and Goetinck, P.F., Proc. Natl Acad. Sci. USA, 1996, 93, 10246-10249. 61. Sorrell, J.M., Carrino, D.A., Baber, B.A., and Caplan, A.I., Anat. Embryol., 1999, 199, 45-56. 62. Sorrell, J.M., Carrino, D.A., and Caplan, A.I., Matrix-, 1993, 13, 351-361. 63. Sorrell, J.M., Lintala, A.M., Mahmoodian, F., and Caterson, B., J. Immunol., 1988, 140, 4263-4270. 64. Sorrell, J. M, Mahmoodian, F., Schafer, LA., Davis, B., and Caterson, B., J. Histochmn?. Cytochem., 1990, 38, 393-402. 65. Sweet, M.B.E., Thonar, E.J.-M.A., and Marsh, J., Arch. Bioclzern. Biopliys., 1979, 198, 439448. 66. Thonar, E.J.-M.A., and Sweet, M.B.E., Arch. Biochem. Biophys., 1981, 208, 535- 547. 67. Turnbull, J.E., Fernig, D.G., Ke, Y., Wilkinson, M.C., and Gallagher, J.T., J. Biol. Chem., 1992,267, 10337-10341. 68. Vogel, K.G., Paulsson, M., and Heineglrd, D., Biochem. J., 1984, 223, 587-597. 69. Widelitz, R.B., Jiang, T.-X., Noveen, A,, Ting-Berreth, S.A., Yin, E., Jung, H . 4 , and Chuong, C.-M., Micro. Res. Tech., 1997, 38, 452-465. 70. Yamagata, Y., Yoneda, M., and Suzuki, S., J. Biol. Chem. 1987,262, 4161-4152. 71. Zhang, Y., Cao, L., Yang, B.L., and Yang, B.B., J. Biol. Chem., 1998,273, 21342-21351. 72. Zimmermann, D.R. and Ruoslahti, E., EMBO J., 1989, 8, 2975-2981.

Carbohydrates in Chemistry and Biology Beat Emst, Gerald W. Hart, Pierre Sinay copyright Q WILEY-VCH Verlag GmbH. 2000

45 Proteoglycans and Hyaluronan in Vascular Disease Thonzas N. Wight

45.1 Introduction Vascular diseases affect arteries and the most prevalent and clinically significant of these is atherosclerosis. Among the diseases of the western world, atherosclerosis is the leading cause of death and morbidity, responsible for almost half of all deaths in the United States, despite the fact that the mortality rate for this disease has been declining since the early 1960s [l].One reason for the decline is the recognition that dietary fat, in the form of cholesterol carrying lipoproteins, contributes to the development of atherosclerotic lesions and that lowering plasma cholesterol reduces the incidence and severity of the disease [2]. The inability to prevent this disease no doubt lies in the complexity of its pathogenesis, the multitude of etiologic factors, and the difficulties in sorting out the interrelationships of the many pathways that lead to the formation of the atherosclerotic plaque. Atherosclerosis is a disease of large and medium-sized arteries. Although any artery may be affected, the aorta, coronary and cerebral arteries are the prime targets of this disease. The disease involves the progressive thickening of the intimal layer of arteries, caused by increases in the number of vascular cells and amount of extracellular matrix (ECM) in the vessels. In addition, components in the blood, such as the cholesterol carrying lipoproteins, deposit in these thickenings and result in plaque formation and lumen narrowing [3,4]. These plaques are likely to rupture causing clot formation and thrombosis, which lead to vessel occlusion and myocardial or cerebral infarcts. Atherosclerosis takes place over several decades of life in different stages and it is likely that every living human adult has some stage of this disease! Documenting the sequence of events in the pathogenesis of any human disease is difficult because the early phases of the disease often go undetected and the disease is only recognized when clinical symptoms appear. Thus, the bulk of information that is available on the pathogenesis of atherosclerosis is derived from animal studies in which initiating events can be controlled and subsequent events closely

744

45 Proteoglycans and Hyaluronan in Vascular Disease

monitored both temporally and spatially. The two most common ways of initiating atherosclerotic lesion development in animals are increasing plasma lipids by either fat feeding or the use of molecular genetics to create high levels of plasma lipids [5] and by vascular injury using devices such as a balloon catheter [6]. Frequently, the best results are obtained from a combination of both approaches which suggests that this disease has a nutritional component that can be controlled and a vascular wall component that is more difficult to control. For the most part, these experimental regimes performed in animals create lesions that resemble those found in humans. However, there are species differences in many of the parameters that contribute to vascular lesion development so caution needs to be exhibited in transferring information received from the animal studies to the human. The ability of arteries to function normally depends upon their structural integrity. In large part, the structural integrity of an artery is determined by the nature of the ECM. The vascular ECM is a reinforced composite of collagen and elastic fibers embedded in a viscoelastic gel composed of proteoglycans, hyaluronan, water and a variety of glycoproteins [7] These components interact through entanglement and cross-linking to form a biomechanically active polymer network that imparts tensile strength, elastic recoil and compressibility to the artery. In addition, ECM components interact with vascular cells to modify their behavior [8]. Alterations in the balance of ECM components, as occurs in atherosclerosis, promotes the events that cause the formation of vascular lesions [7, 81.

45.2 Proteoglycans and Hyaluronan Proteoglycans and hyaluronan are ECM components that undergo significant changes in the formation of the atherosclerotic plaque. Although these molecules constitute a minor component of normal blood vessels, they increase dramatically in the early phases of lesion development and contribute to increases in lesion mass and volume [7, 81. In addition, they interact with lipoproteins and enzymes involved in lipoprotein metabolism to promote lipid retention in the developing lesion [9- 121. They also bind clotting and fibrinolytic factors and thus play a role in the thrombotic events that characterize the formation of the clinically significant advanced atherosclerotic plaque [ 13, 141. Finally, they interact with vascular cells to influence proliferation and migration, which are fundamental cellular events in early atherosclerotic lesions [ 151. The polysaccharide portion of these molecules, for the most part, governs their interactions but the core proteins exhibit selective binding properties as well [16, 171. The early studies to investigate the importance of these molecules in atherogenesis used extensive proteolytic digestion to release the glycosaminoglycans from the lesions for characterization. These studies demonstrated that the principle types of glycosaminoglycans found in vascular lesions were chondroitin 4 and 6 sulfate (C-4-S, C-6-S), dermatan sulfate (DS), heparan sulfate (HS) and hyaluronic acid

45.3 Ve’ersican (CSPG.s)

745

(HA) with trace amounts of heparin (Hep) and keratan sulfate (KS) (see reviews in [9-12, IS]). Furthermore, it was shown that these glycosaminoglycans increased in the early phases of atherosclerosis and that areas prone to develop lesions often contained altered amounts and types of glycosaminoglycans. Lipid accumulation within atherosclerotic plaques was frequently correlated with increases in glycosaminoglycan content and in vitro binding assays showed that glycosaminoglycans could bind B-lipoproteins through ionic interactions. Binding affinities between theses two sets of molecules depended on the type of glycosaminglycan and its size and charge characteristics. In the late 1970s and early 1980s, extraction techniques which preserved the protein moieties of the proteoglycans were used to recover these molecules from atherosclerotic lesions and, together with the use of antibodies which recognized different core proteins, it became clear that the different glycosaminoglycans within the lesions were attached to specific core proteins. These advances, together with the use of recombinant DNA technology, have allowed rapid progress to be made in the last decade in identifying the principle vascular proteoglycans present in vascular lesions and some of the reasons why these molecules accumulate during the formation of the atherosclerotic plaque. Thus, it is now recognized that the bulk of the CS chains present in atherosclerotic lesions are present as the proteoglycan versican. The DS chains are distributed between two highly homologous core proteins and form the proteoglycans decorin and biglycan. Heparan sulfate chains are found on several different core proteins that make up the perlecan and syndecan proteoglycan gene families. Hyaluronan, unlike the sulfated glycosaminoglycans, is not attached to a core protein and exists as a polysaccaride polymer. These proteoglycans and HA are present in defined topographical and spatial patterns in atherosclerotic lesions.

45.3 Versican (CSPGs) The bulk of information regarding the involvement of versican in atherosclerosis comes from immunohistochemical and “in situ” hybridization studies [ 19-25]. This proteoglycan is present throughout the lesions of atherosclerosis and particularly enriched in regions that contain smooth muscle cells [19-251. The synthesis of versican by arterial smooth muscle cells is stimulated following vascular injury and this molecule accumulates in the early proliferative and migratory phases associated with the initiation of the lesion [24, 251. Versican is also prominent at the edges of the necrotic core in more advanced plaques [20] and in close proximity to some of the deposited lipoproteins within the lesions [22, 231 suggesting a possible role for versican in lipid retention. In vitro binding assays show that versican isolated from cultured arterial smooth muscle cells binds to LDL with saturable kinetics and with a binding affinity of 2.3 x 10-’M [26, 271. Estimating the stoichiometry of the interaction reveals at least five LDL particles bound to a single CS chain on versican [27].

746

45 Proteoglycans and Hyuluronan in Vascular Diseuse

One of the versican binding sites in the apoprotein B of LDL has been located at residue 3363 since replacing this charged residue with a neutral amino acid by sitedirected mutagenesis eliminates versican binding to LDL in vitro [26]. Another consequence of lipoprotein retention by versican and other proteoglycans is the influence on lipoprotein oxidation. Oxidized lipoproteins exist throughout the atherosclerotic lesion and contribute to lesion severity [28]. In vitro experiments show that lipoproteins bound to C-6-S rich proteoglycans are more readily oxidized than non-bound lipoproteins and taken up more rapidly by macrophages [29]. On the other hand, C-4-S glycosaminoglycans inhibit copper-mediated LDL oxidation and thus this glycosaminoglycan may act as an antioxidant [30]. The reason for these opposite activities is not known but likely involves the ability of the CS chains to chelate the copper [29]. Oxidation of LDL decreases binding to versican and other vascular proteoglycans [27, 3 I ] but bridging molecules such as lipoprotein lipase promote complex formation between modified and unmodified LDL and versican [27, 321. Chondroitin sulfate proteoglycan-LDL completes are taken up rapidly by macrophages [33, 341 and smooth muscle cells [35]. Such uptake leads to the accumulation of intracellular lipid and the formation of “foam cells” which are characteristically found in atherosclerotic lesions. Internalization of CSPG-LDL complexes takes place via LDL receptor dependent and independent pathways and is accompanied by decreased degradation of the internalized LDL and increased cholesterol ester synthesis. Thus, proteoglycans not only promote the extracellular retention of lipoproteins but the intracellular retention as well. There have been suprisingly few biochemical studies on the nature of versican in atherosclerotic lesions. A large CSPG has been isolated from atherosclerotic lesions whose core protein size ranged from 160 kDa to 245 kDa and contained a mixture of C-4-S and C-6-S glycosaminoglycan chains with C-6-S predominating [36-401. CSPG-lipoprotein complexes have been isolated from human atherosclerotic lesions [41].Analysis of binding constants of the CSPG isolated from lesions reveals that multiple LDL particles can bind to a single CS chain [42-441 similar to what has been shown in vitro [27). Interestingly, vascular injury produces elongated chains on the large CSPG and these chains have an increased capacity to bind lipoproteins [43]. In addition, lipoproteins isolated from patients with an atherogenic lipoprotein profile (i.e. small dense LDL with moderate hypertriglyceridemia and low HDL) have a high affinity for vascular CSPG [45]. A major source of versican in atherosclerotic lesions is the smooth muscle cell [46, 471. The synthesis of versican by arterial smooth muscle cells in culture is highly regulated. For example, growth factors such as PDGF, and TGF beta increase core protein synthesis and cause elongation of the CS chains attached to the versican core protein [48]. Such alterations in the size of the CS chain increases the binding of this proteoglycan to lipoproteins both in vitro [49] and in vivo [43]. Thus, it is likely that growth factor modifications in the synthesis of versican and other proteoglycans may be partly responsible for increased lipid retention in the developing lesion. While these growth factors stimulate versican synthesis, other factors inhibit the production of this proteoglycan. For example, IL- 1 decreases versican synthesis but upregulates decorin synthesis in arterial smooth muscle cell cultures [ 50-5 I].

45.4 Hyaluronan

747

Versican is also a predominant ECM component of restenotic lesions and present in the hypercellular and myxoid regions of these lesions [52-531. The versican rich regions are unusual in that other ECM fibrous components such as collagen are present in low amounts. Interference with versican accumulation following vascular injury using blocking antibodies to TGF beta blocks intimal thickening [54]. In addition, lesion regression induced by altered flow is associated with versican degradation.

45.4 Hyaluronan Versican is a member of a proteoglycan gene family that shares the property of binding to the glycosaminoglycan hyaluronan. This interaction creates high molecular weight aggregates that occupy large hydrodynamic domains in the tissue. Hyaluronan is present throughout the ECM of atherosclerotic lesions [22, 551 and is especially prominent in early lesions where there is active cell proliferation and migration [56]. Hyaluronan is also present in regions of the lesions that contain macrophages and lymphocytes and may serve as a substate for these cells. These cells carry receptors for HA (i.e. CD44) and blocking these receptors by exogenous administration of HA prevents the accumulation of these cells in developing lesions [57]. Since the presence of inflammatory cells contributes significantly to lesion severity, targeting HA may be a useful strategy to reduce or eliminate lesion formation! In addition, HA is prominent with versican in human restenotic lesions that form within months following directional atherectomy [56]. Such HA rich regions would be prone to swelling resulting in tissue expansion and lumen encroachment. Thus, rapid expansion of the restenotic lesion could, in large part, be due to the edematous changes created by hyaluronan and associated molecules such as versican [53]. Loss or breakdown of HA could lead to expulsion of water, causing tissue shrinkage. This conversion may involve a water clogged ECM becoming a cicatrix that shrinks and contracts the arterial wall, causing loss of lumen diameter. It remains to be shown whether vascular remodeling that occurs in restenosis involves changes in the water content of the lesion. In vitro studies show that HA and versican are prominent surrounding proliferating and migrating smooth muscle cells [58] and both of these molecules are upregulated when these cells are stimulated to proliferate and migrate [48, 591. Hyaluronan binds to the surface of smooth muscle cells via at least two HA binding proteins, CD44 [60] and RHAMM (Receptor for Hyaluronan Mediated Motility) [61]. Interference with this binding using either blocking antibodies or short oligosaccharides of HA inhibits proliferation and migration of smooth muscle cells [58]. It is also of interest that other HA binding molecules influence arterial smooth muscle cell phenotype. For example, TSG-6 is a small HA binding protein that is significantly upregulated following arterial injury [62].Furthermore, overexpression of TSG-6 stimulates arterial smooth muscle cell proliferation. It is as yet unclear as

748

45 Proteoglycuns and Hyuluronan in Vusculur Diseuse

to why this protein should confer a growth advantage to arterial smooth muscle cells but this molecule may be involved in stabilizing the HA-rich ECM that is necessary for the proliferation of these cells [%I. Recent studies have shown that HA is not only important outside the cell but enriched inside proliferating smooth muscle cells [63], possibly influencing the availability of factors that are involved in cell cycle traverse. An important aspect of HA biology, when considering the role of this molecule in regulating cell behavior, is the fact that HA fragments, generated during the remodeling of the ECM, have biological activity not possessed by the intact molecule. For example, fragments of HA induce proinflammatory chemokine and cytokine expression by monocytes and thus may play critical roles in regulating the proinflammatory involvement of these cells in the formation of the atherosclerotic plaque [64].In addition, fragments of HA stimulate endothelial migration and proliferation [65-681 and the synthesis of ECM molecules such as type I and VIII collagen by these cells [69], molecules associated with the angiogenic phenotype. In fact, HA fragments promote the formation of new blood vessels in vivo [70, 711 confirming their importance as possible regulators of the neovascularization response, a critical step in the formation of atherosclerotic lesions [72].

45.5 Decorin/Biglycan (DSPGs) Perhaps the second most quantitatively significant group of proteoglycans present in atherosclerotic lesions is the dermatan sulfate containing proteoglycans, decorin and biglycan. These molecules are part of a larger gene family characterized by leucine rich core proteins that possess the capacity to bind other proteins [73]. Morphological, biochemical, immunochemical and immunohistochemical studies have shown these two molecules are located throughout lesions but with different distributions. Decorin tends to be found in association with collagen deposits and is more prominent in primary atherosclerotic lesions compared to restenotic lesions [74]. In addition, decorin is often associated with TGF beta-rich regions of the lesion suggesting a possible relationship between these two molecules [22]. Transfection of arterial smooth muscle cells with the decorin gene and transferring these cells to injured vessels, reduces lesion formation and creates thinner lesions enriched in collagen type 1 [75]. Such results indicate that decorin may play a critical role in the organization of collagen fibrils in developing lesions and could serve to stabilize plaques by increasing collagen fibrillogenesis as plaques develop. However, decorin attached to collagen serves as a binding site for LDL and apoprotein (a) and is thus thought to be another proteoglycan involved in lipid accumulation during atherogenesis [76, 771. Decorin is also found in lesions adjacent to macrophages. Macrophage condition medium is capable of stimulating decorin synthesis by arterial smooth muscle cells and the active agent in the conditioned medium is IL-1 [51]. Biglycan is also located throughout atherosclerotic lesions but is more associated with smooth muscle containing regions of the plaque [21, 22, 741. Biglycan is also present in the macrophage and lipid-filled core of the plaque [23]. In fact, extensive

45.6 PerlecanlSyndecuns (HSPGs)

749

co-localization of biglycan with apoE and apoB containing lipoproteins [23] has stimulated interest in the role of these proteoglycans in the buildup of lipid in the atherosclerotic plaque. In vitro binding studies show that, like versican, biglycan binds to LDL with saturable kinectics and with an association constant of 1.7 x lOP7M 1271. Biglycan also binds lipoprotein lipase and promotes native and oxidized lipoprotein retention within the ECM [27]. In addition to serving as a depot for lipid, biglycan may also influence lipid metabolism within the developing lesion since this proteoglycan binds phospholipase A2 and enhances the activity of this enzyme and the generation of proinflammatory products [78, 791. Vascular endothelial cells are another source of biglycan and decorin. Confluent monolayers of aortic endothelial cells synthesize biglycan but do not synthesize decorin [SO, 811. Resting capillaries are negative for decorin but positive for biglycan [82]. However, when endothelial cells are stimulated to sprout and form tubes in culture, decorin synthesis along with the synthesis of type 1 collagen is upregulated 1811. In fact, viral mediated expression of decorin by endothelial cells cultured on collagen gels promotes tube formation and reduces apoptosis [ 831. Interestingly, overexpression of decorin by bovine endothelial cells also inhibits their migration by promoting fibronectin fibrillogenesis around the migrating cells [ 841. These studies collective point to a pro-angiogenic role of these small DSPGs.

45.6 Perlecan/Syndecans (HSPGs) Perlecan is a high molecular weight HSPG that is localized diffusely throughout atherosclerotic lesions [22]. This proteoglycan is prominent in the subendothelial space and surrounding smooth muscle cells in the lesion but does not, however, colocalize with lipids within the plaque as observed for CSPGs and DSPGs. However, like other proteoglycans, perlecan is upregulated in response to arterial injury in experimental animals 1241. In addition, HSPGs exist as part of the plasma membrane of endothelial and smooth muscle cells. These proteoglycans form a family of four and are termed syndecans [85]. For example, syndecan 4 is present on the surface of endothelial cells and binds antithrombin I11 to help prevent coagulation at the blood vessel surface [ 141. When endothelial cells are stimulated to migrate, perlecan and syndecan expression are decreased [ 861 but syndecan expression is stimulated when migration of these cells is blocked by growth factors such as PDGF AA 1871. Such findings suggest that these HSPGs are negatively correlated with endothelial cell proliferation. Syndecan expression is differentially regulated when arterial smooth muscle cells are stimulated by serum and following vascular injury [88-91]. While their role in the injury response is not understood, these HSPGs may play a role in the proliferative response of the smooth muscle cells to bFGF since cell surface HSPGs act as co-factors for high affinity binding of bFGF to its cell surface receptor [92, 931. Interestingly, removing heparan sulfate from injured vessels by heparanase treatment blocks smooth muscle cell proliferation in response to bFGF 1941.

750

45 Proteoglycans and Hyaluronan in Vascular Disease

Heparan sulfate proteoglycans also play a role in the uptake of lipoproteins by vascular cells [95, 961. This activity is related to the ability of HSPGs to bind lipoprotein lipase, which serves as a bridging molecule between the lipoprotein and the proteoglycan. While HSPGs promote arterial smooth muscle cell proliferation, they may also act as growth inhibitors. There is a vast literature on the antiproliferative nature of heparin on smooth muscle cells and the ability of heparin or HS isolated from blood vessels to inhibit the neointimal response following injury [97, 981. In fact, perlecan expression appears to be mostly limited to post replicative smooth muscle cells [99]. When arterial smooth muscle cells are cultured in the presence of perlecan, replication specific transcription factors such as Oct 1 are down regulated [ 1001. Removing the heparan sulfate chains from the perlecan substrate can eliminate this inhibition. It should also be noted that growth inhibition induced by heparan sulfate containing molecules is frequently associated with alterations in the composition of the ECM. For example, infusion of heparin following vascular injury produces lesions that contain considerably less collagen and elastic fibers than those experimental lesions that form in the absence of infused heparin [ 1011. Such changes may predispose the lesions to rupture due to lack of fibrous proteins.

45.7 Summary Proteoglycans and hyaluronan are ECM molecules that accumulate in lesions of atherosclerosis and contribute to lesion severity in a number of different ways. Not only do these molecules form the structural basis for these lesions to grow, they interact with a variety of other molecules involved in processes fundamental to the development of lesions and, in doing so, function as partial regulators of atherosclerotic events. If these molecules are a common element in the regulation of most of the atherosclerotic events, targeting this component of the lesion may be the best way to control the development of atherosclerosis.

Acknowledgments This review was prepared with grant support from the National Institutes of Health (HL-18645: DK 024561. Special appreciation to Ms Judi Morris for the typing of the manuscript.

References 1. Kannel W, Prevalence, incidence, and mortality of coronary heart disease. In: Atherosclerosis

and Coronary Heart Disease, Fuster V, Ross R and Topol E, Editors. 1996, Lippincott-Raven: Philadelphia. pp. 13-21,

Rejerences

75 1

2. Grundy SM, Lipids, nutrition and coronary heart disease. In: Atherosclerosis and Coronary Heart Disease, Fuster V, Ross R and Topol E, Editors. 1996, Lippincott-Raven: Philadelphia. pp. 45-68. 3. Benditt EP and Schwartz S, Blood vessels in pathology, In: Pathology, Rubin E and Faber J, Editors. 1988, Lippincott Company: Philadelphia. pp. 452-495. 4. Ross R and Fuster V, The pathogenesis of atherosclerosis. In: Atherosclerosis and Coronary Heart Disease, Fuster V, Ross R and Topol E, Editors. 1996, Lippincott-Raven: Philadelphia. pp. 441-460. 5. Breslow J, Plump A and Dammerman M, New mouse models of lipoprotein disorders and atherosclerosis. In: Atherosclerosis and Cbronury Heart Disease, Fuster V, Ross R and Topol E, Editors. 1996, Lippincott-Raven: Philadelphia. pp, 363-378. 6. Schwartz S and Reidy M, Restenosis. An assessment of factors important in arterial occlusion. In: Atherosclerosis and Coronary Heart Disease, Fuster V, Ross R and Topol E, Editors. 1996, Lippincott-Raven: Philadelphia. pp. 701 -714. 7. Wight T, The vascular extracellular matrix. In: Atherosclerosis and Coronary Herirt Disease, Fuster V, Ross R and Topol E, Editors. 1996, Lippincott-Ravcn: Philadelphia. pp. 421440. 8. Wight TN. The extracellular matrix and atherosclerosis. Curr Opin Lipidol 6:326-34, 1995. 9. Radhakrishnamurthy B, Srinivasan SR, Vijayagopal P and Berenson GS. Arterial wall proteoglycans-biological properties related to pathogenesis of atherosclerosis. Eur Heart J 11 SUPPIE:148-57, 1990. 10. Williams KJ and Tabas I. The response-to-retention hypothesis of early atherogenesis. Arterioscler Thromb Vusc Biol 15:551-61, 1995. 11. Hurt-Camejo E, Olssoii U, Wiklund 0, Bondjers G and Camejo G. Cellular consequences of the association of apoB lipoproteins with proteoglycans. Potential contribution to atherogenesis. Arterioscler Thromb Vasc Biol 17:1011-7, 1997. 12. Camejo G, Hurt-Camejo E, Wiklund 0 and Bondjers G. Association of apo B lipoproteins with arterial proteoglycans: pathological significance and molecular basis. Atherosclerosis 139:205-22, 1998. 13. Bourin MC and Lindahl U. Glycosaminoglycans and the regulation of blood coagulation. Biochem J 289:313-30, 1993. 14. Rosenberg RD; Shworak NW, Liu J; Schwartz JJ and Zhang L. Heparan sulfate proteoglycans of the cardiovascular system. Specific structures emerge but how is synthesis regulated? J Clin Invest 99:2062-70, 1997. 15. Wight TN, Kinsella MG and Qwarnstrom EE. The role of protcoglycans in cell adhesion, migration and proliferation. Curr Opin Cell Biol4:793-801, 1992. 16. Jackson RL, Busch SJ and Cardin AD. Glycosaminoglycans: molecular properties, protein interactions, and role in physiological processes. Physiol Rea 71:481-539, 1991. 17. Kjellen L and Lindahl U. Proteoglycans: structures and interactions [published erratum appears in Annu Rev Biochem 1992;61:followingviii]. Annu Rev Biochem 60:443-75, 1991. 18. Wight TN. Vessel proteoglycans and thrombogenesis. Prog Hemost Thromb 5:l-39, 1980. 19. Gutierrez P, O’Brien K, Ferguson M, Nikkari S, Alpers C and Wight TN. Differences in the distribution of versican, decorin, and biglycan in atherosclerotic human coronary arteries. Cardiovasc Puthol6:271-278, 1997. 20. Halpert I, Sires UI, Roby JD, Potter-Perigo S, Wight TN, Shapiro SD, Welgus HG, Wickline SA and Parks WC. Matrilysin is expressed by lipid-laden macrophages at sites of potential rupture in atherosclerotic lesions and localizes to areas of versican deposition, a proteoglycan substrate for the enzyme. Proc Natl Acad Sci USA 93:9748-53, 1996. 21. Lin H, Wilson JE, Roberts CR, Horley KJ, Winters GL, Costanzo MR and McManus BM. Biglycan, decorin, and vcrsican protein expression patterns in coronary arteriopathy of human cardiac allograft: distinctness as compared to native atherosclerosis. J Heart Lung Transplant 151233-47, 1996. 22. Evanko SP, Raines EW, Ross R, Gold L1 and Wight TN. Proteoglycan distribution in lesions of atherosclerosis depends on lesion severity, structural characteristics, and the proximity of platelet-derived growth factor and transforming growth factor-beta. A m J Pathol 152:533-46, 1998.

752

45 Proteoglycans and Hyaluronan in Vascular Disease

23. O’Brien KD, Olin KL, Alpers CE, Chiu W, Ferguson M, Hudkins K, Wight TN and Chait A. Comparison of apolipoprotein and proteoglycan deposits in human coronary atherosclerotic plaques: co-localization of biglycan with apolipoproteins. Circulation 98:519-27, 1998. 24. Nikkari ST, Jarvelainen HT, Wight TN, Ferguson M and Clowes AW. Smooth muscle cell expression of extracellular matrix genes after arterial injury. Am J Pathol 144:1348-56, 1994. 25. Geary R, Nikkari S, Wagner W, Williams J, Adams M and Dean R. Wound healing: A paradigm for lumen narrowing after arterial reconstruction. J Vasc Surg 27:96-108, 1997. 26. Boren J, Olin K, Lee I, Chait A, Wight TN and Innerarity TL. Identification of the principal proteoglycan-binding site in LDL. A single-point mutation in apo-B100 severely affects proteoglycan interaction without affecting LDL receptor binding. J Clin Invest 101:2658-64, 1998. 27. Olin K, Potter-Perigo S, Barrett H, Wight T and Chait A. Lipoprotein lipase enhances the binding of native and oxidized low density lipoproteins to versican and biglycan synthesized by cultured arterial smooth muscle cells. J Biol Chem 274:34629-36, 1999. 28. Steinberg D, Parthasarathy S, Carew TE, Kboo JC and Witztum JL. Beyond cholesterol. Modifications of low-density lipoprotein that increase its atherogenicity [see comments]. N Engl J Med 320:915-24, 1989. 29. Hurt-Camejo E, Camejo G, Rosengren B, Lopez F, Ahlstrom C, Fager G and Bondjers G. Effect of arterial proteoglycans and glycosaminoglycans on low density lipoprotein oxidation and its uptake by human macrophages and arterial smooth muscle cells. Arterioscler Thromh 12:569-83, 1992. 30. Albertini R, Ramos P, Giessauf A, Passi A, De Luca G and Esterbauer H. Chondroitin 4-sulphate exhibits inhibitory effect during Cu2+-mediated LDL oxidation. FEBS Lett 403: 154-8, 1997. 31. Oorni K, Pentikainen MO, Annila A and Kovanen PT. Oxidation of low density lipoprotein particles decreases their ability to bind to human aortic proteoglycans. Dependence on oxidative modification of the lysine residues. J Biol Chem 272:21303-11, 1997. 32. Edwards IJ, Goldberg IJ, Parks JS, Xu H and Wagner WD. Lipoprotein lipase enhances the interaction of low density lipoproteins with artery-derived extracellular matrix proteoglycans. J Lipid Re.? 34: 1 155-63, 1993. 33. Hurt E and Camejo G. Effect of arterial proteoglycans on the interaction of LDL with human monocyte-derived macrophages. Atherosclerosis 67: 1 15-26, 1987. 34. Vijayagopal P. Regulation of the metabolism of lipoprotein-proteoglycan complexes in human monocyte-derived macrophages. Biochem J 301 :675-8 1, 1994. 35. Ismail NA, Alavi MZ and Moore S. Lipoprotein-proteoglycan complexes from injured rabbit aortas accelerate lipoprotein uptake by arterial smooth muscle cells. Atherosclerosis 105:7987, 1994. 36. Salisbury BG and Wagner WD. Isolation and preliminary characterization of proteoglycans dissociatively extracted from human aorta. J Biol Chem 256:8050-7, 1981. 37. Wagner WD, Salisbury GJ and Rowe HA. A proposed structure of chondroitin 6-sulfate proteoglycan of human normal artery and adjacent atherosclerotic plaque. Arteriosclerosis 6:407-17, 1986. 38. Dalferes ER, Jr., Radhakrishnamurthy B, Ruiz HA and Berenson GS. Composition of proteoglycans from human atherosclerotic lesions. Exp Mol Pathol47:363-76, 1987. 39. Cherchi GM, Coinu R, Demuro P, Formato M, Sanna G, Tidore M, Tira ME and De Luca G. Structural and functional modifications of human aorta proteoglycans in atherosclerosis. Matrix 10:362-72, 1990. 40. Coinu R, Cherchi G, Formato M, Demuro P and De Luca G. LDL interaction with proteoglycans isolated from human aortas with different atherosclerotic involvement. Haematologica 781270-6, 1993. 41. Srinivasan SR, Radhakrishnamurthy B, Pargaonkar PS, Berenson GS and Dolan P. Lipoprotein-acid mucopolysaccharide complexes of human atherosclerotic lesions. Biochim Biophys Act0 388~58-70, 1975. 42. Srinivasan SR, Vijayagopal P, Eberle K, Radhakrishnamurthy B and Berenson GS. Lowdensity lipoprotein binding affinity of arterial wall proteoglycans: characteristics of a chondroitin sulfate proteoglycan subfraction. Biochim Biophys Acta 1006:159-66, 1989.

43. Srinivasan SR, Xu JH, Vijayagopal P, Radhakrishnamurthy B and Berenson GS. Injury to the arterial wall of rabbits produces proteoglycan variants with enhanced low-density lipoproteinbinding property. Biochim Biophys Acta 1168:158-166, 1993 44. Vijayagopal P, Figueroa JE, Fontenot JD and Glancy DL. Isolation and characterization of a proteoglycan variant from human aorta exhibiting a marked affinity for low density lipoprotein and demonstration of its enhanced expression in atherosclerotic plaques. at her^sclerosis 127:195-203, 1996. 45. Anber V, Griffin BA, McConnell M, Packard CJ and Shepherd J. Influence of plasma lipid and LDL-subfraction profile on the interaction between low density lipoprotein with human arterial wall proteoglycans. Atherosclero.sis 124:261-71, 1996. 46. Yao LY, Moody C, Schonherr E, Wight TN and Sandell LJ. Identification of the proteoglycan versican in aorta and smooth muscle cells by DNA sequence analysis, in situ hybridization and immunohistochemistry. Matrix Biol 14:213-25, 1994. 47. Lemire JM, Braun KR, Maurel P, Kaplan ED, Schwartz SM and Wight TN. Versican/PG-M isoforms in vascular smooth muscle cells. Arteriosc/er Thromb Vasc Biol 19:1630-9, 1999. 48. Schonherr E, Jarvelainen HT, Sandell LJ and Wight T N . Effects of platelet-derived growth factor and transforming growth factor-beta 1 on the synthesis of a large versican-like chondroitin sulfate proteoglycan by arterial smooth muscle cells. J Biol Chem 266: 17640-7, 1991. 49. Camejo G, Fager G, Rosengren B, Hurt-Camejo E and Bondjers G. Binding of low density lipoproteins by proteoglycans synthesized by proliferating and quiescent human arterial smooth muscle cells. J Biol Chern 268:14131-7, 1993. SO. Ostberg CO, Zhu P, Wight TN and Qwarnstrom EE. Fibronectin attachment is permissive for IL-1 mediated gene regulation. FEBS Lett 367:93-7. 1995. 51. Edwards IJ, Xu H, Wright MJ and Wagner WD. Interleukin-I upregulates decorin production by arterial smooth muscle cells. Arterioscler Thuomh 14:1032-9, 1994. 52. Matsuura R , Isaka N, Imanaka-Yoshida K, Yoshida T, Sakakura T and Nakano T. Deposition of PG-M/versican is a major cause of human coronary restenosis after percutaneous transluminal coronary angioplasty. J Pathol 180:311-6, 1996. 53. Wight TN, Lara S, Riessen R, Le Baron R and Isner J. Selective deposits of versican in the extracellular matrix of restenotic lesions from human peripheral arteries. A m J Pathol 151:963-73, 1997. 54. Wolf YG, Rasmussen LM and Ruoslahti E. Antibodies against transforming growth factorbeta 1 suppress intimal hyperplasia in a rat model. J Clin fnuest 93: 1 172-8, 1994. 55. Levesque H, Girard N, Maingonnat C, Delpech A, Chauzy C, Tayot J, Courtois H and Delpech B. Localization and solubilization of hyaluronan and of the hyaluronan-binding protein hyaluronectin in human normal and arteriosclerotic arterial walls. Athero,sclerosis 105:Sl-62, 1994. 56. Riessen R, Wight TN, Pastore C, Henley C and h e r JM. Distribution of hyaluronan during extracellular matrix remodeling in human restenotic arteries and balloon-injured rat carotid arteries. Circulation 93:1141-7, 1996. 57. Ferns GA, Konneh M, Rutherford C, Woolaghan E and Anggard EE. Hyaluronan (HYALBV 52001 inhibits neo-intimal macrophage influx after balloon-catheter induced injury in the cholesterol-fed rabbit. Atherosclerosis 1 14:157-64, 1995. 58. Evanko SP, Angello JC and Wight TN. Formation of hyaluronan- and versican-rich pericellular matrix is required for proliferation and migration of vascular smooth muscle cells. Arterioscler Thromh Vasc Biol 19:1004-13, 1999. 59. Papakonstantinou E, Karakiulakis G, Roth M and Block LH. Platelet-derived growth factor stimulates the secretion of hyaluronic acid by proliferating human vascular smooth muscle cells. Proc Natl Acad Sci USA 92:9881-5, 1995. 60. Jain M, He Q, Lee WS, Kashiki S, Foster LC, Tsai JC, Lee ME and Haber E. Role of' CD44 in the reaction of vascular smooth muscle cells to arterial wall injury. J Clin Invest 97596.603, 1996. 61. Savani RC, Wang C, Yang B, Zhang S, Kinsella MG, Wight TN, Stern R, Nance DM and Turley EA. Migration of bovine aortic smooth muscle cells after wounding injury. The role of hyaluronan and RHAMM. J Clin Invest 95:1158-68, 1995.

154

45 Proteoylycans and Hyaluronan in Vasculuv Diseuse

62. Ye L, Mora R, Akhayani N, Haudenschild CC and Liau G. Growth factor and cytokineregulated hyaluronan-binding protein TSG-6 is localized to the injury-induced rat neointima and confers enhanced growth in vascular smooth muscle cells. Circ Res 81:289-96, 1997. 63. Evanko S and Wight TN. Intracellular localization of hyaluronan in proliferating cells. J Histochem Cytochem 47:1331-41, 1999. 64. McKee CM, Penno MB, Cowman M, Burdick MD, Strieter RM, Bao C and Noble PW. Hyaluronan (HA) fragments induce chemokine gene expression in alveolar macrophages. The role of HA size and CD44. J Clin Invest 98:2403-13, 1996. 65. West DC and Kumar S. The effect of hyaluronate and its oligosaccharides on endothelial cell proliferation and monolayer integrity. Exp Cell Res 183:179-96, 1989. 66. Slevin M, Krupinski J, Kumar S and Gaffney J. Angiogenic oligosaccharides of hyaluronan induce protein tyrosine kinase activity in endothelial cells and activate a cytoplasmic signal transduction pathway resulting in proliferation. Lab Invest 78:987-1003, 1998. 67. Montesano R, Kumar S, Orci L and Pepper MS. Synergistic effect of hyaluronan oligosaccharides and vascular endothelial growth factor on angiogenesis in vitro. Lab Invest 75:249-62, 1996. 68. Trochon V, Mabilat C, Bertrand P, Legrand Y, Smadja-Joffe F, Soria C, Delpech B and Lu H. Evidence of involvement of CD44 in endothelial cell proliferation, migration and angiogenesis in vitro. Int J Cancer 66:664-8, 1996. 69. Rooney P, Wang M, Kumar P and Kumar S. Angiogenic oligosaccharides of hyaluronan enhance the production of collagens by endothelial cells. J Cell Sci 105:213-8, 1993. 70. Feinberg R and Beebe D. Hyaluronate in vasculogenesis. Science 220:1177-1179, 1983. 71. West DC, Hampson IN, Arnold F and Kumar S. Angiogenesis induced by degradation products of hyaluronic acid. Science 228:1324-1326, 1985. 72. Zhang Y, Cliff WJ, Schoefl GI and Higgins G. Immunohistochemical study of intimal microvessels in coronary atherosclerosis. Am J Pathol 143:164-72, 1993. 73. Iozzo RV. The family of the small leucine-rich proteoglycans: key regulators of matrix assembly and cellular growth. Crit Rev Biochem Mol Biol32:141-74, 1997. 74. Riessen R, Isner JM, Blessing E, Loushin C, Nikol S and Wight TN. Regional differences in the distribution of the proteoglycans biglycan and decorin in the extracellular matrix of atherosclerotic and restenotic human coronary arteries. A m J Pathol 144:962-74, 1994. 75. Fischer JW, Kinsella MG, Slowes MM, Clowes AW and Wight TN. Local expression of bovine decorin by cell mediated gene transfer reduces neointima formation after balloon injury in rats. Circ,Res In Press, 1999. 76. Pentikainen MO, Oorni K, Lassila R and Kovanen PT. The proteoglycan decorin links low density lipoproteins with collagen type 1. J Biol Chem 272:7633-8, 1997. 77. Klezovitch 0, Edelstein C, Zhu L and Scanu AM. Apolipoprotein(a) binds via its C-terminal domain to the protein core of the proteoglycan decorin. Implications for the retention of lipoprotein(a) in atherosclerotic lesions. J Biol Chem 273:23856-65, 1998. 78. Sartipy P, Johansen B, Camejo G, Rosengren B, Bondjers G and Hurt-Camejo E. Binding of human phospholipase A2 type I1 to proteoglycans. Differential effect of glycosaminoglycans on enzyme activity. J Biol Chem 271:26307-14, 1996. 79. Sartipy P, Bondjers G and Hurt-Camejo E. Phospholipase A2 type I1 binds to extracellular matrix biglycan: modulation of its activity on LDL by colocalization in glycosaminoglycan matrixes. Arterioscler Thromb Vusc Biol 18:1934-41, 1998. 80. Jarvelainen H, Kinsella MG, Wight TN, and Sandell LJ. Differential expression of small chondroitin/dermatan sulfate proteoglycans, PG-I/BigIycan and PG-II/Decorin, by vascular smooth muscle and endothelial cells in culture. J Biol Chem 266:23274-23281, 1991. 81. Jarvelainen HT, Iruela-Arispe ML, Kinsella MG, Sandell LJ, Sage EH and Wight TN. Expression of decorin by sprouting bovine aortic endothelial cells exhibiting angiogenesis in vitro. Exp Cell Res 203:395--401, 1992. 82. Bosse A, Schwarz K, Vollmer E and Kresse H. Divergent and co-localization of the two small proteoglycans decorin and proteoglycan-100 in human skeletal tissues and tumors. J Histochem Cytochem 41:13-9, 1993. 83. Schonherr E, O’Connel BC, Schittny J, Robenek H, Fasterman D, Fisher LW, Plenz G, Vischer P, Young MF and Kresse H. Paracrine or virus mediated induction of decorin

References

755

expression by endothelial cells contributes to tube formation and prevention of apoptosis in collagen lattices. Eur J Cell Biol78:44- 55, 1999. 84. Kinsella MG, Fischer JW, Mason DP and Wight TN. Retoviral mediated expression of decorin by macrovascular endothelial cells: effects on cellular migration. J B i d Chem In Press, 1999. 85. Bernfield M, Kokenyesi R. Kato M , Hinkes MT, Spring J, Gallo RL and Lose EJ. Biology of the syndecans: a family of transmembrane heparan sulfate proteoglycans. Annu Rea Cell Bid 8:365-93, 1992. 86. Kinsella MG, Tsoi CK, Jarvelainen HT and Wight TN. Selective expression and processing of biglycan during migration of bovine aortic endothelial cells. The role of endogenous basic fibroblast growth factor. J Biol Chem 272:3 18-25, 1997. 87. Koyama N , Kinsella MG, Wight TN, Hedin U and Clowes AW. Heparan sulfate proteoglycans mediate a potent inhibitory signal for migration of vascular smooth muscle cells. Circ Res 83:305- 13, 1998. 88. Cizmeci-Smith G, Langan E, Youkey J, Showalter LJ and Carey DJ. Syndecan-4 is a primaryresponse gene induced by basic fibroblast growth factor and arterial injury in vascular smooth muscle cells. Arterioscler Tliromh Vasc Biol 17:172-80, 1997. 89. Wang H, Moore S and Alavi MZ. Expression of syndecan-l in rabbit neointima following deendothelialization by a balloon catheter. Atherosclerosis 13 1: 141-7, 1997. 90. Cizmeci-Smith G and Carey DJ. Thrombin stimulates syndecan-l promotor activity and expression of a form of syndecan-1 that binds antithrombin 111 in vascular smooth muscle cells. Arterioscler Tlzromh VUSL. Biol 17:2609- 16, 1997. 91. Cizmeci-Smith G, Stahl RC, Showalter LJ and Carey DJ. Differential expression of transmembrane proteoglycans in vascular smooth muscle cells. J Biol Chem 268: 18740-7, 1993. 92. Conrad E, In: Hepurin-Binding Proteins. 1998, Academic Press: San Diego. pp. 301-349. 93. Gallagher JT, Heparan sulfate proteoglycans: the control of cell growth. In: Extrucellulur Matrix, Comper W, Editor. 1996, Harwood Academic Publishers. pp. 230-245. 94. Silver PJ, Moreau JP, Denholm E, Lin YQ, Nguyen L, Bennett C, Recktenwald A, DeBlois D, Baker S and Ranger S. Heparinase 111 limits rat arterial smooth muscle cell proliferation in vitro and in vivo. Eur J Phurmucol351:79-83, 1998. 95. Fuki IV, Kuhn KM, Lomazov IR, Rothman VL, Tuszynski GP, Iozzo RV, Swenson TL, Fisher EA and Williams KJ. The syndecan family of proteoglycans. Novel receptors mediating internalization of atherogenic lipoproteins in vitro. J Clin Invest 100:161 l --22, 1997. 96. Williams KJ and Fuki IV. Cell-surface heparan sulfate proteoglycans: dynamic molecules mediating ligand catabolism [see comments] [published erratum appcars in Curr Opin Lipidol 1998 Feb; 9(1):80]. Curr Opin Lipidol 8:253-62. 1997. 97. Nikkari ST and Clowes AW, Heparin and heparinoids: control of the intimal response. In: Pharmacoloyic Suppression of Intimal Hyperplusiu, WJ Q-B, Editor. 1993, R.G. Landes: Austin, TX. pp. 69-79. 98. Bingley JA, Hayward IP, Campbell JH and Campbell GR. Arterial heparan sulfate proteoglycans inhibit vascular smooth muscle cell proliferation and phenotype change in vitro and neointimal formation in vivo. J Vusr Sury 28:308-18, 1998. 99. Weiser MC. Belknap JK, Grieshaber SS, Kinsella MG and Majack RA. Developmental regulation of perlecan gene expression in aortic smooth muscle cells. Mutrix B i d 15:331-40. 1996. 100. Weiser MC, Grieshaber NA, Schwartz PE and Majack RA. Perlecan regulates Oct-1 gene expression in vascular smooth musclc cells. Mol Biol Cell 8:999-1011, 1997. 101. Snow AD, Bolender RP, Wight TN and Clowes AW. Heparin modulates the composition of the extracellular matrix domain surrounding arterial smooth muscle cells. A m J Pathol 137:313-30, 1990.

Carbohydrates in Chemistry and Biology Beat Emst, Gerald W. Hart, Pierre Sinay copyright Q WILEY-VCH Verlag GmbH. 2000

46 Functions of Glycosyl Phosphatidylinositols Nikola A . Baumann, Anant K. Menon, and David M. Rancour

46.1 Introduction In this Chapter we review current information and ideas about the cellular functions of glycosyl phosphatidylinositols (GPIs), a family of glycolipid structures defined by the structural motif Manal-4GlcNa1-6myo-inositol-P-lipid. GPIs anchor a variety of functionally diverse hydrophilic molecules (proteins, proteoglycans and complex phosphoglycans) to cell membranes in eukaryotic cells [ 11. These molecules include protozoal coat proteins (e.g. variant surface glycoprotein of Trypanosoma brucei), cell surface receptors (e.g. folate receptor), complement regulatory proteins (e.g. decay accelerating factor), cell surface hydrolases (e.g. 5’-nucleotidase, acetylcholinesterase), and cell adhesion molecules (e.g. NCAM isoforms). In addition, there exist a number of proteins of unknown function that are GPI-anchored, including the normal and scrapie forms of the Prion protein. GPI-anchored proteins have the general structure protein-(ethanolamine-P-6Manal-2Manal-6Mana1-4GlcNal6myo-inositol-P-lipid), where the structure within parentheses represents a minimal GPI moiety and the carboxyl-terminal of the protein is amide-linked to the GPI ethanolamine residue. The biosynthesis of these structures is described in Chapter 25. Recently, novel forms of GPIs have been identified in the yeast cell wall in the form of GPI remnants, derived from GPI-anchored proteins, that form the crossbridges linking mannoproteins to cell wall glucans [2, 31. Macromolecules that are GPI anchored rely on the anchor for cell surface expression. Thus, defects in GPI anchoring can have severe consequences if they cause non-expression of functionally critical macromolecules. A well documented example is paroxysmal nocturnal hemoglobinuria (PNH), a human disease caused by a GPI biosynthesis defect in affected blood cells leading to the non-expression of complement regulatory proteins [4]. The functional diversity of GPI-anchored molecules [5] makes it difficult to identify an evolutionary impetus for GPI anchoring. However, it is becoming increasingly clear that the membrane dynamics of GPI structures are quite different compared to those of ‘conventionally’ anchored macromolecules containing pro-

758

46 Functions o j Glycosyl Phosphatidylinositols

teinaceous transmembrane domains. GPI structures are confined to the lumenal/ exoplasmic leaflet of cellular membranes, typically the plasma membrane where most GPI-anchored macromolecules reside. Yet GPI structures are able to act as sorting determinants in exocytic and endocytic pathways (the sorting machinery being typically confined to the cytoplasmic face of cellular membranes) as well as interact with intracellular components such as the acylated src-family tyrosine kinases. It has been suggested that GPIs are associated with membrane microdomains (lipid rafts) and that it is through this association that they exert their transbilayer functions [6-11, 73, 741. Many cells also contain significant amounts of ‘free’ GPIs, not linked to proteins or glyco-polymers [12]. The function of these molecules is unclear, although there exist controversial proposals that they may be the source of phosphooligosaccharides involved in insulin signal transduction [ 13, 141. The brief overview provided in the preceding paragraphs gives some indication of the cellular functions of GPIs. The purpose of this Chapter is to examine these ideas in more detail. The reader is referred to numerous recent reviews to obtain more comprehensive information as well as a more extensive listing of the primary literature.

46.2 Parasite Coats: Extreme GPI-Anchoring GPIs are used as the primary mode of membrane macromolecule anchoring in the parasitic protozoa. For example, members of the order Kinetoplastida, suborder Trypanosomatina (includes T. brucei, T. cruzi and Leishmania spp.) utilize GPIs to anchor their most abundant cell-surface molecules (reviewed by [ 1, 15, 16, 751. These anchored molecules include glycoproteins and complex phosphoglycans. The sheer abundance of these molecules (>lo5 molecules/cell; in some cases -lo7 molecules/ cell) suggests that GPIs and the anchoring function are important. These surface molecules have been proposed to aid in the parasite’s ability to evade their host’s defense mechanisms, thus ensuring the parasite’s survival. A prominent example is the GPI-anchored variant surface glycoprotein (VSG) of bloodstream forms of the parasitic protozoan, T. brucei, the causative agent of African sleeping sickness and animal trypanosomiasis in sub-Saharan Africa. VSG molecules form a dense coat on the trypanosome surface and are presumed to be indispensable to the survival of the parasite in its mammalian host. Agents that disrupt the GPI anchor will likely compromise the density and stability of the surface coat resulting eventually in cell lysis in the presence of host serum components [17].

46.3 Yeast GPIs and the Cell Wall Studies of GPI-anchored proteins in yeast have provided insight into alternative roles for GPIs. In addition to providing ,membrane anchorage of soluble proteins to

46.4 Paro.xysma1 Nocturnal Henioglobinuriu

159

the plasma membrane, a clear and novel role for GPIs in cell wall biosynthesis has recently been elucidated (reviewed in [2, 3, 181). Cross-linking of mannoproteins to p-1,6 glucan in the cell wall is dependent on the presence of a GPI-remnant at the C-terminus of the mannoprotein [ 191. GPI-anchored mannoproteins destined for cell wall crosslinking arrive at the plasma membrane via the secretory pathway and, upon exposure to the periplasmic space, are transglycosylated resulting in a glycosidic linkage between a GPI-remnant mannose and the non-reducing end of a p-1,6 glucan polymer. The enzyme(s) that catalyzes this reaction is (are) unknown. A GPI-anchor is necessary but not sufficient for this reaction to take place since other GPI-anchored proteins either remain intact and membrane anchored (e.g. Gaslp) or are released in soluble form as a result of the action of an unknown phospholipase ( e g , Gcelp) but not crosslinked to p-1,6 glucan. This variation in processing suggests that there are additional signals that determine the eventual fate of GPI-anchored proteins once they reach the cell surface. Comparative analyses of primary amino acid sequences of GPIanchored proteins which are transglycosylated to p-1,6 glucan and those which are not indicate that amino acids 0-2 to 0 - 5 (where 0 is the amino acid to which the GPI anchor is attached) are important in selection for cell wall crosslinking [20]. Current models suggest that the yeast cell wall is constructed from ‘buildingblocks’ composed of crosslinked mannoprotein, p-1,6 glucan, p-1,3 glucan, and chitin. These blocks are assembled in unknown order and then integrated into the dynamic cell wall, over time becoming completely immobilized due to covalent and non-covalent associations with other components of the wall.

46.4 Paroxysmal Nocturnal Hemoglobinuria (PNH): Disease and Defects in GPI-Anchoring of Proteins A human clinical condition arises from the inability of certain blood cells to express GPI-anchored proteins on their cell surface. This rare disease, paroxysmal nocturnal hemoglobinuria (PNH), is due to somatic mutation of the X-linked PIG-A gene in hematopoietic stem cells resulting in an acquired hemolytic anemia [4]. The PIG-A gene encodes the putative catalytic component of UDP-G1cNAc:phosphatidylinositol N-acetylglucosaminyltransferase (GPI GlcNAcT), the enzyme responsible for catalyzing the first reaction of GPI biosynthesis. Cells defective in PIG-A are either unable to synthesize GPIs or show a marked decrease in the synthesis of GPIs and GPI-anchored proteins. With the decreased ability of PIG-A-defective cells to express GPI-anchored complement regulatory proteins (e.g. CD55 which functions to inhibit the formation or destabilization of the C3 convertase, and CD59 which protects the membrane from attack by the C5-C9 complex of activated complement), affected red blood cells become susceptible to complement-mediated lysis, resulting in the release of heme and hemoglobin into the blood, filtering by the kidney, and excretion in the urine. An interesting paradox has been recognized through studies of PNH [21]. Even though a decrease in GPI biosynthesis makes the blood cells more sensitive to

760

46 Functions of Glycosyl Phosphatidylinositols

complement-mediated lysis, persistence of the defective clone remains. This has led to a proposal that PNH patients must have other factors which select for clonal expansion and maintenance of GPI-deficient blood cells. For example, antibody selection against GPI-anchored CD52 in vitro in some cases produces a PNH phenotype in T and B cells. Interestingly, this GPI-negative phenotype is not caused by a classical PNH mutation in the PIG-A gene but rather was found to be reversible when the GPI-negative B cells were separated from the total population resulting in a redistribution to a mixed population of GPI-positive and GPI-negative B cells [22]. The mechanism for mutant stem cell selection is not known, however, this implies an alternative, negative consequence for the presence of GPIs which has not yet been determined.

46.5 GPIs in the Secretory and Endocytic Pathways GPIs are synthesized in the endoplasmic reticulum (ER) in excess of the amount required for protein anchoring. In many cells the pool of excess (‘free’) GPIs is considerable, reaching levels of approximately lo6 molecules/cell. These GPIs exit the ER and can be found in most cellular membranes, including the plasma membrane [12, 23-26]. Curiously, GPIs accumulate in the exoplasmic leaflet of the plasma membrane over long time periods where they can be detected by exogenous membrane topological reagents [24-261. The mechanism by which non-proteinlinked GPIs are trafficked from the ER to other cellular membranes is the subject of current investigations: available data suggest the possibility of two trafficking pathways, one involving transport vesicles for delivery of GPIs to the cell surface and possibly another non-vesicular route involving inter-membrane transfer of cytoplasmically disposed GPIs at points of membrane contact [23-261. Although the biological function of these lipids is unclear, it is possible that they may play a role in signal transduction [27, 28). GPI-anchored proteins are assembled in the ER via a transamidation reaction in which a GPI moiety is covalently linked to a freshly translocated protein bearing a carboxyl terminal GPI-directing signal sequence [29]. If GPI biosynthetic enzymes or the transamidase are disrupted in any way (as in PNH; see above), or if biosynthetic substrates are depleted [30], anchoring does not occur and the pro-protein is retained in the ER and eventually exported to and degraded by the proteasome [31, 321. Thus, GPI modification is critical for export of this class of proteins from the ER and it has been suggested that GPI may act as a forward transport signal [30, 331. ER-to-Golgi transport of GPI proteins in yeast is dependent on sphingolipid synthesis, consistent with the idea that sphingolipids promote the packaging of GPI protein into transport vesicles [34, 741 or somehow facilitate the fusion of GPIcontaining vesicles with the Golgi membrane [35]. Using inhibitors of ceramide synthesis it is thus possible to specifically block ER export of GPI proteins without affecting transport of soluble or transmembrane proteins [ 341. Extensive sorting of membrane components occurs in the trans Golgi network

46.5 GPIs in the Secretory and Endocytic Puthwuys

761

(TGN). In some polarized epithelial cells, GPI-anchored proteins and many glycosphingolipids (GSLs) are targeted to the apical membrane domain. The GPI anchor serves as an apical transport signal for membrane proteins in at least one strain of the MDCK kidney epithelial cell line and in Caco2 intestinal cells [36]. Investigations into this sorting phenomenon have led to the proposal of a general model for sorting of apically directed proteins and lipids [9, 11, 36-38]. A key ingredient of the sorting mechanism is the proposal that GPI proteins and GSLs, together with cholesterol. associate into membrane domains (lipid rafts) in the TGN. Since GSLs are synthesized in the Golgi and tend to be concentrated in the late secretory pathway, such associations would be characteristic of late rather than early stages of the secretory traffic of GPI proteins. These GPI-protein-GSL-cholesterol-containing lipid rafts serve as segregation devices ensuring the separation of a variety of apical cargo from material to be delivered to the basolateral membrane or retained in the TGN; one or more rafts is then encapsulated into an apical transport vesicle for transport to the apical plasma membrane domain. The raft model is supported by recent studies showing, for example, that manipulation of cellular cholesterol or sphingolipid levels can lead to selective alterations in sorting of apical proteins such as influenza hemagglutinin [39]. However, a number of questions remain, mainly resulting from a widely used technique in which rafts are isolated from cells in the form of detergent resistant membranes (see below). Consistent with the idea of GPI-protein-GSL-rich raft formation is the observation that GPI-anchored proteins traversing the secretory pathway become insoluble in cold Triton X-100 upon reaching the medial Golgi, and remain detergentresistant once at the plasma membrane [40]. These detergent-resistant membranes (DRMs) [lo], or detergent-insoluble glycolipid complexes (DIGS)[9], characterized by the presence of GPI proteins and a high sphingolipid and cholesterol content, have been interpreted to represent a coalescence of pre-existing rafts [41]. Detergent insolubility can be reconstituted in model membrane systems with the appropriate lipid environment including high melting temperature lipids such as saturated acyl chain containing phospholipids, sphingolipids and cholesterol [ 6, 10, 421. However, the precise relationship between DRMs and the rafts proposed in apical sorting remains the subject of some controversy. Furthermore, with respect to the sorting of proteins with transmembrane domains, rafts may represent only one of many devices that respond to a hierarchy of sorting signals in the cytoplasmic tails and ectodomains of the proteins in question. Once at the cell surface, GPI-anchored proteins can undergo endocytosis and subsequent recycling back to the PM (431. Endocytosis of GPI proteins occurs nonselectively via clathrin-coated and non-coated membrane invaginations at rates that are at least five times slower than the rate of signal-mediated internalization of membrane receptors such as the transferrin receptor [43]. However, once in the endosomes, GPI proteins are held back from bulk membrane recycling pathways such that their reappearance at the cell surface occurs approximately three times more slowly than for recycling receptors. Endosomal retention of GPI proteins is cholesterol dependent since depleting cells of cholesterol accelerates GPI protein externalization to a rate comparable to bulk membrane flow. These data are compatible with the idea of membrane rafts playing a role in endocytic sorting, in addi-

162

46 Functions of Glycosyl Phosphutidylinositols

tion to their postulated role in the exocytic pathway. The retention of GPI proteins in endosomes has important implications for the biology of some proteins such as the GPI-anchored folate receptor. It is conceivable that folate is released from its receptor in the acidic milieu of the endosomal compartments and retention of the receptor in these compartments allows efficient delivery of the released folate into the cytoplasm. In another example related to prion diseases, the GPI-anchored cellular scrapie protein is known to be converted to the infectious prion form in an acidic environment. Conversion is more efficient with the GPI-form of the protein than with a transmembrane form, and prion formation is inhibited by cholesterol depletion [44]. These observations suggest a critical role for the GPI-mediated endocytic retention of the scrapie protein in prion formation. The information available at present implies that GPI-anchored proteins are sorted in unique ways in both the exocytic and endocytic pathways. Detergentinsoluble membrane rafts containing cholesterol and sphingolipids may aid in the sorting of GPI-anchored proteins in the trans Golgi network as well as play a role in their slow rate of endocytic recycling.

46.6 Organization of GPI Proteins in the Plasma Membrane Recent investigations of the organization and lateral mobility of a number of membrane components including lipids, lipid-modified proteins, transmembrane proteins and GPI proteins indicate that the plasma membrane is laterally heterogeneous at the submicron level. The precise nature of the heterogeneities, termed membrane microdomains, and the implications of domain organization for the biological function of domain constituents remain matters of considerable debate [ 1 1, 731. Particularly attractive is the notion of ‘smart’ membranes [45] in which membrane components, such as those involved in signal transduction, are temporally and spatially restricted in microdomains so as to speed and optimize the processing of environmental signals (see below). GPI proteins are components of GSL-cholesterol-rich membrane domains (rafts) defined principally by the detergent insolubility criterion described above. Many lipids and transmembrane proteins are excluded from the rafts suggesting a considerable degree of specificity in the association of raft components. In support of this idea, Harder et al. [46] showed that antibody-induced clustering of GPI proteins and GSLs yielded microscopically detectable clusters (patches) that co-localized extensively on the cell surface. The GPI protein patches, however did not colocalize with patches of transferrin receptor, a representative transmembrane protein known to be excluded from detergent-insoluble rafts. In contrast, patches containing the GPI protein placental alkaline phosphatase (PLAP) accumulated the lipid-modified, src-like protein tyrosine kinase, fyn, a molecule restricted to the cytoplasmic leaflet of the plasma membrane. These data led Harder et al. to conclude that GPIproteins and transferrin receptor are somehow segregated and apparently immiscible, whereas GPI-proteins, fyn and GSLs interact at the plasma membrane. The

46.6 Organization q f GPI Proteins in the Plasma Membrane

163

GPI-GSL interactions are confined to the exoplasmic leaflet of the plasma membrane, whereas GPI-fyn or GSL-fyn interactions necessarily occur in transverse fashion across the membrane (see below). Clustering of GPI-anchored proteins at the cell surface has also been detected by other methods including fluorescence resonance energy transfer (FRET) [47] and chemical cross-linking [41]. In the FRET experiments, the extent of energy transfer (measured as fluorescence emission anisotropy) between a fluorescent analog of folk acid bound to GPI-anchored folate receptors was found to be density-independent. Energy transfer could be altered only by photobleaching the folate analog or diluting out the analog with folic acid. These results are consistent with organization of the folate receptor in sub-pixel size domains, estimated to be approximately 70 nm in size. Domain organization of the folate receptor was shown to be cholesteroldependent and dependent on the GPI-anchor: a transmembrane form of the same protein was diffusely distributed. Similar conclusions were reached in studies making use of chemical crosslinking techniques. GPI-proteins could be crosslinked at the cell surface into aggregates consisting of at least 15 molecules and, as with the FRET technique, crosslinking was cholesterol-dependent and specific for GPIanchored proteins. Further support comes from recent single-particle tracking (SPT) experiments designed to determine the lateral mobility of the GPI-protein Thy-1 and the GSL GM 1. SPT showed that for a large fraction of these molecules diffusion was transiently confined for 7-9 seconds to regions averaging 260-300 nm diameter [48]. Depletion of cellular GSLs using a GSL synthesis inhibitor reduced the number of molecules exhibiting confined diffusion and also reduced the size of the confinement zone, suggesting a connection between the SPT-defined domains and detergentinsoluble rafts. Although the experiments described above indicate that GPI-proteins are clustered within GSL-cholesterol-rich domains, other biophysical data indicate that: 1) this picture may represent only one level of cell surface organization of these molecules; 2 ) the lateral diffusion characteristics of GPI-proteins may not be greatly different from those of transmembrane proteins; and 3 ) GPI-proteins may enjoy considerably greater motional latitude than envisaged in the raft model. For example, early work by Edidin and co-workers showed differences in the lateral organization of GPI-proteins versus transmembrane proteins. Edidin and Stroynowski [49] used fluorescence recovery after photobleaching (FRAP) measurements to show that transmembrane proteins, but not GPI-proteins, are restricted to micronsize membrane domains. In other experiments using optical tweezers to move antibody-tagged proteins across the cell surface, Edidin et al. [50] showed that it is possible to move GPI-proteins farther than transmembrane proteins: the barrierfree path at room temperature was found to be almost three-fold greater for GPIproteins than for transmembrane proteins, indicating that GPI-proteins are not confined to the same extent as transmembrane proteins and that the barriers to lateral diffusion are localized to the cytoplasmic side of the membrane. In a related experimental system, Simson et al. [51] showed that both GPI-linked and transmembrane fornis of neural cell adhesion molecule (NCAM) indicated that NCAM molecules on the cell surface can freely diffuse in some regions of the membrane

764

46 Functions of Glycosyl Phosphatidylinositols

but are transiently confined in other regions. The confinement regions were proposed to have a diameter of 300 nm with confinement of both types of NCAM molecules lasting approximately 8 seconds. These data clearly differ from those obtained via the use of optical tweezers [SO]. Contrasting results were also seen with the FRET technique. Kenworthy and Edidin [52]reported that most GPI-anchored 5’-nucleotidase is randomly distributed on the apical surface of MDCK cells. These data differ from those of Varma and Mayor [47] who used FRET somewhat differently and concluded that GPI-proteins were indeed clustered. It remains to be seen whether these contrasting conclusions are a result of methodological differences, or differences in the biological material analyzed (endogenous proteins were used by Kenworthy and Edidin [52]whereas the results of Varma and Mayor [47] were obtained with cells expressing GPI-proteins introduced by gene transfection). Although there are notable exceptions, the bulk of the information available thus far indicates that GPI-proteins are organized at the cell surface in association with cholesterol and GSLs. The precise nature of the association and the size, structure and composition of the domains thus formed remain to be elucidated. It is possible that some of these organizational parameters may depend on subtle details of method and/or biology, and that the domains may differ depending on biological context. The reader is directed to recent surveys of this area [ 10, 11, 38, 53, 73, 741 for further elaboration of these points.

46.7 Association of GPI-Anchored Proteins with Caveolae A number of reports suggest that clusters of GPI-proteins are associated with caveolae, flask-like cell surface invaginations of the plasma membrane found in many-but not all-cells (reviewed in [54,76]).This conclusion is based on immunocytochemistry and immunofluorescence data, as well as the observation that the caveolar coat protein, caveolin, co-isolates with GPI-proteins in raft-like membranes after Triton extraction of cells [ 541. However, considerable controversy surrounds this idea since other reports clearly demonstrate that the perceived association of GPI-proteins with caveolae is the result of inadequate fractionation procedures [ 55, 561 or, in the case of the immunolocalization studies, an artifact of antibodyinduced clustering following weak fixation [6]. Furthermore the implied suggestion that caveolar localization is required for Triton-insolubility of GPI-proteins is clearly not the case since GPI-proteins are Triton-insoluble in cells lacking caveolin and detectable caveolae [57]. While it is clear that a caveolar localization of GPIproteins would have functional implications, the data in support of this idea are far from adequate. The reader is referred to reviews [6, 54, 58, 761 for further details.

46.8 Detergent Insolubility and Signaling via GPI-Proteins Antibody-mediated crosslinking of some cell surface GPI-proteins in lymphoid cells can cause phosphorylation of cellular substrates, increases in intracytoplasmic cal-

46.9 Meiiihrune Release of GPI-Atwhored Proteins

165

cium concentration and downstream events resulting in a mitogenic response indicating that GPI-proteins are able to interact with cytoplasmic signal transduction pathways. For example, antibody-mediated crosslinking of GPI-proteins such as Thy-1, Qa-2 or Ly-6A on thymocytes results in lymphokine production and cell proliferation; transmembrane forms of Qa-2 and Ly-6A cannot mediate cell activation suggesting that the GPI anchor is a critical component of the stimulus. How precisely GPI-protein-mediated cell signaling occurs is unknown, but it has been clearly shown that GPI-proteins confined to the exoplasmic leaflet of the plasma membrane can interact with acylated signalling proteins such as heterotrimeric G proteins and non-receptor protein tyrosine kinases that are restricted to the opposite membrane leaflet. These interactions have been demonstrated in experiments showing co-isolation of GPI-proteins with src-homology protein tyrosine kinases such as lck, fyn, and lyn in detergent-insoluble rafts as well as in immunoprecipitates [8, 59, 601. A recent study [61] showed that clustering of GPI-linked FcyR (a member of the IgG receptor family) with the transmembrane form of FcyR resulted in enhanced calcium flux. The same result was obtained upon clustering the transmembrane form with two other GPI-anchored proteins but not upon clustering the transmembrane form with itself. These data indicate that a GPI-protein is necessary for certain forms of signal transduction but that the nature of the GPI-protein itself is immaterial. This suggests that a special function is being supplied by the GPI itself. Similar conclusions were reached in a study of the effect of lowering cellular cholesterol on the calcium response induced by crosslinking GPI-proteins on T cells [ 621. Calcium responses to GPI-protein crosslinking were markedly reduced in cholesterol-depleted cells whereas no effect was seen on responses driven by the T cell receptor-CD3 pathway. Since the GPI-proteins tested shared no sequence homology, Stulnig et al. [62] concluded that that the GPI anchor itself was functionally critical in the signal transduction process. What this function is remains to be elucidated.

46.9 Membrane Release of GPI-Anchored Proteins The existence of a nutnber of GPI-specific phospholipases (e.g., PI-PLC and GPIPLD [63]) suggests additional possibilities for the biological significance of GPI anchors. An attractive hypothesis is that GPIs serve as metastable anchors that can be readily severed by endogenous GPI-specific phospholipases [ 161. This would result in: 1) the release of a soluble, possibly active form of a once membrane-anchored inactive protein; and 2) a lipid entity (e.g. diacylglycerol, ceramide, phosphatidic acid) that may serve as a second messenger in a signal transduction cascade (see below). Phospholipase-mediated cleavage may also be viewed as a mechanism to down-regulate surface expression of a range of GPI-anchored proteins. In support of these ideas, soluble forms of the GP-2 rat pancreatic granule protein, liver alkaline phosphatase Torpedo 5’-nucleotidase, and Trypanosomu cruzi Ssp4 antigen and trans-sialidase have been reported that appear to result from phospholipase C

766

46 Functions of Glycosyl Phosphatidylinositols

or GPI-PLD-mediated GPI anchor hydrolysis [ 16, 641. However, the precise function of the soluble proteins (versus the function of their membrane-bound counterparts) remains to be established. Another model for membrane release of GPI-anchored proteins comes from observations showing that GPI-linked molecules can be transferred between cells and that the transferred molecules stably insert in the external leaflet of the acceptor cell’s plasma membrane [72]. This GPI transfer may provide a mechanism for expression of foreign proteins on the cell surface [65]. Enrichment of GPI-anchored proteins has been shown in vesicles released from human red blood cells, however the biological significance of such a “shedding” of GPI-anchored proteins during plasma membrane vesiculation remains to be determined [66].

46.10 GPIs as Second Messenger Signaling Molecules GPIs have been implicated to play a role in hormone, growth factor and cytokinemediated signal transduction (reviewed by [13, 14, 27, 671. It has been proposed that hydrolysis of GPI produces a water soluble inositol phosphoglycan (IPGJ which acts as a second messenger in a variety of biological responses. The IPG fragment could be derived by excision from cell surface GPI-proteins through the combined action of a phospholipase and a protease, or via hydrolysis of free GPIs located in either leaflet of the plasma membrane bilayer. However, it is unclear whether the free GPIs described earlier are involved. For IPG molecules released at the cell surface rather than intracellularly, transport into the cell could occur via putative transporters located in caveolae [ 141. Despite the detailed nature of this proposal, there is no good structural information on IPG and clear cut data that support the idea of a role for GPI-derived glycans in signal transduction are sparse. Some of the more suggestive reports are summarized below. More comprehensive accounts may be found in the reviews [ 13, 14, 671. Lazar et al. [68] used K562 human erythroleukemia cell lines expressing human insulin receptor (introduced by gene transfection) to show that while parental cells displayed insulin-sensitive synthesis of glycogen, GPI-defective K562 cell lines were essentially unresponsive. These results suggest that GPIs are required for the stimulation of glycogen synthesis by insulin in these cells. Mohadevi and Hooper [69] studied the effect of insulin action on the GPI-protein membrane dipeptidase in the insulin-responsive 3T3-L 1 adipocytic cell line. Their results show that insulin treatment results in the phospholipase C-mediated release of membrane dipeptidase and the formation of the well characterized second messenger, diacylglycerol. It is possible that the released protein is further hydrolyzed by proteases to generate a glycan fragment that subsequently enters the cell but there are no data to support this idea. The conclusion from these studies is that insulin can activate a cell surface phospholipase C capable of hydrolyzing GPI-protein anchors and generating potential second messengers, components of the general model described above. Free GPIs displayed at the cell surface in mammalian cells are unlikely to be hydrolyzed by this mechanism since they typically contain an

46.11 Summary

767

additional, inositol-linked fatty acid that prevents phospholipase C hydrolysis (see Chapter 25). Earlier work by [70] also provided evidence of insulin-mediated release of a GPI protein. The data presented by these authors led them to conclude that upon insulin addition to BC3H 1 myocytes, the GPI-protein alkaline phosphatase was released into the culture medium as a result of proteolysis, and the resulting GPI remnant was hydrolyzed by a phospholipase giving rise to IPG. Although the data presented constitute the principle ingredients of the model described above, none of the hydrolytic products were structurally analyzed and much remains to be done. In recent studies, Frick et al. [71] synthesized IPG derivatives based on the GPIanchor of the yeast protein Gcelp, a CAMP binding protein. These synthetic IPGs were found to mimic some of the metabolic actions of insulin in rat adipocytes including lipogenesis and activation of glycogen synthase. The glycan core structure of typical GPI anchors was required for maximal insulin-mimetic activity, with some variations possible with respect to the type of residues modifying the terminal mannose residue as well as variations in glycosidic linkages. Although it remains to be seen how these complex charged structures penetrate the plasma membrane to enter the cytosol, their presumed site of action, these types of studies provide perhaps the most convincing evidence for the ability of a structurally defined GPIrelated glycan to elicit an insulin-like response. Despite these suggestive data, many questions remain to be answered. Credible structural information on the IPG moiety is fragmentary at best, the identity of the hydrolytic enzyme(s) responsible for its production remains to be elucidated, and the membrane topology of the precursor, products, and players need to be determined.

46.11 Summary In this Chapter we have attempted to summarize current ideas about the functions of GPIs. GPIs are found in cells as free glycolipids or protein anchors. The function of the free lipids is unknown, but a number of proposals exist that place these lipids in signal transduction pathways. In their role as protein anchors, GPIs can modulate the surface expression and membrane dynamics of proteins and consequently have a direct influence on protein function. This is perhaps most clear in recent observations that indicate that GPI-proteins are clustered at the cell surface in association with detergent resistant membrane rafts. Although there has been tremendous recent progress in understanding the role of GPIs in a variety of cellular functions, clarification of a number of issues is still needed and much remains to be learned before arriving at an appreciation of why cells synthesize GPIs and why it is necessary to modify a wide range of proteins with a GPI anchor. Acknowledgments

The authors would like to thank Jay Bangs, Peter Butikofer, Mark Field and Cedric Simonot for comments on the material presented in this Chapter. Work by the authors is supported by NIH grant GM55427 (to A.K.M.).

768

46 Functions of Glycosyl Phosphatidylinositols

References 1. 2. 3. 4. 5.

M.J. McConville and M.A.J. Ferguson, Biochem. J., 1993, 294, 305-324 P.N. Lipke and R. Ovalle, J. Bacteriol., 1998, 180, 3735-3740 J.C. Kapteyn, H. van den Ende and F.M. Klis Biochim. Biophys. Acta 1999, 1426, 373-383 L. Luzzatto, M. Bessler, and B. Rotoli, Cell, 1997, 88, 1-4 M.C. Field and A.K. Menon, Lipid Modijications of Proteins (M.J. Schlessinger, ed.), CRC Press, 1992, pp83-134 6. F.R. Maxfield and S. Mayor Ado, Exp. Med. Biol. 1997,419, 355-364 7. 0. Nosjean, A. Briolay and B. Roux Biochim. Biophys. Acta 1997, 1331, 153-186 8. P.J. Robinson Ado. Exp. Med. Biol. 1997, 419, 355-364 9. K. Simons and E. Ikonen Nature 1997,387, 569-572 10. D.A. Brown and E. London Annu. Rev. Cell Dev.Bid. 1998,14, 111-136 11. K. Jacobson and C. Dietrich Trends Cell Biol. 1999, 9, 87-91 12. A.K. Menon, N.A. Baumann, W. van’t Hof and J. Vidugiriene Biochem. Soc. Trans. 1997, 25, 861-865 13. M.C. Field Glycobiology, 1997, 7, 161-168 14. P. Stralfors Bioessays 1997, 19, 327-335 15. M.A.J. Ferguson, Phil. Trans. R. Soc. Lond. B, 1997, 352, 1295-1302 16. M.A.J. Ferguson Parusitol. Today 1994, 10, 48-52 17. M.A.J. Ferguson, J.S. Brimacombe, J.R. Brown, A. Crossman, A. Dix, R.A. Field, M.L.S. Giither, K.G. Milne, D.K. Sharma and T.K. Smith Biochim. Biophys. Acta 1999, in press 18. P. Orlean, The Molecular and C ~ l l ~ l Biology ar of’the Yeast Saccharomyces: Cell Cycle and Cell Biology, Volume 3, Cold Spring Harbor Laboratory Press, 1997, pp229-362 19. R. Kollar, B.B. Reinhold, E. Petrakova, H.J.C. Yeh, G. Ashwell, J. Drgonova, J.C. Kapteyn, F.M. Klis, and E. Cabib, J. Biol. Chem., 1997,272, 17762-17775 20. K. Hamada, H. Terashima, M. Arisawa, and K. Kitada, J. Biol. Chem. 1998, 273, 2694626953 21, M. Bessler and P. Hillmen S K ~Hemuto/., . 1998, 35, 149-167 22. V.C. Taylor, M. Sims, S. Brett and M.C. Field Biochem. J. 1997, 322, 919-925 23. W. van? Hof, E. Rodriguez-Boulan and A.K. Menon J. Biol. Chem. 1995,270, 24150-24155 24. N. Singh, L.-N. Liang. M.L. Tykocinski and A.M. Tartakoff J. Bid. Cliem. 1996, 271, 1287912884 25. N.A. Baumann, J. Vidugiriene, C.A. Machamer and A.K. Menon J. Bid. Chem. 2000 in press. 26. M.J. McConville and A.K. Menon. Mol. Memb. Biol. 2000, in press. 27. M.G. Low and A.R. Saltiel Science 1988,239, 268-215 28. S.D. Tachado, P. Gerold, R. Schwarz, S. Novakovic, M. McConville and L. Schofield Proc. Nut1 Acad. Sci. U.S.A. 1997, 94, 4022-4021 29. S. Udenfriend and K. Kodukula Annu. Rev. Biochem. 1995, 64, 563-591 30. T.L. Doering and R. Schekman EMBO J. 1996,15, 182-191 31. M.C. Field, P. Moran, W. Li, G.A. Keller and 1.W. Caras J. Biol. Chem. 1994, 269, 1083010837 32. B. Wilbourn, D.N. Nesbeth, L.J. Wainwright and M.C. Field Biochem J. 1998, 332, 111-1 18 33. M.A. McDowell, D.M. Ransom and J.D. Bangs Biochern. J. 1998, 335, 681-689 34. A. Horvath, C. Siitterlin, U. Manning-Krieg, N.R. Movva and H. Riezman EMBO J. 1994, 13, 3687-3695 35. C. Siitterlin, T.L. Doering, F. Schimmoller, S. Schroder and H. Riezman J. Cell Sci. 1997, /lo, 2703-2714 36. T. Weimbs, S.H. Low, S.J. Chapin and K.E. Mostov Trends Cell Biol. 1997, 7, 393--399 37. K. Simons and G. van Meer Biochemistry 1988, 27, 6197-6202 38. A. Rietveld and K. Simons Biochim. Biophys. Acta 1998. 1376, 467-479 39. P. Keller and K. Simons J. Cell Biol. 1998, 140, 1357-1367 40. D.A. Brown and J.K. Rose Cell 1992, 68, 533- 544 41. T. Friedrichson and T.V. Kurzchalia Nature 1998, 394, 802-805 42. R.J. Schroeder, S.N. Ahmed, Y. Zhu, E. London and D.A. Brown J. Biol. Chem. 1998, 273, 1150-1157

Refirences

769

43. S. Mayor, S. Sabharanjak and F.R. Maxfield EMBO J . 1998,17,4626-4638 44. A. Taraboulos, M. Scott. A. Semenov, D. Avraham, L. Laszlo and S.B. Prusiner J. Cell Biol. 1995, 129, 121-132 45. D. Bray Annu. Rev. Biophys. Bionzol. Struct. 1998: 27, 59-75 46. T. Harder, P. Scheiffele, P. Verkade and K. Simons J. Cell Bid. 1998, 141, 929-942 47. R. Varma and S. Mayor Nature 1998, 394, 798-801 48. E. Sheets, G.M. Lee, R. Simson and K. Jacobson Biochernistry 1997, 36, 12449-12458 49. M. Edidin and I. Stroynowski J. Cell Bid. 1991, 112, 1143-1 150 50. M. Edidin, S.C. Kuo and M.P. Sheetz Science 1991, 254, 1379-1381 51. R. Simson, B. Yang, S.E. Moore, P. Doherty, F.S. Walsh and K.A. Jacobson Biophys. J . 1998, 74, 297-308. 52. A.K. Kenworthy and M. Edidin J. Cell Bid. 1998, 142, 69-84 53. R.E. Brown J. Cell Sci. 1998, 111, 1L9 54. R.G.W. Anderson Annu. Reu Biochem. 1998, 67, 199-225 55. J.E. Schnitzer, D.P. Mclntosh, A.M. Dvorak. J. Liu and P. Oh Science 1995,269, 1435-1439 56. R.-V. Stan, W.G. Roberts, D. Predescu, K. Ihida, L. Saucan, L. Ghitescu and G.E. Palade Mol. Bid. Cell 1997; 8, 595-605 57. A.M. Fra, E. Williamson, K. Simons, and R.G. Parton J. Biol. Chem. 1994,269, 30745-30748 58. R.G. Parton and K. Simons Science 1995, 26Y. 1398-1399 59. I. Stefanova, V. Horejsi, I.J. Ansotegui, W. Knapp and H. Stockinger Science 1991,254, 10161019 60. K.A. Melkonian, A.G. Ostermeyer. J.Z. Chen, M.G. Roth and D.A. Brown J. Bid. Chem. 1999, 274, 3910--3917 61. J.M. Green, A.D. Schreiber and E.J. Brown J.C't4 Bid. 1997, 139, 1209-1217 62. T.M. Stulnig, M. Berger, T. Sigmund, H. Stockinger, V. Horejsi and W. Waldhausl J. Biol. Chern. 1997, 272, 19242-~19247 63. M.G. Low, Lipid Modifications of Pvofeins: A Pructicul Approuch, IRL Press, 1992, pp 117154 64. J.T. Deng, M.F. Hoylaerts. M.E. De Broe and V.O. van Hoof Am J. Physiol. 1996, 271, 6655-6663 65. S. Ilangumaran, P.J. Robinson and D.C. Hoessli Trends Cell Biol. 1996, 6 , 163-~167 66. P. Biitikofer, F.A. Kuypers, C.M. Xu. D.T.Y. Chiu and B. Lubin Blood 1989, 74, 1481-1485 67. D.R. Jones, and I. Varelo-Nieto lnt. J. Biocheni. Cell Bid. 1998, 30, 313-326 68. D.R. Lazar, J.J. Knez. M.E. Medof. P. Cuatrecasas, and A.R. Saltiel Proc. Nut1 Acud. Sci. USA 1994, 91, 9665-9669 69. S. Movahedi and N.M. Hooper Biochen?. J . 1997, 326. 531-537 70. G. Romero, L. Luttrell, A. Rogol, K. Zeller, E. Hewlett and J. Larner Science 1988, 240, 50951 1

71. 72. 73. 74. 75. 76.

W. Frick, A. Bauer, J. Bauer, S. Wied and G. Miiller Biocheni. J. 1998. 336, 163-181 M.G. Low J. Lub. Clin. Med., 1998, 131, 189-191 N.M. Hooper Mol. Mernb. Biol. 1999, 16, 1455156 M. Mutiiz and H. Riezman EMBO J. 2000, 19, 10- 15 M.A.J. Ferguson J. Cell Sci. 1999, 112, 2799--2809 E.J. Smart, G.A. Graf, M.A. McNiven, W.C. Sessa, J.A. Engelman, P.E. Scherer, T. Okamoto and M.P. Lisanti Mol. Cell Bid. 1999, 19, 7289-7304

Carbohydrates in Chemistry and Biology Beat Emst, Gerald W. Hart, Pierre Sinay copyright Q WILEY-VCH Verlag GmbH. 2000

47 Glycosphingolipid Microdomains in Signal Transduction, Cancer, and Development Sen-itiroh Hakomori and Kazuko Handa

47.1 Clustered GSLs as Functional Units An obvious and well-established function of glycosphingolipids (GSLs) is to serve as major cell surface antigens, e.g. allogeneic, developmentally regulated, and tumor-associated antigens (Table 1, item 1). Another role is as receptors for bacterial or viral toxins, mediating the process of microbial infection (Table 1, item 2). A less well-defined function of GSLs is to mediate cell adhesion/recognition through carbohydrate-binding protein (lectin) or carbohydrate-carbohydrate interaction (Table 1, item 3). Each of these GSL functions (antigenicity, receptor activity, or cell binding) is strongly influenced by ceramide composition of GSL, which may affect the degree of GSL clustering at the cell surface. The effect of GSL clustering (i.e. density of GSL per unit membrane protein or phospholipid) on GSL antigenicity has been well studied for tumor-associated GSL antigens. Certain monoclonal antibodies (mAbs) established after immunization of mice with syngeneic tumor cells react with specific GSLs when GSL density is above a certain threshold value, but not when density is below threshold value [ l , 21. Similarly, interaction of Gb3 with verotoxin [3, 41, or GM3-Gg3 interaction as a basis of melanoma cell adhesion to endothelial cells [ 5 ] , are highly dependent on density of GSL presented in the system. Thus, GSL function in general should be considered in terms of clustered groups rather than individual GSL molecules dispersed randomly at the cell surface. GSL clustering has been observed in classic studies by scanning electron microscopy (EM) with freeze-fracture technique using ferritin-labeled or gold sol-labeled anti-GSL mAbs [6-81, and more recently by transmission EM using gold sollabeled anti-GSL mAbs [9, 101. Results indicate that: i) GSL clusters are separate from glycoprotein clusters at the cell surface; ii) GSL clusters occur even at phospholipid liposome surface in the absence of cholesterol, i.e. cholesterol is not an essential component for GSL clustering, and GSLs per se have a strong ability to cluster.

172

47 Glycosphingolipid Microdomains

Table 1. Glycosphingolipid functions. GSL antigens Histo-blood group ABH, Lewis, I/i antigens, Forssman, H-D antigens [38] Developmentally-regulated antigens, e.g. SSEA-1, SSEA-3, SSEA-4 Tumor-associated antigens, e.g. GD2, GD3 (human melanoma), GM3 (mouse melanoma), globo-H (breast carcinoma), Le", Ley, SLe" (gastrointestinal, colorectal, lung carcinoma) [39] GSL receptors Choleratoxin (GMI) 140, 411 Shigella or verotoxin (Gb3) [42, 431 E. coli adhesin and other bacterial adhesins (globo-series) 1441 GSLs in cell adhesion/ recognition i. Mediated by GSL-binding protein GM3 binding protein mediating ovarian granulosa cell interaction [45] GDla, GTlb, GTl p, and GQ1 ba-mediated adhesion to myelin-associated glycoprotein (MAG) [461 Myeloglycan binding to selectin [47] GMl -galectin-1 binding 1481 ii. Mediated bv GSL-GSL interaction GM3-LacCer/Gg3 interaction mediating melanoma adhesion to endothelial cells [ 5 , 211 Gb4-GalGb4/nLcd interaction mediating human teratocarcinoma cell interaction [20] Gal-sulfatide interaction [49] Lex-Lexinteraction mediating mouse embryogenesis [50] GSLs controlling signal transduction: see this Chapter.

This is contrary to the generally accepted idea that sphingolipid microdomains are composed of sphingolipids and cholesterol (for review see [ 111. In fact, the GSL microdomain is insensitive to cholesterol-binding reagents such as filipin and nystatin, and is separable from caveolae (see following two sections).

47.2 GSL Clusters, Associated with Signal Transducers, are Functional Units Separable from Caveolae GSL clusters associated with c-Src, Src family kinases, Ras, Rho, FAK, and/or Csk were found to be insoluble in media containing non-ionic detergents, or buffer with high ionic strength, and to be separable following homogenization of cells in these media followed by gradient centrifugation, e.g. [12]. They were separated as a lowdensity, light-scattering fraction. Types of GSL associated with signal transducers in this fraction, isolated from various types of cells, are listed in Table 2. Based on a number of studies, membrane fraction separated by a similar procedure was claimed to be essentially the same as caveolae (for review see [13, 141, al-

47.3 Cell Adhesion Coupled with Signal Transduction

773

Table 2. GSLs and transducers associated with glycosignaling domain (GSD) of tumor cell lines. Cell line

Major GSLs

Signal transducers associated

melanoma B16

GM3

mammary carcinoma F28-7 human embryonal carcinoma 2102 neuroblastoma Neuro2a promyelocytic leukemia HL60

LacCer

c-Src, Rho A, Ras, FAK, paxillin (no c-Met) c-Src and uncharacterized tyrosine phosphorylated component c-Src, Rho A, FAK

Gb4, GalGb4, n L q GMl , GM3 GM3, n L q

c-Src, Csk, Rho A, Ras (no Trk A; no p75NTR) c-Src and uncharacterized tyrosine phosphorylated component

though this concept turned out to be incorrect. Caveolae, invaginations of plasma membrane, are characterized by the presence of scaffold protein caveolin [ 151, glycosylphosphatidylinositol (GPI) anchored protein [ 121, various signal transducer molecules, and possibly growth factor receptor [16, 171. They play a major role in endocytosis and signal transduction. Our studies indicate clearly that GSL microdomain (representing GSL clusters), containing c-Src and other signal transducers, can be separated from caveolincontaining fraction [ 181. The microdomain may also contain chloroform/ methanolsoluble “proteolipid” protein with low molecular mass (-15 kDa), showing coimmunoprecipitation with GSLs 1481. The contrasting properties of caveolae and GSL microdomain isolated from mouse melanoma B16 cells are summarized in Table 3.

47.3 Cell Adhesion Coupled with Signal Transduction Initiated by GSL Microdomain: Concept of Glycosignaling Domain (GSD) The presence of GSL microdomain facilitates cell adhesion to GSL-coated plate, based on GSL-GSL interaction which is synergistic with adhesion mediated by other receptors (e.g. integrins) 1191. Examples are listed in Table 1, item 3 3 . GSLdependent adhesion initiates signal transduction leading to activation of transcription factors API and CREB in human embryonal carcinoma 2102 cells 1201. GM3-dependent adhesion of B16 melanoma cells to Gg3-coated plate, which enhances cell motility [21], is associated with activation of focal adhesion kinase (FAK) and enhanced GTP binding to Rho A and Ras [22]. Thus, GM3 in melanoma cell microdomain, when stimulated by Gg3 ligand, activates signal transducers leading to enhanced cell motility, although the mechanism remains to be elucidated. The initial stage of transducer change is phosphorylation of c-Src (within three minutes) following GM3-dependent adhesion or addition of anti-GM3

774

47 Glycosphingolipid Microdomains

Table 3. Differences between GSL microdomain (glycosignaling domain) and caveolae in B 16 melanoma cells, in terms of chemical composition and function. Glycosignaling domain

GM3 GlcCer sphingomyelin cholesterol caveolin c-Src, Rho, Ras Ras H PLP (proteolipid protein*) GSL-dependent cell adhesion c-Met receptor** integrin receptor effect of cholesterol binders filipin, nystatin

Caveolae

++t

+ ++ + ++ + ++

t

Jr ++t

++

-

-

no effect on adhesion and signaling

destroy structure, block signaling

* Chloroform-methanol soluble protein with low molecular mass (-1 5 kDa) showing binding affinity to GM3 [51]. ** Major tyrosine kinase receptor expressed in B 16 melanoma cells. Synonyms: hepatocyte growth factor receptor; scatter factor receptor. Present in high-density protein fraction (Fr. 12). mAb DH2. This event is followed by FAK activation [18]. A similar close association of GD3 ganglioside and Src family kinase Lyn was indicated by activation of Lyn in rat cerebellar cells by anti-GD3 mAb R24 within five minutes [23]. GSL microdomain can be separated from caveolin-containing fraction by antiGM3 mAb or anti-caveolin mAb. Fraction reactive with anti-GM3 is associated with c-Src, whereas caveolin-containing fraction does not contain GM3 nor c-Src but does contain Ras. Thus, signal transduction initiated by GM3 stimulation takes place through activation of associated c-Src but not through caveolae components. GM3-dependent adhesion of B16/F10 cells and FAK activation triggered by this adhesion were not affected by filipin or nystatin, cholesterol-binding reagents which abolish caveolae structure and function. This suggests the presence of a microdomain, clearly distinguishable from caveolae, which was termed “glycosignaling domain” (GSD) [ 181. Proteolipid proteins showing close association with GM3 are presumably present in GSD, contributing to maintenance of GSD structure and modulating signaling. Possible organization of GSD and its functional notion are illustrated in Figure 1.

47.4 Role of GSLs in Control of Growth Factor and Hormone Receptors: Possible Relationship with GSL Microdomain We observed that Fab antibody directed to GM3 arrests growth of GM3-expressing cells at the G1 phase of cell cycle [24], and that exogenous addition of GM3 arrests

B I

e Ab

lectin

C

/

signal Figure 1. Proposed organization of glycosignaling domain (GSD) in relation to its function. A: GSLs, which consist of ceramide and oligosaccharide having axes perpendicular to each other, tend to cluster in phospholipid bilayer to form “GSL patches.” Clusters of glycoproteins (Gp) are separate from GSL patches, as indicated by electron microscopy with freeze-fracture technique. Transducer molecules (TDa, TDb) are associated closely with GSLs. A hydrophobic “proteolipid” protein (PLP) (soluble in chloroform/methanol) spans both membrane layers, has affinity to GSLs, and may have affinity to transducers. B: GSL patches are the site where weak cell adhesion occurs initially, through GSL-GSL interaction. This process is enhanced synergistically with the effect of integrins (I), adhesion proteins (Ap), and sugar-binding proteins (S). C: Various ligands (complementary GSLs, antibodies (Ab), and lectins) to GSL patches stimulate associated transducers, alter conformation of transducers, and initiate signal transduction.

776

47 Glycosphingolipid Microdomains

fibroblast growth factor (FGF)-dependent cell growth through inhibition of FGF internalization [25]. Since then, a series of studies focused on effect of gangliosides on growth factor or hormone receptors revealed that GM 1 inhibits platelet-derived growth factor (PDGF) tyrosine kinase [26], and GM3 inhibits epidermal growth factor (EGF) tyrosine kinase [27]. Insulin receptor tyrosine kinase is inhibited by sialosylparagloboside [28], while Trk A tyrosine kinase in PC12 cells is promoted strongly by GM 1 [29]. Glucosylceramide may induce proliferation of keratinocytes [30], and perhaps other cell types [31]. Lactosylceramide (LacCer) in arterial smooth muscle cells enhances Ras-GTP binding to promote mitogen-activated protein kinase (MAPK) activation leading to proliferation [32]. LacCer also mediates TNFa-induced intercellular adhesion molecule-1 (ICAM-1) expression and promotes neutrophil adhesion [33]. Effects of GSL in modulation of signal transduction were reviewed recently [ 341. Despite a few claims that growth factor receptors are associated with caveolae, and the fact that tyrosine kinase phosphorylation of receptor in caveolae is rapidly induced upon PDGF or EGF stimulation of cells [ 16, 171, growth factor receptors are absent or barely detectable in GSL microdomain. A major question therefore remains: How do GSLs affect growth factor receptor function? We have no information as to whether: i) GSLs interact with growth factor receptors within caveolae, separate from glycosignaling domain; or ii) growth factor receptors are transiently translocated into glycosignaling domain to have their function modulated. Possible interactions of receptors with GSLs, inside and outside GSL microdomain, are illustrated schematically in Figure 2.

47.5 Functional Role of Developmentally-Regulated and Tumor-Associated GSLs Synthesis and expression of GSLs are altered dramatically during development and oncogenic transformation [35]. High expression of a tumor-associated GSL antigen in tumors, but not in mature differentiated normal cells, represents retrograde expression of the antigen, i.e. mimics its expression in immature or fetal stages of development. Another feature of tumor cells, particularly those with high malignancy, is the loss of cellular polarity. Normal, non-transformed cells, with the exception of blood cells, have a clearly polarized structure: i) surface membrane structure involved in cell-to-cell or cell-to-extracellular matrix adhesion; ii) surface structure open to the external environment and involved in endocytosis or exocytosis. Structure (i) in epithelial cells, termed “basolateral surface,” is rich in adhesion receptors including integrins (ii), termed ‘apical surface,” has much higher level of sphingolipids,

47.5 Role of Developrnentully-Reguluted and Tumor-Associated GSLs

Ill

2 phosphorylation

receptors cluster close to GSL microdomain, cause signaling

3

$ 4

interaction of GSLs with receptors leads to signal modification Figure 2. Proposed interaction of growth factor receptors with GSLs in microdomain. GSL microdomain does not contain growth factor receptors or integrin receptors. Receptors may be dispersed or present in loci separate from GSLs in resting state (step 1). However, function of these receptors is susceptible to GSLs, particularly gangliosides. Receptors may migrate to the GSL microdomain upon cell stimulation (step 2), and some of them enter the microdomain (step 3), whereupon receptor function is influenced by GSLs.

GPI-anchored proteins, and perhaps growth factor receptors [ 361. Structure (ii) is obviously enriched in GSLs and GSD. The majority of tumor cells derived from epithelial cells are characterized by GSD distribution all over the cell surface, and no structural distinctions at different sites on the cell surface. Many tumor cells are characterized by the accumulation of tumor-associated GSL antigens, and by large GSD with multiple signal transducers, In contrast, normal adult cells have lower quantity of the specific GSL, and fewer kinds of transducer; e.g. Src family kinase is often found in normal cells [ 2 3 ] .Efficiency of signal transduction through GSD may be lower in normal than in tumor cells, although extensive further study is needed. The concept of transformation-dependent loss of polarity, and associated changes in GSD, is illustrated in Figure 3. As an example, GM3 is expressed at much lower level in normal melanocytes than in B16 melanoma cells, and is regarded as a melanoma-associated antigen. High GM3 expression provides for high-density clusters of GM3 associated with at least four signal transducers (c-Src, Ras, Rho, FAK), conferring highly susceptible status for signal transduction initiated by GM3. This mechanism is outlined in the preceding section. Other known examples of tumor-associated GSLs associated with signal transducers are listed in Table 2. In order to assess the functional role of GSLs in GSD, reagents which destroy

778

47 Glycosphingolipid Microdomains normal cells in normal tissues

A

cancer cells

1

BM

3

ECM

B 1. normalGSD

limited or inhibitory signaling

stirnulatory signaling

Figure 3. Proposed difference in GSD between normal and tumor cells. A: Normal epithelial cells in epithelial tissue (panel 1) and fibroblasts in connective tissue (panel 3) are characterized by polarized structure, i.e. GSD (indicated by black spots) is localized on the free (apical) surface or at the lateral surface where one cell contacts the next, but not on the surface in contact with basement membrane (BM) or extracellular matrix (ECM). In cancer cells, polarized structure is lost, i.e. GSD is distributed on all surfaces (panels 2, 4). B: GSD is qualitatively different in normal cells (panel 1) compared to cancer cells (2). In normal cells, types of transducers (TD) associated with GSD (G) are limited, and signal transduction through cell-cell interaction is limited or inhibitory. In cancer cells, many types of transducer (TDa, TDb, TDc) are associated with GSD: which are larger than those in normal cells, and cell-cell interaction induces stimulatory signaling to enhance cell motility and proliferation.

structure and function of GSD in tumor cells may be useful. Such reagents may stop tumor growth and the metastastic process, since GSD initiates signal transduction and may mediate tumor cell adhesion to target cells to initiate metastasis. As an example, GM3 highly expressed in GSD of mouse melanoma B16 mediates B16 cell adhesion to endothelial cells [ 5 ] . A preliminary experiment along this line, using synthetic lyso-GM3, indicate that this reagent, at subtoxic dose, is able to disrupt GSD structure and function, inhibit GM3-dependent cell adhesion, and block activation of FAK kinase. Synthetic lactosylsphingosine or dilactosylsphingosine had no such effect, indicating that the effect of lyso-GM3 is specific [37]. References 1. Nores GA, Dohi T, Taniguchi M, Hakomori S: Density-dependent recognition of cell surface GM3 by a certain anti-melanoma antibody, and GM3 lactone as a possible immunogen: Requirements for tumor-associated antigen and immunogen. J. Zmmunol. 1987; 139: 3171-3176.

References

119

2. Hakomori S, Nudelman ED, Levery SB, Solter D, Knowles BB: The hapten structure of a developmentally regulated glycolipid antigen (SSEA-1) isolated from human erythrocytes and adenocarcinoma: A preliminary note. Biochem. Biophys. Rex Commun. 1981; 100: 1578- 1586. 3. Boyd B, Magnusson G, Zhiuyan Z, Lingwood CA: Lipid modulation of glycolipid receptor function: Availability of Gal(a1-4)Gal disaccharide for verotoxin binding in natural and synthetic glycolipids. Eur. J. Biochem. 1994; 223: 873-878. 4. Kiarash A, Boyd B, Lingwood CA: Glycosphingolipid receptor function is modified by fatty acid content: Verotoxin 1 and verotoxin 2c preferentially recognize different globotriaosyl ceramide fatty acid homologues. J. Biol. Chem. 1994; 269: 11138-1 1146. 5. Kojima N, Shiota M, Sadahira Y, Handa K, Hakomori S: Cell adhesion in a dynamic flow system as compared to static system: Glycosphingolipid-glycosphingolipid interaction in the dynamic system predominates over lectin- or integrin-based mechanisms in adhesion of B16 melanoma cells to non-activated endothelial cells. J. Biol. Chem. 1992; 267: 17264-17270. 6. Tillack TW, Allietta M, Moran RE, Young WWJ: Localization of globoside and Forssman glycolipids on erythrocyte membranes. Biochim. Biophys. Actu 1983; 733: 15-24. 7. Rock P, Allietta M, Young WWJ, Thompson TE, Tillack TW: Organization of glycosphingolipids in phosphatidylcholine bilayers: Use of antibody molecules and Fab fragments as morphologic markers. Biochemistry 1990; 2 9 8484-8490. 8. Rock P, Allietta M, Young WWJ, Thompson TE, Tillack TW: Ganglioside G M and ~ asialoG M Iat low concentration are preferentially incorporated into the gel phase in two-component, two-phase phosphatidylcholine bilayers. Biochemistry 1991; 30: 19-25. 9. Rahmann H, Rosner H, Kortje K-H; Beitinger H, Veybold V: Ca2+-gangliosideinteraction in neuronal differentiation and development. In: Svennerholm L, Asbury AK, Reisfeld RA, et al., eds. Biological function of gungliosides (Progress in Brain Research, Vol. 101). Amsterdam: Elsevier, 1994; 127-145. 10. Sorice M, Parolini I, Sansolini T, et al.: Evidence for the existence of ganglioside-enriched plasma membrane domains in human peripheral lymphocytes. J. Lipid Res. 1997; 38: 969-980. 11. Simons K , Ikonen E: Functional rafts in cell membranes. Nature 1997; 387: 569-572. 12. Brown DA, Rose JK: Sorting of GPI-anchored proteins to glycolipid-enriched membrane subdomains during transport to the apical cell surface. Cell 1992; 68: 533-544. 13. Parton RG, Simons K: Digging into caveolae. Science 1995; 269: 1398-1399. 14. Brown DA, London E: Structure of detergent-resistant membrane domains: Does phase separation occur in biological membranes? Biochem. Biophys. Res. Commun. 1997; 240: 1-7. 15. Okamoto T, Schlegel A, Scherer PE, Lisanti MP: Caveolins, a family of scaffolding proteins for organizing “preassembled signaling complexes” at the plasma membrane. J. Biol. Chem. 1998; 273: 5419-5422. 16. Liu P, Ying Y, KO Y-G, Anderson RGW: Localization of platelet-derived growth factorstimulated phosphorylation cascade to caveolae. J. Biol. Chem. 1996; 271: 10299-10303. 17. Mineo C, James GL, Smart EJ, Anderson RGW: Localization of epidermal growth factorstimulated Ras/Raf-1 interaction to caveolae membrane. J. Biol. Chem. 1996; 271: 1193011935. 18. Iwabuchi K, Handa K, Hakomori S: Separation of “glycosphingolipid signaling domain” from caveolin-containing membrane fraction in mouse melanoma B16 cells and its role in cell adhesion coupled with signaling. J. Biol. Chem. 1998; 273: 33766-33773. 19. Kojima N, Hakomori S: Synergistic effect of two cell recognition systems: Glycosphingolipidglycosphingolipid interaction and integrin receptor interaction with pericellular matrix protein. Glycobioloyy 1991; 1: 623-630. 20. Yu S, Withers DA, Hakomori S: Globoside-dependent adhesion of human embryonal carcinoma cells, based on carbohydrate-carbohydrate interaction, initiates signal transduction and induces enhanced activity of transcription factors API and CREB. J. Biol. Chem. 1998; 273: 2517-2525. 21. Kojima N, Hakomori S Cell adhesion, spreading, and motility of GMj-expressing cells based on glycolipid-glycolipid nteraction. J. Bid. Chem. 1991; 266: 17552-17558. 22. Iwabuchi K, Yamamura S, Prinetti A, Handa K, Hakomori S: GM3-enriched microdomain involved in cell adhesion and signal transduction through carbohydrate-carbohydrate interaction in mouse melanoma B16 cells. J. Biol. Chem. 1998; 273: 9130-9138.

780

47 GIycosphingolipid Microdomains

23. Kasahara K, Watanabe Y, Yamamoto T, Sanai Y: Association of Src family tyrosine kinase Lyn with ganglioside G D in ~ rat brain: Possible regulation of Lyn by glycosphingolipid in caveolae-like domains. J. Biol. Chem. 1997; 272: 29947-29953. 24. Lingwood CA, Hakomori S: Selective inhibition of cell growth and associated changes in glycolipid metabolism induced by monovalent antibodies to glycolipids. Exp. Cell Res. 1977; 108: 385-391. 25. Bremer EG, Hakomori S: GM1 ganglioside induces hamster fibroblast growth inhibition in chemically-defined medium: Ganglioside may regulate growth factor receptor function. Biochem. Biophys. Rex Commun. 1982; 106: 711-718. 26. Bremer EG, Hakomori S, Bowen-Pope DF, Raines EW, Ross R: Ganglioside-mediated modulation of cell growth, growth factor binding, and receptor phosphorylation. J. Biol. Chem. 1984; 259: 6818-6825. 27. Bremer EG, Schlessinger J, Hakomori S: Ganglioside-mediated modulation of cell growth: Specific effects of GM3 on tyrosine phosphorylation of the epidermal growth factor receptor. J. Biol. Chem. 1986; 261: 2434-2440. 28. Nojiri H, Stroud MR, Hakomori S: A specific type of ganglioside as a modulator of insulindependent cell growth and insulin receptor tyrosine kinase activity: Possible association of ganglioside-induced inhibition of insulin receptor function and monocytic differentiation induction in HL60 cells. J. Biol. Chem. 1991; 266: 4531-4537. 29. Mutoh T, Tokuda A, Miyada T, Hamaguchi M, Fujiki N: Ganglioside GMI binds to the Trk protein and regulates receptor function. Proc. Natl Acad. Sci. USA 1995; 92: 5087-5091. 30. Holleran WM, Ginns EI, Menon GK, et al.: Consequences of p-glucocerebrosidase deficiency in epidermis: Ultrastructure and permeability barrier alterations in Gaucher disease. J. Clin. Invest. 1994; 93: 1756-1764. 31. Shayman JA, Deshmukh GD, Mahdiyoun S, et al.: Modulation of renal epithelial cell growth by glucosylceramide: Association with protein kinase C, sphingosine, and diacylglycerol. J. Biol. Chem. 1991; 266: 22968-22974. 32. Bhunia AK, Han H, Snowden A, Chatterjee S: Lactosylceramide stimulates Ras-GTP loading, kinases (MEK, Raf), p44 mitogen-activated protein kinase, and c-fos expression in human aortic smooth muscle cells. J. Biol. Chem. 1996; 271: 10660-10666. 33. Bhunia AK, Arai T, Bulkley G, Chatterjee S: Lactosylceramide mediates tumor necrosis factora-induced intercellular adhesion molecule-1 (ICAM-1) expression and the adhesion of neutrophil in human umbilical vein endothelial cells. J. Biol. Chem. 1998; 273: 34349-34357. 34. Yates AJ, Rampersaud A: Sphingolipids as receptor modulators: An overview. In: Ledeen RW, Hakomori S, Yates AJ, et al., eds. Sphingolipids as signaling nzodulutors in the nervous system, vol845. New York, NY: New York Acad Sci, 1998; 57-71. 35. Hakomori S, Kannagi R: Glycosphingolipids as tumor-associated and differentiation markers. J. Natl Cancer Inst. 1983; 71: 231-251. 36. Simons K, van Meer G: Lipid sorting in epithelial cells. Biochemistry 1988; 27: 6197-6202. 37. Iwabuchi K, Zhang Y, Handa K, Hakomori S: Lyso-GM3 specifically disrupts structure and function of GM3-enriched glycosignaling domain in B16 melanoma cells [Abstract]. FASEB J. 1999; in press. 38. Hakomori S, Young WWJ: Glycolipid antigens and genetic markers. In: Kanfer JN, Hakomori S, eds. Hundbook of lipid research 3: Sphingolipid biochemistry. New York: Plenum Press, 1983; 381-436. 39. Hakomori S: Cancer-associated glycosphingolipid antigens: Their structure, organization, and function. Acta Anatomica 1998; 161: 79-90. 40. Holmgren J, Lonnroth I, Mansson J-E, Svennerholm L: Interaction of cholera toxin and membrane GM 1 ganglioside of small intestine. Biochemistry 1975; 72: 2520-2524. 41. Fishman PH, Brady RO: Biosynthesis and function of gangliosides: Gangliosides appear to participate in the transmission of membrane-mediated information. Science 1976; 194: 906-91 5. 42. Jacewicz M, Clausen H, Nudelman E, Donohue-Rolfe A, Keusch GT: Pathogenesis of Shigella diarrhea: XI. Isolation of a Shigella toxin-binding glycolipid from rabbit jejunum and HeLa cells and its identification as globotriaosylceramide. J Exp Med 1986; 163: 1391-1404. 43. Lingwood CA, Law H, Richardson S, et al.: Glycolipid binding of purified and recombinant Escherichia coli produced verotoxin in vitro. J Biol Chem 1987; 262: 8834-8839.

Rejevences

78 1

44. Karlsson K-A: Animal glycosphingolipids as membrane attachment sites for bacteria. Annu. Rev. Biochem. 1989; 58: 309-350. 45. Hattori M, Horiuchi R, Hosaka K, Hayashi H, Kojima I: Sialyllactose-mediated cell interaction during granulosa cell differentiation. J. Bid. Chem. 1995; 270: 7858-7863. 46. Yang LJ-S, Zeller CB, Shaper NL, et al.: Gangliosides are neuronal ligands for myelinassociated glycoprotein. Proc. Nut1 Acad. Sci. USA 1996; 93: 814-818. 47. Handa K, Stroud MR, Hakomori S: Sialosyl-fucosyl poly-LacNAc without the sialosyl-LeX epitope as the physiological myeloid cell ligand in E-selectin-dependent adhesion: Studies under static and dynamic flow conditions. Biochemistry 1997; 36: 12412- 12420. 48. Kopitz J, von Reitzenstein C, Burchert M, Cantz M, Gabius HJ: Galectin-1 is a major receptor for ganglioside G M l , a product of the growth-controlling activity of a cell surface ganglioside sialidase, on human neuroblastoma cells in culture. J. Biol. Chem. 1998; 273: 11205-1 1211. 49. Stewart RJ, Boggs JM: A carbobydrate-carbohydrate interaction between galactosylceramidecontaining liposomes and cerebroside sulfate-containing liposomes: Dependence on the glycolipid ceramide composition. Biochemistry 1993; 32: 10666-10674. SO. Eggens I, Fenderson BA, Toyokuni T, Dean B. Stroud MR, Hakomori S: Specific interaction between LeX and Le" determinants: A possible basis for cell recognition in preimplantation embryos and in embryonal carcinoma cells. J. Biol. Chem. 1989; 264: 9476-9484. 51. Handa K : unpublished data.

Carbohydrates in Chemistry and Biology Beat Emst, Gerald W. Hart, Pierre Sinay copyright Q WILEY-VCH Verlag GmbH. 2000

48 The Primary Cell Walls of Higher Plants Jocelyn K. C, Rose, Malcolm A. O’Neill, Peter Albersheim, and Alan Darvill

48.1 Introduction (What is a Cell Wall?) All plant cells are surrounded by a wall. In growing cells, meristematic cells and the cells in succulent tissues, such as fruit, the cells are surrounded by a polysaccharide rich extracellular matrix (0.1 pm-l pm thick) termed the primary wall (see Figure 1). Primary walls were for many years considered to provide mechanical strength by being rigid inert structures. This view has gradually changed and the primary wall is now regarded as a complex and dynamic organelle that influences many aspects of plant growth and development. Several years ago a debate occurred between plant scientists concerning the relationship of a plant cell to its wall-is the wall an “extracellular” matrix or “intracellular but extraprotoplasmic”? A proponent of the “extracellular matrix”, stated that

“. . . whereas removal of the plasma membrane kills a cell, removal of its cell wall does not. For this reason, we cannot consider the cell wall an essential part of a plant cell, the basic unit of plant life” [95]. This is true since the wall can be removed and the wall-less cell (a protoplast) does not die. However, the protoplast cannot divide and is not a fully functioning cell. Higher plants without walls, if they existed at all, would be rather unappealing entities. How different our world would be. No more the majesty of the rain forest or the beauty of the meadow in springtime-Wordsworths’ “host of golden daffodils” would not be “fluttering and dancing in the breeze”. The primary wall as seen through a low-power microscope is a thin, featureless layer. With greater magnification and more sophisticated techniques certain details can be observed. However, there is a limit to what can be observed through a microscope. Walls can be isolated and components analyzed. However, a wall, unlike an enzyme, does not have a specific activity that can be assayed and the primary wall, unlike other organelles such as mitochondria or chloroplasts, is not delimited

784

48 The Primary Cell Walls of Higher Plants

Figure 1. (A). A diagrammatic representation of a plant cell. The cytoplasm is bounded by the plasma membrane which is itself surrounded by the primary wall. The region where the primary walls of two cells abut one another is called the middle lamella. Air spaces are often present particularly at the junction of cell corners. Abbreviations used-Pla = plastid; Chl = chloroplast; Mit = mitochondrion; Nuc = nucleus; Vac = vacuole; Cyt = cytoplasm; Pdm = plasmodesmata. (B). A simplified model of a possible arrangement of the major wall components. Wall proteins have been omitted for clarity. Abbreviations used: HG = homogalacturonan; RG-I = rhamnogalacturonan I; RG-I1 = rhamnogalacturonan 11; B-ester = borate ester cross-link of RG-11; XG xyloglucan

by a distinct boundary. An isolated “wall” is not a wall-it is the wall material that remains after a series of extractions. We predict that as more cell biologists, chemists, biochemists, and molecular biologists take up the challenge our knowledge of the components, properties and functions of cell walls will increase dramatically. We briefly describe in this Chapter what is known about the structures of the major components of primary walls, how these components are organized, and some of the potential functions of primary walls in plant growth and development.

48.2 Purification of Cell Walls and Isolation of Wall Components Primary cell walls have been isolated from suspension-cultured plant cells, and from stems, leaves, fruits, seeds, and storage organs of plants [ 11. Suspension-cultured

48.2 Puvijiication of Cell Wulls und Isolation o j Wull Cor?zponents

185

plant cells can be obtained as an homogeneous tissue and thus can be a convenient, homogeneous source of walls. In contrast, the tissues of plant organs are composed of a mixture of different cell types and the composition of a wall preparation isolated from such tissue is a mixture of the wall types present. Moreover, wall composition changes during cell growth and development (e.g. during fruit ripening) and this must be considered when comparing walls isolated from plants at different stages of development. The isolation of cell walls involves procedures that remove most, if not all, of the cytoplasmic contents, while minimizing both the solubilization of wall components and the fragmentation of wall components by endogenous wall-degrading enzymes [ 11. The procedures used include homogenizing plant tissue in aqueous buffers/ detergents at low temperature followed by treatments with phenol/acetic acid/ water, with a-amylase or aqueous DMSO, and with organic solvents including chloroform/methanol. Less pure preparations of cell walls can be obtained by homogenizing plant tissue in aqueous 80% ethanol or methanol. “Walls” prepared in this manner are often contaminated with intracellular components including mitochondria, chloroplasts, nucleic acids, proteins, polyphenols, pigments, and starch. Numerous chemical and enzymic procedures have been used to solubilize the components of isolated primary walls [ 11. Pectic polysaccharides are solubilized by hot water, phosphate buffers, chelating reagents (ammonium oxalate, EDTA, or CDTA), and dilute alkali (50 mM sodium carbonate). Hemicelluloses are solubilized by treating walls with alkali (IM and 4M KOH, 4M KOH Na borate), 8M urea, or with guanidine thiocyanate. However, chemical treatments almost always solubilize a mixture of wall components and do not remove all of the target components. This difficulty can in part be overcome by using purified endo-glycanases to solubilize wall polymers [2]. These enzymes cleave specific glycosidic bonds and thereby solubilize a particular class of wall components. For example, endopolygalacturonase solubilizes pectic polysaccharides whereas endoglucanases or endoxylanases solubilize oligosaccharide fragments of xyloglucans and arabinoxylans, respectively. Significant amounts ( 10%) of pectin, hemicellulose, and glycoprotein are often present even after a wall has been exhaustively treated with enzymes and chemicals. This has led to the suggestion that some wall components are covalently linked to or entangled with one another. The amount of a polysaccharide solubilized is, in large part, dependent on the tissue irrespective of the method of solubilization. More importantly, the solubilized material is rarely, if ever, a single polysaccharide type. Thus, chemical and chromatographic procedures are required to separate and purify the individual classes of solubilized polymers [I]. Pectic polysaccharides have been fractionated by the formation of their insoluble quaternary ammonium salts, and by size-exclusion and anion-exchange chromatographies. Hemicelluloses have been fractionated by alcohol precipitation, by the formation of copper, barium, and iodine complexes, and by size-exclusion, and anion-exchange chromatographies [I]. Various combinations of these procedures are typically used to obtain homogeneous polysaccharide preparations.

+

786

48 The Primary Cell Walls of Higher Plants

48.3 The Structural Components of the Primary Cell Wall 48.3.1 Cell Walls and the Diversity of Flowering Plants

The angiosperms (flowering plants; Division Magnoliophyta) occupy more habitat types, have the greatest number of species (-235,000) of any plant group, and range in size from the small aquatic duckweed ( WoZfJia, 1-2 mm) to large trees (Eucalyptus, 100 m) [3]. Angiosperms have been divided into two major classes: the dicotyledons (Magnoliopsida) and monocotyledons (Liliopsida). The dicotyledons, which comprise 170,000 species, are divided into six sub-classes (Magnolidae, Hamamelidae, Caryophillidae, Dilleniidae, Rosidae, and Asteridae) whereas the monocotyledons, which comprise -65,000 species, are divided into five subclasses (Alismatidae, Arecidae, Commelinidae, Zingiberidae, and Liliidae). Perhaps not unexpectedly, the primary walls of only a limited number of plant species have been analyzed. The species examined are typically of economic importance (e.g. tomato and corn), convenient to grow (e.g. suspension-cultured cells), or suitable for genetic manipulation (e.g. Arabidopsis). Some general features of the composition and structure of primary walls have emerged. It has been proposed that there are two general types of primary wall based on the structure and amounts of hemicellulosic polysaccharides and the relative amounts of the pectic polysaccharides [4].Type I walls, which typically contain xyloglucan and 20-35% pectin, are found in all dicotyledons (e.g tomato, apple, soybean) and in some monocotyledons (Liliidae e.g. onion and garlic). Type I1 walls, which contain glucuronarabinoxylan and
-

48.3.2 The Structural Components of the Primary Wall The cumulative results of studies over the last 30 years have led to the hypothesis that the primary walls of angiosperms and gymnosperms are composed of cellulose, the hemicelluloses xyloglucan and arabinoxylan, and the pectic polysaccharides homogalacturonan, rhamnogalacturonan I (RG-I), and rhamnogalacturonan I1 (RG-11), albeit in varying proportions [6]. For example, xyloglucan and arabinoxylan account for 20% and 5%1, respectively, of suspension-cultured dicotyledon walls whereas in cultured maize cells they account for 5% and 40%, respectively, of the wall. Homogalacturonan, RG-I, and RG-I1 together account for 20-35% of the walls of dicotyledons and non-graminaceous monocotyledons but account for
48.3 The Structural Components of the Primary Cell Wall

787

Cellulose microfibril

Figure 2. The 1,4-linked p-D-glucosyl residues that associate to form cellulose. Adjacent glucosyl residues are rotated 180" relative to each other. In naturally-occurring cellulose (Cellulose I) the reducing end of each glucan chain is at the same end of the microfibril. The formation of intra- and inter-molecular hydrogen bonds (* a a ) between the glucan chains is in large part responsible for the rigid structure of cellulose. In primary walls approximately 36 glucan chains associate to form a microfibril which have diameters of -15 nm. The highly ordered and partially crystalline microfibrils provide much of the tensile strength of the primary wall.

naturally occurring cellulose (cellulose I) the glucan chains are termed parallel since their reducing ends are all at the same end of the microfibrils (Figure 2). Xyloglucans also have a backbone composed of 1,4-1inked p-D-glucosyl residues. Approximately 75% of these residues are substituted at C-6 with a-D-Xylp, p-DGalp-(1,2)-a-~-Xylp,and a-~-Fucp-( 1,2)-P-~-Galp-( 1,2)-a-~-Xylp (Figure 3A). The galactosyl residues are often O-acetylated at C-6. Structural variation is apparent in xyloglucans from different species. For example, fucose has not been detected in the xyloglucans of Poaceae [7] and Solanaceae [8]. The xyloglucans of the Solanaceae, including both tobacco and tomato, contain a-~-Ara-( 1,2)-a-~-Xylp-side chains, although only tomato xyloglucan contains P-Araf-( 1,3)-a-AraJ1(I ,2)-u-~-Xylpside chains [8]. Arabinoxylans (Figure 3B) have a backbone composed of 1,4-linked S-D-Xylp residues-some of the Xylp residues may be O-acetylated [7]. Araf residues are or linked to position C-2 and/or C-3 of the backbone. In primary walls B-D-G~C~A 4-0-Me P-D-G~c~A residues are also attached to the xylan backbone and these polysaccharides are called glucuronarabinoxylans. The Araf side chains of arabinoxylan and glucuronarabinoxylan may also contain ester-linked phenolic acids such as ferulic acid [9] that are potential sites for cross-linking by oxidative coupling (Figure 3B and C ) .

188

48 The Primary Cell Walls of Higher Plants a-Fucp

A

4

p-Galp c OAc a-Xylp

a-Xylp

4

4

4 4

a-Xylp

6 6 6 P-Glcp(1-4)-P-Glcp(1-4)-P-Gl~p( 1-4)-P-Gl-(

1-4)-

a-Araf

4-O-Me-P-GlcpA

4

4

3 2 p-xylp(l-4)-p-xylp(l-4)-p-xylp( 1-4)-p-xylp(l-4)-p-xylp-(1-4)-p-xylp-( 12 AXG, OAc a-Araf B

t

t

c1,

C H , O H oI c H ,

0I

FMk OH

C-0

I

OH

0 I

AXG, Figure 3. (A). A partial structure of a xyloglucan. The 1,4-linked p-D-glucan backbone is substituted at C-6 with mono-, di-, and trisaccharides in a rather regular pattern. The xyloglucans present in the walls of the Solanaceae, Poaceae, and seed xyloglucans are not fucosylated. (B). A partial structure of a glucuronoarabinoxylan. The 1,4-linked P-D-xylan backbone is substituted with arabinosyl and (4-@Me) glucuronosyl residues, and 0-acetyl groups. Some of the arabinosyl residues are substituted with ester-linked phenolic acids such as ferulic acid. (C). Ferulic acid residues may be oxidativeiy coupled and thereby form inter- and intra-molecular cross links.

Homogalacturonan is a linear chain of 1,4-1inked a-D-GalpA residues in which many of the carboxyl groups are methyl esterified (Figure 4A). Homogalacturonan may, depending on the plant source, also be partially 0-acetylated [lo]. Rhamnogalacturonan I comprises a group of closely related pectic polysaccharides that contain a backbone of the repeating disaccharide 4)-a-~-GalpA-( 1,2)-a-~Rhap (1 in Figure 4B). Some of the backbone GalpA residues may be 0-acetylated on C-2 and/or C-3 but there is no evidence that the GalpA residues are methyl esterified [ 101. Some 20-80% of the Rhap residues are substituted at C-4 with neutral and acidic oligosaccharides, depending on the plant source and tissue type. The carbohydrate chains attached to the rhamnogalacturonan backbone are composed of linear and branched a-r.-Arafand P-D-Galp residues (2 and 3 in Figure 4B), although their relative proportions may differ depending on the plant source. Side chains attached by Arafto C-4 of Rhap in the RG-I backbone contain only Araf

48.3 The Structurul Components of the Primary Cell Wu11

A

Unesterified GalpA

189

Methyl esterified GalpA

COCH,

O-acetylated GalpA 6

1 4)-a-GalpA-( 1-2)-a-Rhap-( 1-4)-a-GalpA-( 1-2)-a-Rhap-( 1-

2

P-Galp-( 1-6)-p-Galp-( 1-6)-P-Galp-( 1-4)-a-Rhap-( 12

t

3

a-Araf-( 1-5)-a-Araf-( 1-2)-a-Araf-( 1-3)-P-Galp-( 1-4)-a-Rhap-( 1-

Jd

t’

4 4-O-Me-P-GlcpA-( 1-6)-Galp-( 1-

O-2-a-Araf-( 1-5)-a-Araf-( 1-

HO

5

Figure 4. (A). A partial structure of homogalacturonan. In primary walls between SO and 80% of the GalpA are esterified. (B). A partial structure of rhamnogalacturonan I (1); the backbone of RGI is composed of the disaccharide repeating unit -+4)-a-~-GalpA-( l+2)-a-~-Rhap-(l + . The GalpA residues are often 0-acetylated. Some of the rhamnosyl residues are substituted at C-4 with oligosaccharides composed predominantly of Araf and Galp residues (2 and 3). Some side chains contain (4-0-Me)GlcpA (4). In the Chenopodiaceae, the side chains are esterified with phenolic acids such as ferulic acid ( 5 ) . Oxidative coupling of the phenolic acids may be a way in which pectins are cross-linked.

residues. In contrast, side chains attached to the backbone by Galp may contain only Galp residues or a mixture of Galp and Arafresidues. The Araf residues are always on the exterior of these side chains. Small amounts (lL2%) of CX-L-FUC~, B-DGlcpA, and 4-0-Me P - D - G ~ c ~residues A may also be present (4 in Figure 4B). The number of glycosyl residues in the side chains is variable and may range from one to more than twenty glycosyl residues. The RG-I side chains of some plants may be esterified with phenolic acids (5 in Figure 4B) such as ferulic acid [9]. Rhamnogalacturonan I1 is a low molecular mass (-5-10 kDa) pectic polysaccharide [ 101. Unlike RG-I, RG-I1 does not have a backbone of the repeating disaccharide 4)-a-~-GalpA-( 1,2)-a-~-Rhap. Rhamnogalacturonan I1 is composed of eleven different glycosyl residues (see Figure 5A) including the unusual sugars 3-Ccarboxy-5-deoxy-~-~-xylose (aceric acid; AcefA), Apif, 2-0-Me Xylp, 2-0-Me

790

48 The Primary Cell Wulls of Higher Plants

A

a-Rhap

i (11) +'

KdoD

+

pAraf

p-Araf

i

+

Dhap

(111)

3 3 3 I)a-GalpA-(1-4)a-GalpA-(1-4)%-GalpA-(1-4)-a-GalpA-(1-rl)-a-GalpA-(1-4)-a-GalpA-(1-4)-a-GalpA-( 1-4)-a-GalpA-(1-

PApif- 4 3

I

4

4

p-GalpA

+ 3 0-Rhap2 +a-GalpA

BRhaD

4

4 a-Fuy, 3

+2Me-a-Xylp

(1) OAc 2Mea-Fucp Rhap?+-Araf

B

4

+2 a-Galp

$ pGlcpA2 +a-Galp

.*2 a-Rhap+2

i

a-Arap

Backbone of one RG-II molecule 4)-a-GalpA-(l-4)-a-GalpA-(l-4)-a-GalpA-( 1-4)-a-GalpA-(1-4)-a-GalpA-( 1-4)-a-GaIpA-(1-4)-a-GalpA-(l-4)-a-GalpA-(l. I I

4 __*

Borateesterified Apiosyl residues

o/Bb

HO

OH

Unesterified

HO

OH

Apiosylresidues

+

0 4)a-G~lpA-(l-4)-a-GalpA-(l-4)-a-GalpA-(1-4)-u-GalpA-(l-4)-a-GalpA-(1-4)-a-~l~-(l-4)-a-GalpA-(l-4)-a-GalpA-(l-

Backbone of second RG-II molecule Figure 5. (A). A partial structure of rhamnogalacturonan 11. The backbone of RG-11, released by endopolygalacturonase treatment of primary walls, is composed of between 8 and 15 4-linked a-D GalpA residues. Four side chains (I-IV) are known to be attached to the backbone, although their locations relative to one another are not known. (B). The 1.2 borate-diol cross-link of RG-11. A single 1.2 borate-diol ester is believed to cross-link two RG-I1 monomers. The cross-link is located in each monomer between the apiosyl residues of the 2-0-Me xylose-containing side chain (side chain I in A).

Fucp, 2-keto-3-deoxy-~-manno-octulosonic acid (Kdo), and 2-keto-3-deoxy-~-lyxoheptulosaric acid (Dha). Apiose and aceric acid are the only two-branched sugars known to exist in plants. The backbone of RG-I1 contains at least seven 1,6linked a-D-GalpA residues some of which may be methyl esterified. Two of the backbone GalA residues are substituted at C-3 with two structurally different disaccharides (see I1 and 111 in Figure 5A). Two structurally different octasaccharides are attached to C-2 of two other backbone GalA residues (see I and IV in Figure 5A). Despite the complexity of RG-11, its structure is conserved in the walls of all higher plants. Moreover, RG-I1 is now known to exist in the primary cell wall predominantly as a dimer cross-linked by a specifically located borate ester (Figure 5B) [ l l , 121. There is some debate as to whether arabinans, galactans, and arabinogalactans are present in the wall as individual polymers or covalently linked to RG-I or other

48.4 Biosynthesis of Wull Components

79 I

wall components [ 101. Type I1 arabinogalactans, which have a 1,3-linked P-D-Galp backbone that is substituted at C-6 with Araf-containing side chains, are often covalently linked to hydroxyproline-containing proteins such as arabinogalactan proteins (AGPs). Arabinogalactans containing a 1,4-1inked p-D-Galp backbone that is substituted at C-3 with Arafresidues, which are referred to as Type I arabinogalactans, are associated with RG-I. The primary walls of specialized tissues may contain other types of polysaccharide. For example, the primary walls of reproductive tissue ( e g apples and pine pollen) contain xylogalacturonans. In these polysaccharides P-u-Xylp residues are attached to C-3 of a 1,4-linked a-D-galacturonan backbone. Pollen tube walls contain callose, a 1,3-linked b-D-glucan. In those plants studied callose is present at the site of new wall formation during cell division but is removed when the wall has formed. Small amounts of callose have also been reported to be present particularly in the regions of the wall penetrated by plasmodesmata. The walls of Lemna and sea grasses contain apiogalacturonans-P-D-Api f residues are attached either as a single Apif residue or as the disaccharide P-~-Apif-(1,3’)-P-~-Apifto C-2 of a 1,4linked am-galacturonan backbone [lo]. The primary walls of the Solanaceae and Bromeliads have been reported to contain galactoglucomannans [ 13, 141. The primary walls of cereal coleoptiles contain 3- and 4-linked P-D-glucans [7]. Glycoproteins, particularly those rich in hydroxyproline, serine, arabinose, and galactose, collectively called hydroxyproline-rich glycoproteins (HRGPs), are components of the primary walls of many plants whereas glycine-rich proteins may only be present in the walls of specialized cells [ 151. Numerous non-structural proteins are also present in primary walls. Many of these proteins are enzymes involved in wall metabolism, whereas other proteins (e.g. expansins) are proposed to have a role in wall expansion, although their mode of action has not been determined [ 161. The primary walls of grasses typically contain phenolic acids such as ferulic and coumaric acids that are linked as esters to polysaccharides [7]. The walls of grasses often contain silicon, although the biological significance is not known [ 171. Dicotyledon primary walls, with the exception of the Chenopodiaceae, contain few if any phenolic esters and silicon is rarely present in more than trace amounts.

48.4 Biosynthesis of Wall Components Hundreds of enzymes are likely to be required for the biosynthesis of cell wall polymers and their precursors, and a large number of proteins may also be required to regulate wall synthesis. Cellulose, the predominant polysaccharide in most primary walls, is synthesized by plasma membrane-bound enzyme complexes [ 181. In contrast, xyloglucan, glucuronoarabinoxylan, methyl esterified homogalacturonan, RG-I, RG-11, and wall glycoproteins are synthesized in the Golgi apparatus 119, 201. The newly synthesized polysaccharides are transported from the Golgi apparatus in vesicles that migrate to and fuse with the plasma membrane. The polysaccharides are released into the extracellular space and then incorporated into the

792

48 The Primary Cell Walls of Higher Plants

wall. In cultured cells and possibly all plant tissues, a proportion of the pectin and hemicellulose is not incorporated but is secreted as water-soluble polysaccharide. Polysaccharides are synthesized from energetically-activated nucleoside-diphosphate(NDP)-monosaccharides. In growing tissues, the NDP-monosaccharides (e.g. NDP-GalpA, -Rhap, -Galp, -AraJ; -GlcpA, -Fucp, -Xylp, -Apif) are synthesized from UDP-Glc (or GDP-Man) via a nucleotide sugar interconversion pathway. NDP-monosaccharides are also generated, to a lesser extent, by a salvage pathway in which monosaccharide- 1-phosphates are converted to their corresponding NDPmonosaccharides [21]. Kdo may be formed by the condensation of phosphoenolpyruvate (PEP) and D-arabinose and then converted to its CMP-derivative by CMP-Kdo synthase [22]. The formation of Dha, aceric acid, and the methyletherified glycosyl residues (2-0-Me Xylp, 2-0-Me Fucp, and 4-0-Me GlcpA) present in cell wall polysaccharides has not been investigated. The synthesis and interconversion of NDP-monosaccharides is catalyzed by enzymes located predominantly in the cytosol [21]. The NDP-monosaccharides are transported by NDP-monosaccharide translocators into the Golgi apparatus where they become available to glycosyl transferases. Glycosyl transferases catalyze the transfer of a glycosyl residue from an NDP-monosaccharide, in an anomeric- and linkage-specific manner, to mono-, oligo-, or polysaccharides. It is not known if non-cellulosic wall polysaccharides are synthesized by multi-component synthases, nor have the factors that initiate and terminate polysaccharide biosynthesis been determined. There is limited evidence showing that plant cell wall polysaccharides are not assembled from lipid-linked oligosaccharide sub-units as are bacterial polysaccharides (201. At least four glycosyl transferases are required for the synthesis of fucosylated xyloglucans. A 1,4-1inked P-D-glucan synthase, an a-D-xylosyl, a p-D-galactosyl, and a a-L-fucosyl transferase. a and P-L-arabinosyl transferases are required for the formation xyloglucans in some plants. P-D-glucan synthase, a-D-xylosyl, P-D-galactosyl, and a-L-fucosyl transferase activities have been detected in impure membrane preparations [23] but only the a-L-fucosyl transferase has been purified to homogeneity and the corresponding gene cloned [24]. The Galp residues of xyloglucans are often 0-acetylated but no 0-acetyl transferases have been isolated and characterized. The biosynthesis of arabinoxylans and glucuronarabinoxylans has not been studied in detail [4]. A large number (>50) of glycosyltransferases and other types of transferase are required to synthesize pectic polysaccharides [20]. The activities of several glycosyltransferases have been identified in cell-free membrane preparations from plants. However, only a few of these enzymes have been partially characterized and shown to be involved in pectin biosynthesis. These include D-galacturonosyl transferase, galactosyl transferase, arabinosyl transferase, and apiosyl transferase [20]. An enzyme that catalyzes the methylesterification of homogalacturonan (homogalaturonan methyltransferase) has been partially characterized [20].There are also reports that plants contain methyltransferases that catalyze the methylesterification of RG-I and RG-I1 [25]. N o 0-acetyl transferase involved in pectin biosynthesis has been characterized nor have the enzymes that catalyze the addition of phenolic acids to pectins [20].

48.6 Cellulose-Xyloglucan Interactions

793

Cellulose is synthesized from UDP-Glc by plasma membrane-localized multi subunit enzyme complexes [18, 261. It is believed that the “cellulase synthases” themselves are responsible for the highly ordered structure of cellulose since the formation of parallel glucan chains is entropically unfavorable. Despite considerable effort, high rates of cellulose synthesis have not been achieved using cell-free preparations. Indeed, attempts to elucidate the mechanism of cellulose synthesis or to purify the cellulose synthase enzyme complex have proved largely unsuccessful [27, 281. However, two genes, termed CelAs, that have structural homology with cellulose synthase genes from the bacterium Acetohircter xylinum have recently been identified in cotton [30]. A mutation in a homologous gene in Arabidopsis results in impaired cellulose microfibril synthesis [29]. A family of at least 15 related genes has now been identified in Arabidopsis. The functions of these genes are not known although it is likely that they are involved in cellulose biosynthesis in different organs, tissues, and cell types, in primary or secondary walls, and in the formation of other P-glucans.

48.5 Organization of the Plant Primary Cell Wall The organization of the primary cell wall has been described in various models that contain essentially the same components interacting to form a complex macromolecular super-structure. However, the extent and nature of the interactions remains largely undetermined. Early models hypothesized that a monolayer of xyloglucan hydrogen-bonded to, and covered the surface of, the cellulose microfibrils. The xyloglucans are described as being covalently linked to arabinogalactans that themselves are covalently linked to pectins and structural proteins [311. More recent interpretations of wall architecture have emphasized the predominance of non-covalent associations between matrix polymers (see Figure 1C), and envisioned the xyloglucan-cellulose, pectin, and structural protein domains as independent but interacting networks 14, 32, 331.

48.6 Cellulose-Xyloglucan Interactions Early observations suggested that xyloglucans associate non-covalently with cellulose [34, 351 by hydrogen-bonding between their respective P-glucan backbones. The binding of xyloglucan to cellulose in vitro is specific and is not affected by the presence of other cell wall glucans, arabinogalactdn, or pectin [36]. Studies with the cellulose-synthesizing bacterium Acetohacfer xylinunz suggest that xyloglucan can compete with cellulose glucan chains in vitro during the synthesis and assembly of microfibrils, since the diameter of the microfibrils decreases when the bacterium is grown in the presence of xyloglucan [37]. In plants it is believed that the association

794

48 The Primary Cell Walls of Higher Plants

between xyloglucan and cellulose microfibrils occurs at or near the site of cellulose synthesis and the assembly of microfibrils from 0-glucan chains may be influenced by the presence of xyloglucan [36]. The results of in vitro binding studies and computer modeling techniques suggest that the conformation, solubility, and the pattern and degree of side-chain substitution of xyloglucan modulate its adsorption to cellulose [38, 391. The glucan backbone of xyloglucan most likely adopts a flattened 21 helical structure (two glucosyl residues per turn of the helix), similar to that of cellulose [38, 40, 411. The terminal Xyl residues and the Fuc-Gal-Xyl side chains may fold onto one face of the xyloglucan backbone thereby freeing the other face of the backbone for binding to cellulose [31, 381. The Ara-Xyl side chains of xyloglucans from solanaceous plants have also been suggested to fold onto one face of the backbone [8]. In contrast, the Gal-Xyl side chains have been reported to sterically hinder the xyloglucan backbone and reduce binding to cellulose [38]. Indeed, xyloglucans substituted with Fuc-Gal-Xyl side chains have a greater binding rate and affinity for cellulose than xyloglucans substituted with Gal-Xyl side chains [42]. The cellulose-binding affinities of fucosylated xyloglucans and arabinosylated xyloglucans have not been compared. The results of several studies have led to the suggestion that in muro, xyloglucan not only coats but also forms cross-links between the cellulose microfibrils (see Figure 1C) [43, 441. In addition there is evidence that a small amount (
48.7 Interactions Between Pectins and Other Cell Wall Components Research over the last two decades has established that methyl esterified homogalacturonan, RG-I, and RG-I1 are the predominant components of the primary

48.7 Interactions Between Prctins und Other Cell Wull Components

795

wall pectic matrix [6, lo]. Information on how these polysaccharides are linked to each other or linked to other components in the wall is largely lost when they are solubilized by chemical or enzymic treatments [4, 5 I]. Nevertheless, the results of recent studies [ 11, 121 have provided evidence that pectins are indeed covalently cross-linked together (see Figure IC). The results of numerous studies have established that RG-I1 is present in the primary wall predominantly as a dimer that is cross-linked by a 1:2 borate-diol ester [I I , 121. A single borate ester cross-links two of the four apiosyl residues present in the dimer ([52] see Figure 5B). In vitro studies have shown that in the presence of boric acid and certain cations, two RG-I1 monomers rapidly self-assemble to form a dimer [52]. Moreover, the structure of RG-I1 itself may determine the location of the borate ester, since the location of the cross-link is the same in naturally occurring and in vitro-formed dimers. Thus, RG-I1 is the first example of a plant cell wall pectic polysaccharide that self-assembles to form structurally identical dimers [ 521. The specificity and cation-dependence of this cross-linking suggest that there are distinct structural requirements for dimer formation and this may explain why the structure of RG-I1 is highly conserved in vascular plants [12, 531. It is not known whether dimer formation in plunta results from spontaneous self-assembly or is an enzymically catalyzed process. The boron requirement and wall pectin content of many plants are correlated [54-561. There is increasing evidence that a function of boron essential for normal plant growth and development is to covalently cross-link wall pectins (see Figure 5A). The structural organization of a boron cross-linked pectin network is likely to be a factor that contributes to the physical and biochemical properties of the wall because boron deficiency results in abnormal walls. For example, the results of recent studies provide evidence that the pore size of pectin-rich cell walls is determined in large part by borate ester cross-linking of RG-I1 [57]. Homogalacturonan and RG-I1 are likely to be covalently linked since they both have backbones composed of 1,4-linked a-D-GalpA residues and they are both solubilized by treating walls with endopolygalacturonase (EPGase). There are no glycosyl residues in RG-I that are susceptible to EPGase treatment although EPGase solubilizes RG-I. However, no oligosaccharides composed of the disaccharide repeating unit of the RG-I backbone have been isolated covalently linked to fragments of homogalacturonan [59]. In primary walls a significant proportion (50-80%) of the GalpA residues in homogalacturonan are esterified [ 5 8 ] . There is increasing evidence that the distribution of methyl esters (e.g. random or block-wise), the degree of methyl esterification, and structure of pectins differ in a cell and tissue-specific manner and that the pectin structure is developmentally regulated [60]. This may be due in part to the presence in the wall of pectin-modifying enzymes such as pectin methyl esterase, exo- and endo-polygalacturonases, arabinanases, and galactanases. The de-esterification of pectin by methyl esterases increases its net negative charge and may result in the increase ionic cross-linking by calcium ions, which itself may alter the rheological properties of the wall. Similarly, modification of pectin structure by glycanases may also result in the alteration of the physical properties of the wall. A recent report suggested that differences in the localization of pectins within the wall may occur in

796

48 The Primary Cell Walls of Higher Plants

part as a result of a phase separation of structurally distinct pectic polymers [61]. Phase separation is a process whereby mixtures that contain high concentrations of structurally distinct polymers segregate into different domains, each of which is enriched with one type of polymer. There have been numerous claims that some pectins are cross-linked by a baselabile bond. In the walls of the Chenopodiaceae (e.g. spinach and sugar beet) some of the pectic arabinosyl and galactosyl residues are esterified with ferulic or coumaric acid and these phenolic residues are potential sites for cross-linking via oxidative coupling [9, 621. Indeed, dehydrodiferuloyl cross-links are formed when beet or spinach pectin is treated in vitro with peroxidase [63]. However, compelling evidence has not been obtained for the existence in muro of intermolecular dehydrodiferuloyl cross-linked pectins [9]. This is in part due to the difficulty of distinguishing inter- and intramolecular cross-links in such molecules. Homogalacturonan has been reported to contain, in addition to methyl esters, other esters that may cross-link homogalacturonan to itself or to other wall polymers [64]. Homogalacturonan and RG-I have also been reported to be covalently linked to cellulose, xyloglucan, xylan, glucuronoxylans, and/or structural glycoproteins [65-671, but additional structural evidence is required to substantiate these claims. Indeed, such cross-links, if they exist, have proved to be extremely difficult to isolate and characterize.

48.8 Glycoproteins in the Cell Wall Several classes of proteins have been identified in plant walls and termed “structural proteins”, implying an absence of enzymatic function (reviewed in [ 151). These include hydroxyproline-rich glycoproteins (HRGPs), proline-rich proteins (PRPs), glycine-rich proteins (GRPs) and arabinogalactan proteins (AGPs). Primary walls also contain numerous enzymes, and still other proteins with no known function [681. HRGPs were the first structural wall proteins to be identified and were named “extensins”, although subsequent evidence suggested that they have a role in wall rigidification and slowing of cell expansion, rather then facilitating growth [ 151. The “warp-weft’’ model of Lamport [69] placed particular emphasis on covalent crosslinkages between structural proteins, where a lattice-work of extensin anchored the cellulose microfibril layers in place. Compelling evidence for the covalent or ionic cross-linking of extensin to wall polysaccharides has not been obtained, although the peroxidase-catalyzed formation of intramolecular isodityrosine cross-links in extensins has been demonstrated [ 151. The AGPs are a sub-class of the HRGPs that typically comprise a core protein rich in hydroxyproline/proline, serine and alanine and that are highly glycosylated, often containing more than 90% carbohydrate by weight. Various glycan side chains including an arabino-3,6-galactan have been identified. The presence of putative GPI membrane anchors in some AGPs has led

48.9 Heterogeneity in the Primary Cell Wall

791

to the suggestion that AGPs may form a link between the cell wall and the cytosol and thus have the potential to act as signaling molecules [70]. Proline-rich proteins and GRPs are only lightly glycosylated and are apparently insolubilized within the walls of specific cell and tissue types, where they have been suggested to play various roles in development [ 151.

48.9 Heterogeneity in the Primary Cell Wall

Relatively few plant species have been studied with respect to their cell wall composition or structure. However, it is possible to generalize regarding the presence of specific polysaccharides and the relative proportions of cell wall components within broad taxonomic groups [4, 61. Compositional variation is also evident within a single plant species and examples have been reported where walls from specific tissues or cell types exhibit distinct characteristics that reflect their unique physiological functions. Recent advances in probing wall structure by immunocytochemical methods and by Fourier transform infra-red (FTIR) spectroscopy have revealed differences in cell wall organization among adjacent cells of apparently similar type within the same tissue, such as pericarp cells in a ripening fruit [71]. Distinct structural domains have also been observed within the transverse or longitudinal plane of a single cell wall [58].This is likely a consequence of both differential targeting and deposition of polysaccharides to different faces of the cell and also distinct metabolism of the polysaccharides within the cell wall. For example, immunocytochemical analyses of the cell walls from the pericarp of ripening tomato fruit suggested that pectin modification and the coincident secretion of pectin modifying enzymes, occurs in distinct domains in a block-wise fashion [71]. Thus, each cell of a tissue behaves in an apparently autonomous manner, exhibiting a high degree of regulation of cell wall architecture and metabolism independent of its neighbor, despite being symplastically interconnected through plasmodesmata. The use of monoclonal antibodies to specific cell wall epitopes and FTIR spectroscopy has reinforced the view that cell walls are dynamic structures and exhibit considerable temporal and spatial variation [60, 721. Determining the interrelationship and functional significance of structural networks within the cell wall, their interaction with the symplast, and the basis and importance of cell wall domains and their structural differences are major challenges for plant cell biologists. When the structures of the primary walls of a single cell can be determined, the walls will be described in the context of their positions, developmental states, and as the end point of signal transduction pathways. Together these features will provide the information needed to describe plant morphology, development, and the interactions of plants with other organisms and the their environment in much greater detail than is now possible.

798

48 The Primary Cell Walls of Higher Plants

48.10 Function and Metabolism of Plant Primary Cell Walls Early botanists referred to the obvious structural role of walls as forming “little boxes” around the cells. This was most apparent in the substantial thickenings of secondary walls in mature tissues. However, over the last few decades an appreciation has developed for the complexity and multi-functional nature of cell walls. The primary wall is now regarded as a complex, dynamic organelle rather than an inert “extracellular” matrix. It is now apparent that cell walls influence, and are integral to, many aspects of plant growth and development.

48.10.1 Mechanical Support Mechanical strength is a property generally associated with secondary walls; these walls contain substantial deposits of cellulose microfibrils and lignin, a highly crosslinked phenolic polymer that is synthesized and covalently linked into the wall at the cessation of cell expansion. Secondary walls allow plants to exploit the aerial environment and account for the ability of eucalyptus and redwood trees to reach heights of more than 100 m. The primary wall must provide sufficient structural support during plant growth to resist the internal turgor pressure of the cell. This property of walls is generally attributed to the tensile strength of the cellulose-hemicellulose network. The walls of non-lignified cells that are exposed to greater mechanical forces, such as guard cells, typically contain more cellulose microfibrils to provide greater tensile strength. However, it is difficult to ascertain the contributions of the cellulose-xyloglucan network vis-a-vis the pectic networks to the overall tensile strength of the wall. Furthermore, it is likely that there is considerable variation in the contributions of the two networks in different cells and tissues and between species. Indeed, there may be considerable developmental flexibility in the mechanisms plants use to achieve mechanical strength in their cell walls. For example, in dichlorobenzonitrileadapted cells, which contain little if any of the cellulose-xyloglucan network, the pectin matrix and extensin appear to provide sufficient mechanical strength to allow the cells to grow and divide [73-751.

48.10.2 Regulation of Cell Expansion A plant cell may have to accommodate a 1,000-10,000-fold increase in volume and in length from its size at the time it is formed until the time it is fully expanded. The wall remains more or less the same thickness during expansion of the cell. Thus cell expansion requires the synthesis of new wall material. The expansion itself is driven by the cell turgor pressure, typically 0.3-1 MPa, and results in wall stresses of 10100 MPa [16]. The loosening of wall structure required to accommodate the expanding protoplast involves insertion of new wall material that must be strictly regulated to avoid stress-induced blebbing or rupture of the plasma membrane and

48.10 Function and Metabolism of Plant Primury Cell Walls

799

cellular disruption. Controlled cell wall extension may also involve regulated restructuring and modification of wall architecture, and the disruption of covalent bonds as well as non-covalent associations. Turgor-driven cell expansion is a consequence of three superimposed but independent phenomena; viscoelastic deformation, wall stress relaxation, and wall synthesis. Viscoelastic effects are reversible and occur whenever walls are placed under tension by turgor pressure. Viscoelasticity is a intrinsic mechanical property of the primary wall that is not mediated by enzymic or chemical reactions [76, 771. In contrast, wall stress relaxation which is associated with expansive growth is believed to involve the chemical or enzymic irreversible loosening of load-bearing bonds in the wall. As a consequence, turgor pressure and water potential are reduced. Water is then taken up by the cell with the result that turgor is restored and there is a net increase in cell volume. Wall stress relaxation is regulated by many hormonal and environmental stimuli that are believed to effect the mechanical properties of the wall, principally through the action of enzymes that loosen load-bearing bonds. This allows polymer slippage and wall expansion as water is taken up into the cell. The simultaneous incorporation of newly-synthesized polysaccharides into the wall and the formation of new cross-links is likely to be required to control the rate and extent of wall expansion. The principal load-bearing structure in the primary wall is believed to be the cellulose-hemicellulose network and synthesis of the polysaccharides together with assembly and disassembly of this network may be fundamental to cell expansion and differentiation. Hydrolytic enzymes including endo-l,4-P-glucanases (EGases) that cleave load-bearing xyloglucan crosslinks in dicot walls (see Figure 6) and xylanases that cleave glucuronoarabinoxylans in grass walls, may mediate the breaking of load-bearing bonds. In contrast, a class of transglycosylases called XETs (also termed EXTs) may mediate the both the hydrolysis and reformation of new cross-links and the incorporation of new polymers into the network. Xyloglucan endotranglycosylases catalyze the endo-cleavage of xyloglucan and the ligation of the newly-generated reducing end to the non-reducing end of another xyloglucan molecule (see Figure 6). Distinct subclasses of XETs and EGases may have a role in restructuring the cellulose-xyloglucan framework not only during cell expansion but also during fruit ripening and cellular differentiation where wall disassembly occurs [ 501. The involvement of exoglycanases, such as P-galactosidase, has similarly been proposed. Exoglycanases that remove the carbohydrate side chains from the xyloglucan backbone may modulate the action of EGases and XETs and thereby indirectly affect the tensile strength of the wall. However, compelling evidence is lacking for a role of exoglycanases in wall expansion. A family of proteins called expansins have been suggested to act as a catalyst for cell expansion [78]. Expansins appear to exhibit no hydrolytic activity but instead may disrupt hydrogen bonds between cellulose and hemicelluloses (Figure 6). However, the precise substrates and mechanism of action of expansins are not known. Expansins, unlike XETs, EGases, and pectinases, have been reported to cause isolated dead tissue under tension to elongate faster by promoting wall stress relaxation. Nevertheless, elongation growth may involve disassembly and restructuring of the wall and require the synergistic action of several classes of proteins to

800

48 The Primary Cell Walls of Higher Plants

a-fucosldase

Rhamnogalacturonases Endo and exo arabinanas Endo and exo galactanase

Pectin methyl esterasa' Pectin lyases

Figure 6 . The potential sites for modification of Type I primary cell walls by enzymes and/or expansins. The activities of the enzymes/expansins are associated with numerous aspects of plant growth and development including cell expansion, fruit ripening, and organ abscission.

maintain wall tensile strength (Figure 6). Qualitative as well as quantitative differ, ences in polysaccharide synthesis and deposition may also be important in the process of differentiation or development. Distinct differences in the composition of the primary wall and the synthesis of specific wall polysaccharides during cell expansion have been documented in several grass species (reviewed in [4]). Cell expansion can in certain tissues and plants be regulated by environmental stimuli such as light, water stress, and temperature and by plant hormones, including auxin, giberellic acid, ethylene and brassinosteroids. Numerous reports describe correlations between these regulatory factors and changes in the structure or synthesis of primary wall polysaccharides and of enzyme and gene activities that act on wall components. Such correlative data suggest possible critical components that determine the rate and extent of cell expansion, as well as rate limiting mechanisms governing wall stress relaxation. For example, auxin is known to promote the solubilization and depolymerization of xyloglucan in addition to the expression of genes encoding EGases, XETs and expansins. Auxin also promotes tissue acidification, although there is controversy as to whether this alone is sufficient to account for growth induction in vivo. 48.10.3 Morphogenesis and Differentiation The synthesis of primary cell walls influences not only the rate of cell expansion but also the direction of cell expansion. Most plant growth is asymmetric, giving rise

48.10 Function and Metabolism of Plant Primary Cell Walls

801

to distinct highly differentiated cell types, tissues, and organs. Such asymmetry is associated with the disproportionate deposition of cell wall materials and may also involve selective wall restructuring and wall relaxation along specific planes of the cell surface. In non-elongating cells, cellulose microfibrils do not have a distinct pattern of orientation around the cell. However, at the onset of cell enlargement, the microfibrils undergo re-orientation, forming spiral helices wound along the longitudinal axis of the cell. Such an arrangement of microfibrils is believed to resist the considerable tangential tension but provides relatively little resistance in the longitudinal axis and thus facilitates differential cell elongation. Thus, in general, the orientation of the microfibrils is believed to contribute to the direction of cell elongation, while wall synthesis and the modification of load-bearing structures in the primary wall dictate the rate of expansion. Directional cell expansion and the consequent asymmetric growth that leads to the formation of plant organs such as leaves and flowers is likely to require both regulated wall synthesis and wall loosening. For example, localized control of tissue expansion may have a role in leaf formation since the application of expansin to tomato seedling apical meristems has been reported to result in the formation of “leaf-like” structures [79]. Moreover, the expression of one expansin gene is elevated in the cells of leaf primordia and may therefore have a role in cell differentiation [80]. An EGase gene has recently been shown to be associated with early carpel development and differentiation [81]. Such data suggest that localized wall loosening and/or wall modification are required for organogenesis, although additional studies are required to substantiate these claims. Modification of the architecture of the primary wall is a prominent feature of fruit ripening and organ abscission, both of which involve terminal cellular differentiation but not cell expansion. Fruit ripening typically involves substantial changes in texture that are believed to result from a combination of a loss of turgor, primary wall disassembly and intercellular wall loosening. In many species (e.g. tomato and melon) the fruit softening process has been associated with depolymerization and solubilization pectins and xyloglucan. Depolymerization of pectins in the middle lamella is believed to contribute to cell separation while depolymerization of the cellulose-xyloglucan network appears to be one of the early events in the softening of several fruit species [82, 831. During ripening the activities of EGases, XETs, polygalacturonase, pectin methyl esterase, and several exo-glycanases typically increase, and ripening-related expansin genes have also been identified [ 841. Similar enzymic activities and patterns of gene expression have been correlated with organ abscission and other examples of cellular differentiation. A pattern therefore emerges of a highly dynamic primary cell wall that plays an important role in regulating growth, morphogenesis and cell differentiation. The regulation and coordination of wall synthesis and disassembly during such processes is not understood and provides a considerable future challenge for plant developmental biologists. 48.10.4 Plant Cell Wall Oligosaccharides in Defense and Cell Signalling

The cuticle and epidermal cell wall are the first obstacles that most pathogens must overcome to gain access to its host. To penetrate the outer the surface of a plant,

802

48 The Primary Cell Walls of Higher Plants

pathogens must either take advantage of wounds or natural openings in the plant surface or produce a range of enzymes (cutinases, cellulases, hemicellulases and pectinases) that fragment the primary wall [85].The endoglycanases that hydrolyze cell wall polysaccharides generate low-molecular mass oligosaccharides. These oligosaccharides are a carbon source for the pathogen but they also act as signal molecules (elicitors) that are perceived by, and elicit host defense responses in, plant cells. For example, oligogalacturonides with a degree of polymerization (DP) of 1015 are generated by endopolygalacturonase and pectin/pectate lyase fragmentation of homogalacturonan and induce rapid and long-term defense responses in plant cells and tissues. The rapid (1-2 min) responses, which include membrane H+/K+/ Ca2+ flux, membrane depolarization, membrane protein phosphorylation and an oxidative burst, are transient (1-4 h) and may initiate long-term responses. The long-term defense responses include the synthesis of callose, cellulose, lignin, suberin, and structural cell wall proteins, the synthesis of antimicrobial compounds, and the synthesis of proteinase inhibitors. In addition, plant cells synthesize enzymes (endo-l,3-glucanases and chitinases) that hydrolyze the glucan and chitin present in fungal cell walls and release oligosaccharides that themselves elicit defense response in plants [86]. The mechanisms by which plant cells perceive oligosacchairde elicitors are poorly understood, although biological activity is known to be correlated with the DP of the elicitor, its glycosyl-residue composition, and glycosyl sequence. Putative receptors or high-affinity binding proteins have been identified for oligogalacturonides, and oligosaccharide fragments of chitin and P-glucans. However, evidence is lacking for the nature of the downstream elements in the signal cascade that leads to the initiation of plant defense responses [86]. Several cell wall-degrading endoglycanases have been purified that can generate elicitors from host or pathogen cell walls in vitro. However, the generation and turnover of elicitors in vivo is likely to be more complex and include factors such as the localized concentration of the elicitor and its stability in the apoplast. The generation and stability of elicitors is further complicated by the observation that both plants and fungal pathogens secrete proteins that selectively combine with and inhibit the activities of cell wall-degrading endoglycanases. For example, plants synthesize polygalacturonase inhibitor proteins that inhibit the activity of fungal polygalacturonases. These inhibitor proteins may slow the degradation of the plant cell wall and increase the half-life of endogenous oligogalacturonide elicitors [87]. Conversely, some fungal pathogens secrete glucanase inhibitor protein that bind to and inhibit specific plant P-1,3-glucanases; enzymes that the plant secretes to degrade the cell walls of the fungus [88]. Several plant cell-wall-derived oligosaccharides have been identified that have plant growth regulating activities. They have activities that are comparable to those of auxin, kinetin, and gibberellin, leading to the suggestion that these oligosaccharides are a class of endogenous plant growth regulators [89]. For example, a xyloglucan-derived nonasaccharide, at an optimum concentration of 1 nM, inhibits auxin-induced growth promotion of pea stems [90]. Fucosylated xyloglucanderived oligosaccharides have also been shown to inhibit endogenous growth and growth stimulated by low pH, gibberellic acid and fusicoccin [911. Oligosaccharides derived from galactoglucomannan have been reported to inhibit auxin-induced

-

48.12 Bioteclznoloyy and Future Directions

803

growth in vitro [92]. In addition to elicitation of defense responses and regulation of growth, several developmental processes, including fruit ripening and organ morphogenesis, are regulated by pectic and hemicellulosic oligosaccharides [ 891. The in uivo mechanisms that generate these oligosaccharides, their perception by plant cells, and the subsequent downstream signaling events that lead to morphogenetic changes in plant tissues remain to be determined [89].

48.11 Intercellular Transport and Storage The transport of materials between adjacent plant cells is either symplastic or apoplastic. Symplastic transport involves the passage of molecules through numerous complex pores, called plasmodesmata, which directly connect two cells [93]. In contrast, apoplastic transport requires that molecules cross the plasma membrane of one cell, pass through the wall of that cell, cross the middle lamella, pass through the wall of the neighboring cell, and finally cross the plasma membrane of the second cell. The structure of the cell wall is an important determinant of both the type of molecules that may move through the apoplast and the rate at which this occurs. Factors such as molecular size and shape, hydrophobicity, and electrostatic charge are all likely be important factors in regulating apoplastic intercellular transport. Several molecules including sugars and the plant hormone auxin have been reported to move between cells through the apoplast. The regulation of apoplastic transport may provide a mechanism by which molecules at specific locations within the cell walls, or within the walls of particular tissues and organs, are compartmentalized. Pectins and borate ester cross-linking of pectins have been suggested to play an important role in regulating wall pore size [57]and pectin metabolism may therefore be integral to the movement of many classes of macromolecules including cell wall enzymes within the apoplast. The cell wall functions as a storage organelle by regulating the availability, movement and uptake of ions, water and other molecules into the cytosol. The cell wall is a polyanion and binding of cations to wall polysaccharides may be important for regulating the influx of essential ions through the plasma membrane. The polyanions of the wall may immobilize potentially harmful ions including aluminum, cadmium, and lead. Furthermore, certain seeds secrete polysaccharides including mannans, galactomannan, and xyloglucan into the periplasmic space to be used as storage reserve and subsequently mobilized when the seed germinates.

48.12 Biotechnology and Future Directions in the Commercial Applications of Plant Primary Cell Walls The components of plant cell walls are utilized by a number of industries, including those associated with the production of food additives, paper and pulp, and the

804

48 The Primary Cell Walls of Higher Plants

production of fiber and fabrics. These industries are closely associated with the production and application of polysaccharide-degrading enzymes and the commercial production of such enzymes is itself an industry. Several food quality traits are governed to a large extent by primary wall structure and composition. The textural properties and the post-harvest shelf-life of fruit and vegetables are in large part determined by endogenous processes in which the cell wall undergoes selective disassembly. Thus, it is important to develop an understanding of the mechanism of tissue deterioration in ripening fruits and vegetables and identifying strategies that may delay the softening process and enhance resistance to post-harvest microbial infection. There is a increasing awareness that primary wall structure, biosynthesis, and turnover are targets for manipulation by genetic engineering. The generation of plant mutants with altered wall composition, although still in its infancy, has considerable potential for dissecting the mechanisms of wall synthesis and modification [94]. In addition, plants with genetically modified primary walls may have increased resistance to pathogens, produce fruits and vegetables with altered mechanical and textural properties, or provide a source of animal feeds with increased digestibility. Specific manipulation of cell wall structures and mechanism of wall modification may lead to plants with new growth and developmental characteristics. In this way plants may be “designed” to fit specific horticultural and agronomic requirements. The structures and rheological properties of pectins and hemicelluloses can, in principle, be modified by genetic manipulation and thereby increase their industrial value. Such advances will also facilitate the development of novel carbohydratebased plant products that are likely to influence numerous areas of human development including health, nutrition, fabrics and material sciences. Acknowledgment The financial support of the United States Department of Energy (Grants DE-FG0296ER20220 and DE-FG05-93ER20097) is gratefully acknowledged. References 1. R.R. Selvendran. M.A. O’Neill. Isolation and analvsis of cell walls from ulant material. In D. Glick (Ed.) Methods of Biochemical Analysis Voi32, John Wiley and Sons, London, 1985, UP 25-153 2. W.S.York, A.G. Darvill, M. McNeil, T.T. Stevenson, P. Albersheim. Isolation and characterization of plant cell walls and cell wall constituents. Methods Enzymol, 1985, 118, 3-40 3.- S.B. Jones, A.E. Luchsinger, Plant Sytematics, McGraw-Hill Inc, New York. 1986 4. N.C. Carpita, D.M. Gibeaut. Structural models of primary-cell walls in flowering plants consistency of molecular structure with the physical properties of the wall during growth. Plant J , 1993,3, 1-30 5. J.T. Thomas, M. McNeil, A.G. Darvill, P. Albersheim. Structure of Plant Cell Walls. XIX. Isolation and characterization of wall polysaccharides from suspension-cultured Douglas fir cells. Plunt Physiol, 1987, 83, 659-671 6. P. Albersheim, A.G. Darvil, M.A. O’Neill, H.A. Schols, A.G.J Voragen. An hypothesis: The same six polysaccharides are components of the primary cell walls of all higher plants. In

References

805

J. Visser and A.G.J.Voragen (Eds.). Pectins and Pectinases. Elsevier Science, Amsterdam, 1996. pp 47-55 7. N.C. Carpita. Structure and biogenesis of the cell walls of grasses. Ann Rev Plunt Physiol Plunt Mol Biol, 1996, 47, 445-476 8. W.S. York, V.S.K. Kolli, R. Orlando, P. Albersheim, A.G. Darvill. The structures of arabinoxyloglucans produced by solanaceous plants. Curbohydr Res, 1996, 285, 99-128 9. T. Ishii. Structure and function of feruloylated polysaccharides. Plunt Sci, 1997, 127, 111-127 10. M.A. O’Neill, P. Albersheim, A.G. Darvill. The pectic polysaccharides of primary cell walls. In P.M. Dey (Ed.) Methods in Plunt Biochemistry V o l 2 Carbohydrates. Academic Press, London 1990. pp 415-441 11. M. Kobayashi, T. Matoh, J. Azuma. Two chains of rhamnogalacturonan-11 are cross-linked by borate-diol ester bonds in higher plant cell-walls. Plant Physiol, 1996, 110, 1017--1020 12. M.A. O’Neill, D. Warrenfeltz, K. Kates, P. Pellerin, T. Doco, A.G.Darvil1, P. Albersheim. Rhamnogalacturonan 11, a pectic polysaccharide in the walls of growing plant cells, forms a dimer that is covalently cross-linked by a borate ester. J Biol Chem, 1996, 271, 22923-22930 13. S. Eda, Y. Akiyama, K. Kato, A. Ishizu, J. Nakano. A galactoglucomannan from cell walls of suspension-cultured tobacco (Nicotianu tubacum) cells. Curbohydr Res, 1985, 137, 173-181 14. B.G. Smith, P.J. Harris. Polysaccharide composition of unlignified cell walls of pineapple [Annus comosus (L.) Merr.] fruit. Plant Physiol, 1995, 107, 1399-1409 15. G.I. Cassab. Plant cell wall proteins. Annu Rev Plant Physiol Plant Mol Biol, 1998, 49, 281309 16. D.J. Cosgrove. Relaxation in a high-stress environment: the molecular bases of extensible cell walls and cell enlargement. Plunt Cell, 1997, 9, 1031--1041 17. E. Epstein. The anomaly of silicon in plant biology. Proc Nutl Acad Sci USA 1994, 91, 11L17 18. D.P. Delmer, Y. Amor. Cellulose biosynthesis. Plant Cell, 1995, 7, 987-1000 19. D.M. Gibeaut, N.C. Carpita. Biosynthesis of plant-cell wall polysaccharides. FASEB J , 1994, 8, 904-915 20. D. Mohnen. Biosynthesis of pectins and galactomannans. In B.M. Pinto (Ed.) Comprehensive Natural Products Chemistry Vol 3 Carbohydrates und their derivutives including tunnins, cellulose, and related lignins. Elsevier, Amsterdam, 1999, pp 497-527 21. D.S. Feingold. Aldo (and keto) hexoses and uronic acids. In FA Loewus and W Tanner (Eds.). Encyclopedia of Plant Physiology Vol13A. Springer Verlag, Berlin. 1982. pp 3-76 22. R.L. Doong, J.E. Gander, R.J. Ganson, R.A. Jensen. The cytostolic isoenzyme of 3-deoxy-~arabino-heptulosonate 7-phosphate synthase in Spinacia oleracea and other higher plantsextreme substrate ambiguity and other properties. Physiol Plunta, 1992, 84, 351-360 23. A. Faik, C. Chileshe, J. Sterling, G. Maclachlan. Xyloglucan galactosyl- and fucosyltransferase activities from pea microsomes. Plant Physiol, 1997, 114, 245-254 24. R.M. Perrin, A.E. DeRocher, M. Bar-Peled, W . Zeng, L. Norambuena, A. Orellana, N.V. Raikhel, K. Keegstra. Xyloglucan fucosyltranseferase, an enzyme involved in plant cell wall biosynthesis. Science, 1999, 284, 1976-1979 25. T. Bourlard, P. Pellerin, C. Morvan. Rhamnogalacturonan-I and rhamnogalacturonan-I1 are pectic substrates for flax-cell methyl transferases. Plant Physiol Biochem, 1997, 35: 623-629 26. Y. Kawagoe, D.P. Delmer. Recent progress in the field of cellulose synthesis. Trends Glycosci Clycotechnoll998, 10, 291-305 27. D.P. Delmer, P. Ohana, L. Gonen, M. Benziman. In-vitro synthesis of cellulose in plants-still a long way to go. Plant Physiol, 1993, 103, 307-308 28. K. Kudlicka, R.M. Brown. Cellulose and callose biosynthesis in higher-plants. I. Solubilization and separation of 1,3-P-glucan and 1,4-~-glucansynthase activities from mung bean. Plant Physiol, 1997, 115, 643-656 29. T. Arioli, L.C. Peng, A.S. Betzner, J. Burn, W. Wittke, C.W. Herth, C. Camilleri, H. Hofte, J. Plazinski, R Birch, A. Cork, J. Glover, J. Redmond, R.E. Williamson. Molecular analysis of cellulose biosynthesis in Arabidopsis. Science; 1998, 279, 717-720 30. J.R. Pear, Y. Kawagoe, W.E. Schreckengost, D.P. Delmer, D.M. Stalker. Higher plants contain homologs of the bacterial celA genes encoding the catalytic subunit of cellulose synthase. Proc Nutl Acud Sci USA, 1996, 93 12637- 12642 31. K. Keegstra, K.W. Talmadge, W.D. Bauer, P. Albersheim. Structure of plant cell walls. 111.

806

48 The Primary Cell Walls of Higher Plants

A model of the walls of suspension-cultured sycamore cells based on the interconnections of the macromolecular components. Plant Physiol, 1973, 51, 188-197 32. J.E. Varner, S-L. Lin. Plant cell wall architecture. Cell, 1989, 56, 231-239 33. L.F Talbott, P.M. Ray. Molecular size and separability features of pea cell wall polysaccharides: implications for models of primary wall structure. Plant Physiol 1992, 98, 357368 34. G.O. Aspinall, J.A. Molloy, J.W.T.Craig. Extracellular polysaccharides from suspensioncultured sycamore cells. Can J Biochem, 1969,47, 1063-1070 35. W.D. Bauer, K.W. Talmadge, K. keegstra, P. Albersheim. The structure of plant cell walls. 11. the hemicellulose of the walls of suspension-cultured sycamore cells. Plant Physiol, 1973, 51, 174-187 36. T. Hayashi, M.P.F. Marsden, D.P. Delmer. Pea xyloglucan and cellulose. V. Xyloglucancellulose interactions in vitro and in vivo. Plant Physiol, 1987, 83, 384-389 37. R.H. Atalla, J.M. Hackney, I. Uhlin, N.S. Thompson. Hemicelluloses as structure regulators in the aggregation of native cellulose. Zntl J Biol Macromol, 1993, 15, 109-1 12 38. S. Levy, W.S. York, R. Stuikeprill, B. Meyer, L.A. Staehelin. Simulations of the static and dynamic molecular-conformations of xyloglucan-the role of the fucosylated side-chain in surface-specific side-chain folding. Plant J , 1991 1, 195-21 5 39. V.L. Finkenstadt, T.L. Hendrixson, R.P. Millane. Models of xyloglucan binding to cellulose microfibrils. J Carhohydv Chem, 1995, 14, 601-61 1 40. K. Ogawa, T. Hayashi, K. Okamura. Conformational analysis of xyloglucan. Intl J Biol Macromol, 1990, 12, 218-222 41. I.E.P. Taylor, E.D.T. Atkins. X-ray diffraction studies on the xyloglucan from tamarind (Tamarindus indica) seed. FEBS Lett, 1986, 181, 300-302 42. S. Levy, G. Maclachlan. L.A. Staehelin. Xyloglucan side-chains modulate binding to cellulose during in-vitro binding assays as predicted by conformational dynamics simulations. Planr J , 1997, 11, 373-386 43. T. Hayashi. Xyloglucans in the primary cell wall. Annu Rev Plant Physiol Plant Mol B i d , 1989, 40, 139-168 44. M.C. McCann, B. Wells, K. Roberts. Direct visualization of cross-links in the primary plant cell wall. J Cell Sci, 1990, 96 323-334 45. H.G. Edelmann, S.C. Fry. Factors that affect the extraction of xyloglucan from the primarycell walls of suspension-cultured rose cells. Carhohydr Res, 1992, 228, 423-43 1 46. M. Pauly, P. Albersheim, A. Darvill, W.S. York. A model for the cellulose/xyloglucan network in the cell walls of higher plants. Plant J , 2000, in press 47. M.C. McCann, M. Hammouri, R. Wilson, P. Belton, K. Roberts. Fourier transform infrared microspectroscopy is a new way to look at plant cell walls. Plant Physiol, 1992, 100, 1940-1947 48. S.E.C. Whitney, J.E. Brigham, A.H. Drake, J.S.G. Reid, M.J. Gidley. In vitro assembly of cellulose-xyloglucannetworks: Ultrastructural and molecular aspects. Plant J , 1995, 8 , 49 1-504 49. K. Baba, Y. Sone, A. Misaki, T. Hayashi. Localization of xyloglucan in the macromolecular complex composed of xyloglucan and cellulose in pea stems. Planr Cell Physiol, 1994, 35, 439444 50. J.K.C. Rose, A.B. Bennett. Copperative dissasembly of the cellulose-xyloglucan network of plant cell walls; parallels between cell expansion and fruit ripening. Trends Plant Sci,1999, 4 , 176- 183 51. P.W. Needs, N.M Rigby, I.J. Colquhoun, S.G. Ring. Conflicting evidence for non-methyl galacturonoyl esters in Daucus carota. Phytochemistry, 1998, 48, 71-77 52. T. Ishii, T. Matsunaga, P. Pellerin, M A . ONeill, A. Darvill, P. Albersheim. The plant cell wall polysaccharide rhamnogalacturonan IT self-assembles into a covalently cross-linked dimer. J Biol Chem, 1999, in press 53. T. Matoh, M. Kobayashi. Boron and calcium, essential inorganic constituents of pectic polysaccharides in higher plant cell walls. J Plant Res, 1998, 111, 179-190 54. H.N. Hu, P.H. Brown, J.M. Labavitch. Species variability in boron requirements is correlated with cell-wall pectin. J Exp Bot, 1996; 47, 227-232 55. T. Matoh, S. Kawaguchi, M. Kobayashi. Ubiuity of a borate rhamnogalacturonan-I1 complex in the cell-walls of higher plants. Plant Cell Physiol, 1996, 37, 636-640

References

807

56. P.H. Brown, H. Hu. Does boron play only a structural role in the growing tissues of higher plants? Plant Soil, 1997, 196, 211-215 57. A. Fleischer, M.A. O’Neill, R. Ehwald. The pore size of non-graminaceous plant cell walls is rapidly decreased by borate-ester cross-linking of the pectic polysaccharide rhamnogalacturonan 11. Plant Physiol, 1999, 121, 829-838 58. M.C. McCann, K. Roberts. Changes in cell wall architecture during cell elongation. J. Exp. Bot, 1994, 45, 1683--1691 59. M.A. O’Neill, A.G. Darvill, P. Albersheim P (1999) Pectic substances. In Encyclopedia of Life Sciences, MacMillan Reference, London, In press 60. M.C. McCann, K. Roberts, R.H. Wilson, M.J. Gidley, D.M. Gibeaut, J.B. Kim, N. Carpita. Old and new ways to probe plant cell-wall architecture. Can J Bot, 1995, 73, SlO3-Sl13 61. A.J. McDougall, N.M. Rigby, S.G. Ring. Phase separation of plant cell wall polysaccharides and its implications for cell wall assembly. Plant Physiol, 1997, 114, 353-362 62. S.C. Fry. Phenolic components of the primary cell wall. Biochem J , 1982,203,493-504 63. F. Guillon, J-F. Thibault. Oxidative cross-linking of chemically and enzymatically modified sugar-beet pectin. Carhohydr Polymers, 1990, 12, 353-374 64. J.B. Kim, N.C. Carpita. Changes in eslerification of the uronic-acid groups of cell-wall polysaccharides during elongation of maize coleoptiles. Plant Physiol, 1992, 98, 646-653 65. G.B. Seymour, I.J. Colquhoun, M.S. DuPont, K.R. Parsley, R.R. Selvendran. Composition and structural features of cell wall polysaccharides from tomato fruits. Phytochernistry, 1990, 29, 725-732 66. M.A. Coimbra, N.M. Rigby, R.R. Selvendran, K.W. Waldron. Investigation of the occurrence of xylan-xyloglucan complexes in the cell walls of Olive pulp (Oleu europaea). Curbohydr Polymers, 1995, 27, 277-284 67. J. Fu, A.J. Mort. More evidence for covalent attachment of xyloglucan to the rhamnogalacturonan of cotton suspension culture cell walls. Plant Physiol, 1995, 108 (Suppl), 123 68. D. Robertson, G.P. Mitchell. J.S. Gilroy, C. Gerrish, G.P. Bolwell, A.R. Slabas. Differential extraction and protein sequencing reveals major differences in patterns of primary cell wall proteins from plants. J Biol Chem 1997, 272, 15841-15848 69. D.T.A. Lamport. The primary cell wall: a new model. In Young RA and Rowell RM (Eds.) Cellulose: Structure, Modifications und H~vdrolysis.John Wiley, New York 1986, pp 77-90 70. C. Schultz, P. Gilson, D. Oxley, J. Youl, A. Bacic. GPI-anchors on arabinogalactan-proteins: implications for signaling in plants. Trends Plant Sci, 1998, 3, 426-431 71. N.M. Steele; M.C. McCann, K. Roberts. Pectin modification in cell walls of ripening tomatoes occurs in distinct domains. Plant Physiol, 1997, 114, 373-381 72. G. Freshour, R.P. Clay, M.S. Fuller, P. Albersheim, A.G. Darvill, M.G. Hahn. Developmental and tissue-specificstructural alterations of the cell wall polysaccharides of Arahidopsis thaliancr roots. Plant Physiol, 1996, 110, 1413- 1429 73. E. Shedletzky, M. Shmuel, D.P. Delmer, D.T.A. Lamport DTA. Adaptation and growth of tomato cells on the herbicide 2,6-dichlorobenzonitrile leads to production of unique cell walls virtually lacking a cellulose-xyloglucan network. Plant Physioll990, 94, 980-987 74. B. Wells, M.C. McCann, E. Shedletzky, D.P. Delmer, K. Roberts. Structural features of cell walls from tomato cells adapted to grow on the herbicide 2,6-dichlorobenzonitrile.J Microscop 1994, 173, 155-164 75. R.P. Sabba, N.A. Durso, K.C. Vaughn. Structural and immunocytochemical characterization of the walls of dichlorobenil-adapted BY-2 tobacco cells. Znt J Plant Sci,1999, 160, 275-290 76. D.J. Cosgrove. Wall extensibility: its nature, measurement and relationship to plant cell growth. New Phytol, 1993, 124, 1-23 77. T.E. Proseus, K.E. Ortega, J.S. Boyer. Separating growth from elastic deformation during cell enlargement. Plant Physiol, 1999, 119, 775-784 78. D.J. Cosgrove. Cell wall loosening by expansins. Plant Physiol, 1998, 118, 333-339 79. A.J. Fleming, S. McQueen-Mason, T. Mandel. C. Kuhlemeier. Induction of leaf primordia by the cell-wall protein expansin. Science, 1997, 276, 1415-1418 80. D. Reinhardt, F. Wittwer, T. Mandel, C. Kuhlemeier. Localized upregulation of a new expansin gene predicts the site of leaf formation in the tomato meristem. Plant Cell, 1998, 10, 1427- 1437

808

48 The Primary CXI Walls of Higher Plants

81. M-H. Yung, R. Schaffer, J. Putterill. Identification of genes expressed during early Arabidopsis carpel development by mRNA differential display: Characterization of A TCEL2, a novel endo1,4-P-~-glucanase.Plant J, 1999, 17, 203-208 82. C.B.S. Tong, K.C. Gross. Glycosyl-linkage composition of tomato fruit cell wall hemicellulosic fractions during ripening. Physiol Plant, 1988, 74, 365-370 83. J.K.C. Rose, K.A. Hadfield, J.M. Labavitch, A.B. Bennett. Temporal sequence of cell wall disassembly in rapidly ripening melon fruit. Plant Physiol, 1998, 117, 345-361 84. J.K.C. Rose, H.H Lee, A.B. Bennett. Expression of a divergent expansin gene is fruit-specific and ripening-regulated. Proc Natl Acad Sci USA, 1997, 94, 5955-5960 85. J.D. Walton. Deconstructing the cell wall. Plant Physiol, 1994, 104, 1113-1 118 86. F. Cote, M.G. Hahn. Oligosaccharins- structures and signal-transduction. Plant Mol Biol, 1994,26, 1379-1411 87. B.J. Cook, R.P. Clay, C.W. Bergmann, P. Albersheim, A.G. Darvill. Fungal polygalacturonases exhibit different substrate degradation patterns and differ in their susceptibility to polygalacturonase inhibitor proteins. Mol Plant Microbe Int, 1999, 12, 703-71 1 88. K.S. Ham, S.C. Wu, A.G. Darvill, P. Albersheim. Fungal pathogens secrete an inhibitor protein that distinguishes isoforms of plant pathogenesis-related endo-p-glucanses. Plant J , 1997, 11, 169-179 89. A. Darvill, C. Augur, C. Bergmann, R.W. Carlson, J.J. Cheong, S. Eberhard, M.G. Hahn, V-M. Lo, V. Marfa, B. Meyer, D. Mohnen. M.A. O’Neill, M.D. Spiro, H. Vanhalbeek. Oligosaccharins-oligosaccharides that regulate growth, development and defense responses in plants. Glycobiology, 1992, 2, 181-198 90. W.S. York. A.G. Darvill. P. Albersheim. Inhibition of 2.4-dichloro~henoxvacetic acidstimulated elongation of pea stem segments by a xyloglucan oligosaccharide. Plant Physiol, 1984. 75. 295-297 91. R.A. Creelman, J.E. Mullet. Oligosaccharins, brassinolides, and jasmonates: Nontraditional regulators of plant growth, development, and gene expression. Plant Cell, 1997, 9, 1211-1223 92. 0. Auxtova-samajova, D. Liskova, D . Kakoniova, M. Kubackova, S. Karasonyi, L. Bilisics. Inhibition of auxin stimulated short-term elongation growth of pea stem segments by galactoglucomannan-derived oligosaccharides. J Plant Physiol, 1996, 147, 61 1-613 93. L.A. Mezitt, W.J. Lucas. Plasmodesmal cell-to-cell transport of proteins and nucleic acids. Plant Mol Biol, 1996,32, 251-273 94. W-D. Reiter. Arabidopsis thaliana as a model system to study synthesis, structure, and function of the plant cell wall. Plant Physiol Biochem, 1998, 36, 167-176 95. A. Staehelin. What is a plant cell? A response. Plant Cell, 1991, 3, 553

Carbohydrates in Chemistry and Biology Beat Emst, Gerald W. Hart, Pierre Sinay copyright Q WILEY-VCH Verlag GmbH. 2000

49 Glycolipids and Bacterial Pathogenesis Ciifsoovd A . Lingwood

49.1 Introduction Cell surface glycolipids provide the preferred interface between multicellular hosts and their unicellular pathogens. Preferred, that is, by the microbial parasite. For the most part, this interface occurs in the context of an epithelial cell boundary within either the GI, respiratory or reproductive tract. The high glycolipid content typical of epithelial cells would imply some, as yet unappreciated, function to offset the retention of this apparent ‘invitation to colonize’. Karlsson has suggested that glycolipid receptors are of low affinity to function as a second mode of microbial cell attachment, downstream from recognition of a higher affinity, tissue-specific receptor [ 1, 21 and that this second receptor function mediates closer attachment to the membrane for internalization, etc. While there is significant evidence for multiple Carbohydrate receptors for pathogenic bacteria, there is as yet no evidence for the sequential use of such species. The ‘thermodynamics’ of transition to a receptor of lower affinity would be complex. However, the lower affinity of binding claimed for glycolipid receptors may be merely a reflection of the assay used to detect them. In the in vivo situation, glycolipid-ligand interactions can be of extremely high affinity. Unlike the peripheral sugars of more extended glycolipid carbohydrate structures, the glycolipid sugar sequences closely apposing the plasma membrane are restricted to glycolipids and thus offer bacteria the potential for specific targeting of these structures. Three properties may have influenced the selection of such short carbohydrate chain glycolipids as preferred mediators of bacterial pathogenesis. The first is the modulatory effect of the lipid moiety of a glycolipid and its hydrophobic environment on GSL receptor function. The sorting of cell surface lipids in the endocytic and retrograde transport pathways, highlights the second property of glycolipids which may prove attractive for bacterial pathogens, particularly those seeking to invade cells. Thirdly, the role of GSL degradation products as second messenger

8 10

49 Glycolipids and Bacterial Pathogenesis

receptor function

second messenger signal transduction

\

oo I

H

YH

OH

im

GLUCOSE

II

0

-intracellulartargeting ? -modification of ncepor functioll

Figure 1. Functional domains of a glycosphingolipid

signal transducers may offer the opportunity to modify host cell signaling pathways to microbial advantage. Indeed a glycosphingolipid can be considered in three domains (Figure 1). The first is the carbohydrate, which acts as receptor for ligand recognition; the second is the fatty acid (possibly sphingosine chain) which modulates the carbohydrate receptor function; and the third is the sphingosine (and ceramide) breakdown products which act as lipid second messengers.

49.2 Modulation of Glycolipid Receptor Function A theme that is becoming apparent in glycolipid receptor function is the ability of the lipid moiety and the membrane environment to modify the receptor function of given GSL sugar sequence [3]. Many different pathogenic bacteria have been shown to bind to the same carbohydrate sequence in in vitro binding assays e.g. GSL of the ganglioseries [4-8]. Since these different organisms show different host tissue specificity, and cause different aspects of cellular pathology, a reasonable hypothesis is that the local environment of a given glycolipid adds a particular nuance to its glycolipid receptor function. Karlsson’s early demonstration of a propensity of a particular species of propioni bacteria to bind selectively to lactosyl ceramide isoforms containing hydroxylated fatty acids within the ceramide moiety [9] provides a good example of such an effect. The same group has showed similar binding specificity for the gastric pathogen Helicobacter pylori in vitro also [lo]. Karlsson has presented some thermodynamic modeling studies to explain the preferential recognition of highly hydroxylated lactosylceramide [ 101. These studies indicate a potential hydrogen bond between the 2’ hydroxyl of the hydroxylated fatty acid and the 6 0 H of the glucose. This hydrogen bond restricts the angle around the anomeric linkage to result in the exposure of different epitopes within the lactose disaccharide for recognition at the cell surface. This effect is similar to

49.2 Modulation of Glycolipid Receptor Function

81 1

the hydrogen bond reported in the crystal structure of cerebroside between the sphingosine nitrogen and the anomeric oxygen [ 111. Such direct bonding interactions are the more extreme end of a spectrum of potentially modulatory effects of the lipid moiety and its microenvironment on the carbohydrate of glycolipids for ligand binding. Nyholm has calculated for glucosyl ceramide that the relative plane of plasma membrane in relation to a membrane bound glycolipid can have a major effect in restricting the allowable conformations around the anomeric linkage 1121. The relative plane of the membrane is determined by a balance of the characteristics of the hydrocarbon chain of the glycolipid and that of the phospholipid matrix in which it is embedded. This approach was effectively used to explain the differential binding of Esrherichia coli bearing varients of the P pili which bind the gal al-4gal disaccharide [13, 141 found in globoseries glycolipids. The different P pili bound the disaccharide preferentially in the context of different globoseries glycolipids and this varied according to whether the GSL was presented on a tlc plate or within liposomes or red cells [ 151. This concept has also been particularly useful in considering the role of glycolipid receptor recognition in verotoxin (VT) mediated pathology. This E. coli derived subunit toxin is the cause of the hemolytic uremic syndrome (HUS), a microangiopathy primarily of the pediatric renal glomerulus [16).VT binds specifically to globotriaosyl ceramide (Gbs). The receptor is present in the pediatric, but not the adult renal glomerulus [17]. Changes in both the phospholipid and glycolipid hydrocarbon chain length affect verotoxin/Gb3 binding [IS]. Changes in either of these may thus affect the sugar orientation. This, however, is likely to be a simplistic example of a more complicated modulation. Increasing the Gb3 fatty acid chain length resulted in an increase in toxin receptor binding up to a maximum chain length of C:20-22 [ 191 and unsaturation further increased binding. Further chain length decreased receptor recognition. This result is certainly compatible with Nyholm’s theoretical calculations. However, when looking at the effect of increasing Gb3 fatty acid chain length on another member of the verotoxin family, VT2c, a completely different spectrum of effects were seen in that the C: 18 species, particularly the unsaturated form, was selectively recognized [19]. If the lipid modulatory effect was solely based on the relative plane of the membrane to restrict the freedom of rotation around the anomeric linkage then the same effect on the binding of both toxins would be expected. Binding of VT1 and VT2c to a panel of deoxy galabiose glycolipids indicated that these two toxins recognized different epitopes within the terminal galabiose disaccharide [20]. This suggests therefore that, as the fatty acid chain length is increased, the different epitopes recognized by these two toxins are differentially exposed. Molecular modeling of the Gb3 binding site within the VT1 pentamer indicated two potential Gb3 binding sites (sites 1 and 11) which accommodate different conformers around the anomeric oxygen [20] as defined by Nyholm’s calculations. Site I, within the cleft between adjacent B subunit monomers, was the most favoured for VTl; however access to this site was restricted in VT2c and we proposed that site 11, a depression in the B subunit surface apposing the target cell surface, was preferentially used in this case [20], and thus that these two toxins bound different conformers of Gb3.

812

49 Glycolipids and Bacterial Pathogenesis

We have, however, been unable to demonstrate a difference in conformation of the carbohydrate moiety of C:l8:l and C:22:1 fatty acid containing Gb3 by deuterium NMR [21]. The Gb3 binding sites we defined by molecular modeling [20] equate approximately to the binding sites recently defined by co-crystallography of the VTl subunit pentamer with the lipid-free Gb3 oligosaccharide [22]. However, the orientation of the sugar within these binding sites is different in the cocrystal and site I1 was the major site occupied. In addition, a third site for the sugar in association with tryp 34, was found in these studies. However, the binding affinity for the lipid-free sugar is approximately six orders of magnitude less than that defined for the intact Gb3 glycolipid and the relevance of this cocrystal structure to the physiological mechanism of Gb3 recognition has yet to be established. NMR studies of a similar VT1Gb3 oligosaccharide complex did not find evidence of a third binding site [23] but nevertheless confirmed the location of binding at site I1 as the major site of sugar occupancy, with site I accounting for up to 15% of the bound oligosaccharide. The importance of the lipid moiety of Gb3 in the binding of verotoxin is further illustrated by the differential effects of different mechanisms of glycolipid immobilization on verotoxin recognition as a function of the lipid component. By tlc overlay synthetic galabiosyl glycolipids containing a single acyl chain are strongly recognized, but in the context of a lipid bilayer these structures are ineffective receptors [24]. In contrast, the same oligosaccharide sequence conjugated to a double acyl chain are poorly recognized on tlc overlay but are as effective as natural Gb3 when presented in a lipid bilayer format [24]. The effect of the lipid moiety on glycolipid receptor function may yet prove merely one example by which the microenvironment of the carbohydrate structure recognized has an influence on the receptor function of that sugar sequence. This possibility is illustrated by our finding that following oxidative cleavage of the sphingosine double bond of various glycolipids and coupling the resulting acids to BSA, generates neoglycoconjugates whose receptor function is quite distinct from that of the original native glycolipid. Specifically we were looking at the glycosphingolipid binding of HIV/gpl20 and demonstrated that glycolipids which were not recognized in their native state became highly effective receptors for gp120 when coupled to BSA 1251.

49.3 Stress Response and Glycolipid Receptors Growth conditions both in vivo and in vitro may affect glycolipid recognition via the stress response. We have found that both cognate and stress-induced, prokaryote and eukaryote hsp70s bind to sulfogalactosyl ceramide (SGC or sulfatide) and sulfogalactosyl glycerolipid (SGG) in vitro [26] and evidence is mounting that such proteins are expressed on the cell surface to mediate cell attachment. Hsp70 recognition of sulfatide is also an example in which the lipid moiety of the GSL plays a modulatory role in receptor function. Although the 3’ sulfate is necessary for glycolipid recognition, the 3’ sulfogalactose sugar is without inhibitory activity for this interaction [27].

49.4 Subcellular Gb3 Truflckiny

8 13

Helicobacter pylori is a microorganism for which many glycolipid and other receptors have been described [28].These include the recognition of sulfatide (3’ sulfogalactosyl ceramide) [29, 301 and sulfatide has been shown to be important in the binding of H. pylori to specific cell lines [31]. Studies from our group, however, have set this in vitro recognition of sulfatide in a different physiological context. Earlier studies on germ cell interaction had indicated that heat-shock protein on the surface of male germ cells could mediate recognition of sulfated galactolipids by these cells [26, 321. In addition we showed that mycoplasma species, infection with which can be associated with infertility, also bound specifically to sulfated glycolipids [33, 341. This interaction could be inhibited by antimycoplasma hsp70 antibodies [26]. We therefore postulated that surface hsp70s could mediate the attachment of both eukaryotic and prokaryotic cells to eukaryotic sulfogalactolipids. Thus, we considered the recognition of sulfatide by H. pylori in a different light. In our hands, untreated H. pylori did not bind to sulfatide. Considering the acidic niche of H. pylori colonization within the stomach, we theorized that it might be possible for H. pylori to undergo an acid-induced stress response to express surface hsp70s and thereby acquire the ability to bind to sulfatide. We were successful in being able to demonstrate the induction of sulfatide binding following brief exposure to low pH and that this induction could be inhibited by inhibitors of protein synthesis or in the presence of anti-hsp70 antibodies [6]. We were also able to show that a bonafide heat-shock would also induce sulfatide recognition in H. pylori. We subsequently showed that following a heat-shock, Hemophilus influenzae was also induced to recognize sulfatide [35] and that this recognition could also be inhibited in the presence of antihsp70 antibodies. Our studies with recombinant hsc70 confirmed the ability to bind sulfatide and indicated that the sulfatide binding site was within the N-terminal ATPase containing domain [27]. Thus the ability of bacteria in culture to bind to sulfatide may be a reflection of the degree of stress these organism experience in vitro. Other organisms for which sulfatide binding has been reported include Bordetella pertussis [36], Mycoplasma pneumoniae [37], M. hyopneumoniue [75]. While cell surface hsp70s may perform an adhesion function to mediate sulfatide recognition, we have found that intracellular hsp70s also bind sulfatide in in vitro assays [26]. In this context it is of interest to note that cytosolic sulfatide has now been reported [38]. Several other examples of cell surface heat-shock proteins have been reported both in eukaryotes [39, 401 and prokaryotes [41]. Several of these have been implicated in adhesion, suggesting that cell surface hsp70-mediated sulfatide adhesion may be a common, evolutionary conserved mechanism for intracellular recognition.

49.4 Subcellular Gb3 Trafficking Not only is the lipid moiety important in modulating the sugar moiety of a glycolipid for ligand binding but it may also play a role in intracellular trafficking. We have noted that multi-drug resistant variants of Gb3-containing tumor cells

8 14

49 Glycolipids and Bacterial Pathogenesis

are particularly sensitive to verotoxin in vitro [42]. This increased sensitivity is correlated with an alteration in the intracellular targeting of fluorescent labeled toxin following receptor mediated endocytosis [43]. Most cell surface glycolipids are targeted to the endocytic pathway to endosomes and lysosomes once internalized [44]. In contrast, Gb3 is transported back through the secretory pathway (Golgi, ER) by a process of retrograde transport. In highly toxin sensitive cells, internalized verotoxin is targeted to the endoplasmic reticulum, nuclear envelope and even within the nucleus itself. In the less sensitive parental cell lines, the toxin is transported to the Golgi only. Analysis of the Gb3 fatty acid composition of several such cell lines indicated that the retrograde nuclear targeting was associated with a greater preponderance of Gb3 isoforms containing short fatty acid chains (C:16, C: 18), whereas cells which target the toxin to the Golgi had a preponderance of longer chain (C:22, C:24) fatty acid containing Gb3 isoforms [43]. Thus, the shorter chain species were associated with higher sensitivity. We also showed that CD19, a transmembrane B cell marker which shows amino acid similarity to the verotoxin B subunit at its extracellular N terminus [45], was targeted intracellularly to the ER/nuclear envelope following cell surface ligation in Gb3 positive cells to initiate apoptosis [46]. CD19 was internalized in Gb3 negative cells but not to the ER/nuclear envelope and ligation of CD19 did not result in apoptosis in Gb3 negative cells. Thus transit from the plasma membrane to the ER/ nucleus may be a retrograde trafficking route for short fatty acid chain Gb3 isoforms. Cotreatment with VT1 B subunit prevented antibody ligation-mediated CD19 internalization and protected cells against antiCD19 induced apoptosis [46]. This suggests that VT1 B internalization of Gb3 prevents CD19 internalization in Gb3 positive cells. Thus Gb3 may provide a retrograde transport pathway from the cell surface to the nucleus. MDRl, the efflux pump involved in multidrug resistance of neoplastic cells, may play a direct role in this glycolipid trafficking pathway. Transfection of MDCK cells with the MDRl gene resulted in the upregulation of short chain fatty acid (C:16, C:18) containing Gb3 (GC and LC were also upregulated) and a marked increase in verotoxin sensitivity [47] which is prevented in the presence of MDRl inhibitors. MDRl has been previously shown to be capable of translocating glucosyl ceramide (GC) across the lipid bilayer [48]. GC is synthesized on the cytosolic surface of the Golgi but the transferases involved in the synthesis of more complex GSL are within the Golgi lumen [49]. We have proposed that MDRl provides a mechanism for the translocation of cytosolic GC to the lumenal side of the Golgi membrane to provide substrate for LC and subsequently Gb3 synthesis. This serves to emphasize our contention that verotoxin itself forms the basis of a new approach to Gb3 positive neoplastic disease [50]. The differential trafficking of Gb3 fatty acid isoforms is, to some degree, in contrast to the effect of Gb3 fatty acid chain length on verotoxin binding, since the shorter fatty acid chain Gb3 isoforms are, in fact, less effective at toxin binding but more effective in mediating cell killing. This suggests that the differential routing of the Gb3 fatty acid isoforms is based on distinct properties from those which modulate carbohydrate receptor function.

49.6 Gi.yrosphingo1ipid.sund Siynul Transduction

8 15

Several precedents exist for the intracellular sorting of lipids to different membranes according to fatty acid chain length [51, 521.

49.5 Model for Lipid Sorting Based on Chain Length The basis for the selection of shorter fatty acid chain containing Gb3 isoforms for ER and nuclear targeting may depend solely on thermodynamic principles. A gradient exists for the cholesterol content of membranes from the cell surface through the Golgi/ER to the nuclear envelope. This higher cholesterol content of the plasma membrane increases the dimensions of the bilayer, such that the sorting of glycolipid fatty acid isoforms may depend only on their ability to equilibrate in the membrane of the appropriate transbilayer width. Such a sorting proposal has been proposed for the retention of resident Golgi proteins within Golgi membranes [53, 541. Since cell surface glycolipids are internalized via pinching off of membrane vesicles, either caveoli or clathrin coated pits, and the formation of vesicles will tend to reduce the transbilayer width, this process could select for shorter fatty acid containing lipids (Figure 2). According to the radius of curvature of the vesicle being formed, shorter chain species will be preferentially accommodated within the vesicle as illustrated. Thus the finding that the short chain species are preferentially associated with ER/ nuclear targeting may represent a sorting mechanism based on the number of vesicular buddings and fusions that need to occur for vesicular trafficking between cell surface, endosomes, TGN, Golgi cisterni and eventual ER targeting. In this light, it is of interest to consider the data of Karlsson [lo], who showed that preferential binding of Helicohacter pylori to lactosyl ceramide species containing hydroxylated fatty acids, only occurred in membranes containing cholesterol, thus suggesting that different carbohydrate epitopes can be presented for a single glycolipid species as it transits through a trafficking pathway in which the cholesterol content is decreasing. Both verotoxin and cholera toxin (which binds to the ganglioside G M 1) follow a retrograde transport pathway inside the cell which is independent of pH gradients [55]to the Golgi. Although cholera toxin will bind to the lipid free oligosaccharide of G M I , substitution of the G M 1 oligosaccharide on different lipid species results in a variation in efficacy of reconstitution of cholera toxin responsiveness [56],suggesting that these different lipid-bound G M 1 oligosaccharide species are handled differently inside the cell.

49.6 Glycosphingolipids and Signal Transduction Bacterial attachment to host cells has been shown to subvert host/cell signal transduction pathways [57-60]. This can be to promote the process of bacterial attach-

8 16

49 Glycolipids and Bacterial Pathogenesis

AB = CD

R’tends to R*

Figure 2. Vesicular budding results in a reduction in bilayer width. Consider a plasma membrane (or indeed any membrane) from which a vesicle is about to bud. The distances A, B and C, D on opposite sides of the membrane are initially equal. With the initial formation of the invagination, the length of membrane A, B of necessity becomes less than that C, D. The radius of invagination is R’ for the membrane surface to be within the budding vesicle and R2 for that of the external vesicle surface. R2 - R ’ = t, the thickness of the membrane bilayer. As the vesicle is formed and A approaches B, R’ tends to R 2 and therefore t tends to 0. Thus the formation of a vesicle theoretically restricts the bilayer thickness of the resulting vesicle. This restricted dimension can best be accommodated by lipid species with shorter hydrocarbon chains. The degree to which this “pressure” is exerted depends upon the length of R’, ie., the size of the vesicle budding. The smaller the vesicle the greater the tendency for bilayer narrowing.

ment itself, alter host cell transcription or induce apoptosis. This latter case is of particular interest since degradation products of glycosphingolipids have been shown to be capable of inducing and mediating apoptosis in several cell systems [61, 621. Lysoglycolipids and sphingosine (and dimethyl sphmgosine) inhibit protein kinase C [63] to induce apoptosis [64, 651. Apoptosis can be induced by increased levels of ceramide [66, 671 via activation of the caspase PARP cleavage pathway [68]. Sphingosine-1-phosphate,in contrast has been found to mediate many mitotic signals [69] and can act to balance ceramide signaling [64]. There is a vast literature on sphingolipid signaling [ 6 1, 701. Several examples of the induction of apoptosis in host cells following attachment of bacteria which recognize glycolipid receptors have been reported [71, 721. Addition of bacterial sphingomyelinase is a procedure commonly used in cell biology to increase eukaryotic cell ceramide levels in signaling studies [73].

References

8 17

The expected connection between bacterial binding to cell surface glycolipids, internalization, glycolipid degradation and subsequent signal generation has recently been made [73]. In what is likely to be a precedent-setting study. in an extension of their previous work [60, 741, Svanborg’s group showed that attachment PapG adhesin bearing E. coli to cells resulted in the upregulation of intracellular ceramide to induce cytokine production. This effect was not mediated by activation of host cell plasma membrane neutral sphingo~nyelinasebut rather proposed to be derived from the degradation of the PapG glycolipid receptor itself. In conclusion, glycosphingolipids provide a channel for interaction between pro and eukaryotes and thereby the possibility of exogenous intervention. With little modification in basic structure, simple GSLs may provide a library of addresses within the cell and a road map to reach them. With some ingenuity, bacterial GSL ligands may eventually permit the direction of traffic on such signaling pathways. Acknowledgments

Studies from my laboratory have been supported by MRC grants MT13073, MT 12559 and NIH grant R01 DK52098. References 1. Karlsson, K., Angstrom, J., Bergstrom, J. and Lanee, B., Microbial interaction with animal cell

surface carbohydrates, APMIS, 1992, 100:71-83. 2. Karlsson, K. A., Microbial recognition of target-cell glycoconjugates, Curr Opin Sfruct Biol. 1995. 5:622-635. 3. Lingwood, C. A,, Aglycone Modulation of Glycolipid Receptor Function, Glycoconj J , 1996, 13:495-503. 4. Krivan, H. C., Roberts, D. D. and Ginsburg, V., Many pulmonary pathogenic bacteria bind specifically to the carbohydrate sequence GalNAcP1-4 Gal found in some glycolipids, Proc Natl Acud Sci USA, 1988, 85:6157-6161. 5. Krivan, H., Nilsson, B., Lingwood, C. A. and Ryu, H., Chlumydiu trachomutis and Chhmydia pneumoniue bind specifically to phosphatidylethanolamine in HeLa cells and to GalNacpl4GaIp1-4Glc sequences found in asialo-GM 1 and asialo-GM2, Biochem Biophys Res Commun, 1991, 175:1082-1089. 6. Huesca, M., Borgia, S., Hoffman, P. and Lingwood, C. A., Acidic pH changes receptor binding of Helicobacter pylori: a binary adhesion model in which surface heat-shock (stress) proteins mediate sulfatide recognition in gastric colonization, Infect Immun, 1996, 64:2643-2648. 7 Busse, J., Hartmann, E. and Lingwood, C. A,, Receptor Affinity Purification of a LipidBinding Adhesin from Haemophilus influenzue, J Infect Dis, 1996, 175:77-83. 8 Yu, L., Lee, K. K., Hodges, R. S., Paranchych, W. and Irvin, R. T., Adherence of Pseudomnonas ueruginosu and Cundidu alhicans to glycosphingolipid (Asialo-GM1) receptors i s achieved by a conserved receptor-binding domain present on their adhesins, Infect Immun, 1994. 6215213-5219. 9. Stromberg, N . and Karlsson, K. A,, Characterization of the binding of Propionibacterium qranulosurn to glycolipids adsorbed on to surfaces, J Biol Chem, 1990,265:11244--11250. 10. Angstrom, J., Teneberg, S., Milh, M. A , , Larsson, T., Leonardsson, 1.. Olsson, B.-M., Halvarsson; M. O., Danielsson, D., Naslund, I., Ljungh, A,, Wadstrom, T. and Karlsson, K.A,, The lactosylceramide binding specificity of Helirobacter pylori, Glycobiology, 1998, 8:297309.

8 18

49 Glycolipids and Bacterial Pathogenesis

1 1. Pascher, I. and Sundell, S., Molecular arrangements in sphingolipids. The crystal structure of cerebroside, Chem Phys Lipids,1977, 20:175-191. 12. Nyholm, P.-G. and Pascher, I., Orientation of the saccharide chains of glycolipids at the membrane surface: Conformational analysis of the glucose-ceramide and the glucose-glyceride linkages using molecular mechanics (MM3), Biochemistry, 1993, 32: 1225-1234. 13. Kihlberg, J., Hultgren, S., Normark, S. and Magnusson, G., Probing the combining site of the PapG adhesin of uropathogenic Escherichia coli bacteria by synthetic analogues of galabiose, J A m Chem Soc, 1989,111:6364-6368. 14. Nilsson, U., Striker, R., Hultgren, S. and Magnusson, G., PapG adhesin from E. coli 596 recognizes the same saccharide epitope when present on whole bacteria and as isolated protein, Bioorgun Med Chem, 1996, 4:1809-1817. 15. Stromberg, N., Nyholm, P.-G., Pascher, I. and Normark, S., Saccharide orientation at the cell surface affects glycolipid receptor function, Proc Nutl Acad Sci USA, 1991, 88:93409344. 16. Arbus, G. S., Association of verotoxin-producing E. coli and verotoxin with hemolytic uremice syndrome, Kid Intl, 1997, SI:S91-S96. 17. Lingwood, C. A,, Verotoxin-binding in human renal sections, Nephron, 1994, 66:21-28. 18. Arab, S. and Lingwood, C. A,, Influence of phospholipid chain length on verotoxin/ globotriaosyl ceramide binding in model membranes: comparison of a surface bilayer film and liposomes, Glycoconj J , 1996, 13:159-166. 19. Kiarash, A., Boyd, B. and Lingwood, C. A,, Glycosphingolipid receptor function is modified by fatty acid content: Verotoxin 1 and Verotoxin 2c preferentially recognize different globotriaosyl ceramide fatty acid homologues, J Biol Chem, 1994, 269:11138-11146. 20. Nyholm, P. G., Magnusson, G., Zheng, Z., Norel, R., Binnington-Boyd, B. and Lingwood, C. A,, Two distinct binding sites for globotriaosyl ceramide on verotoxins: molecular modelling and confirmation by analogue studies and a new glycolipid receptor for all verotoxins, Chem Biol, 1996, 31263-275. 21. Jones, D., Lingwood, C. A., Barber, K. R. and Grant, C. W. M., Globoside as a Membrane Receptor: A Consideration of the Oligosaccharide Communication with the Hydrophobic Domain, Biochemistry, 1997, 36:8539-8547. 22. Ling, H., Boodhoo, A,, Hazes, B., Cummings, M., Armstrong, G., Brunton, J. and Read, R., Structure of the Shiga toxin B-pentamer complexed with an analogue of its receptor Gb3, Biochemistry, 1998, 37:1777-17788. 23. Shimizu, H., Field, R. A,, Homans, S. W. and Donohue-Rolfe, A., Solution structure of the complex between the B-subunit homopentamer of verotoxin VT-I from Escherichia coli and the trisaccharide moiety of globotriaosylceramide, Biochemistry, 1998, 37: I 1078-1 1082. 24. Boyd, B., Zhiuyan, Z., Magnusson, G. and Lingwood, C. A,, Lipid modulation of glycolipid receptor function: Presentation of galactose al-4 galactose disaccharide for Verotoxin binding in natural and synthetic glycolipids, Eur J Biochem, 1994, 2235373-878. 25. Mylvaganam, M. and Lingwood, C. A., A convenient oxidation of natural glycosphingolipids to their “ceramide acids” for neoglycoconjugation. Bovine serum albumin-glycoceraide acid conjugates as investigative probes for HIV coat protein gpl20-glycosphingolipid interactions, J Biol Chem, 1999,274:20725-20732. 26. Boulanger, J., Faulds, D., Eddy, E. M. and Lingwood, C. A,, Members of the 70kDa heat shock protein family specifically recognize sulfoglycolipids: role in gamete recognition and mycoplasma related infertility, J Cell Physiol, 1995, 165:7-17. 27. Mamelak, D. and Lingwood, C., Expression and sulfogalactolipid binding of the recombinant testis-specific cognate heat shock protein 70, Glycoconj J , 1997, 14:715-722. 28. Lingwood, C. A,, Oligosaccharide receptors for bacteria: a view to a kill, Curr Biol, 1998, 2:695-700. 29. Saitoh, T., Sugano, K., Natomi, H., Zhao, W., Okuzumi, K., Iwamori, M. and Yazaki, Y., Glycosphingolipid receptors in human gastric mucosa for Helicobacter pylori, Eur J Gastroenterol Hepatol, 1992, 4:S49-S53. 30. Saitoh, T., Natomi, H., Zhao, W., Okuzumi, K., Sugano, K., Iwamori, M. and Nagdi, Y., Identification of glycolipid receptors for Helicobacter pylori by TLC-immunostaining, FEBS Lett, 1991, 282:385-387.

References

8 19

31. Kamisago, S., Iwamori, M., Tai, T., Mitamura, K., Yazaki, Y. and Sugano, K., Role of sulfatides in adhesion of Helicobacterpylori to gastric cancer cells, Inj2cf Immun, 1996,64:624-628. 32. Lingwood, C. A., Protein-glycolipid interactions: (I) Binding of specific germ cell proteins to sulfatoxygalactosylacylalkylglycerol, the major glycolipid of mammalian spermatogenic cells, Cun J Biochem Cell Biol, 1985, 63:1077-1085. 33. Lingwood, C. A., Quinn, P. A,, Wilansky, S., Nutikka, A,, Ruhnke, H. and Miller, R. B., Common sulfoglycolipid receptor for mycoplasma involved in infertility, B i d Reprod> 1990, 43:694-697. 34. Lingwood, C. A,, Schramayr, S. and Quinn, P., The male germ cell specific sulfogalactoglycerolipid is recognized and degraded by mycoplasma which induce male infertility, J Cell Physiol, 1990, 142:170-176. 35. Hartmann, E. and Lingwood, C. A., Brief heat shock induces a long-lasting alteration in the glycolipid receptor binding specificity of Hemophilus infiuenzue., Injkct Imrnun, 1997, 65:17291733. 36. Brennan, M. J., Hannah, J. H. and Leininger, E., Adhesion of Bordetellupertussis to sulfatides and to the GalNAcfl4Gal sequence found in glycosphingolipids, J Biol Chem, 1991,266:18827-18831. 37. Krivan, H., Olson, M., Barile, M., Ginsburg, V. and Roberts, D., Adhesion of Mycoplusmu pneumoniae to sulfated glycolipids and inhibition by dextran sulfate, J Biol Chem, 1988, 264:9283--9288. 38. Berntson, Z., Hansson, E., Ronnback. L. and Fredman, P., Intracellular sulfatide expression in a subpopulation of astrocytes in primary cultures, J Neurosci Res, 1998, 52:559-568. 39. Foltz, K. R., Partin, J. S. and Lennarz, W. J., Sea urchin receptor for sperm: sequence similarity of binding domain and Hsp 70, Science, 1993, 259:1421-1425. 40. Multhoff, G., Botzler, C., Jenner, L., Schmidt, J., Ellwart, J. and Issels, R., Heat shock protein 72 on tumor cells. A recognition structure for natural killer cells, J Immunol, 1997, 158:4341-4350. 41. Raulston, J. E., Davis, C. H., Schmiel, D. H., Morgan, M. W. and Wyrick, P. B., Molecular characterization and outer membrane association of a Chlamydia fruchomutis protein related to the hsp70 family of proteins, J Biol Chem, 1993,268:23139-23147. 42. Farkas-Himsley, H., Rosen, B., Hill, R., Arab, S. and Lingwood, C. A,, Bacterial colicin active against tumour cells in vitro and in vivo is verotoxin 1, Proc Nut1 Acad Sci, 1995, 9269967000. 43. Arab, S. and Lingwood, C., Intracellular targeting of the endoplasmic reticulum/nuclear envelope by retrograde transport may determine cell hypersensitivity to Verotoxin: sodium butyrate or selection of drug resistance may induce nuclear toxin targeting via globotriosyl ceramide fatty acid isoform traffic, J Cell Physiol, 1998, 177:646-660. 44. Kok, J. and Hoekstra, D., Glycosphingolipid trafficking in the endocytic pathway., Curr Topics Mernbr, 1994,40: 45. Maloney, M. D. and Lingwood, C. A,, CD19 has a potential CD77 (globotriaosyl ceramide)binding site with sequence similarity to verotoxin B-subunits: Implications of molecular mimicry for B cell adhesion and enterohemorrhagic Escherichiu coli pathogenesis, J Exp Med, 1994, 180:191-20 1. 46. Khine, A. A,, Firtel, M. and Lingwood, C. A,, CD77-dependent retrograde transport of CD19 to the nuclear membrane: Functional Relationship between CD77 and CD19 during germinal center B-cell apoptosis, J Cell Physiol, 1998, 1761281-292. 47. Lala, P., Ito, S. and Lingwood, C. A,, Transfection of MDCK cells with the MD Rl gene results in a major increase in globotriaosyl ceramide and cell sensitivity to verocytotoxin, J Biol Chem. in press. 48. van Helvoort, A,, Smith, A,, Sprong, H., Fritzsche, I., Schinkel, A,, Borst, P. and van Meer, G., MDRl P-Glycoprotein is a lipid translocase of broad specificity, while MDR3 Pglycoprotein specifically translocates phosphatidyl choline, Cell, 1996, 87:507-517. 49. Lannert, H., Gorgas, K., Meianer, I., Wieland, F. T. and Jeckel, D., Functional organization of the Golgi apparatus in glycosphingolipid biosynthesis, J Biol Chem, 1998, 273:2939-2946. 50. Arab, S., Russel, E., Chapman, W., Rosen, B. and Lingwood, C., Expression of the Verotoxin receptor glycolipid, globotriaosylceramide, in Ovarian Hyperplasias, Oncol Res, 1997, 9 5 5 3 563.

820

49 Glycolipids and Bacterial Pathogenesis

51. Bertho, P., Moreau, P., Morre, D. and Cassange, C., Monensin blocks the transfer of very long chain fatty acid containing lipids to the plasma membrane of leek seedlings. Evidence for lipid sorting based on fatty acyl chain length, Biochim Biopys Acta, 1991, 1070:127-134. 52. Longmuir, K. and Haynes, S., Evidence that fatty acid chain length is a type I1 cell lipidsorting signal, A m J Physiol, 1991, 26O:L44-L5 1. 53. Nilsson, T., Slusarewicz, P., Hoe, M. H. and Warren, G., Kin recognition. A model for the retention of Golgi enzymes, FEBS Lett, 1993, 330:1-4. 54. Gleeson, P. A,, Teasdale, R. D. and Burke, J., Targeting of proteins to the Golgi apparatus, Glycoconj J , 1994, 11:381-394. 55. Schapiro, F., Lingwood, C. A , , Furuya, W. and Grinstein, S., pH-independent targeting of glycolipids to the Golgi complex, A m J Physiol, 1998, 274:319-332. 56. Pacuszka, T., Bradley, R. M. and Fishman, P. H., Neoglycolipid analogues of ganglioside G M I as functional receptors of cholera toxin, Biochemistry, 1991, 30:2563-2570. 57. Hoepelman, A. I. M. and Tuomanen, E. I., Consequences of Microbial Attachment: Directing Host cell functions with adhesins, Znfect Znimun, 1992, 60:1729-1733. 58. Finlay, B. and Cossart, P., Exploitation of mammalian host cell functions by bacterial pathogens, Science, 1997, 276:718--725. 59. Segal, E., Lange, C., Covacci, A,, Tompkins, L. and Falkow, S., Induction of host signal transduction pathways by Helicohacter pylori, Proc Nut1 Acad Sci USA, 1997, 94:7595-7599. 60. Svensson, M., Lindstedt, R.: Radin, N. S. and Svanborg, C., Epithelial glycosphingolipid expression as a determinant of bacterial adherence and cytokine production, lnfect Zmmun, 1994, 62:4404-4410. 61. Spiegel, S., Foster, D. and Kolesnick, R., Signal Transduction through lipid second messengers, Curr Opin Cell Biol, 1996, 8:159-167. 62. Merrill, A,, Hannun, Y. and Bell, R., Sphingolipids and their metabolites in Cell regulation, Sphingolipids Part A: Functions and Breakdown Products, 1993, 25:1-24. 63. Hannun, Y. A. and R.M., B., Lysosphingolipids inhibit protein kinase C: Implications for the sphingolipodoses, Science, 1987,235:670-674. 64. Cuvillier, O., Pirianov, G., Kleuser, B., Vanek, P., Coso, O., Gutkind, J. and Spiegel, S., Suppression of ceramide mediated programmed cell death by sphingosine-1-phospate, Nature, 1996,381:800-803. 65. Ohta, H., Sweeney, E. A., Masamune, A., Yatomi, Y., Hakomori, S.-I. and Igarashi, Y., Induction of apoptosis by sphingosine in human leukemic HL-60 cells: A possible endogenous modulator of apoptiotic DNA fragmentation occurring during phorbol ester-induced differentiation, Cancer, 1995, 55:69 1-697. 66. Jarvis, W. D., Kolesnick, R. N., Fornari, F. A., Traylor, R. S., Gewirtz, D. A. and Grant, S., Induction of apoptotic DNA damage and cell death by activation of the sphingomyelin pathway, Proc Nut1 Acad Sci USA, 1994, 91:73-77. 67. Jarvis, W. D., Grant, S. and Kolesnick, R. N., Ceramide and the induction of apoptosis, Clin Canc Res, 1996,2:1-6. 68. Mizushima, N., Koike, R., Kohsaka, H., Kushi, Y., Handa, S., Yagita, H. and Miyasaka, N., Ceramide induces apoptosis via CPP32 activation, FEBS Lett, 1996, 39.2267-271, 69. Zhang, H., Desai, N. N., Olivera, A., Seki, T., Brooker, G . and Spiegel, S., Sphingosine-lphosphate, a novel lipid, involved in cellular proliferation, J Cell Bid, 1991, 114:155-167. 70. Hannun, Y ., Functions of ceramide incoordinating cellular responses to stress, Science, 1996, 274: 1855-1859. 71. Chen, Y. and Zychlinsky, A,, Apoptosis induced by bacterial pathogens, Microb Path, 1994, 17:203-212. 72. Jones, N. L., Shannon, P. T., Cutz, E., Yeger, H. and Sherman, P. M., Increase in proliferation and apoptosis of gastric epithelial cells early in the natural history of Helicobacter pylori infection, A m J Pathol, 1997, 151:1695-1703. 73. Hedlund, M., Duan, R.-I., Nilsson, A. and Svanborg, C., Sphingomyelin, glycosphingolipids and ceramide signalling in cells exposed to P-fimbriated Escherichia coli, Mol Microhiol, 1998, 29: 1297-1 306. 74. Hedlund, M., Svensson, M . and Nilsson, A,, Role of the ceramide-signaling pathway in cytokine responses to P-fimbriated Esrherichiu coli, J Exp Med, 1996, 183:1037-1044.

Carbohydrates in Chemistry and Biology Beat Emst, Gerald W. Hart, Pierre Sinay copyright Q WILEY-VCH Verlag GmbH. 2000

50 Glycobiology of Viruses Hildqard Geyer and Rudolf Geyer

50.1 Summary Enveloped viral particles usually contain one or more different types of surface proteins the majority of which is modified by the addition of N-linked carbohydrate chains, 0-linked carbohydrates or both. Such glycoproteins play a key role in infection by mediating the first contact with host cell receptors, virus binding and subsequent fusion of viral and cellular membranes, thus initiating infection of the cell. They direct virus morphogenesis at the budding site and may also exhibit receptor-destroying enzyme activity necessary for virus release. Many of these proteins represent major targets of the host’s humoral immune response leading to the production of neutralizing antibodies. Reported biological functions of respective carbohydrate constituents include, amongst others, initiation and/or maintenance of polypeptide folding into the biologically active conformation, protection of the polypeptide chain from proteolytic attack and modulation of the physical properties as well as the antigenicity or immunogenicity of these glycoproteins. For their biosynthesis, the host cell’s enzymatic machinery and compartmentalization mechanisms are used. Virus-encoded glycoproteins can be, therefore, considered as attractive models for studying intracellular folding, processing, targeting and transport of glycoproteins in general. Furthermore, questions regarding putative functions of glycoprotein-glycans as well as host-cell dependent influences may be investigated. In this Chapter, general aspects of viral glycoproteins are briefly addressed. In addition, the glycobiology of three virus systems, Friend murine leukemia viruses, Marburg virus and hepatitis €3 virus is dicussed.

50.2 General Aspects Many animal and human viruses are surrounded by a lipid envelope in which one or more viral glycoproteins are integrated (cf, Table 1). These glycoproteins can be

_

_

GP GP94, GP43, GP18

Marburg virus Borna disease virus

Influenza A virus Influenza C virus

Orthomyxoviridae Influenzavirus A,B Influenzavirus C

HA, NA HEF, CM2

3; c, m

G

Lyssuvirus Filoviridae Filo virus Bornaviridae

19; c, h, m 13(GP94),4(GP43), (GP18)'; c, h, m

2; c

F, HN F, H N F, H F, G, SH

Sendai Virus Newcastle disease virus Measles virus Human respiratory syncytial virus G

18(S), 9(HE); c

S, M, HE

Mouse hepatitis virus

2(E1), 2(E2); c, m 3(E1), 3(E2); c, h, m

N-glycans (number; type)

Characteristics of glycosylationh

5(EI), 1 l(E2): (c)', m

Surface glycoproteins"

E l , E2

~

8(Er"'), 1(El), 4(E2); c

~

Ern", E 1, E2

_

Tick-borne encephalitis virus Bovine viral diarrhea virus Hepatitis C virus

_

El, E2 E l , E2

_

Sindbis virus Rubella virus

Virus (example)

Vesicular stomatitis virus Rabies virus

Rhabdoviridae Vesiculotiirus

Hepucivirus Coronaviridae Coronavirus Param yxoviridae Parumyxovirus Rubuluvirus Morbillivirus Pneumovirus

Pestivirus

Togaviridae Alphuvirus Rubivirus Flaviviridae Flavivirus

Family/Genus

~

Table 1. Classification of enveloped viruses of vertebrates with emphasis on families and genera containing surface glycoproteins.

~~~

+++

+++ (G)

+ (E2)

0-glycosylation

Epstein-Barr virus

Human cytoniegalovirus

Varicella-Zoster virus

Cytomegaloeirus

Vuricelloz'irub

Herpes simplex virus 1

Hepatitis B virus

gp350/220, gpl10. gp85, gp55/80, gp35 gpUL4, gpUL33, gpUL55, gpUL74, gpUL75, gpULl00, gpULll5

GPI, GP2

Lymphocytic choriomeningitis virus

Rous sarcoma virus Murine leukemia virus Mouse mammary tumor virus Mason-Pfizer monkey virus HTLV 1 HIV-I' Human foamy virus

G I , G2 G1, G2 GI. G2

Bunyamwera virus Rift valley fever virus Hantaan virus

Lymphocryptocirus

HTLV/BLV' Lenticirus Spunzuvirus Hepadnaviridae Orthohepadnaoirus Herpesviridae Simplexvirus

D-type virus

Retroviridae ALSV' Mammalian C-type virus B-type virus

Runyaviridae Bunyacirus Phlebovirus Huntavirus Arenaviridae Arenavirus

4(gp46), l(&P21) 24(gp120), 5(gp41): c. h. m 14(gpl30/gp47); m

6(GPl), 3(GP2); c

2(G1), 2(C2); c, h, m 1(G1),3(G2); c, h. m 4(G1), l(G2); m

(+IC

7

Vaccinia virus

Poxviridae Orthopoxvirus

A33R, A34R, A56R, B5R

Surface glycoproteins”

2(A33R), 1(A34R), 5(A56R), 3(B5R); c

N-glycans (number; type)

Characteristics of glycosylationb

+ (A56R)

0-glycosylation

The number of surface glycoproteins may differ among viruses of the same genus. Type of glycosylation as determined by inhibition of glycosylation or glycan processing, susceptibility towards glycosidases, reactivity towards lectins and/or structural analysis. Since glycan types may depend on the host cells used as well as the source employed for glycoprotein analysis (e.g., lysates of virus-infected cells, supernatants of infected cells or mature virions), different structural types may occur. Numbers indicate potential N-glycosylation sites in glycoproteins as well as amounts of 0-linked sugar chains of the virus example listed; they are based on [50]and protein sequence database search. Putative N-glycosylation sites may vary among different viruses in the same genus and even among different isolates of the same virus. If not analysed in detail, the degree of U-glycosylation is classified as weak (+) or high (+++). Letters (c, h, m) designate the presence of complex, hybrid or high-mannose type N-glycans, respectively. Contradicting results. In contrast to M of mouse hepatitis virus, M of other coronaviruses may carry N-linked glycans. No classical N-glycosylation sites (N-X-S/T), but sensitive to endoglycosidase and N-glycanase. ALSV, avian leukosis-sarcoma virus; HTLV, human T-cell lymphotropic virus; BLV, bovine leukemia virus; HIV, human immunodeficiency virus

Virus (example)

Family/Geaus

Table 1 (continued)

2

7

B

3

% 0

0

x2

cr, 0

P

N

00

50.2 General Aspects

A

825

B

Figure 1. Architecture of Friend murine leukemia virus. (A) Electron microscopic picture of FMuLV particles (kindly donated by H. Frank, Max-Planck-Institute for Virus Research, Tubingen, Germany). (B) Schematic representation of the virion.

often observed as spike-like projections covering the surface of the virus particle (see, for example, Figure IA). Enveloped viruses acquire their membrane by budding at an appropriate membrane of their host cell which is in many cases the plasma membrane, but endoplasmic reticulum (ER) or Golgimembranes may be also utilized. Viral surface glycoproteins represent typically type I membrane proteins which are characterized by a cleaved N-terminal signal peptide, a C-terminal hydrophobic anchor sequence and a small C-terminal cytoplasmic domain, but type 2 proteins having an uncleaved N-terminal signal-anchor sequence or type 3 proteins with multiple membrane-spanning domains also exist. The major proportion of such glycoproteins is usually situated on the external side of the viral envelope forming the glycosylated ectodomain [ 1 , 21. 50.2.1 Functions of Viral Surface Glycoproteins Most viral envelope glycoproteins are essential in that they are required during the viral life cycle, either for virion assembly, secretion and/or infectivity [ l , 31. To infect a cell, the virion must attach to the cell surface, penetrate into the cell and become sufficiently uncoated to make its genome accessible for transcription. Virus envelope glycoproteins are usually associated with receptor binding and/or membrane fusion. For some viruses (e.g. rhabdoviruses) a single membrane glycoprotein is responsible for both functions, whereas for others like, for instance, paramyxoviruses, receptor binding and fusion activities reside in different glycoproteins. Attachment of an enveloped virus to susceptible cells requires specific binding of, at least, one viral envelope glycoprotein to a cell surface receptor. Complex viruses, however, such as vaccinia and herpes simplex virus may exhibit different types of molecules for receptor binding. Furthermore, such envelope glycoproteins may display several domains, each of which interacting with a different receptor. The

826

50 Glycobiology of’ Viruses

binding properties determine to a great extend the viral tropism. Penetration of enveloped viruses occurs by fusion of the virion envelope with either the plasma membrane (e.g. in the case of paramyxo-, herpes-, retro- or coronaviruses) or with cytoplasmic membranes after endocytosis (like orthomyxo-, rhabdo-, bunya-, arena- or togaviruses). In any case, activation of viral fusion domains usually involves conformational rearrangement of the crucial surface molecules, which can be triggered either by cleavage of biosynthetic precursor glycoproteins after binding or by the low pH of endosomes [4]. In addition to receptor binding and membrane fusion, viral envelope glycoproteins may also exhibit receptor-destroying enzymatic activities acting, for example, as sialidase (e.g. in the case of influenza A viruses) or sialate-9-0-acetylesterase (influenza C virus and some coronaviruses). Viral envelope glycoproteins are important targets of the host’s antibody response. They may carry dominant B-cell epitopes for the production of protective neutralizing antibodies. Consequently, virus glycoproteins or the respective antibodies are essential in active or passive immunization and may become increasingly important in immunotherapy of viral infections. To evade the immune response, viruses are able to generate mutant glycoproteins comprising subtle amino acid substitutions in hypervariable regions of the molecules, including addition or deletion of glycosylation sites as exemplified by influenza viruses or HIV (see following Chapters). This process can give rise to an antigenic drift which, for example, is observed in human influenza subtypes or persistent infections of HIV and hepatitis C virus.

50.2.2 Biosynthesis All viruses make use of the host cell machinery for the synthesis, folding, transport and modification of their envelope glycoproteins [ I , 3 , 5 , 61. Glycoprotein maturation may include transfer, trimming and processing of N-linked oligosaccharide side chains, the addition of 0-linked sugars to certain Ser or Thr residues, the interaction with cellular chaperones to achieve correct folding as well as defined oligomerization. The absence of virus-encoded enzymes involved in glycosylation indicates that functional virus glycoproteins can be formed simply by using the repertoire of oligosaccharide structures specified by a given host cell. Thus, the host cell mostly determines the spectrum of glycan structures found in viral glycoproteins. Utilization of potential glycosylation sites and the pattern of oligosaccharides displayed at specific sites is, however, also determined by the primary amino acid sequence, polypeptide folding, and the accessibility of the attached oligosaccharide chains to the various processing enzymes [7].

50.2.3 Function of Carbohydrate Substituents

A universal role of glycoprotein-glycosylation is still obscure [8]. There is, however, evidence that glycan chains of viral glycoproteins may fulfill, at least, two major functions. Inside the cell, i.e. during biosynthesis, they can assist in the folding of

50.2 General Aspects

827

proteins by mediating contact of nascent glycoproteins to lectin-like chaperones as has been demonstrated, for instance, for the N-glycans of HIV, hepatitis B (see below), vesicular stomatitis or influenza virus envelope glycoproteins [3, 6, 91. Outside the cell, e.g., on host cell or virus surfaces, they can provide specific recognition structures for interaction with a variety of external ligands or modulate such interactions, thus influencing viral tropism and pathogenicity. Generating or masking antigenic epitopes, they may also influence the antigenicity of virion surface structures and lower the ability of the immune system to ward off the viral infection [lo]. Differences in sialylation, for instance, may render a virus more or less susceptible to binding by the hepatic asialoglycoprotein receptor and, thus, influence its clearance. Oligosaccharide substituents can protect the glycoprotein against proteolysis and degradation and may also interfere with cleavage-dependent maturation events which are often a prerequisite for deciphering the fusion activity of a viral glycoprotein. Furthermore. clusters of 0-linked chains are thought to strongly influence the three-dimensional structures of the surface glycoproteins in that they favour extended, rod-like and possibly, solvent accessible structures rather than highly folded ones - a mechanism which could be important in stabilizing viral surface spikes and/or in mediating virus-cell interactions.

50.2.4 Oligosaccharide Diversity Viral genomes often display significant sequence diversity within a given virus population. As a consequence, encoded virus proteins are often heterogeneous in sequence. Resulting changes in the spatial arrangement of the protein or in the interactions between carbohydrate side chains adjacent to one another may influence the processing of oligosaccharides at specific locations of the viral glycoprotein [ 71. The conservation of certain glycosylation sites. however, points to an important role in determination or maintenance of structural and functional integrity of the molecule [6]. Oligosaccharides at highly conserved N-glycosylation sites appear to be important for folding and/or assembly of the glycoproteins into their biologically active forms or for protection of the molecule by masking proteolytically sensitive sites. In more variable regions of the glycoproteins, amino acid exchanges may cause the addition or deletion of glycosylation sites. Although possibly not necessary for viral survival, such differences may influence the properties of the glycoprotein and enable viruses to get adapted to different environments. To understand the life cycles of enveloped viruses and the role of the carbohydrate moieties linked to their surface glycoproteins, a detailed knowledge of the carbohydrate structure of these glycoproteins is required. Characterization of glycosylation types mostly relies on enzymatic removal of N- or 0-linked oligosaccharides or inhibition of glycosylation processing in combination with gel electrophoresis (cf. Table 1). Detailed structural analyses have been carried out in few cases only, the results of which are summarized in Table 2. Since this minireview is not intended to give a comprehensive overview of all viral glycoproteins analyzed so far, three viral systems will be highlighted which are of great biological and biomedical importance: murine leukemia viruses, Marburg virus and hepatitis B virus.

828

50 Glycobiology of Viruses

Table 2. Carbohydrate substituerits of viral envelope glycoproteins established by structure analysis. Basic structures and modifications

Viral gIycoproteins",b

High-mannose type N-dvcans' Mana6 1 Mana6 1 Mana3 1 Manp4GlcNAc Mana3 1

Sendai virus HN,F; NDV FOd; MBGV GP; Influenza virus HA; F-MuLV gp7 1; F-MCFV gp70; F-SFFVp &p55/ gp65; HIV-I gp120; HIV-2 gp130; SIV gP130

Hybrid type N-glycans" Mana6 1 Mana6 1 Manu3 Manp4GlcNAc GaIp4GlcNAcp2Mana31 - Mana6 + Gala3 - Manu6 - Mana3 - Mana6 + Gala3 - Manah Mana3 + Fuca3 Neu5Aca3/6

'

~

+

MBGV GP; F-MuLV gp71; HIV-1 gp120 F-MuLV gp7 1 F-MuLV gp7 1 Sendai virus F F-MuLV gp7 1 Sendai virus F F-MuLV gp71: F-SFFVp gp65

Complex type N-glycans GaIp4GlcNAcp2Mana61 Fuca6 1 Manp4GlcNAcp4GlcNAc Galp4GlcNAcp2Mana3 -

FUCU~

+ GlcNAcP4" + Gala3

+ GlcNAcP4' + Gala3 Fucct6 + GlcNAcP4' Fuca6 + Gala3 Fuca6 + GlcNAcj34' + Gala3 ~

~

~

-

(1 -2) GalP4

(1-2) Ga@4 + GlcNAcP4' 2 Gal04 - (1-2) GlcNAcP2 + (1-2) F U C U ~ (1-2) Neu5Aca3/6 -

-

+

Sendai virus HN,F; Influenza virus HA; F-MnLV gp71; F-MCFV gp70; F-SFFVp gp65; HIV-1 gp120; HIV-2 gp130; SIV gp130: HBV S,M,L Influenza virus HA; F-MCFV gp70; HIV-1 gp120; HBV S,M,L MBGV GP; Influenza virus HA; F-MuLV gp71; HIV-1 gp120; HIV-2 gp130 F-MuLV gp7 1 F-MuLV gp7 1 Influenza virus HA F-MuLV gp7 1 F-MuLV gp71 Influenza virus HA; F-SFFVp gp65; HIV-I gp120 Influenza virus HA Influenza virus HA Sendai virus HN,F F-MuLV gp71; F-SFFVp gp6.5; HIV-I gp120, HIV-2 gp130; SIV gp130 HBV S,M,L

- Fucuh + (1-2) Neu5Aca3/6 Galp4GlcNAcfl6/41 Influenza virus HA; F-MuLV gp71; Galp4GlcNAcp2Mana6/3 1 F U C U1 ~ F-MCFV gp70; F-SFFV, gp65; HIV-I Manp4GlcNAcp4GlcNAc gp120; HIV-2 gp130; SIV gp130 Calp4GlcNAcp2Mana3/6 1 F U C ~ Influenza virus HA; F-MCFV gp70; HIV- 1 gp 120 ~

50.2 General Aspects

829

Table 2 (Continued) Basic structures and modifications -

CalP4

+ GlcNAcP4' + Gala3 - Fucu6 + Galu3 + (1-3) NeuSAca3/6 Galp4GlcNAcp6 1 Galp4ClcNAcp2Mana6 1 Fuca6 1 Manp4GlcNAcp4GlcN Ac Galp4GlcNAcp2Mana3 1 GaliMGlcNAcp4 - Fucct6 + GlcNAcP4" + Gala3 - Fuca6 + Gala3 + Galfi4GlcNAcP3 +(1-4) Neu5Aca3/6

Viral glycoproteinsa,b HIV-2 gp130; SIV gp130 MBGV CP; HIV-1 gp120 ; HIV-2 gp130 F-MuLV gp7 I F-MuLV gp7 1 VSV G; F-MuLV gp71; F-SFFVp gp6.5; HIV-I gp120; HIV-2 gp130; SIV gp130 MBGV GP; F-MuLV gp7 1; F-MCFV gp70; F-SFFVp gp6.5; HIV-I ~ 1 2 0 ; HIV-2 gp 130; SIV gpl30 HIV-I gp120 MBGV GP; HIV-I gp120; HIV-2 gp130 F-MuLV gp71 F-MuLV gp71 HIV-1 ~ 1 2 0HIV-2 ; gp130; SIV ~ 1 3 0 ; F-MuLV gp71; F-SFFVp ~ 6 5 HlV-1 gp120; HIV-2 gp130; SIV gp130

0-glycans Galp3GalNAc - GalP3 + GlcNAcP6 + (1-2) Galp4GlcNAcP6 + (1-2) Neu5Aca3/6

+ Neu5,9Acz

MBGV GP; F-SFFVp gp65; HBV M; HBV M MBGV GP MBGV GP MHV M; F-MuLV gp71; F-SFFVp ~ 6 5 ; HBV M, MHV M

Abbreviations used: MHV, mouse hepatitis virus; NDV, Newcastle disease virus; VSV, vesicular stomatitis virus; MBGV, Marburg virus; F-MuLV, Friend murine leukemia virus; F-MCFV, Friend mink cell focus-inducing virus; F-SFFVp, polycythemia-inducing strain of Friend spleen focus-forming virus; H(S)IV, human (simian) immunodeficiency virus; HBV, hepatitis B virus hData taken from: MHV [Sl], Sendai virus [52]; NDV [53]; VSV [54]; MBGV [36]; Influenza virus [ 5 5 , 561; F-MuLV 1161; F-MCFV [19]; F-SFFVp [28, 30, 311; HIV-1 122, 23, 571; HIV-2 [21, 581; SIV [59]; HBV [49, 601 Mostly released by endo-P-N-acetylglucosaminidase H except for Sendai virus glycoprotein glycans, which were liberated by hydrazinolysis. In this case, hybrid type glycans carried an additional Fuca6-residue at the innermost GlcNAc Uncleaved precursor glycoprotein Bisecting GlcNAc-residues

830

50 Glycobiology of Viruses

The glycobiology of influenza virus and HIV will be separately reviewed in the following Chapters of this issue.

50.3 Examples 50.3.1 Friend Murine Leukemia Virus Complex Murine leukemia viruses are replication-competent mammalian C-type retroviruses. The Friend virus complex consists of at least two components, a replicationcompetent murine leukemia virus (F-MuLV) and a replication-defective spleen focus-forming virus (F-SFFV). In the absence of F-SFFV, F-MuLV may cause different types of disease, ranging from slowly developing lymphatic leukemia to progressive erythroleukemia, when injected into newborn susceptible mice. On the other hand, F-SFFV induces an acute erythroleukemia in adult mice and is, therefore, primarily responsible for the erythroproliferative effect of the Friend virus complex, whilst F-MuLV provides the helper functions required for F-SFFV replication. Furthermore, polytropic viruses (also called mink cell focus-inducing viruses, MCFVs), generated by recombination during replication of MuLV in mice, are assumed to accelerate the progression of disease [ l l , 121. The surface of replication-competent F-MuLV and F-MCFV is studded with knob-like projections consisting of surface (SU) glycoprotein, gp70/71, and transmembrane (TM) pl2E subunits (Figure l), both of which are encoded by the viral envelope gene ( e m )as a common precursor, gPr90, which is then proteolytically processed. The retroviral envelope glycoprotein (Env) mediates the first contact with host cell receptors, thus specifying host-range and interference properties, whereas the non-glycosylated TM subunit functions as an anchor for gp70/7 1 by non-covalent interaction and/or intersubunit disulfide linkage [ 131. In addition, p12E is involved in the fusion of viral and cellular membranes [ 14, 151. F-MuLV gp7 1 (F-MCFV gp70) contains eight (seven) potential N-glycosylation sites all of which are occupied by carbohydrate side-chains [16]. Studies using specific inhibitors revealed that N-glycosylation and proper processing of the highmannose type N-glycans appear to be essential for efficient transport of MuLV glycoproteins to the plasma membrane, proteolysis of gPr90 and incorporation into virions, demonstrating that N-glycosylation of the F-MuLV envelope glycoprotein is important for the production of infectious virus. Carbohydrate structure analyses of F-MuLV gp71 [16] revealed that each Nglycosylation site is characterized by a distinct pattern of high-mannose, hybrid or diantennary, triantennary and tetraantennary complex type N-glycans (cf. Table 2). In particular, the fourth attachment site from the N-terminus (Asn-302),representing the first one in the constant region of MuLV Envs, was found to carry the majority of highly processed triantennary and tetraantennary oligosaccharide species. In addition, several Ser and Thr residues, situated either in the proline-rich region of the molecule or in close proximity to Asn-302, were shown to be 0-glycosylated. This

50.3 Examples

831

finding is in so far remarkable as mutational analyses, involving Asn-302 as well as adjacent amino acids, suggest that this region plays an important role in the folding of gPr90 as well as for the stability of the interactions between TM and SU subunits [17]. Although the N-glycan attached to Asn-302 appears to be not essential for viral infectivity, elimination of this glycosylation site significantly impairs intracellular processing and transport of gPr90. Obviously, efficient maturation of the precursor molecule requires this position to be glycosylated. Since glycosylation sites homologous to Asn-302 are conserved in many retroviral envelope glycoproteins, glycosylation at this position seems to be of general functional importance [181. Carbohydrate analyses of F-MCFV Env again demonstrated the presence of high-mannose and complex type glycans [191. Similar to F-MuLV gp71, F-MCFV gp70 was found to be glycosylated in a site-specific manner with regard to the distribution of high-mannose and complex type oligosaccharide chains. Expression of the virus in mouse embryo fibroblasts (NIH3T3) or mink lung cells further revealed that, in particular, the complex type glycans of the viral glycoprotein displayed host cell specific differences with respect to oligosaccharide branching, sialylation and substitution by additional Gala-residues (Table 2). Similar cell type-specific variations in glycosylation were observed in many other viruses as, for instance, in HIV-1, HIV-2 and Marburg virus glycoproteins [20-241. F-SFFV-induced erythroleukemia in mice is either associated with a polycythemia (F-SFFVp) or a slight anemia (F-SFFVA)[ l l , 121. In both cases, the presence and expression of the viral envelope gene was shown to be essential and sufficient for induction of the disease. The precise mechanism by which SFFV causes erythroleukemia is still obscure. Available evidence, however, indicates that the SFFV enu product interacts with the erythropoietin receptor (EpoR) at the cell surface causing splenomegaly and abnormal erythroblast proliferation in susceptible mice [25, 261 which is assumed to lead to leukemic transformation after additional cytogenetic changes [27]. F-SFFV env encodes a glycoprotein with four (F-SFFVA)or five (F-SFFVp) potential N-glycosylation sites, the fifth of which is considered to be unglycosylated [ 1 I]. The primary env product formed, gp55, exhibits exclusively high-mannose type N-glycans [28, 291 and is not incorporated into virions, but accumulates mainly in the rough ER of the host cell. Only a small fraction of F-SFFVp gp55 molecules is processed in the Golgi apparatus, yielding a second glycoform of this protein, gp65, which carries complex type N-glycans as well as partially sialylated O-linked sugar chains [30, 311 and can readily be detected on the cell surface. Since mitogenic activation of EpoR by the F-SFFVp env product is assumed to require the interaction of the two proteins at the cell surface, plasma membrane expression of gp65 is considered to be a prerequisite for the leukemogenicity of F-SFFVp. Mutational analyses, in which either single or combinations of different glycosylation sites of F-SFFVp Env were inactivated [32], revealed that the viruses remained fully pathogenic when the N-terminal glycosylation sites 1 and/or 2 were mutated. Elimination of glycosylation sites 3 and 4, located in the C-terminal part of the molecule, however, rendered the virus apathogenic, independent of mutations at other sites. Carbohydrate analyses demonstrated that the primary env products of

832

50 Glycobiology of Viruses

pathogenic glycosylation mutants were, at least in Rat-1 cells, processed similar to the wild-type, whereas mutant glycoproteins lacking glycosylation sites 3 and 4 were found to be arrested at the level of the primary translation product comprising exclusively high-mannose type N-glycans [311. Hence, intracellular processing of F-SFFVp Env appears to correlate with both the in vivo pathogenicity of the respective glycosylation mutants and their ability to trigger erythropoietin-independent cell proliferation [32, 331. Noteworthy, ultrastructural localization studies showed that all enu products studied, including the one lacking all glycosylation sites, could be detected at the cell surface. Thus, the intracellular transport of wild-type and mutant F-SFFVp Envs is obviously independent from their glycosylation status, at least in Rat-1 cells [61]. Neither the intracellular maturation of F-SFFVp Env nor the formation of secondary enu products seem to be necessary for the transport to the plasma membrane. The question as to whether this is also true for the natural target cell(s) of SFFV remains to be investigated. Likewise, it is still unclear, whether the impaired interaction of mutated F-SFFVp Env with EpoR is a consequence of improper folding of the polypeptide and/or due to immature glycan structures. In summary, murine leukemia virus glycoproteins provide excellent examples for a site-specific glycosylation and a direct correlation of their glycosylation status with viral pathogenicity, thus highlighting the functional importance of the respective glycoprotein-glycans. 50.3.2 Marburg Virus (MBGV) Marburg virus and the closely related Ebola virus (EBOV) are typical representatives of the family Filoviridae in the virus order Mononegavirales. Filoviruses can cause an extremely severe form of hemorrhagic fever in humans and nonhuman primates with fatality rates of 30% for MBGV and up to 90% in the case of EBOV. Until today, neither the natural host(s) of these viruses nor the factors influencing their evolution and ecology have been identified. Although outbreaks have always been self-limited so far, the potential of these viruses to be transmitted by aerosol and the lack of immunoprophylaxis make these infections a matter of high concern in biomedical science [341. MBGV particles have a characteristic filamentous form with surface spikes consisting of only one glycoprotein, GP, which is inserted into the viral membrane as a homotrimer. GP is a typical type I transmembrane protein and the only glycosylated structural protein of the virion. Amino acid sequence determination revealed that its central region is variable and extremely hydrophilic carrying the bulk of the N- and 0-glycosylation sites whereas the two conserved, external parts of the molecule are assumed to be linked by intramolecular disulfide bridges [34, 351. MBGV GP contains 19 potential N-glycosylation sites as well as several clusters of Ser and Thr residues which might serve as 0-glycosylation sites. Oligosaccharide analyses revealed that the sugar chains of MBGV GP account for about 50% of its total molecular mass [35]. Most of the N-linked glycans are complex type species comprising, in particular, tri- and tetraantennary structures [36]. The major part of the

50.3 Examples

833

sugar moiety, however, represents O-linked species of neutral mucin-type (see Table 2). Similar high proportions of 0-glycans have been reported only for a few other viral glycoproteins, namely the respiratory syncytial virus G protein and some glycoproteins of Herpesviridae (cf. Table 1). Analytical data suggest that MBGV G P contains in fact 19 N-linked and 30 O-linked carbohydrate chains [35]. Another interesting feature of GP glycosylation is its unusual sialylation pattern. Depending on the host cell line, oligosaccharide chains either completely lack terminal sialic acid or are substituted, at best, by minute amounts of a(2-3)-linked sialic acids which are mostly attached to O-glycans [20]. The fact that GP is the only surface protein of the virion suggests a function in mediating binding to cellular receptors and fusion with cellular membranes. One of the main target organs of MBGV is the liver, and recent studies have identified the asialoglycoprotein receptor, present on hepatocytes, as a receptor candidate [ 371. This receptor recognizes with high preference tri- and tetraantennary N-linked glycans with terminal Gal residues, i.e. unsialylated species. Since MBGV can also infect many cells not expressing the asialoglycoprotein receptor like endothelial cells or cells of the mononuclear phagocytic system [38], the virus is assumed to use additional cellular receptors, as well. Fatal filovirus infections usually end with a high viremia and no evidence for an immune response. Although GP is assumed to be the major antigenic molecule of the virion, its interaction with the host’s immune system might be modulated by the high content of carbohydrates. In the case of influenza virus hemagglutinin, it has been demonstrated that the bulk of sugars covers antigenic epitopes. Since most carbohydrate substituents appear as “self” to the immune system, this concentrated glycosylation may reduce the potential of large areas of GP to act as an immunogenic target. In addition, the central region of the molecule displays an extremely high percentage of amino acid variations which possibly reflect the adaptation of filoviruses to natural host selective pressures [ 391. In conclusion, MGBV G P represents a viral glycoprotein displaying a high degree of N - and, in particular, O-glycosylation. Precise functions of these glycans, however, remain to be investigated.

50.3.3 Hepatitis B Virus (HBV) HBV is the human-pathogenic member of the Hepadnaviridae family of viruses which infects over 300 million people worldwide, causing acute and chronic hepatitis, liver cirrhosis and hepatocellular carcinoma [40]. A series of similar viruses has been isolated from a variety of animal species, including woodchucks (WHV), which serve as important models for HBV [41,42].The HBV genome encodes three related envelope (g1yco)proteins termed large (L), middle (M) and small (S) proteins, which together represent the “hepatitis surface antigen” (HBsAg). The three proteins are produced from a single open reading frame through alternative translation start sites (Figure 2B). L consists of a S and a preS domain (preS1 plus preS2), M contains the preS2 and the S domain, and S consists only of the S domain. All three proteins comprise four to five putative transmembrane w-helices (Figure 2C)

834

50 Glycobiology of Viruses

C

Filaments

B

L 1081119aa 5 5 a a

226 aa

P39 I Q N 2

N

M

9p33 I Q P S

S

p25 I gp28

Figure 2. The human hepatitis B virus (HBV) and its envelope proteins. (A), Virions and subviral particles. The virion consists of a DNA containing nucleocapsid which is surrounded by a lipid envelope containing the envelope (g1yco)proteinsL, M and S. Spheres are mainly composed of S with small amounts of M and L. Filaments contain more L and less S. (B), Schematic organization of the HBV surface proteins. L consists of 3 domains: the preS1, preS2 and S domain, M lacks the preSl sequence and S contains only the S domain. N - and 0-glycosylation sites are marked, parentheses indicate partial glycosylation. The apparent molecular masses of the respective gene products are also given in kDa. P and gp designate unglycosylated and glycosylated polypeptides. (C), Postulated topology of M. The current model, based on theoretical considerations, envisages four to five transmembrane passages. The two N-glycosylation sites carrying diantennary glycans and the 0-glycosylation site at Thr-37 ( f ) are indicated.

(v)

and a common potential N-glycosylation site at Asn-146 in the S region which is differently utilized. Interestingly, Asn- 146 is highly glycosylated in L (>90%), moderately in S (about 40%) and weakly in M (about 20%) [41] suggesting a different accessibility of the S domains to cotranslational glycosylation in the three proteins. Obviously, in L, M, and S varying proportions of the respective peptide region are exposed to the lumen of the ER [43]. Only the M protein is additionally N-glycosylated at Asn-4 within its preS2 region, whereas this site is never modified in the L protein. This divergence can also be explained by a different cotranslational topology of L and M. The preS2 domain of M is immediately translocated to the ER lumen, whereas the preS domain of L contains a cytosolic retention sequence which makes this Asn residue inaccessible for glycosylation [40, 43, 441. The HBV envelope (g1yco)proteins are not only secreted as constituents of infectious virions but also as nucleic acid-free subviral particles, which are secreted in vast excess as compared to virions. Two kinds of subviral particles are found: spheres, consisting mainly of S and only small amounts of M or L, and filaments, which contain a greater proportion of L (Figure 2A). Most of the circulating pool

50.3 Examples

835

of HBsAg in the serum of infected hosts are spheres. Subviral particles, though noninfectious, are highly immunogenic when injected into non-infected recipients and efficiently induce a neutralizing anti-HBs antibody response [40]. All three envelope (g1yco)proteins are important in the viral life cycle. The L chains are thought to carry the major receptor recognition domain on the preSl sequence. L and S have been shown to be necessary for virion secretion, while production of subviral particles is obviously driven by S alone. The role of M was enigmatic for a long time since it has been reported that M-deficient viruses can be also produced and secreted [44]. It remained open, however, whether these mutated virions would be able to induce acute or chronic hepatitis in susceptible hosts. The current model of virion formation involves the attachment of preassembled nucleocapsids to the cytosolic parts of the envelope proteins, which have been inserted into the ER membrane. Virions bud into the lumen of the ER, thereby acquiring their envelopes, and are transported through the Golgi apparatus to the plasma membrane. Recent studies provided evidence for a particular role of N-glycosylation and glycan trimming in the efficient secretion of infectious virions, enlightening especially the crucial role of the N-glycans linked to Asn-4 of the M glycoprotein [9]. Site-directed mutagenesis, knocking out this sequon, indicated that the preS2 glycans of M are absolutely required for the release of virus [45]. Using glycosylation inhibitors, it could be further demonstrated that glycosylation and the first steps in the processing pathway of respective oligosaccharides are essential for virion, but not subviral particle, formation. Consequently, HBV was not secreted when cells were incubated with tunicamycin or inhibitors of a-glucosidases I and I1 [46]. Subviral particle egress, however, was not impaired. It could be further demonstrated that M associated with the lectin-like chaperone calnexin and that this interaction was strictly dependent on the presence of a glycan at Asn-4 of M whereas the glycan at Asn-146 was not involved [47]. Since calnexin is known to act as a folding device for many glycoproteins, prevention of this interaction by inhibitors of a-glucosidases is assumed to cause misfolding of M [48]. Since HBV buds into the ER, this misfolding may impair envelopment of the viral capsid and, consequently, secretion of virions. Further processing of the oligosaccharide to the diantennary structure found in the mature molecule (Table 2) is presumably of no direct biological significance. Conversely, subviral particle release is not dependent on N-glycosylation and glycan processing events, probably due to a different secretory pathway. Beyond N-glycosylation, M is partially modified within its preS2 domain by 0linked carbohydrate chains representing T,, T and sialosyl T-antigens [47, 491 (see Table 2). The C-terminal half of the preS2 region of M (amino acids 27-47) is remarkably rich in Ser and Thr residues. Starting from patient-derived HBsAg particles, 0-glycans could be assigned to a single threonine residue, Thr-37 [49]. On the contrary, the M protein of WHV is assumed to carry two or three 0-linked substituents [411. The biological significance of the 0-glycosylation remains unknown as yet. It is clearly unrelated to the secretion of M and subviral particles, but may rather support HBV infectivity or influence immunogenicity. Noteworthy, the L (glyco)protein, though harbouring the same preS2 amino acid sequence, is completely devoid of 0-linked chains. In agreement with recent findings that WHV viremia may be suppressed by u-

836

50 Glycobiology o j Viruses

glucosidase I inhibitors in experimentally infected woodchucks [42], M glycoprotein may be considered as a promising target for antiviral therapy of hepatitis B. Acknowledgments We thank H. Feldmann, R. Friedrich and W.H. Gerlich for fruitful collaboration, W. Willems for critical reading of the manuscript and the Deutsche Forschungsgemeinschaft for financial support (SFB 272 and 535). References 1. R. W. Compans, P. C. Roberts, Viruses as model systems in cell biology, Methods Cell B i d , 1994, 43, 3-42. 2. R. W. Compans, Virus entry and release in polarized epithelial cells, Curr. Top. Microbiol. Immunol., 1995,202,209-219. 3. R. W. Doms, R. A. Lamb, J. K. Rose, A. Helenius, Folding and assembly of viral membrane proteins, Virology, 1993, 193, 545-562. 4. M. Lanzrein, A. Schlegel, C. Kempf, Entry and uncoating of enveloped viruses, Biochem. J., 1994, 302, 313-320. 5. S. Gillam, The Jeanne Manery Fisher Memorial Lecture 1994. Molecular biology of rubella virus structural proteins, Biochem. Cell B i d , 1994, 72, 349-356. 6. D. Einfeld, Maturation and assembly of retroviral glycoproteins, Curr. Top. Microhiol. Immunol., 1996,214, 133-176. 7. I. T. Schulze, I. D. Manger, Viral glycoprotein heterogeneity-enhancement of functional diversity, Glycoconjugate J., 1992, 9, 63-66. 8. A. Varki, Biological roles of oligosaccharides: all of the theories are correct, Glycobiology, 1993,3, 97- 130. 9. A. Mehta, N . Zitzmann, P. M. Rudd, T. M. Block, R. A. Dwek, Alpha-glucosidase inhibitors as potential broad based anti-viral agents, FEBS Lett., 1998, 430, 17-22. 10. K. Munk, E. Pritzer, E. Kretzschmar, B. Gutte, W. Garten, H. D. Klenk. Carbohydrate masking of an antigenic epitope of influenza virus haemagglutinin independent of oligosaccharide size, Glycobiology, 1992, 2, 233-240. 11. D. Kabat, Molecular biology of Friend viral erythroleukemia, Curr. Top. Microbiol. Zmmunol., 1989, 148, 1-42. 12. S . K. Ruscetti, Erythroleukaemia induction by the Friend spleen focus-forming virus, Baillieres Clin. Haematol., 1995, 8, 225-247. 13. D. J. Opstelten, M. Wallin, H. Garoff, Moloney murine leukemia virus envelope protein subunits, gp70 and PrlSE, form a stable disulfide-linked complex, J. ViroL, 1998, 72, 6537-6545. 14. A. Rein, J. Mirro, J. G. Haynes, S. M. Emst, K. Nagashima, Function of the cytoplasmic domain of retroviral transmembrane protein: pl5E-p2E cleavage activates the membrane fusion capability of the murine leukemia virus Env protein, J. Virol., 1994, 68, 1773-1781. 15. A. Pinter, in H. Hanafusa, A. Pinter, M. E. Pullman (Eds.): Retroviruses and disease, Academic Press, New York 1989, pp. 20-39. 16. R. Geyer, J. Dabrowski, U . Dabrowski, D. Linder, M. Schluter, H.-H. Schott, S . Stirm, Oligosaccharides at individual glycosylation sites in glycoprotein 7 1 of Friend murine leukemia virus, Eur. J. Biochem., 1990, 187, 95-110. 17. Z. Li, A. Pinter, S. C. Kayman, The critical N-linked glycan of murine leukemia virus envelope protein promotes both folding of the C-terminal domains of the precursor polyprotein and stability of the postcleavage envelope complex, J. Virol., 1997, 71, 7012-7019. 18. R. H. Felkner, M. J. Roth, Mutational analysis of the N-linked glycosylation sites of the SU envelope protein of Moloney murine leukemia virus, J. Virol., 1992, 66, 4258-4264. 19. H. Geyer, R. Kempf, H.-H. Schott, R. Geyer, Glycosylation of the envelope glycoprotein from a polytropic murine retrovirus in two different host cells, Eur. J. Biochem., 1990, 193, 855-862.

References

837

20. H. Feldmann, S. T. Nichol, H. D. Klenk, C. J. Peters, A. Sanchez, Characterization of filoviruses based on differences in structure and antigenicity of the virion glycoprotein, Virology, 1994, 199, 469-473. 21. S. Liedtke, R. Geyer, H. Geyer, Host-cell-specific glycosylation of HIV-2 envelope glycoprotein, Glycoconj. J., 1997, 14, 785-793. 22. T. Mizuochi, M. W. Spellman, M. Larkin, J. Solomon, L. J. Basa, T. Feizi, Carbohydrate structures of the human-immunodeficiency-virus (HIV) recombinant envelope glycoprotein gp120 produced in Chinese-hamster ovary cells, Biochem. J., 1988, 254, 599-603. 23. T. Mizuochi, T. J. Matthews, M. Kato, J. Hamako, K. Titani, J. Solomon, T. Feizi, Diversity of oligosaccharide structures on the envelope glycoprotein gp 120 of human immunodeficiency virus 1 from the lymphoblastoid cell line H9. Presence of complex-type oligosaccharides with bisecting N-acetylglucosamine residues, J. Bid. Chem., 1990, 265, 8519-8524. 24. T. Mizuochi, M. W. Spellman, M. Larkin, J. Solomon, L. J. Basa, T. Feizi, Structural characterization by chromatographic profiling of the oligosaccharides of human immunodeficiency virus (HIV) recombinant envelope glycoprotein gp120 produced in Chinese hamster ovary cells, Biomed. Chromatoyr., 1988, 2, 260- 270. 25. J. P. Li, H. 0. Hu, Q. T. Niu, C. Fang, Cell surface activation of the erythropoietin receptor by Friend spleen focus-forming virus gp55, J. Virol., 1995, 69, 1714-1719. 26. E. Gomez-Lucia, Y. Zhi, M. Nabavi, W. Zhang, D. Kabat, M. E. Hoatlin, An array of novel murine spleen focus-forming viruses that activate the erythropoietin receptor, J. Virol., 1998. 72, 3742-3750. 27. Y. Ben-David, A. Bernstein, Friend virus-induced erythroleukemia and the multistage nature of cancer, Cell, 1991, 66, 831-834. 28. K.-H. Strube, H.-H. Schott, R. Geyer, Carbohydrate structure of glycoprotein 52 encoded by the polycythemia-inducing strain of Friend spleen focus-forming virus, J. Bid. Chem. 1988, 263, 3762-3771. 29. J. Volker, H. Geyer, R. Geyer, Glycosylation of glycoprotein 5 5 encoded by the anaemiainducing strain of Friend spleen focus-forming virus, Glycoconj. J., 1994, I I , 133- 139. 30. K.-H. Strube, R. Geyer, Carbohydrate structure of glycoprotein 65 encoded by the polycythemia-inducing strain of Friend spleen focus-forming virus, Eur. J . Biochern., 1989, 179, 441 -450. 3 1. A. Freis, S. Rau, R. W. Friedrich, R. Geyer, Glycosylation pattern and processing of envelope gene products encoded by glycosylation mutants of Friend spleen focus-forming virus, Glycobiology, 1993, 3, 465-473. 32. S. Rau, R. Geyer, R. W. Friedrich, The role of gp55 N-glycosylation in pathogenesis of Friend spleen focus- forming virus, J. Gen. Virol., 1993, 74, 699-705. 33. Y. Wang, S. C. Kayman, J. P. Li, A. Pinter, Erythropoietin receptor (EpoR)-dependent mitogenicity of spleen focus-forming virus correlates with viral pathogenicity and processing of env protein but not with formation of gp52-EpoR complexes in the endoplasmic reticulum, J. Virol., 1993, 67, 1322-1327. 34. H. Feldmann, H. D. Klenk, Marburg and Ebola viruses, Adv. Virus Rex, 1996, 47, 1-52. 35. C. Will, E. Miihlberger, D. Linder, W. Slenczka, H . D. Klenk, H. Feldmann, Marburg virus gene 4 encodes the virion membrane protein, a type I transmembrane glycoprotein, J. Virol., 1993, 67, 1203-1210. 36. H. Geyer, C. Will, H. Feldmann, H. D. Klenk, R. Geyer, Carbohydrate structure of Marburg virus glycoprotein, Glj~cobiology,1992, 2, 299-3 12. 37. S. Becker, M. Spiess, H. D. Klenk, The asialoglycoprotein receptor is a potential liver-specific receptor for Marburg virus, J. Gen. Virol., 1995, 76, 393-399. 38. H. J. Schnittler, H. Feldmann, Molecular pathogenesis and filovirus infections: Role of macrophages and endothelial cells, Curr. Top. Microbiol. Immunol., 1998, 235, 175~-204. 39. A. Sanchez, S. G. Trappier, U. Stroher, S. T. Nichol, M. D. Bowen, H. Feldmann, Variation in the glycoprotein and VP35 genes of Marburg virus strains, Virology, 1998, 240, 138-146. 40. M. Kann, W. H. Gerlich, in L. Collier, A. Balows, M. Sussman (Eds.): Topley & Wilsons Microbiology and Microbial Injections, Vol. l , Arnold. London, Sidney, Auckland 1998, pp. 745774. 41. T. K. Tolle, D. Glebe, M. Linder, D. Linder, S. Schmitt, R. Geyer, W. H. Gerlich, Structure

838

50 Glycobiology of Viruses

and glycosylation patterns of surface proteins from woodchuck hepatitis virus, J. Virol., 1998, 72, 9978-9985. 42. T. M. Block, X. Lu, A. S. Mehta, B. S. Blumberg, B. Tennant, M. Ebling, B. Korba, D. M. Lansky, G. S. Jacob, R. A. Dwek, Treatment of chronic hepadnavirus infection in a woodchuck animal model with an inhibitor of protein folding and trafficking, Nut. Med., 1998, 4, 610-614. 43. R. Prange, R. E. Streeck, Novel transmembrane topology of the hepatitis B virus envelope proteins, EMBO J , 1995, 14, 247-256. 44. V. Bruss, D. Ganem, The role of envelope proteins in hepatitis B virus assembly, Proc. Natl Acad. Sci. U. S. A . , 1991, 88, 1059-1063. 45. A. Mehta, X. Lu, T. M. Block, B. S. Blumberg, R. A. Dwek, Hepatitis B virus (HBV) envelope glycoproteins vary drastically in their sensitivity to glycan processing: evidence that alteration of a single N-linked glycosylation site can regulate HBV secretion, Proc. Nut/ Acad. Sci. U S. A , , 1997,94, 1822-1827. 46. X. Lu, A. Mehta, R. A. Dwek, T. Butters, T. Block, Evidence that N-linked glycosylation is necessary for hepatitis B virus secretion, Virology, 1995, 213, 660-665. 47. M. Werr, R. Prange, Role for calnexin and N-linked glycosylation in the assembly and secretion of hepatitis B virus middle envelope protein particles, J. Virol., 1998, 72, 778-782. 48. X. Lu, A. Mehta, M. Dadmarz, R. A. Dwek, B. S. Blumberg, T. M. Block, Aberrant trafficking of hepatitis B virus glycoproteins in cells in which N-glycan processing is inhibited, Proc. Nut1 Acad. Sci. U. S. A . , 1997, 94, 2380-2385. 49. S. Schmitt, D. Glebe, K. Alving, T. K. Tolle, M. Linder, H. Geyer, D. Linder, J. PeterKatalinic, W. H. Gerlich, R. Geyer, Analysis of the preS2 N- and 0-linked glycans of the M surface protein from human hepatitis B virus, J. Biol. Chern., 1999, 274, 11945-11957. 50. B. N. Fields, D. M. Knipe, P. M. Howley, Fields Virology, Vol. I , 2, Lippincott-Raven Publishers, Philadelphia 1996. 51. H. Niemann, R. Geyer, H.-D. Klenk, D. Linder, S. Stirm, M. Wirth, The carbohydrates of mouse hepatitis virus (MHV) A59: Structures of the 0-glycosidically linked oligosaccharides of glycoprotein E l , EMBO J., 1984, 3, 665-670. 52. H. Yoshima, M. Nakanishi, Y. Okada, A. Kobata, Carbohydrate structures of HVJ (Sendai virus) glycoproteins, J. Biol. Chem., 1981, 256, 5355-5361. 53. S. Diabate, R. Geyer, S. Stirm, Structure of the major oligosaccharides in the fusion glycoprotein of Newcastle disease virus, Eur. J. Biochem., 1984, 139, 329-336. 54. C. L. Reading, E. E. Penhoet, C. E. Ballou, Carbohydrate structure of vesicular stomatitis virus glycoprotein, J. Biol. Chem., 1978, 253, 5600--5612. 55. R. Geyer, S. Diabate, H. Geyer, H. D. Klenk, H. Niemann, S. Stirm, Carbohydrates of influenza virus. Structure of the oligosaccharides linked to asparagines 406 and 478 in the hemagglutinin of fowl plague virus, strain Dutch, G/ycoconj. J., 1987, 4 , 17-32. 56. W. Keil, R. Geyer, J. Dabrowski, U. Dabrowski, H. Niemann, S. Stirm, H. D. Klenk, Carbohydrates of influenza virus. Structural elucidation of the individual glycans of the FPV hemagglutinin by two-dimensional ' H n.m.r. and methylation analysis, EMBO J., 1985, 4, 27112720. 57. H. Geyer, C. Holschbach, G. Hunsmann, J. Schneider, Carbohydrates of human immunodeficiency virus. Structures of oligosaccharides linked to the envelope glycoprotein 120, J. Biol. Chem., 1988,263, 11760-1 1767. 58. S. Liedtke, M. Adamski, R. Geyer, A. Pfiitzner, H. Riibsamen-Waigmann, H. Geyer, Oligosaccharide profiles of HIV-2 external glycoprotein: Dependence on host cells and virus isolates, Glycobiology, 1994, 4, 477-484. 59. H. Holschbach, J. Schneider, H. Geyer, Glycosylation of the envelope glycoprotein gp130 of simian immunodeficiency virus from sooty mangabey (Cercocebus atys), Biochern. J., 1990, 267, 759-766. 60. B. L. Gillece-Castro, S. J. Fisher, A. L. Tarentino, D. L. Peterson, A. L. Burlingame, Structure of the oligosaccharide portion of human hepatitis B surface antigen, Arch. Biochem. Biophys, 1987,256, 194-201. 61. A. Pahmeier, J. Roth, R. Geyer, unpublished.

Carbohydrates in Chemistry and Biology Beat Emst, Gerald W. Hart, Pierre Sinay copyright Q WILEY-VCH Verlag GmbH. 2000

51 The Glycobiology of Influenza Viruses Stephen J. Stray and Gillian M. Air

51.1 Introduction The myxoviruses were originally named for their propensity to bind to mucin. All members of the family Orthomyxovirida: (influenza viruses) and most of the family Paramyxoviridz bind to sialic acid on cell surfaces to initiate infection. However, the viral membrane-anchored spike proteins are themselves glycosylated, and terminal sialic acid causes self-aggregation unless it is removed by a viral “receptor destroying activity”. The presence of both receptor binding and destroying activities in the same particle is unique to viruses of the ortho- and paramyxovirus families and presents a biological conundrum in understanding why the viruses have retained use of a receptor molecule which is also expressed on their own glycoproteins. The orthomyxoviruses are now defined as enveloped viruses having segmented negativesense single-stranded genomic RNA and replicating in the nucleus. They are divided into three serological types: influenza A, B and C. Influenza epidemics have occurred seasonally for millennia. The influenza pandemic of 1918-1 9 19: responsible for at least 30 million deaths worldwide, remains the single most lethal biological event in the course of recorded history. Influenza A and B virions contain two major integral membrane glycoproteins which interact with cellular glycans: hemagglutinin (HA), which binds cell surface sialic acid and also mediates fusion of host and viral membranes, and neuraminidase (NA, EC 3.2.1.18), which cleaves sialic acid moieties from oligosaccharides. The influenza C receptor binding activity recognizes 9-0-acetylated sialic acids, and the receptor destroying activity is sialate 9-0-acetylesterase (EC 3. I . 1.53). Receptor binding, destruction, and fusion functions are provided by a single glycoprotein,

HEF. The segmented genome allows for interchange of cognate segments between viruses of the same type (reassortment). Reassortment of type A viruses is thought to be important in generating pandemic strains by introducing into the human population viruses with novel HA and sometimes NA. As HA and NA are the

840

51 The Glycobiology of Influenza Viruses

main antigenic determinants of the virus, the resulting virus is not recognized by antibodies induced by earlier viruses and thus evades the host immune system (antigenic shift). The high error rate of the viral RNA-dependent RNA polymerase allows selection of mutations which are resistant to existing antibodies (antigenic drift), allowing repeated infections of the same host by related viruses. Influenza A viruses infect other animals and birds; these species may be sources of novel HA and NA genes for introduction into human influenza A [ 1). At least 15 HA subtypes (HlLH15) and nine NA subtypes (Nl-N9) have been isolated. To date only Hl-H3 and NlLN2 have been found in human influenza A viruses, but epizootic outbreaks have occurred where humans have been infected with animal or bird viruses. A recent example was the lethal epidemic in Hong Kong in 1997. The virus isolated from human casualties was an avian H5N1 influenza but it had not acquired the capacity to transmit from human-to-human up until the time it was eradicated by total chicken slaughter in December 1997 [2]. Human influenza virus was isolated in 1933 when a virus was successfully transmitted to a ferret and then inadvertently retransmitted to a human subject [3]. Hirst demonstrated both receptor binding and destruction activities by observing that influenza virus caused chicken red blood cells to agglutinate at low temperatures, and on warming disaggregate by a mechanism which rendered then incapable of a second round of agglutination [4]. Extensive studies by Burnet and colleagues [ 5 , 61 lead to the identification of neuraminic acid as the HA ligand and NA substrate, with the structure being definitively determined by Gottschalk in 1957 [7].The role of sialic acid in infection is still not resolved; Fazekas de St. Groth proposed in 1947 that sialic acid recruits virus to sites on the cell resulting in ingestion or “viropexis” [a]. We have reexamined the issue of influenza receptor usage and propose that a secondary receptor may be used, perhaps in a two-stage entry process [9]. The receptor for influenza C is assumed to be 9-0-acetyl sialic acid, based on the esterase activity of its receptor-destroying enzyme [ 101. In vivo, 9-0-acetyl sialic acids are added selectively to glycoproteins and glycolipids in a limited range of cell types, although these cell types are reasonably widely distributed [ 111, including mucosal epithelia. A putative receptor protein, a 40 kDa mucin heavily modified by 9-0-acetyl sialic acid, is present on apical surfaces of permissive MDCK cells and is rapidly and constitutively endocytosed [ 121. X-ray crystal structures of type A (H3) HA, type C HEF protein and type A and B NA proteins have given unique insights into sialic acid binding and cleavage and guided the design of inhibitors based on knowledge of interactions between the proteins and their substrates.

51.2 Receptor Binding Proteins: Influenza A HA and Influenza C HEF The HA proteins of type A and B, influenza virus and the HEF protein of type C are trimeric type 1 membrane glycoproteins. Each is synthesized on ER-associated ribosomes as a precursor protein of 60-70 kDa which undergoes cotranslational N-

51.2 Receptor Binding Proteins: Infuenzu A H A and Infuenzu C HEF

841

linked (but not 0-linked) glycosylation. The precursor (HA0 or HEFo) is processed to remove the N-terminal signal peptide (presumably by signal peptidase) and by trypsin-like proteases to cleave HA0 (HEFo) into HA, (HEF1) and HA2 (HEF2). HA1 contains the sialic acid binding pocket and thus is presumably responsible for recruitment of the virus to the cell surface; in HEF the esterase enzyme activity is also located in HEFI. HA2 and HEFZ contain the transmembrane domains and the hydrophobic fusion peptide thought to penetrate the host cell membrane (Figure I). Type A HAS have 3-9 N-linked glycosylation sites; HA, is glycosylated in all strains, while HA2 of some strains contains no carbohydrate. Compositional analysis revealed that glycosylation is somewhat site-specific; in egg-grown HI and H3 viruses the site at the subunit interface has a simple carbohydrate chain (GlcNAc and mannose only), while more exposed chains have complex type carbohydrate of varying compositions, including fucose residues [ 13, 141. Glycan structures of HI HA have been determined in detail from virus grown in Madin Darby bovine kidney cells (MDBK) [15]. Sites toward the tip of the spike exhibit mainly diantennary complex carbohydrate, while the site closest to the viral membrane has tetra- (and some tri-) antennary glycans. All chains, except that at the subunit interface, contained fucose and most were terminated with ul-3 linked galactose. This was interpreted as being due to removal of sialic acid to generate acceptors for galactosylation, but may reflect low activity of sialyltransferase in this cell line. Size and complexity of carbohydrate substituents is probably governed by accessibility to glycosyltransferases in the folded state; the presence of fucose (egg and MDBKderived protein) and a-linked galactose (MDBK cells) is presumably not specific to viral proteins but a common feature of carbohydrate in these cells. The nascent fusion cleavage site is exposed on the surface of the HA0 trimer allowing processing by cellular or extracellular proteases. The range of proteases able to cleave HA0 contributes to the pathogenicity of influenza A viruses: changes that insert or expose extra basic amino acids in the cleavage loop allow a broader range of proteases to cleave HA0 and thus increases the range of cell types in which the virus can replicate. Such mutations have resulted in avian viruses of increased lethality [I61 and were also seen in the 1997 H5N1 human influenza outbreak in Hong Kong [ 171. Influenza virions appear to be taken up via a vesicle-mediated pathway [18]. As the virus-containing endosome is acidified, the HA undergoes a conformational change to allow fusion of viral with host cell membrane. X-ray crystallographic and electron microscope data suggest that the HA, subunits of the trimer open as petals in a flower to reveal the hydrophobic fusion peptide, which undergoes enormous change in conformation driven by rearrangement of helices (“spring-loaded”) to allow penetration of the host cell membrane [19]. Type A HAS of different subtypes and type B HA share approximately 30% sequence identity, but there is only 12%1sequence identity with type C HEF. However, the type A H3 HA and type C HEF three-dimensional structures are highly conserved [20, 211. HA and HEF are modular three domain structures comprised of the receptor-binding domain, the fusion/transmembrane domain and a third domain which in HEF provides the esterase activity. In HA, this domain appears to

842

51 The GEycobio1og.y of Influenza Viruses

a

nding

influenza C HEF

b HEF 40

HA 10

73

150

310

50

127

287

366

432

Figure 1. Structural homology of influenza A hemagglutinin (HA) and influenza C hemagglutininesterase-fusion (HEF) glycoproteins. (a) X-ray crystal structures of showing locations of receptor binding, esterase and fusion domains. The fusion domain is composed of polypeptide sequences designated F1, F2 and F3; the HEF esterase domain is composed of polypeptide sequences El and E2 (unique to HEF) and E’ which is structurally conserved between HEF and HA. (b) Schematic of HA and HEF sequences showing locating of polypeptides contributing to conserved structural features. Domain boundaries arc approximate; figure is not to scale. Adapted from [21] using coordinate file HEFl kindly supplied by Dr. Don Wiley and the program Look v2.0.

play a structural role only, although a second sialic acid binding site was found in the trimer interface associated with this domain [22]. The apparent mutability of surface residues while internal scaffolding structures are retained is critical to influenza biology in several respects; not only does this allow evolutionary insertion or

51.2 Receptor Binding Proteins: Injuenza A H A and In$uenzu C H E F

843

deletion of functionality in its glycoproteins, but also allows the virus to escape antibody responses by mutation of residues in surface loops without deleterious effects on global structure [23]. This is demonstrated by the comparison of HA and HEF and the ability to transfer the hemagglutinating activity of an avian virus NA to the framework of a non-hemagglutinating mammalian virus NA [24].

51.2.1 Structure of Receptor Binding Domain and Mechanism of Sialic Acid Recognition The receptor binding domain of both HA and HEF is composed of antiparallel sheets folded over each other in a “Swiss roll” conformation. The receptor binding sites for both proteins are constituted by loops inserted into this structural framework. Sialic acid recognition is mostly by van der Waals interactions and some hydrogen bonds so binding is driven by shape complementarity. Different HAS have different substrate preferences, based mainly upon the linkage between sialic acid and galactose; Receptor preference maps to amino acids in the binding pocket, especially residue 226 in H3 HA1. The early “Hong Kong” flu H3 HA X-31 (A/Aichi/2/68 high-growth reassortant), specific for Neu5Aca2-6Gal- has leucine at this position, whereas a variant which preferentially binds NeuSAca2-3Gal- has glutamine. Surprisingly, the molecular basis of the ligand discrimination is still unclear. The Gln may contribute a further hydrogen bond as well as altering the shape of the binding pocket [25]. Interactions with other residues of the sugar chain are also possible; shallow surface channels appear to accommodate sugar chains terminated in either a2-3 or a2-6-linked sialic, acid, although the position of the sialic acid is slightly different in each case as predicted by the alterations seen in the binding site structures [26]. Remarkably, the most chemically distinct feature of sialic acid, the C 1 carboxylate, is not exploited in substrate recognition by charge-charge interaction with basic residues. This observation may be very significant in the role of HA in infection; the ligand site may bind other substrates in addition to sialic acid, or excessively tight ligand binding may interfere with an important aspect of HA function such as transient binding and release of sialic acid as the virus “browses” the cell surface in search of a secondary receptor [9]. In HEF interactions with sialic acid, this may allow a general interaction of the virus with cell surface sialic acids such as Neu5Ac while searching for its specific ligand Neu5,9Ac2. The HEF crystal structure accounts for the selectivity of the HEF binding site for Neu5,9Acz due to the interaction of the methyl group of the C9 acetate with a hydrophobic pocket formed by two phenylalanine rings and a proline residue, and hydrogen bonds between the ester and carbonyl oxygen to tyrosine and arginine respectively. Consistent with biochemical data showing a lack of linkage preference by HEF, the loop analogous to the loop in HA containing residues implicated in linkage selectivity is truncated in HEF. Mutagenesis analysis of sequences in type B HA analogous to the binding pocket helix found in H3 HA and HEF suggests that this structure has a different

844

51 The Glycobiology of Influenza Viruses

conformation and function in B HA and may contribute strongly to type B HA antigenicity [27]. 51.2.2 HEF Esterase Domain and Mechanism of Cleavage The esterase domain of HEF is composed of three polypeptide segments: two Nterminal to the receptor binding domain (El and E’, structurally homologous to the vestigial domain in HA); the third (E2) is present as an insertion between the receptor binding and fusion domains of HEF. Non-cleavable Neu5,9Aq analogues can act as receptors for influenza C binding, fusion, and entry, but drug inhibition of esterase activity prevents both infection and hemolysis [28, 291. Inhibitors may act to lock the esterase domain so that pH-induced structural alterations at the fusion domain pocket cannot be propagated along the molecule to HEFl . A catalytic triad in HEF consisting of serine 57 (S71 in precursor sequence) histidines 354 or 355 (H368 and H369 in precursor sequence) and aspartic acid 247 (D261 of precursor sequence) was proposed on the basis of mutagenesis data [30]. However, the crystal structure reveals that Asp247 is remote from the esterase site. Mutations at this residue presumably perturb the global structure and illustrate the perils of interpreting mutagenesis results without good controls for structural integrity. In the crystal structure, the active site contains Ser57, His355, and Asp352 in the appropriate geometry to form a charge relay network to abstract a proton from Ser57, which is positioned for nucleophilic attack of the carbonyl carbon for hydrolysis of the C9 acetate ester bond. His354 probably plays a role in maintaining the correct geometry of the active site although it is too far from the substrate to interact directly.

51.3 Influenza NA (types A and B) The NA protein of types A and B is a homotetrameric type I1 transmembrane protein having a short N-terminal cytoplasmic domain, a hydrophobic signal/ anchor domain, an extended “stalk” domain, and globular “head” bearing the enzyme active site and antigenic determinants. Type A NAs of different subtypes share about 50% sequence homology, while there is approximately 25% sequence homology between NAs of types A and B. Studies of oligosaccharides of A/Tokyo/3/ 67 (N2) heads from egg-grown virus revealed that four of five potential N-linked glycosylation sites were occupied; the two sites closest to the subunit interfaces had only simple carbohydrates containing GlcNAc, mannose, and trace amounts of galactose and fucose. The other two more exposed sites had complex glycan chains containing GlcNAc, mannose, galactose, fucose, and, for the most exposed site on the surface, GalNAc [31]. B/Lee/40 NA has six potential glycosylation sites, two in the stalk region and four in the head, of which two are occupied. B/Lee/4O NA glycans were shown to contain GlcNAc, mannose, glucose, and fucose [32].

51.3 Influenza N A (types A and B )

845

NA heads are readily prepared from whole virus by proteolysis and generally retain their tetrameric structure and enzymatic activity. X-ray crystal structures of NAs of both type A and B viruses are now available, as are several bacterial sialidase structures. There is strong conservation of the basic “P-propeller” structural motif (six topologically identical four-stranded p sheets arrayed around a central axis) between the NA monomer and the core of the bacterial sialidases. The active site is a pocket in the top of the monomer. The P-propeller fold is shared by numerous other glycohydrolases with widely varying substrates [33]. As for HA and HEF, the residues involved both in antibody interactions and in substrate binding and hydrolysis are found in loops which lie outside core topological elements. Substrate hydrolysis occurs at a site protected from antibody binding while antigenic sites are located on the surface of the molecule, as predicted in the “canyon hypothesis” [34] (Figure 2). In addition to the high structural conservation between sialidases, many amino acids in the active site are invariant among influenza NAs (five of these are conserved between influenza NA and bacterial sialidases). These are identically positioned in the NA structures determined to date and are described as the “first shell” or “second shell”. First shell amino acids directly bond to sialic acid in the active site, while second shell residues are thought to ensure correct active site geometry. 51.3.1 Mechanism of Sialic Acid Cleavage The mechanism of sialidases is still controversial. Crystal structures of type B NA complexed with sialic acid showed the presence of a flattened sugar ring in a similar conformation to that found in crystals complexed with 2-deoxy-2,3-dehydro-Nacetylneuraminic acid (NeuSAc2en, DANA), a competitive inhibitor of NA and bacterial sialidases [35], and it was suggested that the bound species was the oxocarbonium ion transition state [36] stabilized by strong interactions with active site residues. Attack by water of the transition state oxocarbonium ion in solution would yield both c1 and P anomer products as either face could undergo attack. However, N2 NA has been shown by NMR to release its product as the a anomer (rapidly converted to the more stable p anomer in solution) [37]. Presumably an incoming water molecule has access only to the face of the bound intermediate sugar ring exposed by the polysaccharide leaving group, thus the product would have the same c1 anomeric configuration as the original polysaccharide. Kinetic isotope effects support this model [ 381. By contrast, Sulmonellu typhimurium sialidase releases sialic acid as the P anomer, suggesting that hydrolysis by this enzyme proceeds via a different mechanism, although a similar intermediate is possible if a water molecule is bound with the substrate and attacks the intermediate from below (reviewed in [39]). Mutagenesis of B/Lee/40 NA based on the crystal structure revealed the critical role of the hydroxyl group of tyrosine 409, positioned to interact with the positively charged C2 carbon postulated in the carbocation intermediate. Conservative substitution of phenylalanine at this position yielded enzyme that was correctly folded

a

b

116

$3374 Figure 2. Structure of neuraminidase and its substrate binding site. (a) NA monomer of A/tern/ Australia/G70c (N9) showing location of active site (stippled) and epitope (circles) for neutralizing antibody NC41 (adapted from [ 5I]). Amino acids contacting neutralizing monoclonal antibody NC41 are indicated. Dark blue circles: heavy chain contacts; light blue circles: light chain contacts; purple circles: residues contacting both heavy and light chains; red circles: contact residues critical for antibody binding [52].The four-fold intersubunit axis is at the bottom right corner. (b) Sialic acid bound in active site of B/Lee/40 NA, showing interaction with first shell amino acid sidechains. Tyr 409 is responsible for stabilizing the charge on the hydrolytic transition state via interactions between its hydroxyl oxygen and the sialosyl oxocarbonium ion intermediate [39, 431.

51.4 Function

of

Viral Receptor Destroyirzy Enzymes

841

and transported but inactive [39]. Mutation of aspartate 149 to glutamate did not change the broad pH range of the enzyme but drastically reduced specific activity; thus this residue does not act as the previously-postulated proton donor for hydrolysis. Recent data indicating that sugar ring structures are elastic rather than rigid suggests that structural transitions via planar intermediates may be much more energetically favorable than previously thought [40]. The planar intermediate has been exploited in the development of inhibitors of NA and of virus growth; several inhibitors based on planar molecules have been shown to inhibit in the micromolar range [41], for review see [42]. Detection of Neu5Ac2en as a product of cleavage by NA and other sialidases [43, 441 may indicate an additional mechanism, akin to the dehydration mechanism of the hyaluronic acid and proteoglycan lyases, which generates a planar reaction product with a double bond in the sugar ring.

51.4 Function of Viral Receptor Destroying Enzymes Receptor destroying functions are required at least in part because sialic acids are incorporated into the glycoproteins and glycolipids of virions. Failure to remove sialic acids from the virion or derivatization of the virion with sialic acid in vitro causes the formation of large aggregates of virus particles with a corresponding reduction in infectivity [45-481. Elegant studies to determine the sialic acid linkage specificity of type A HA and NA indicate that the specificity of HA and NA of natural viruses is not necessarily matched, but that the specificities drift over time until both HA and NA recognize sialic acids in the same linkage [49], the same phenomenon is observed with reassortant viruses generated in the laboratory to have known HA and NA specificities [50]. Despite 65 years of study of influenza virus, the reason ortho- and paramyxoviruses have evolved to bind cells via molecules which are incorporated into virions and which therefore must be removed is still unclear. Sialic acid may be a convenient means for type A and B viruses to initiate contact with the cell surface in order to find a rarer receptor required in a subsequent stage of entry. Influenza C may have adapted to recognize 9-O-acetyl sialic acids which may act as markers for cell types in which the virus has adapted to replicate. Sialic acids in general may identify cells capable of rapidly synthesizing and transporting glycoproteins and thus be a useful means for the virus to select cells capable of producing large amounts of viral glycoprotein required for assembly of progeny virions to propagate infection.

Acknowledgments Work in the authors’ laboratory was supported by grants AI-18203 and AI-31888 from the National Institutes of Health (USA).

848

51 The Glycobiology of Influenza Viruses

References 1. Laver WG, Webster RG. (1973) Studies on the origin of pandemic influenza. 111. Evidence implicating duck and equine influenza viruses as possible progenitors of the Hong Kong strain of human influenza. Virology, 51, 383-391. 2. Shortridge KF, Zhou NN, Guan Y, Gao P, Ito T, Kawaoka Y , Kodihalli S, Krauss S, Markwell D, Murti KG, Norwood M, Senne D, Sims L, Takada A, Webster RG. (1998) Characterization of avian H5N1 influenza viruses from poultry in Hong Kong. Virology 252, 331342. 3. Smith W, Andrewes GH, Laidlaw PP. (1933) A virus obtained from influenza patients. Lancet 1933 (ii), 66-68. 4. Hirst GK. (1941) The agglutination of red cells by allantoic fluid of chick embryos infected with influenza virus. Science 94, 22-23. 5. Fazekas de St. Groth S. (1948) Destruction of Influenza Virus Receptors in Mouse Lung by an Enzyme from V. Cholera. A m t . J. Exp. Biol. 26, 29-36, 6. Stone JD. (1947) Enzymic Modification of the Reaction between Influenza Virus and Susceptible Tissue Cells. Nature 159, 78 1. 7. Gottschalk A. (1957) The specific enzyme of influenza virus and Vibrio cholerae. Biochim. Biophys. Acta 23, 645-646. 8. Fazekas de St. Groth S. (1948) Viropexis, the Mechanism of Influenza Virus Infection. Nature 162, 294-295. 9. Stray SJ, Cummings RD, Air GM. Influenza infection of disialylated cells. Glycobiology in press. 10. Rogers GN, Herrler G, Paulson JC, Klenk H-D. (1986) Influenza C virus uses 9-O-acetyl-Nacetylneuraminic acid as a high affinity receptor determinant for attachment to cells. J. Biol. Chern. 261, 5947-5951. 11. Klein A, Krishna M , Varki NM, Varki A. (1994) 9-0-acetylated sialic acids have widespread but selective expression: analysis using a dual function probe derived from influenza C hemagglutinin-esterase. Proc. Natl Acad. Sci. USA 91, 7782-7786. 12. Zimmer G, Klenk H-D, Herrler G. (1995) Identification of a 40-kDa Cell Surface Sialoglycoprotein with the Characteristics of a Major Influenza C Virus Receptor in a Madin Darby Canine Kidney Cell Line. J. Biol. Chern. 270, 17815-17822. 13. Ward CW, Dopheide TA. (1981) Evolution of the the Hong Kong influenza A subtype. Structural relationships between the hemagglutinins from A/duck/Ukraine/l/63 (Hav7) and the Hong Kong (H3) hemagglutinins. Biochern. J. 195, 337-340. 14. Basak S, Pritchard DG, Bhown AS, Compans RW. (1981) Glycosylation sites of the influenza viral glycoproteins: characterization of tryptic glycopeptides from the A/USSR (H IN 1) hemagglutinin glycoprotein. J. Virol. 37, 549-558. 15. Mir-Shekari SY, Ashford DA, Harvey DJ, Dwek RA, Schulze IT. (1997) The glycosylation of the influenza A virus hemagglutinin by mammalian cells. J. Biol. Chern. 272, 4027-4036. 16. Webster RG, Rott R. (1987) Influenza virus A pathogenicity: the pivotal role of hemagglutinin. Cell 50, 6665-666. 17. Subbarao K, Klimor A, Katze J, Reguery H, Lim W, Hall H, Perdue M, Swayne D, Bender C. Huang J, Hemphill M, Rowe T, Shaw M, Xu X, Fukuda K, Cox N . (1998) Characterization of an avian influenza A (H5N1) virus isolated from a child with a fatal respiratory illness. Science 279, 393-395. 18. M a t h KS, Reggio H, Helenius A, Simons K. (1981) Infectious Entry of Influenza Virus in a Canine Kidney Cell Line. J. Cell Biol.91, 601-613. 19. Wharton SA, Calder LJ, Ruigrok RWH, Skehel JJ, Steinhauer DA, Wiley DC. (1995) Electron microsopy of antibody complexes of influenza virus haemagglutinin in the fusion pH conformation. EMBO J. 14, 240-246. 20. Wilson IA, Skehel JJ, Wiley DC. (1981) Structure of the hemagglutinin membrane glycoprotein of influenza virus at 3 A resolution. Nature 289, 366-373. 21. Rosenthal PB, Zhang X, Formanowski F, Fitz W, Wong C-H, Meier-Ewert H, Skehel JJ, Wiley DC. (1998) Structure of the haemagglutinin-esterase-fusion glycoprotein of influenza C virus. Nature 396, 92-96.

Refeevcnces

849

22. Sauter NK, Click GD, Crowther RL, Park SJ, Eisen MB, Skehel JJ, Knowles JR, Wiley DC. (1992) Crystallographic detection of a second ligand binding site in influenza virus hemagglutinin. Proc. Nut1 Acud. Sci. USA 89, 324-8. 23. Wiley DC, Wilson IA, Skehel JJ. (1981) Structural identification of the antibody-binding sites of Hong Kong influenza haemagglutinin and their involvment in antigenic variation. Nuture 289, 373-378. 24. Nuss JM, Air GM. (1991) Transfer of the hemagglutinin activity of influenza virus neuraminidase subtype N9 into an N2 background. Virology 183, 496-504. 25. Rogers GN, Paulson JC, Daniels RS, Skehel JJ, Wilson IA, Wiley DC. (1983) Single amino acid substitutions in influenza hiemagglutinin change receptor binding specificity. Nature 304, 76-78. 26. Eisen MB, Sabesan S, Skehel JJ, Wiley DC. (1997) Binding of the Influenza A virus to cellsurface receptors: structures of five hemagglutinin-sialyloligosaccharide complexes determined by X-ray crystallography. Virology 232, 19-3 I . 27. Rivera K, Thomas H, Zhang H, Bossart-Whitaker P, Wei X, Air GM. (1995) Probing the structure of influenza B hemagglutinin using site-directed mutagenesis. Virology 206, 787-795. 28. Strobl B, Vlasak R. (1993) The receptor-destroying enzyme of influenza C virus is required for entry into target cells. Virology 192, 679-682. 29. Herrler G, Crop HJ, Brossmer R. (1995) A synthetic sialic acid analogue that is resistant to the receptor-destroying enzyme can be used by the influenza C virus as a receptor determinant for infection of cells. Biochem. Biophvx Rex Comm. 216, 821-827. 30. Pleschka S, Klenk HD, Herrler G. ( 1 995) Thc catalytic triad of the influenza C virus glycoprotein HEF esterase: characterization by site-directed mutagenesis and functional analyisis. J. Gen. Virol. 76, 2529-2537. 31. Ward CW, Murray JM, Roxburgh CM, Jackson DC. (1983) Chemical and antigenic characterization of the carbohydrate side chains of an Asian (N2) influenza virus neuraminidase. Virology 126, 370-375. 32. Allen AK, Skehel JJ, Yuferov V. (1977) The amino acid and carbohydrate composition of the neuraminidase of B/Lee/40 influenza virus. J. Gen. Virol. 37, 625-628. 33. Pons T, Olmea 0, Chinea G, Beldarrain A, Marquez G, Acosta N , Rodriguez L, Valencia A. (1998) Structural model for family 32 of glycosyl-hydrolase enzymes. Proteins 33, 383-389. 34. Rossmann MG, Rueckert RR. (1987) What does the molecular structure of viruses tell us about viral functions? Microhiol. Sci.4, 206-214. 35. Janakiraman MN, White CL, Laver WG, Air GM, Luo M. (1994) Structure of influenza virus neuraminidase B/Lee/40 complexed with sialic acid and a dehydro-analog at 1.8 A resolution: implications for the catalytic mechanism. Biochemistry 33, 8 172-8 179. 36. Lentz MR, Webster RG, Air GM. (1987) Site-directed mutation of the active site of influenza neuraminidase and implications for the catalytic mechanism. Biochemistry 26, 535 1-5358. 37. Chong AK, Pegg MS, Taylor NR, von Itzstein M. (1992) Evidence for a sialosyl cation transition-state complex in the reaction of sialidase from influenza virus. Eur. J. Biochenz. 207, 33543. 38. Tiralongo J, Pegg MS, von Itzstein M. (1995) Effect of substrate aglycon on enzyme mechanism in the reaction of sialidase from influenza virus. FEBS Lett. 372, 148-150. 39. Ghate AA, Air GM. (1998) Site-directed mutagenesis of catalytic residues of influenza virus neuraminidase as an aid to drug design. Eur. J . Biochem. 258, 320-331. 40. Marszalek PE, Oberhauser AF, Pang Y-P, Fernandez JM. (1998) Polysaccharide elasticity governed by chair-boat transitions of the glucopyranose ring. Nature 396, 661L664. 41. Singh S, Jedrzcjas MJ, Air GM, Luo M, Laver WG, Brouillette WJ. (1995) Structure-based inhibitors of influenza virus sialidase. A benzoic acid lead with a novel interaction. J. Med. Chem. 38, 3217-3225. 42. Air GM, Ghate AA, Stray SJ. (1999) Influenza neuraminidase as a target for antivirals. Adu. Virus Res. 54, 375-402. 43. Burmeister WP, Henrissat B, Bosso C, Cusack S, Ruigrok RW. (1993) Influenza B virus neuraminidase can synthesize its own inhibitor. Structure 1, 19-26. 44. Air GM, Laver WG. (1995) Red cells bound to influenza virus N9 neuraminidase are not released by the N9 neuraminidase activity. Vfroloqy 211, 278-284.

850

51 The Glycobiology of’Znjluenza Viruses

45. Palese P, Tobita K, Ueda M, Compans RW. (1974) Characterization of temperature-sensitive influenza virus mutants defective in neuraminidase. Virology 61, 397-410. 46. Lakshmi MV, Schulze IT. (1978) Effects of Sialation of Influenza Virions on their Interactions with Host Cells and Erythrocytes. Virology 88, 314-324. 47. Liu C, Eichelberger MC, Compans RW, Air GM. (1995) Influenza type A virus neuraminidase does not play a role in viral entry, replication, assembly or budding. J. Virol. 69, 1099-1 106. 48. Hofling K, Brossmer R, Klenk H-D, Herrler G. (1996) Transfer of an esterase-resistant receptor analog to the surface of influenza C virions results in reduced infectivity due to aggregate formation. Virology 218, 127-133. 49. Baum LG, Paulson JC. (199 I ) The N2 neuraminidase of human influenza virus has acquired a substrate specificity complementary to the hemagglutinin receptor specificity. Virology 180, 1015. 50. Kaverin NV, Gambaryan AS, Bovin NV, Rudneva IA, Shilov AA, Khodova OM, Varich NL, Sinitsin BV, Makarova NV, Kropotkina EA. (1998) Postreassortment changes in influenza A virus hemagglutinin restoring functional HA-NA match. Virology 244, 3 15-32 1. 51. Air GM, Laver WG. (1989) The neuraminidase of influenza virus. Prot. Struct. Funct. Genet. 6, 341-356. 52. Nuss JM, Whitaker PB, Air GM. (1993) Identification of critical contact residues in the NC41 epitope of a subtype N9 influenza virus nenraminidase. Prot. Strucf. Funct. Genet. 15, 121-132.

Carbohydrates in Chemistry and Biology Beat Emst, Gerald W. Hart, Pierre Sinay copyright Q WILEY-VCH Verlag GmbH. 2000

52 Glycobiology of AIDS Ten Feizi

52.1 Abstract The envelope glycoprotein of HIV-1 is heavily N-glycosylated, with a diverse array of sequences, some of which are relatively constant (high mannose chains) and others (complex type chains) differ according to the host cells in which the virus is produced. The presence of the N-glycans is important for infectivity. This is almost certainly due to their involvement in the correct folding of the newly synthesized viral envelope glycoprotein. This review extends an earlier discussion [ 11 of the roles of specific oligosaccharide sequences in viral behavior. An example is included of the critical role of a carbohydrate antigen on retroviruses of non-primate animals and the corresponding antibodies in human serum in inactivation of viral infection. This principle may explain the barrier to horizontal transmission of many animal viruses to humans. The question is raised whether a similar principle affects the transmissibility of HIV-1 among humans of differing blood groups.

52.2 Introduction The causative agent of acquired immune deficiency syndrome (AIDS) is an enveloped virus (the human immuno-deficiency virus, HIV-l), a member of the retrovirus family (see Chapter 50 in this volume) [2]. An extensive part of the virion surface is predicted to be covered by carbohydrate as 50%0of the molecular mass of the major envelope glycoprotein is carbohydrate in the form of oligosaccharide chains attached to the polypeptide backbone. In an earlier review of the glycosylation of HIV-1 [ 11 we discussed, first, the evidence that all 24 sites on the envelope glycoprotein, gp 120, are glycosylated with a diverse array of sequences of high-mannose type and complex types. Second, we

852

52 Glycobiology of AIDS

B

U -

Endocvtosis

Host macrophage

Figure 1. Schematic representation of HIV- 1 depicting oligosaccharides of the envelope glycoprotein gp 120 (A); and binding of high mannose type N-glycans to the macrophage endocytosis receptor (B). (Taken from [ I ] with permission.)

discussed the evidence that, as with other virus infections, inhibitors of N-glycosylation or of N-glycan processing have an anti-viral effect on HIV-1. This is most likely due to the misfolding of the newly synthesized non-glycosylated HIV-1 gp 160, the membrane-bound precursor glycoprotein that is cleaved to form gp 120 and the integral membrane protein gp 41. Evidence has indeed been forthcoming for N-glycosylation-dependent interaction of gp 160 with the chaperone proteins calnexin and calreticulin which can assist the folding of newly synthesized, N glycosylated proteins [ 3 , 41. Third, we cited clues to the importance of the presence of N-glycosylation on gp 120 in the infectivity and cyto-pathogenicity of HIV-1. Fourth, we considered the theoretical possibility that, apart from the occurrence of N-glycosylation per se, the expression of specific oligosaccharide sequences on the envelope glycoprotein may influence viral behavior. For example, it may favor dissemination to particular body compartments, or confer tropism for particular cells types (Figure I), or elicit pathological effects such as cytokine stimulation, through

52.3 The Repertoire of N-Glycuns on the Envelope Glycoprotein of IIIV

853

interactions with various endogenous carbohydrate-binding proteins. With the knowledge that there is much shedding of gp 120 from the viral particles, we further proposed that the free as well as the virus-associated gp 120 may bind to membrane-associated lectins of various cell types, and lead to tissue damage (bystander lesions) in the absence of infection as a result of antibody-mediated or cell-mediated immune attack that is primarily directed at antigenic determinants of the viral glycoprotein. In this review, I highlight the evidence for an even wider repertoire of oligosaccharide sequences on the envelope glycoprotein of HIV-1, which further illustrates the host cell-directed variability of viral glycosylation. I dwell further on the possibility that the oligosaccharide repertoire on the envelope glycoprotein may influence the viral behaviour. Because of its possible relevance to HIV-1, I include an example of the protective effect of natural antibodies (xeno-antibodies) in human sera directed at a carbohydrate antigen (xeno-antigen) on the glycoproteins of certain retroviruses of non-primate animals. Here the xeno-antigen is a specific oligosaccharide sequence lacking in humans, and is a further example of biological sequelae of host-cell dirccted viral glycosylation. I discuss the possible relevance of this knowledge, to the transmissibility of HIV infections among individuals of differing blood groups.

52.3 The Repertoire of N-Glycans on the Envelope Glycoprotein of HIV of Human Immunodeficiency Virus Produced in Different Cell Types High-mannose type chains and complex type di, tri- and tetra-antennary N-glycans with or without a bisecting N-acetylglucosamine, and with or without repeated N-acetyllactosamine units (poly-N-acetyllactosamine) or terminal sialic acid were identified in detailed investigations of the carbohydrate sequences on gp 120 of HIV-1 produced in chronically infected T lymphocytes (H9 cells) [ 5 , 61 and transfected Chinese hamster ovary (CHO) cells [7].The range of neutral and sialyl forms were particularly variable, and dependent on the cells in which the virus or the glycoprotein was produced. The oligosaccharide sequences that were characterized in detail were reviewed in [ l ] and are enumerated in Chapter 50 in this volume [2]. The marked influence of the host cell on the details of the glycosylation of viral glycoproteins has been well demonstrated in the case of the envelope glycoprotein of another human retrovirus, HIV-2 [8, 91. The oligosaccharide profiles were compared for the envelope glycoproteins of different virus isolates that differ in their biological properties and the amino acid sequences of their env genes. When the virus isolates were propagated in the same host cells, their N-glycan patterns were similar, whereas there were marked differences when propagated in different host cells. For example, backbone sequences of poly-N-acetyllactosamine-type were a characteristic feature of macrophage-derived viral glycoproteins. The HIV-1 gp 120

854

52 Glycohiology of AIDS

produced in insect cells was found to have a repertoire of high mannose type but no complex type N-glycans [ 10, 1 I]. The presence of sulfate 6-linked to the N acetylglucosamine of complex type chains was reported on gp 120 as well as gp 160 of HIV-1 produced in the human lymphoblastoid cell line Molt-4 [12]. On the N glycans of gp 120 from other mammalian cell lines, there was evidence for the occurrence of fucose at a location other than the core region, susceptible to digestion with wl-3 and 111-4 fucosidase [13].

52.4 Evidence for the Occurrence of 0-Glycans on the Envelope Glycoproteins of HIV-1 Produced in Certain Cell Lines There have been several investigations of the presence of 0-glycans in the envelope glycoprotein of HIV-1. Detailed structural data are lacking however. The common 0-linked sequence Galpl-3GalNAc (with or without terminal sialic acid, also known as the sialyl-T and T antigens, respectively), could not be detected on natural gp 120 derived from infected H9 cells, or on recombinant gp 120 from CHO cells when the glycoproteins were treated with sialidase and exposed to w2-3Galpl3GalNAc: sialyltransferase in the presence of radiolabelled CMP-sialic acid (G. Hart and T. Feizi, cited in [ 11). However, the specificity of this sialyltransferase is such that one would not be able to detect longer 0-linked sequences with po1y-Nacetyllactosamine backbones [ 141, nor would one detect the truncated 0-linked sequence GalNAcw 1-0-SerlThr, and its sialyl analogue (Tn and sialyl-Tn antigens). The possibility that there may be some 0-linked oligosaccharides on gp 160 and gp 120 was first raised on the basis of the observation that, following extensive deglycosylation of these glycoproteins derived from a T lymphoblastoid cell line, VB, using peptide N-glycanase-F (PNFase-F), their pl values were lower than predicted. A slight shift in PI toward a more basic pH was observed following treatment of the PNGase-F-treated glycoprotein with sialidase [ 151, suggesting the presence of small amounts of residual sialyl oligosaccharides. The finding of N-acetylgalactosamine on gp 120 produced by the T cell line MOLT3 was in accord with this conclusion [16]. The presence of small amounts of 0-linked chains on certain preparations of gp 120 was also suggested by chromatographic experiments after alkaline borohydride degradation experiments of H-glucosamine labeled glycoprotein [ 171. Immuno-precipitation experiments with hybridoma-derived antibodies that bind to 0-linked oligosaccharide sequences GalNAcwl-0-Ser/Thr [ 181 or to GalP13GalNAcal-O-Ser/Thr [ 191 revealed a glycoprotein of approximately 120 kDa in HIV-1. Detailed information on the binding specificities of these antibodies is awaited. In particular, it will be important to rule out cross-reactions with other oligosaccharides of the type observed with certain antibodies that bind sialyl-Tn [20]. Provided such cross-reactions with other oligosaccharides can be ruled out, and the presence of antigenically cross-reactive glycoplipids associated with this glycoprotein are also excluded, these data may be taken as evidence for the presence

52.5 Oligosarcharides ~ f g 120 p and gp 41

855

of O-linked oligosaccharides on gp 120, a feature shared with the envelope glycoproteins of several other retroviruses. It had been suggested that serine or threonine residues in the V3 polypeptide loop of the gp 120 (this is a known target for neutralizing antibodies) may be glycosylated with the short chain O-glycans, Tn or sialosyl-Tn, and that the above mentioned antibodies might exert their neutralizing effects by binding to such 0glycans. However, mutant viruses with deletions of the candidate 0-glycosylation sites on the V3 loop did not lose their sensitivity to the anti-Tn and anti-sialosyl Tn [21]. Thus the determinants recognized by these antibodies are located outside the v 3 loop of gp 120. From all of the foregoing, it is reasonable to conclude that small amounts of 0glycans may occur on the HIV-1 glycoproteins produced in certain cell lines, and these would depend on the presence of the appropriate N-acetylgalactosamine: polypeptide N-acetylgalactosaminyl transferases in the host cells [22].

52.5 Oligosaccharides of gp 120 and gp 41 at N-Glycosylation Sites and Their Possible Influence on Viral Infectivity The influence on viral infectivity of each of the 24 N-linked glycosylation sites of gp 120 on a molecular clone of HIV-1 was investigated in a site-directed mutagenesis study in which the consensus Asn-Xaa-Ser/Thr sequence was replaced either by Gln-Xaa-Ser/Thr or His-Xaa-Ser/Thr [23]. Most of the individual consensus sites were found to be dispensable for infection of a CD4-positive cell line. However: six mutant viruses with mutations at the amino terminal half of gp I20 showed delayed growth kinetics compared with the wild-type virus. The authors rightly considered the possibility that diminished infectivity of these mutants was due to amino acid substitutions per se rather than the lack of glycosylation, and with one of the mutants at least, normal growth kinetics were restored after additional amino acid changes. Thus the impaired infectivity of this mutant was due to amino acid changes rather than to the lack of carbohydrate. Three sets of investigations have similarly addressed the influence of N-linked oligosaccharides on the transmembrane glycoprotein gp 41 of HIV-1 on viral infectivity. In the first of these [24], severe impairment of viral infectivity towards a CD4-positive cell line resulted from one mutation. In the second, variable effects on infectivity toward different CD4-positive cells were observed particularly with respect to fusion activity [25]. A third series of more recent investigations [26-291 have been focussed on the impaired fusion activity and have reached the conclusion that the glycosylation site at position 621 is the most important for fusion activity [29], and that gp 41 lacking N-glycans becomes arrested in the Golgi apparatus [28]. Here, also the observed effects could have been a direct result of lack of saccharides or they could be the result of a critical change in the protein folding and conformation.

856

52 Glycobioloyy of AIDS

52.6 gp 120 Glycosylation Can Influence Antigenicity and Immunogenicity At least three mechanisms may be considered whereby the carbohydrate chains could influence immunogenicities and antigenicities of the envelope glycoprotein of HIV-1: i) by expression of carbohydrate sequences that are autoantigenic or isoantigenic; ii) by masking (shielding) the antigenicity of the polypeptide backbone; and iii) by influencing presentation of the glycoprotein to antigen-presenting cells. The subject of the immunogenicities and antigenicities of oligosaccharides on glycoproteins has been reviewed previously [30, 311. Some examples of antigenic determinants on O-linked chains have been discussed above. Other examples of antigenic oligosaccharides include linear and branched poly-N-acetyllactosamine backbone sequences (Ii antigens) that occur on 0- and N-glycans, and to which autoantibodies develop typically following infection with M y c o p l a m a pneumoniae and Epstein Barr virus, respectively [32].Peripheral substitution of the poly-N-acetyllactosamine sequences can generate the major blood group antigens, A, B, and H and also xenoantigens (this topic is further discussed below). The immunogenicity of the peptide domains of gp 120 may be masked by glycosylation. Evidence for this has been obtained using both patient sera and experimental antisera raised to a series of overlapping gp 120 peptides [33]. Thus, as was first shown for the hemagglutinin of influenza virus [34] and more recently for the simian immunodeficiency virus, SIV-1 [35]: that variable glycosylation of the envelope glycoprotein is a mechanism for antigenic variation, and potentially a means of viral escape from the host’s immune response. The effects of shortened oligosaccharides on gp 160 on immunogenicity have also been investigated [36]. Among antisera raised to preparations of the glycoprotein which had been treated with various glycosidases, the intensities of binding to non-glycosylated peptides were variable. Some antisera showed equivalent binding, others showed greater binding and others still showed less binding than did antisera raised to the glycosylated protein. Structural evidence has now been obtained for the existence of an immunologically ‘silent’ face of gp 120 protein that is protected against antibody responses by the dense array of oligosaccharides [37, 381. There is evidence that N-linked carbohydrates may limit the antigenic recognition of gp 120 by T lymphocytes. Among T cell clones derived from persons immunized with a recombinant non-glycosylated segment of gp 120, some 20% could respond only to non-glycosylated polypeptides and not to the glycosylated glycoprotein [ 391. On the other hand, in experiments where gp 120 was intentionally desialylated to expose terminal galactose residues, and then tested for its ability to elicit antigenspecific stimulation of T cell lines and clones (these T cells were specific for the nonglycosylated gp 120-peptides), a marked enhancement of stimulation was observed relative to the stimulation by the untreated gp 120 [40]. The stimulation was elicited in the presence of antigen-presenting monocytes and dendritic cells but not B-

52.7 Succhurides Recognized by Curhoh~rtrate-bindingProteim and Antibodies

857

lymphocytes. Competition experiments using soluble galactose-terminating saccharides as inhibitors indicated that the proliferation enhancing effect was a result of increased uptake of the desialylated glycoprotein into the antigen presenting cells via galactose-binding proteins at their surface.

52.7 Saccharides Recognized by Carbohydrate-binding Proteins, and Antibodies as Potential Neutralization Epitopes on the Envelope Glycoprotein of HIV-1 There have been several reports that carbohydrate-binding proteins and antibodies that recognize oligosaccharides on the envelope glycoprotein of retroviruses can inhibit (neutralize) viral growth. Instances documented for HIV-1 include: i) mannose-recognizing proteins and antibodies; ii) 0-glycan recognizing antibodies; and iii) antibodies to blood group A.

A fourth principle included in the present discussion on account of its topicality is the neutralizing effect on animal retroviruses of xeno-antibodies to the Gala 1-3Gal sequence which occur in human sera and those of Old World primates. 52.7.1 Lectins and Antibodies with Mannose-related Specificities

As reviewed earlier [l,411, several plant lectins, including those that bind preferentially to high mannose type oligosaccharides, and others that bind complex-type chains inhibit HIV-1 infection and syncytium formation. We suggested that, as diverse oligosaccharides are implicated, these inhibition effects may be due to a nonspecific steric hindrance of the gp 120-CD4 interaction rather than the blocking of specific oligosaccharide ligands in these interactions [ 11. Some evidence has since been presented that in the case of the plant lectin concanavalin A (Con A) which binds to mannose-terminating oligosaccharides, there is no interference with the gp 120-CD4 interaction [42].This finding led the authors to suggest that the binding of Con A may interfere with some later event that leads to virus entry. However, in a study with conglutinin 1431, a bovine serum protein that binds oligosaccharides that terminate with mannose, or N-acetylglucosamine residues 144, 451 of the type that occur on HIV-1 15-71, an inhibition of gp 120 binding to the CD4 receptor was observed. These contrasting effects on the gp 120-CD4 interaction cannot be readily explained. It could be that there is a greater steric hindrance with conglutinin, which has a higher molecular weight and a broader saccharide-binding specificity than Con A. Naturally occurring antibodies with specificities related to mannosyl oligosaccharides (isolated by affinity chromatography on mannan adsorbent) have also been

858

52 Glycobiology o j AIDS

shown to bind to gp 160, gp 120 and gp 41 [46]. These antibodies did not neutralize HIV- 1 infection. Whereas these antibodies were isolated and tested for binding at ambient temperature, the infectivity experiments were of necessity carried out at 37°C. I raise the possibility that these natural antibodies may be of the ‘coldreactive’ type, and in common with some other natural autoantibodies directed at carbohydrates [32], they may act at temperatures below 37”C, hence their lack of neutralizing activity. By contrast, elicited hyperimmune antisera to yeast mannans have been reported to inhibit infection [47]. These antibodies would be predicted to have higher affinities than the naturally occurring autoantibodies. 52.7.2 Antibodies to O-Glycan Sequences

In an immunological study [ 191, hybridoma-derived antibodies that bind to 0linked oligosaccharides: GalNAcal -0-Ser/Thr, (anti-Tn: Leu35), or to NeuAca23GalNAcal-O-Ser/Thr (anti sialyl-Tn: TKH2, and B72.3) or to Galpl-3GalNAcwlO-Ser/Thr (anti-T: HH8) were tested for their ability to neutralize HIV-1 infection in vitro. Neutralization was recorded with two IgG antibodies, TKH2 and B72.3. In a further report from the same laboratory [ 181 two other antibodies assigned as antiTn, unlike antibody Leu35, were reported to neutralize HIV-1 infection. 52.7.3 Antibodies to Blood Group A

Other immunological studies by Hansen et a1 [19, 481 suggested that when HIV-1 infection takes place in peripheral blood mononuclear cells (PBMC) of persons of blood group A, there occurs an induction of the expression of blood group A antigen (PBMC normally do not express the blood group A or B antigens), and that this may decorate the gp 120 produced by these cells. Of great interest was the observation that an anti-blood group A antibody, AH1 6, inhibited the infectivity of HIV-1 isolates obtained by passage through the PMBC of donors of blood group A, but not of blood group B or 0 donors [48]. Further biochemical and molecular biological studies would be required to fully interpret the basis of PBMC binding to antibody AH16, including details of its specificity, as it was originally reported to recognize the blood group A sequence based on the type 1 backbone (Galpl3GlcNAc) [49]: GalNAcal-3Gal~I-3GlcNAc11,2 Fucw Hematogenous cells in the human are generally thought to express type 2 (Galpl4GlcNAc) backbones exclusively [50].It will be important to determine whether indeed there occurs in vivo a de novo appearance of blood group antigens on infected PBMC. Also, it is possible that among HIV-1 infected tissues in vivo, there are those that normally express the major blood group antigens. Such tissues would

52.7 Succhurides Recognized by Curho~iylyu'vute-binding Proteins and Antibodies

859

be predicted to produce viral glycoproteins decorated with these antigens. Their presence on the viral glycoprotein could be an important factor that influences the horizontal transmissibility of infection in humans. The naturally occurring isoantibodies (anti-A in the sera of blood group B persons, anti-B in group A, and both anti-A and anti-B in group 0 persons) could be envisaged to have an inhibitory effect on the growth of virus that is decorated with these antigens. To my knowledge there have been no investigations of the transmissibility of HIV-1 (or of other virus infections) in relation to the blood group status of virus donor and virus recipient. This certainly deserves examination in view of the striking anti-retroviral effects of xeno-antibodies discussed below.

52.7.4 Xeno-antibodies to Galal3Gal Sequence It has long been recognized that non-primate-derived C-type retroviruses are inactivated in human serum by an antibody-dependent mechanism involving the triggering of the classical complement pathway. And yet, in the absence of human serum, such viruses can infect and transform human cells in vitro [51]. The biochemical basis of the protective effect of human serum has now been clearly elucidated. As documented earlier [ 521, the viral glycoproteins produced in murine cells contain N-glycans capped with the Galal-3Gal sequence. Not only is this sequence lacking in humans due to the lack of the 1x1-3 galactosyltransferase, but there are in human serum substantial amounts of naturally occurring antibodies (xenoantibodies) to this sequence [53]. It has been shown that the viral inactivation is mediated by these serum antibodies [51, 541. It is thought that, thereby, they provide a barrier for horizontal transmission of retroviruses, from animal species that express the a-galactosyl epitope, to humans and to other Old World primates.

52.7.5 Potential Medical Relevance It can be envisaged that, where the viral inoculum is relatively small, and the recipient is immunocompetent with high levels of xeno-antibodies, effective protection will occur. There is major concern however regarding the risks of retroviral infection in the field of organ xeno-transplantation. This is because the possible use of pig organs and tissues as xenografts is being actively considered [55]. An endogenous C-type retrovirus from cells of the domestic pig was shown, in the absence of human serum, to be capable of infecting and replicating in a range of human cell lines. After replicating in Gala1-3Gal-negative human cells, the progeny virions were found to become highly resistant to inactivation by human serum [55]. This finding has tempered enthusiasm for the use of porcine organs for transplantation. It is considered possible that the immunosuppressed state of the transplant recipients could predispose them to pig-to-human infection. Immuno-suppression aside, knowing the heterogeneity of glycosylation and the selective pressure that the serum antibodies would exert, it can be envisaged that with time xeno-antigen, negative virions could emerge from the transplanted organs.

860

52 Glycobiology of AIDS

As in the case of malignant tumors [56], therapeutic strategies have been contemplated targeted at the carbohydrates in HIV-1 [19]. Clearly the blood group antigens which are abundantly expressed on red cells and on many epithelial tissues would be unsuitable targets. The Tn and sialyl-Tn antigens are potential targets because of their limited expression in healthy tissues [ 181. A major concern, however, is a report that enhancement of HIV-l infectivity occurred in a monocytic cell line, in the presence of an anti-Tn [57].

52.8 Does Viral Oligosaccharide Display Influence Tissue Tropism? There have been extensive investigations of the determinants of macrophagetropism among isolates and molecular clones of HIV-1. These have been largely focussed on protein sequences of the envelope glycoprotein [58-631. The general approach has been to generate chimeric viruses by inserting segments of the envelope sequence from macrophage-tropic strains into backgrounds of non-macrophagetropic viruses. An account of these studies is beyond the scope of this review, but suffice it to say that regions distinct from the CD-4 binding carboxy-terminal domain are involved. In particular the V3 loop domains of gp 120 of macrophage-tropic and of microglia-tropic viruses are important in conferring upon non-macrophagetropic viruses the ability to infect macrophages and also microglial cells derived from brain. A V3 loop domain insert is not sufficient however for high levels of virus production in the macrophage, and there are indications that additional domain(s) amino terminal to this domain may be required to promote infection in these cells. The precise mechanism by which a particular gp 120 sequence may favor entry of HIV strains into macrophages or microglial cells is unknown. It has been suggested [62] that subsequent to CD4 binding, an apposition of the amino terminal envelope region and the surface of the host cell may be necessary to trigger viral entry and infection of macrophages and glial cells. It is clear from the foregoing that new experimental strategies are needed to elucidate the mechanisms involved in macrophage tropism. In the context of this carbohydrate-oriented review, I pose the following questions and elaborate a hypothesis in the hope of stimulating new thoughts on mechanisms of macrophage tropism, and tissue tropism in general. Could it be that among the ‘missing links’ in macrophage-tropism and other tissue tropisms are: i) oligosaccharide display on the envelope glycoprotein? ii) host-cell-associated lectins as depicted in Figure l? By the term oligosaccharide display, I refer not only to the presence of specific oligosaccharide sequences at the virus surface (some of which will be host-cell typespecific) but also to the presence or absence of glycosylation at particular sites and,

52.8 Does Vim1 Oligosacchuride Display Injuence Tissue Tropism? a

b

C

d

e

f

86 1

Figure 2. Schematic representation of a hypothetical glycosylation site on the envelope glycoprotein gp 120 of HIV-I. Different degrees of accessibility are depicted for the same N-glycan chain (ManyGlcNAcz) in a-e, or lack of oligosaccharide in f, resulting from differences in the protein sequence or protein conformation. In each case the glycosylation ‘phenotype’ of the protein is different. A carbohydrate-binding protein in the body, e.g. the macrophage endocytosis receptor, may bind with differing affinities to the ohgosaccharide in these different presentations. Other factors, such as the copy number of the oligosaccharide ligand per mole of protein, and the degree of clustering of the ligand would be additional determinants of avidity of binding. Obvious differences in phenotype would result from the presence of different oligosaccharide sequences at this site (not shown here).

more importantly, the accessibility of the same oligosaccharide at a given glycosylation site (determined by the amino acid sequence of the envelope glycoprotein either in the vicinity of the site or at some distance from it). I propose the term oligosaccharide ‘phenotype’ for this property. It is depicted schematically in its simplest form in Figure 2: a single oligosaccharide at a given glycosylation site. Although all five glycosylation phenotypes (depicted as a-e in Figure 2) contain the same oligosaccharide, a carbohydrate-binding protein may bind to each phenotype with differing affinities, as further discussed below. Carbohydrate-mediated interactions have been detected involving gp 120 derived from T lymphotropic virus IIIB isolate of HIV, and the isolated endocytosis receptor from human placental macrophages [64]. From knowledge of the specificity of this receptor [65, 661, it was predicted [64] that binding of the highest affinity would occur with envelope variants that have oligosaccharide phenotypes with a relative prominence and abundance of high-mannose type oligosaccharides, or with oligosaccharides terminating in N-acetylglucosamine or fucose residues recognized by this lectin. Furthermore, the oligosaccharide(s) in question may need to be numerous or in a clustered state for binding of sufficient high avidity to occur. Whether such carbohydrate-mediated interactions have a role in the initiation of macrophage infection is not yet known. However, there is a precedent for the display of highmannose type oligosaccharides as a key factor in the binding or lack of binding of endogenous lectins to a specific glycosylated protein. I refer here to the interactions of two serum lectins, conglutinin and mannan binding protein, whose carbohydrate binding specificities resemble that of the macrophage endocytosis receptor [44, 45, 671. Of the two serum lectins, only conglutinin shows carbohydrate-mediated bind-

862

52 Glycobiology of AIDS

ing to the complement glycopeptide iC3b, which contains high mannose type oligosaccharides [68]. The iC3b is derived proteolytically from the major serum complement glycoprotein, C3, during the complement cascade. Neither conglutinin nor mannan-binding protein binds to this parent glycoprotein, C3. These are clear examples of biological specificities [67, 681 which depend on the one hand on oligosaccharide display [76] in the context of a specific carrier protein, and on the other hand, the subtle differences in carbohydrate-binding specificities of the two proteins.

52.9 Concluding Remarks Few would argue that the envelope glycoprotein of HIV-1 needs to be glycosylated for the gp 120-CD4 interaction and successful infection of the CD4-positive T lymphocyte. The question is whether the glycosylation status of newly shed virus is a determinant of infectivity, tropism and disordered host-cell function, for example, in endothelial cells [69]. As the viral glycosylation may change rapidly after inoculation, it would be necessary to design ‘acute phase’ experiments to examine interactions between newly shed virus or newly acquired infected cells, and adhesive carbohydrate-binding proteins of the host cells. The E- and P-selectins should be considered among these. The carbohydrate-mediated activities of these proteins at the initial stages of leukocyte extravasation in inflammation is a subject of much current interest [70-721. Not only viral envelope oligosaccharides but also oligosaccharides on infected leukocytes harboring the virus are potential ligands for these proteins. Enormous ramifications can be envisaged with respect to viral dissemination and disease pathogenesis in HIV-1 infection if it can be shown that these adhesion proteins bind with high avidity to certain glycosylation variants of HIV-1 or to T-lymphocytes infected with HIV-1. For example, memory T cells (CD45RA- antigen type) of the type that show high avidity carbohydrdte-mediated binding to E-selectin [73, 741 are CD4-positive, and there is evidence to suggest that upon antigenic activation in vivo, T cells of this type are the main targets of HIV-1 infection [75]. Binding of HIV-l-infected CD45 RA- cells to the E-selectin molecule can be envisaged as an important mechanism for extravasation and tissue spread of HIV-1, and a novel mechanism for the diminished numbers of circulating CD4 cells in patients with AIDS. This would amount to the hijacking of the normal inflammatory pathways involving these endothelial adhesion molecules by cell free virus particles or by cells harboring the infective agent. Acknowledgment

The author is supported by program grant G9601454 from the Medical Research Council.

Rejerences

863

References I. T. Feizi, M. Larkin, AIDS and glycosylation. Glycohiology 1990, I , 17-23. 2. H. Geyer, and R. Geyer, Glycobiology in Viruses. Curr. Biol. 1999, in press. 3. A. Otteken, B. Moss, Calreticulin interacts with newly synthesized human immunodeficiency virus type 1 envelope glycoprotein, suggesting a chaperone function similar to that of calnexin. J. Biol. Chem. 1996, 271,97-103. 4. A. Mehta, N. Zitzmann, P. M. Rudd, T. M. Block, R. A. Dwek, Alpha-glucosidase inhibitors as potential broad based anti-viral agents. FEBS Lett. 1998, 430, 17-22. 5 . H. Geyer, C. Holschbach, G. Hunsmann, J. Schneider, Carbohydrates of human immunodeficiency virus. Structures of oligosaccharides linked to the envelope glycoprotein 120. J. Bid. Chem. 1988,263, 11760-1 1767. 6. T. Mizuochi, T. J. Matthews, M. Kato, J. Hamako, K. Titani, J. C . Solomon, T. Feizi, Diversity of oligosaccharide structures on the envelope glycoprotein gp120 of human immunodeficiency virus 1 from the lymphoblastoid cell line H9. J. Bid. Chem. 1990,265, 85198524. 7. T. Mizuochi, M. W. Spellman, M. Larkin, J. C . Solomon, L. J. Basa, T. Feizi, Carbohydrate structures of the human-immunodeficiency-virus (HIV) recombinant envelope glycoprotein gp120 produced in Chinese-hamster ovary cells. Biochem. J. 1988,254, 599-603. 8. S. Liedtke, M. Adamski, R. Geyer, A. Pfutzner, W. H. Rubsamen, H. Geyer, Oligosaccharide profiles of HIV-2 external envelope glycoprotein: dependence on host cells and virus isolates. Glycohiology 1994, 4, 477-484. 9 S. Liedtke, R. Geyer, H. Geyer, Host-cell-specific glycosylation of HIV-2 envelope glycoprotein. Glycoconj. J. 1997, 14, 785-793. 10 J . C . Yeh, J. R. Seals, C . I. Murphy, H . van Halbeek, R. D. Cummings, Site-specific N-glycosylation and oligosaccharide structures of recombinant HIV- 1 gp120 derived from a baculovirus expression system. Biochemistry 1993, 32, 11087--11099. 11 T. D. Butters, B. Yudkin, G. S. Jacob, I. M. Jones, Structural characterization of the N-linked oligosaccharides derived from HIVgpl20 expressed in lepidopteran cells. Glycoconj. J. 1998, IS, 83-88. 12 A. Shilatifard, R. K. Merkle, D. E. Helland, J. L. Welles, W. A. Haseltine, R. D. Cummings, Complex-type N-linked oligosaccharides of gp 120 from human immunodeficiency virus type 1 contain sulfated N-acetylglucosamine. J. Virol. 1993, 67, 943-952. 13 A. Bolmstedt, M. Biller, J. E. Hansen, J. P. Moore, S. Olofsson, Demonstration of peripheral fucose units in N-linked glycans of human immunodeficiency virus type 1 gp 120: effects on glycoprotein conformation. Arch. Virol. 1997, 142, 2465-2481. 14 T. Feizi, H. C. Gooi, R. A. Childs, J. K. Picard, K. Uemura, L. M. Loomes, S. J. Thorpe, E. F. Hounsell, Tumour-associated and differentiation antigens on the carbohydrate moieties of mucin-type glycoproteins. Biochenz. Soc. Trans. 1984, 12, 591-596. 15. B. S. Stein, E. G. Engleman, Intracellular processing of the gp160 HIV-I envelope precursor. Endoproteolytic cleavage occurs in a cis or medial compartment of the Golgi complex. J. Biol. Chem. 1990,265, 2640-2649. 16. R. K. Merkle, D. E. Helland, J. L. Welles, A. Shilatifard, W. A. Haseltine, R. D. Cummings, gp160 of HIV-I synthesized by persistently infected Molt-3 cells is terminally glycosylated: evidence that cleavage of gpl60 occurs subsequent to oligosaccharide processing. Arch. Biochem. Biophys. 1991, 290, 248-257. 17. J. E. Hansen, H. Clausen, S. L. Hu, J. 0. Nielsen, S. Olofsson, An 0-linked carbohydrate neutralization epitope of HIV-I gp 120 is expressed by HIV-I env gene recombinant vaccinia virus. Arch. Virol. 1992, 126, 11-20, 18. J. E. Hansen, C. Nielsen, M. Arendrup, S. Olofsson, L. Mathiesen, J. 0 . Nielsen, H. Clausen, Broadly neutralizing antibodies targeted to mucin-type carbohydrate epitopes of human immunodeficiency virus. J. Virol. 1991, 65, 6461-6467. 19. J.-E.S. Hansen, H. Clausen, C . Nielsen, L.S. Teglbjaerg, L.L. Hansen, C.M. Nielsen, E. Dabelsteen, L. Mathiesen, S.-I. Hakomori, J . 0 . Nielsen, Inhibition of human immunodeficiency virus (HIV) infection in uitro by anticarbohydrate monoclonal antibodies: peripheral glyco-

864

52 Glycobiology ojAZDS

sylation of HTV envelope glycoprotein gp120 may be a target for virus neutralization. J. Virol. 1990,64,2833-2840. 20. M. Larkin, W. Knapp, M. S. Stoll, H. Mehmet, T. Feizi, Monoclonal antibodies VIB-E3, 1B5 and HB9 to the leucocyte/epithelial antigen CD24 resemble BA-1 in recognizing sialic aciddependent epitope(s). Evidence that VIB-E3 recognizes NeuAca2-6Gal sequences. Clin. Exp. Immunol. 1991,85, 536-541. 21. J. E. Hansen, B. Jansson, G. J. Gram, H. Clausen, J. 0. Nielsen, S . Olofsson, Sensitivity of HIV-I to neutralization by antibodies against 0-linked Carbohydrate epitopes despite deletion of 0-glycosylation signals in the V3 loop. Arch. Virol. 1996, 141, 291-300. 22. H. Clausen, E. P. Bennett, A family of UDP-GalNAc: polypeptide N-acetylgalactosaminyltransferases control the initiation of mucin-type 0-linked glycosylation. Glycobiology 1996, 6, 635-646. 23. W. R. Lee, W. J. Syu, B. Du, M. Matsuda, S. Tan, A. Wolf, M. Essex, T. H. Lee, Nonrandom distribution of gp120 N-linked glycosylation sites important for infectivity of human immunodeficiency virus type 1. Proc. Natl Acad. Sci. U.S.A. 1992, 89, 2213-2217. 24. W. R. Lee, X. F. Yu, W. J. Syu, M. Essex, T. H. Lee, Mutational analysis of conserved Nlinked glycosylation sites of human immunodeficiency virus type 1 gp41. J Virol. 1992, 66, 1799-1 803. 25. D. A. Dedera, R . L. Gu, L. Ratner, Role of asparagine-linked glycosylation in human immunodeficiency virus type 1 transmembrane envelope function. Virology 1992, 187, 377-382. 26. E. Fenouillet, I. Jones, B. Powell, D. Schmitt, M. P. Kieny, J. C. Gluckman, Functional role of the glycan cluster of the human immunodeficiency virus type 1 transmembrane glycoprotein (gp41) ectodomain. J. Virol. 1993, 67, 150-160. 27. E. Fenouillet, J. C. Gluckman, I. M. Jones, Functions of HIV envelope glycans. Trends Biochem. Sci. 1994, 19, 65-70. 28. E. Fenouillet, I. M. Jones, The glycosylation of human immunodeficiency virus type 1 transmembrane glycoprotein (gp41) is important for the efficient intracellular transport of the envelope precursor gp160. J. Gen. Virol. 1995, 76, 1509-1514. 29. C. Perrin, E. Fenouillet, I. M. Jones, Role of gp41 glycosylation sites in the biological activity of human immunodeficiency virus type 1 envelope glycoprotein. Virology 1998,242, 338-345. 30. T. Feizi, Demonstration by monoclonal antibodies that carbohydrate structures of glycoproteins and glycolipids are onco-developmental antigens. Nature 1985, 314, 53-57. 31. T. Feizi, R. A. Childs, Carbohydrates as antigenic determinants of glycoproteins. Biochem. J. 1987,245, 1-11. 32. T. Feizi, The blood group Ii system: a carbohydrate antigen system defined by naturally monoclonal or oligoclonal autoantibodies of man. Immunol. Commun. 1981, 10, 127-156. 33. D. Davis, D. M. Stephens, C. Willers, P. J. Lachmann, Glycosylation governs the binding of antipeptide antibodies to regions of hypervariable amino acid sequence within recombinant gp120 of human immunodeficiency virus type 1. J. Gen. Virol. 1990, 71,2889-2898. 34. J. J. Skehel, D. J. Stevens, R. S. Daniels, A. R. Douglas, M. Knossow, I. A. Wilson, D. C. Wiley, A carbohydrate side chain on hemagglutinins of Hong Kong influenza viruses inhibits recognition by a monoclonal antibody. Proc. Natl Acad. Sci. U , S A 1984, 81, 1779-1783. 35. B. Chackerian, L. M. Rudensey, J. Overbaugh, Specific N-linked and 0-linked glycosylation modifications in the envelope V1 domain of simian immunodeficiency virus variants that evolve in the host alter recognition by neutralizing antibodies. J. Virol. 1997, 71, 7719-7727. 36. A. Benjouad, J. C. Gluckman, H. Rochat, L. Montagnier, E. Bahraoui, Influence of carbohydrate moieties on the immunogenicity of human immunodeficiency virus type 1 recombinant gp160. J. Virol. 1992, 66, 2473-2483. 37. R. Wyatt, P. D. Kwong, E. Desjardins, R. W. Sweet, J. Robinson, W. A. Hendrickson, J. G. Sodroski, The antigenic structure of the HIV gp120 envelope glycoprotein [see comments]. Nature 1998,393, 705-71 1. 38. R. Wyatt, J. Sodroski, The HIV-1 envelope glycoproteins: fusogens, antigens, and immunogens. Science 1998,280, 1884-1888. 39. P. Botarelli, B. A. Houlden, N. L. Haigwood, C. Servis, D. Montagna, S. Abrignani, Nglycosylation of HIV-gpl20 may constrain recognition by T lymphocytes. J Immunol. 1991, 147, 3128-3132.

References

865

40. F. Manca, Galactose receptors and presentation of HIV envelope glycoprotein to specific human T cells. J Iinmunol. 1992, I48, 2278-2282. 41. J. Favero, Lectins in AIDS research. G/.ycohioloyy 1994, 4, 387-396. 42. L. Gattegno, A. Ramdani, T. Jouault, L. Saffar, J. C. Gluckman, Lectin-carbohydrate interactions and infectivity of human immunodeficiency virus type 1 (HIV-I). AIDS Res. Hum. Retroviruses 1992, 8, 27-37. 43. 0. Andersen, A. M. Sorensen, S. E. Svehag, E. Fenouillet, Conglutinin binds the HIV-1 envelope glycoprotein gp 160 and inhibits its interaction with cell membrane CD4. Scand. J. Immunol. 1991, 33, 81-88. 44. R. W. Loveless, T. Feizi, R. A. Childs, T. Mizuochi, M. S. Stoll, Oldroyd R.G., P. J. Lachmann, Bovine serum conglutinin is a lectin which binds non-reducing terminal Nacetylglucosamine, mannose and fucose residues. Biochem. J. 1989,258, 109-1 13. 45. T. Mizuochi, R. W. Loveless, A. M. Lawson, W. Chai, P. J. Lachmann, R. A. Childs, S. Thiel, T. Feizi, A library of oligosaccharide probes (neoglycolipids) from N-glycosylated proteins reveals that conglutinin binds to certain complex type as well as high-mannose type oligosaccharide chains. J. Biol. Chem. 1989,264, 13834-13839. 46. T. Tomiyama, D. Lake, Y. Masuho, E. M. Hersh, Recognition of human immunodeficiency virus glycoproteins by natural anti-carbohydrate antibodies in human serum. Bioclzem. Biophys. Res. Commun. 1991, 177, 279-285. 47. W. E. Muller, M. Bachmann, B. E. Weiler, H. C. Schroder, G. Uhlenbruck, T. Shinoda, H. Shimizu, H. Ushijima, Antibodies against defined carbohydrate structures of Candidd albicans protect H9 cells against infection with human immunodeficiency virus-I in vitro. J. Arquir. Immun. Defic. Syndr. 1991, 4, 694-703. 48. M. Arendrup, J. E. Hansen? H. Clausen, C. Nielsen, L. R. Mathiesen, J. 0. Nielsen, Antibody to histo-blood group A antigen neutralizes HIV produced by lymphocytes from blood group A donors but not from blood group B or 0 donors. AIDS 1991, 5, 441-444. 49. K. Abe, S. B. Levery, S. Hakomori, The antibody specific to Type 1 chain blood group A determinant. J. Zmmunol. 1984, 132, 1951-1954. 50. W. M. Watkins, Biochemistry and genetics of the ABO, Lewis and P blood group systems. Ado. Hum. Gen. 1980,10, 1-136, 379 -385. 51. R. P. Rother, W. L. Fodor, J. P. Springhorn, C. W. Birks: E. Setter, M. S. Sandrin. S. P. Squinto, S. A. Rollins, A novel mechanism of retrovirus inactivation in human serum mediated by anti-alpha-galactosyl natural antibody. J. E.xp. Med. 1995, 182, 1345-1355. 52. H. Geyer, R. Kempf, H. H. Schott, R. Geyer, Glycosylation of the envelope glycoprotein from a polytropic murine retrovirus in two different host cells. Eur. J. Biochem. 1990, IY3, 855862. 53. U. Galili, S. B. Shohet, E. Kobrin, C. L. Stults, B. A. Macher, Man, apes, and Old World monkeys differ from other mammals in the expression of alpha-galactosyl epitopes on nucleated cells. J. Bid. Chem. 1988, 263, 17755- 17762. 54. Y. Takeuchi, C. D. Porter, K. M. Strahan, A. F. Preece, K. Gustafsson, F. L. Cosset, R. A. Weiss, M. K. Collins, Sensitization of cells and retroviruses to human serum by (alpha 1-3) galactosyltransferase. Nuture 1996, 379, 85-88. 55. C. Patience, Y. Takeuchi, R. A. Weiss, Infection of human cells by an endogenous retrovirus of pigs [see comments]. Nut. Med. 1997, 3, 282-286. 56. S. Ilakomori, Aberrant glycosylation in cancer cell membranes as focused on glycolipids: overview and perspectives. Cuncer Res. 1985, 45, 2405-2414. 57. J. E. Hansen, C. Nielsen, H. Clausen, L. R. Mathiesen, J. 0. Nielsen, Effect of anticarbohydrate antibodies on HIV infection in a monocytic cell line (U937). Antiviral Res. 1991, 16, 233-242. 58. S. S. Hwang, T. J. Boyle, H. K. Lyerly, B. R. Cullen, Identification of the envelope V3 loop as the primary determinant of cell tropism in HIV-I. Science 1991, 253, 71-74. 59. W. A. O’Brien, Y . Koyandgi, A. Namazie, J. Q. Zhao, A. Diagne, K. ldler, J. A. Zack, I. S. Chen, HIV-1 tropism for mononuclear phagocytes can be determined by regions of gp120 outside the CD4-binding domain. Nature 1990, 348, 69--73. 60. T. Shioda, J. A. Levy, M. C. Cheng, Macrophage and T cell-line tropisms of HIV-1 are determined by specific regions of the envelope gp120 gene. Nuture 1991, 34Y, 167-169.

866

52 Glycobiology of AIDS

61. P. Westervelt, H. E. Gendelman, L. Ratner, Identification of a determinant within the human immunodeficiency virus 1 surface envelope glycoprotein critical for productive infection of primary monocytes. Proc. Natl Acad. Sci. U.S.A. 1991, 88, 3097-3101. 62. N. E. Sharpless, W. A. O’Brien, E. Verdin, C. V. Kufta, I. S. Chen, D. M. Dubois, Human immunodeficiency virus type 1 tropism for brain microglial cells is determined by a region of the env glycoprotein that also controls macrophage tropism. J. Virol. 1992,66, 2588-2593. 63. P. Westervelt, D. B. Trowbridge, L. G. Epstein, B. M. Blumberg, Y. Li, B. H. Hahn, G. M. Shaw, R. W. Price, L. Ratner, Macrophage tropism determinants of human immunodeficiency virus type 1 in vivo. J. Virol. 1992, 66, 2577-2582. 64. M. Larkin, R. A. Childs, T. J. Matthews, S. Thiel, T. Mizuochi, A. M. Lawson, J. S. Savill, C. Haslett, R. Diaz, T. Feizi, Oligosaccharide-mediated interactions of the envelope glycoprotein gp120 of HIV-1 that are independent of CD4 recognition. AIDS 1989,3, 793-798. 65. P. D. Stahl, J. S. Rodman, M. J. Miller, P. H. Schlessinger, Evidence for receptor-mediated binding of glycoproteins, glycoconjugates and lysosomal glycosidases by alveolar macrophages. Proc. Natl Acad. Sci. U.S.A. 1978, 75, 1399-1403. 66. V. L. Shepherd, Y. C. Lee, P. H. Schlesinger, P. D. Stahl, L-Fucose-terminated glycoconjugates are recognized by pinocytosis receptors on macrophages. Proc. Natl Acad. Sci. USA 1981, 78, 1019- 1022. 67. T. Feizi, Cell-cell adhesion and membrane glycosylation. Curr. Opin. Struct. Biol. 1991, I , 766770. 68. D. Solis, T. Feizi, C. T. Yuen, A. M. Lawson, R. A. Harrison, R. W. Loveless, Differential recognition by conglutinin and mannan-binding protein of N-glycans presented on neoglycolipids and glycoproteins with special reference to complement glycoprotein C3 and ribonuclease B. J. B i d Chem. 1994,269, 11555-1 1562. 69. A. Lafeuillade, M. C. Alessi, M. Poizot, I, N. C. Boyer, C. Zandotti, R. Quilichini, L. Aubert, C. Tamalet, V. Juhan, I, J. A. Gastaut, Endothelial cell dysfunction in HIV infection. J. Acquir. Immun. Dejic. Syndr. 1992, 5, 127- 131. 70. S. D. Rosen, C . R. Bertozzi, Leukocyte adhesion: Two selectins converge on sulphate. Curr. Bid. 1996, 6, 261-264. 71. P. R. Crocker, T. Feizi, Carbohydrate recognition systems: functional triads in cell-cell interactions. Curr. Opin. Struct. Biol. 1996, 6 , 679-691. 72. D. Vestweber, J. E. Blanks, Mechanisms that regulate the function of the selectins and their ligands. Physiol. Rev. 1999, 79, 18 1-2 13. 73. L. J. Picker, T. K. Kishimoto, W. C . Smith, R. A. Warnock, E. C . Butcher, ELAM-1 is an adhesion molecule for skin-homing T cells. Nature 1991, 349, 796-799. 74. Y. Shimizu, S. Shaw, N. Graber, K. J. Horgan, G. A. Van Seventer, W. Newman, Activationindependent binding of human memory T cells to adhesion molecule ELAM-1. Nature 1991, 349, 799-802. 75. F. Miedema, M. Tersmette, R. A. W. Van Lier, AIDS pathogenesis: a dynamic interaction betwwen HIV and the immune system. Immunol. Today 1990, 11, 293-297. 76. D. Solis, M. Bruix, L. Gonzalez, T. Diaz-Mauriiio, M. Rico, J. Jimenez-Barbero, T. Feizi, earrier protein-modulated recognition of Mans N-glycan by the carbohydrate-binding protein. (Submitted for publication).

Carbohydrates in Chemistry and Biology Beat Emst, Gerald W. Hart, Pierre Sinay copyright Q WILEY-VCH Verlag GmbH. 2000

53 Glycobiology of Protozoan and Helminthic Parasites Richard D. Cummings, and A . Kwame Nyame

53.1 Introduction Billions of people world-wide are infected with different parasitic helminths (worms) and protozoans [ 11. Parasites flourish in tropical environments and areas lacking proper hygiene, and affect many people in developing countries and travelers to those countries. Surprisingly, no effective vaccines or preventatives exist against the major parasitic infections and treatments of infected people and animals are irregular in endemic areas and may proceed without definitive diagnosis. Since drug treatments only have short-term effects, people in endemic areas are often reinfected. Using the tools of modern cell and molecular biology and immunology, it is hoped that a better understanding will be gained about the cellular mechanisms by which parasites invade their hosts and thrive. It is further hoped that this new understanding will lead to new diagnostics, therapies and effective vaccines for the major parasitic diseases. In this regard there is a growing realization that complex carbohydrates are key components of many important antigens identified in protozoan and helminthic parasites and that carbohydrate-binding proteins are perhaps critical in the invasion process. This review will attempt to highlight some of the recent advances in this fascinating area of research, with special emphasis on the structures of parasite-derived glycans, their expression, the immune responses to them, and the involvement of parasite- and host-derived lectins in the infection process. Unfortunately, detailed glycobiology is known only for a few of the major parasites but, if the current information can be generalized, glycoconjugates and carbohydrate-binding proteins probably play a major role in the life cycle and infectivity of all parasites.

53.2 General Classification of Parasites Parasites are organisms that exist at the expense of their hosts. They infect humans and animals and may be generally divided into several broad categories; bacteria,

868

53 Glycobiology of Protozoan and Helminthic Parasites

viruses, fungi, protozoans (single-celled animals), helminths (worms: trematodes, cestodes and nematodes), pentastomes, acanthocephalans and arthropods. This review will be limited to the protozoan and helminthic parasites. The protozoan parasites include those that cause malaria (Plasmodium sp.), Chagas’ disease (Trypanosoma cruzi), sleeping sickness ( Trypanosoma brucei gambiense and T.b. rhodesiense), Leishmaniasis (Leishmania sp.), and amoebic dysentery (Entamoeba histolytica). Hundreds of different protozoan parasites, spanning six separate phyla have been identified to date. The parasitic helminths include trematodes, such as Fasciola hepatica and Schistosoma sp., cestodes, such as Taenia solium and Echinococcus sp., and nematodes, such as Ascaris lumhricoides and Toxocara sp. Altogether, more than 300 different species of helminths have been recorded in human infections.

53.3 The Major Protozoan Parasites 53.3.1 Malaria Malaria is caused by several species in the Plasmodium genus, but P. falciparum is the most virulent. The malaria organism has a complex life cycle involving an invertebrate host (Anopheles sp.) and a vertebrate host (reptile, bird or mammal). Many of the stages in the life-cycle of the malarial parasite appear to involve glycoconjugates [2]. When the mosquito bites its vertebrate host sporozoites are released through its saliva and they quickly enter the blood stream and rapidly home within hours to the liver parenchyma. This adhesion and invasion of hepatocytes involves a circumsporozoite (CS) protein on the parasite that recognizes carbohydrate-associated ligands (particularly heparan sulfate) on the basolateral domain of the hepatocyte [3, 41. The CS protein is just one of many lectins expressed by the malarial parasite (Table 1). The CS protein contains a cell adhesive sequence termed region 11-plus [5], which is conserved in all malarial species and is required for CS protein binding to hepatocytes [6]. Thrombospondin is a 420-kDa glycoprotein in animals for the attachment, migration, and proliferation of many different cell types. It has several cell adhesive domains and shares homology with the region 11-plus domain of the CS protein [7]. Curiously, heparan sulfate is ubiquitous on animal cells, yet sporozoites attack primarily hepatocytes. It now appears that the specific interactions of sporozoites with heparan sulfate in hepatocytes may occur in association with a specific hepatocyte-expressed low density lipoprotein receptor-related protein (LRP) [8]. Thus, a related set of heparan sulfate proteoglycans in liver hepatocytes may participate in initial binding of malaria sporozoites and lipoprotein remnants [9, lo]. Interestingly, sporozoites can adhere to the parental Chinese hamster ovary (CHO) cell line, but not to mutants defective in glycosaminoglycan synthesis [ 111. Adhesion of malaria-infected blood cells to chondroitin sulfate has been demonstrated in a variety of systems, including adhesion under flow conditions [ 121 and direct adhesion to immobilized chondroitin sulfates [for example, see 13-16]. It has also been

merozoite merozoite sporozoite infected erythrocyte membrane trypomastigote

promastigotes promastigote tachyzoite merozoites trophozoite trophozoite sporozoite lectin

trophozoite

Plasmodium fulciparum

Leishmania species

Toxoplusina gondii Surcocysti.~muris Giurdiu lumbliu Cryptosporidium purvum Acunthumoebu kerutitis

Entumoebu histolytica

Trypanosornu cruzi

Stage

Parasite

Gene B toxoplasma lectin adhesin taglin 148 kDa Gal/GalNAc 136 kDa mannose-binding protein Gal/GalNAc lectin

Gdl/GalNAc

NeuSAcu2+3Gal NeuSAcw24R Heparan sulfate Heparan sulfate NeuSAca2+3Gal Heparan sulfate GlcNH2, GalNH2, and sialic acid Lipophosphoglycan? (LPG) Heparan sulfate Sulfated glycans Man-6-phosphate? a-Methyl-mannoside Gal/GalN Ac Man

EBA-175 MSA-1 circumsporozoite protein (CS) PEMPl trans-sialidase penetrin

HA

Ligdnd

Carbohydrate-binding protein

Table 1. Examples of carbohydrate-binding proteins in parasitic protozoans.

2

P.

4

5.

5

Y

k,

'i.

Q

870

53 Glycobiology of Protozoan and Helminthic Parasites

shown that a tetradecasaccharide fraction from chondroitin sulfate A blocks P. fulciparum sequestration in the microvascular endothelium of various organs [ 171. Thus, both heparan sulfate and chondroitin sulfate may contribute to adhesion of malaria parasites to cells. Within the hepatocyte, the sporozoite undergoes a metamorphosis into a feeding trophozoite, that ingests the cytosol of the infected cell. Once the trophozoite matures, it undergoes schizogony, in which numerous daughter nuclei and formed, and a schizont phase is reached, where the schizonts mature into merozoites. The merozoites leave the liver and enter the circulation where they attach to and enter circulating erythrocytes and initiate the erythrocytic cycle. There the merozoite reverts to a trophozoite, where it again ingests cytoplasm, and causes a characteristic ring formation in the erythrocyte known as the “signet ring stage”. Pigmented deposits of heme and protein (hemozoin) are generated by the parasite and sequestered. Such sequestration is important for survival of the parasite, because the heme derivative ferriprotoporphyrin IX (hematin) is toxic to the trophozoites, due to its ability to inhibit plasmodia1 proteases. Interestingly, chloroquine, which is used to treat malaria, raises the pH of food vacuoles in the parasite, inhibiting their degradation. Chloroquine also binds to hematin, thus preventing its sequestration. Rosetting between P. faleiparum-infected erythrocytes and uninfected erythrocytes is associated with the occurrence of severe malaria. This rosetting is inhibited by heparan sulfate and heparatinases (heparan sulfate-degrading enzymes), indicating a possible involvement of heparan sulfate as the so-called rosetting receptor [ 181. Dextran sulfate 500K, dextran sulfate 5K, sulfatides, fucoidan, and heparin, but not by chondroitin sulfate A, also block merozoite adhesion to parasitized erythrocytes in vitro [19]. cDNA cloning has been reported for the adhesive ligand generated by P.fakiparum and termed the P. fulcipurum erythrocyte membrane protein 1 (PfEMPl), which contains clusters of glycosaminoglycan-binding motifs and binds glycosaminoglycans 120, 211. PfEMPl is predicted to contain 2,228 amino acids with an estimated molecular weight of 260 kDa. A recombinant form of PfEMPl was shown to functionally recognize heparan sulfate and bind to normal erythrocytes. GAG-binding motifs are predicted to contain clusters of basic amino acid residues in the consensus sequences XBBXBX or XBBBXXBX (B = basic residue: Lys, Arg or His) 1221, although more complex motifs have also been found [23]. The GAG binding proteins in malaria parasites are probably important in vivo, since evidence indicates that chondroitin 4-sulfate administration intravenously in squirrel monkeys interferes with in vivo cytoadherence of parasitized erythrocytes [24]. For invasion of erythrocytes, sialic acid on the erythrocyte surface, probably on the major sialoglycoprotein glycophorin, is also critical for binding and invasion of those cells by the merozoites. Removal of sialic acid by neuraminidase treatment blocks merozoite-erythrocyte interactions. Erythrocyte glycophorin has been implicated in this recognition phenomenon, since individuals lacking glycophorin A (Ena-) or glycophorin B (MkMk) are resistant to invasion. P.faleiparum merozoites bind to parental CHO cells expressing human glycophorin A, but not to glycophorin A expressed in L e d CHO cells, which lack the ability to sialylate glycoconjugates [25],suggesting that invasion of erythrocytes may be mediated by sialic

53.3 The Major Protozoan Parasites

871

acid. There appear to be several sialic acid-binding proteins in merozoites, and two of them that have so far been identified are EBA-175 and MSA-1, which are immunologically distinct and may have different functions. EBA- 175 interactions with erythrocytes may occur by multiple mechanisms, one involving a sialic acid-dependent interaction [26], and another that is sialic acidindependent. Both MSP-1 [27] and EBA-175 antigen [2S] have peptide domains that recognize erythrocyte glycophorin independent of a sialic acid on glycophorin. It has been proposed that initial binding of EBA-175 to erythrocytes is sialic aciddependent and that binding induces a conformational change in EBA-175 that exposes a peptide region of EBA175 that binds in a sialic acid-independent manner and enhances fusion [2S]. Hapten inhibition experiments indicate that both EBA and MSP-1 bind sialic acid in a linkage-specific manner. Other carbohydrate-binding proteins, such as those that recognize GlcNAc-containing glycans, may also be important [29]. Some recent studies have reported that microgametes, which are released by a process of exflagellation from male gametocytes, also recognize and bind to erythrocytes in a process that appears to be dependent on surface sialic acid and erythrocyte glycophorin A [30]. Upon maturation, the trophozoites undergoes schizogony in their asexual cycle to form daughter merozoites, which are released by a ruptured erythrocyte, and continue the cycle of merozoite invasion of new erythrocytes. After many such schizogonic cycles, the merozoites develop into sexually differentiated male and female gametocytes that continue the sexual life cycle in the midgut of the mosquito upon ingestion in a new blood meal. Interesting new in vitro assays suggest that the ookinetes of P. gullinaceum within the midgut of its mosquito host Aedes aeyypti may bind to midgut epithelium via adhesion to a carbohydrate ligand in the host cells [31]. Surprisingly, sialic acid can inhibit this adhesion, although insects do not appear to synthesize sialic acid. In addition, chitin-derived small oligosaccharides containing only GlcNAc inhibit parasite development in the mosquito, when added to the bloodmeal of the insect [32].After a number of asexual generations, some merozoites in infected erythrocytes differentiate into macrogametocytes and microgametocytes that are released by infected erythrocytes. These gametocytes in the blood of the vertebrate host have a short half life and die unless ingested by the mosquito, where the life cycle is continued. Interestingly, the gametocytes can bind to infected and uninfected red cells of the host, but the function of this interaction is unknown. Human malaria gametocytes exhibit interesting host specificity. For example, P. falciparum microgametes can bind human erythrocytes, but not chicken erythrocytes and the binding is blocked by neuraminidase treatment. In contrast, avian malaria, P. gallinaceum microgametes bind chicken, but not human erythrocytes [ 301. Surprisingly, recent evidence suggests that P. falciparum-infected erythrocytes can roll on activated human endothelial cells expressing P-selectin in a Ca2+ and sialic acid-dependent manner [33]. The nature of the ligand(s) on the infected red cells are not known, but it is intriguing to consider that the ligands may be specific parasite-encoded ligands related to those on human neutrophils known to be important for P-selectin recognition [34]. Some of the glycoconjugates synthesized by Plasmodium may also be important

872

53 Glycobiology of Protozoan und Helrninthic Parasites PO&!H,-CH,-NH-CO-Protein

R3

‘

Lc*H:

6myo-Inositol- 1 - r ;

R4

c=o c=o acid

acid

Plasma Membrane

“Inside”

In T. brucei VSG; R,, &, and R, = OH, R, = aGal,, ; R, = ah4an; and the lipid is a diacyl-glycerol In T. crwi 1G7; R, = M a n , R,, R, and R, =OH;R, = ethanolamine or 2-aminoethylphosphonate In L major PSP; R,, R2. R, and R, = OH and the lipid is alkyUacyl-glycerol (Lipid structures may vary depending on the species and the stage of the life-cycle)

Figure 1. The structures of glycosylphosphatidylinositol (GPI) anchored protein in Trypanosoma sp. The conserved trimannosyl core, phosphoethanolamine linkage to protein and GlcN linkage to PI, are common to all GPI structures in eukaryotes. The specific modifications of this core in trypanosomes with carbohydrates, acyl groups, and ethanolamine or 2-aminoethylphosphonate are indicated.

in the host invasion steps. Plasmodium generates both free and protein-bound glycosylphosphatidylinositol lipids (GPI) (Figure l), which are structurally different, and whose expression is developmentally regulated [35]. The free GPIs from Plasmodium exhibit bioactivity and can elevate expression of some host adhesion molecules, such as I-CAM-1, V-CAM-1 and E-selectin in human umbilical vein endothelial cells, via a tyrosine phosphorylation cascade [ 361. The mannose-rich core region of free GPIs from P. falcipurum can activate protein tyrosine kinase in macrophages and the GPI-associated diacylglycerol can independently activate protein kinase C [37]. Thus, GPIs may constitute a kind of outside-in signaling system with a high degree of bioactivity. Malaria parasites derived from cells of New World monkeys and non-primate animals, which express a specific glycosyltransferase in their Golgi apparatus termed the al,3-galactosyltransferase,can acquire the so-called al,3-Gal antigen on surface glycoconjugates [38]. These surface a1,3-Gal structures are recognized as foreign antigens and destroyed by circulating anti-al,3-Gal IgG that normally

53.3 The Major Protozoan Parasites

813

occurs in the blood of Old World monkeys and humans [ 3 2 ] . This interesting glycosylation difference may have implications for evolutionary inactivation of the CI 1,3-galactosyltransferasegene in human and Old World primates and the overall elicitation of other blood group antigens.

53.3.2 Trypanosomiasis Trypanosoma cruzi is the cause of Chagas’ disease, which is a serious disease in South America affecting multiple organ systems. T. brucei garnbiense and T.6. rhodesiense cause African sleeping sickness in humans. T.b. brucei does not infect humans, but causes a disease in native antelopes and other African ruminants called Nagana. T. cruzi is sequestered in host cells, where it can partly evade the immune system, but T. brucei lives free in the host blood and lymph, where it is constantly exposed to host immunity. The African trypanosomes are transmitted by the tsetse fly (Glossina sp.). The life cycle of the T. brucei group can be traced from the intestinal tract of the flies (trypomastigotes + epimastigotes t metacyclic trypomastigotes) to the blood of the mammalian hosts (slender trypomastigotes + intermediate trypomastigotes + stumpy trypomastigotes). The cycle is reinitiated during the insect’s bloodmeal. In contrast, T. cruzi is transmitted by the blood-sucking reduviid bugs of the genera Triatoma, Panstronyylus and Rhodnius. As the bugs bite their mammalian hosts, they deposit feces containing the metacyclic forms of T. cruzi. The parasites enter the blood through the damaged skin and penetrate mainly cardiac muscle cells eventually. There, they change from amastigote 4 epimastigote + trypomastigote forms (replicating only as the amastigote). The trypomastigotes leave the muscle and enter the blood, where they may reinvade muscle or circulate without division until another insect ingests them during blood meal. In the insect the ingested trypomastigotes change to epimastigotes before reaching their metacyclic state. To evade host immunity, T. brucei expresses high levels of a glycosylphosphatidylinositol (GP1)-anchored variable surface glycoprotein (VSG) on its plasma membrane (approximately ten million copies per cell) [39].The GPI anchor is a complex glycolipid linked to the C-terminus of the VSG and many other less-abundant trypanosome glycoproteins (Figure 1). The VSG appears critical for successful infection in animal hosts and it undergoes an extraordinary sequence alteration over time due to expression of new VSG genes. The organism can express up to 1,000 different VSGs, but only one is expressed at a time. The new variants are not recognized by pre-existing antibodies, thus allowing a temporary evasion of immunity. When immunity arises to these variants resulting in their destruction, other variants expand that are not recognized by existing immunity, and the cycle continues until death of the host. However, antigen switching is not dependent on immunity and may be more relevant to parasite competition [40]. When the trypanosome is ingested during a bloodmeal, it initially sheds its VSG coat. Later the VSG coat is re-expressed, thus making the parasite competent to infect mammalian hosts. Other trypanosomes also express GPI anchors, like all eukaryotes, but the specific core modifications differ for each species [41]. While the exact functions of GPJ anchors

874

53 Glycobiology of Protozoan and Helminthic Parasites

are not yet clear, GPI anchors in T. cruzi are required for effective development and parasitism. Induction of a glycosylphosphatidylinositol (GPI) deficiency in T. cruzi by the heterologous expression of T. brucei GPI-phospholipase C depresses surface glycoprotein expression and decreases virulence in infected animals 1421. Cell free systems using washed trypanosome membranes from T. brucei and radiolabeled intermediates have been critical to working out the biosynthetic pathway for GPI anchors, which is largely shared by protozoans and higher animals [4345; and reviewed in 411. GlcNAc is first transferred from UDP-GlcNAc to phosphatidylinositol (PI), forming GlcNAc-PI, which is subsequently de-N-acetylated to generate GlcN-PI. Three mannose residues are added using the donor dolichol phosphoryl-mannose to form Man3GlcN-PI. This intermediate is acylated on the inositol moiety prior to addition of ethanolamine-phosphate, donated from phosphatidylethanolamine, to form glycolipid C’. Following this, the acyl-inositol is deacylated to form the GPI glycolipid A’. This glycolipid undergoes fatty acid remodeling in which sn-1 and sn-2 fatty acids are removed and replaced with myristate to generate the VSG precursor, glycolipid A. Interestingly, there are several myristate exchange reactions that function in an editing pathway to ensure myristolyation of the final GPI anchor. The addition of the GPI anchor to the C-terminal domain of glycoproteins occurs post-translationally and involves a C-terminal GPI-directing signal sequence (10-30 residues at the C-terminus that are largely hydrophobic), which directs the specific attachment of a GPI anchor [46]. The removal of the C-terminal peptide and addition of the GPI anchor is catalyzed in the ER by an unidentified enzyme termed a “transamidase” that catalyzes a transamidation reaction [47-491. As discussed for Plasmodium, sialic acid residues are also critical for successful attachment of circulating T. cruzi trypomastigotes to host cells [50]. The interesting adhesion molecule for this attachment was identified as a GPI-anchored transsialidase (TS) on the surface of the protozoa 151-541. TS can act as a conventional sialidase, but it also has the remarkable ability to remove sialic acid from glycoconjugates of the host and transfer it to its own glycoconjugates in u2-3 linkage to penultimate galactosyl residue in O-glycans (55, 56; also see below). These observations explain the anomaly that the parasites are sialylated, but they lack a sialyltransferase. The TS arises from a family of -80 genes, which share a common N-terminal domain (including the catalytic portion of the enzyme) and a variable number of 12-amino acid repeats in the C-terminal domain 1571. Although not all of these genes appear to encode a functional trans-sialidase activity, their expression is generally correlated with virulence 158, 591. The TS activity is specific for sialic acid linkage (u2,3) and for attachment to terminal Gal residues. The sialic acids appear to protect from complement activation and antibody formation. Interestingly, antibodies induced to the TS by immunization with cDNA for the enzyme provide protective immunity to challenged animals [60]. The O-glycans of T. cruzi surface glycoproteins are the acceptors for the TS action 1611. The structures of the underlying O-glycans have been studied in detail and are highly unusual, but most share a common terminal pl,3-galactosyl residue. The 0-glycans from the 38/43 kDa surface glycoproteins of T. cruzi epimastigote were found to all contain an unusual O-linked GlcNAc core structure and B1-6

53.3 The Major Protozoan Parusites

875

linkages of Gal residues to this core 1621. The structures included GalJ’p1-4 (Galppl-6)GlcNAc-01; Galp~1-3Galp~1-6(Galf~1-4)GlcNAc-ol; [(Galppl-3) (Galppl-2)Galp~1-6](Galfpl-4)GlcNAc-ol; and [(Galpp1-3)(GalpP1-2)Galpp1-61 (Galp~1-2Galf~l-4)GlcNAc-ol. Interestingly, 0-glycans in the Y-strain of T. cruzi have a somewhat related structure in that 0-linked GlcNAc cores are present but, there are remarkable differences otherwise [63].One series contained PI-3 linkages of Gal to the GlcNAc-ol core and included Galppl-3GlcNAc-01; Galppl-6 (Galppl-3)GlcNAc-01; and Galp~l-2Galp~l-6(Galp~l-3)GlcNAc-ol, whereas the other series had a pl-4 linkage of Gal to the GlcNAc-ol core and included Galppl4GlcNAc-01, Galppl-6(Galppl-4)GlcNAc-ol,and Galpp 1-2Galppl-6(Galpp1-4) GlcNAc-01. Additionally, the unusual GalfP1-4GlcNAc-01 with its galactopyranose was also observed in the epimastigote mucin of the G strain of T. cruzi [611. The exact mechanism by which T. cruzi enters mammalian host cells is not clear. It is possible that lectins expressed by host cells or the parasite may be important. Trypomastigotes express a surface heparin-binding protein of -60 kDa termed penetrin, which interacts with heparan sulfate molecules of host cells 164, 651. There is also evidence that mannose-containing glycoproteins on amastigotes, which are absent from trypomastigotes, may be ligands for human serum mannose-binding proteins [66], which may facilitate their entry into phagocytic cells and other cell types. In addition to the obvious value of studying trypanosomes for insights into their biology, studies of N-glycan biosynthesis in trypanosomes have been particularly fruitful in defining the general functions of glycoconjugates in glycoprotein biosynthesis. Trypanosomes are unable to synthesize Dol-P-Glc and, therefore, synthesize a lipid-linked oligosaccharide donor that is Man6,7,,,9GlcNAc2-P-P-Dol, depending on the species [reviewed in 671. The oligosaccharyltransferase of trypanosomes, unlike the mammalian enzyme, efficiently transfers these non-glucosylated donors oligosaccharides to Asn residues in newly synthesized glycoproteins. However, it was found that trypanosomes have a glucosidase 11-like activity, which was shown to be required for deglucosylation of N-glycans glucosylated by the UDP-Glc: glycoprotein glucosyltransferase, an activity that was first detected in those parasites. Glucosylation and deglucosylation of newly synthesized proteins now appears to be an important aspect of chaperone-assisted (calnexin/calreticulin) glycoprotein folding and exit from the ER in all eukaryotes 168, 691.

53.3.3 Leishmaniasis Visceral Leishmaniasis (Kala-azar) is caused by the protozoan parasite Leishmania donovani; other species cause mucocutaneous Leishmaniasis. The disease is transmitted by sand flies, which are the intermediate host or vector. When the flies bite an infected individual they ingest amastogotes forms of the parasite, which undergo transformation in the midgut or hindgut of the fly into promastigotes forms. There, the promastigotes multiply and move to the esophagus and pharynx of the fly, where they can exit into the mammalian host during feeding. Once in the mamma-

876

53 Glycobiology of Protozoan and Helminthic Parasites

-Repeatingunit

Cap

Glycan Core

PI-Anchor

Gala1

I I

Po;

Mana1,Z

I

Gal~l,4Manal-PO~-[6GI1I~1,4Manal-P-],-6Galal,6Galal,3Gal~l,3Manal,3Manal,4GlcNal,6

I

Inositol-PO;-CH,

LPG from L. donovani

I

?"-OH CHz-O-[CHzlz,,zW,

Glcp1,3

I

[Manal,Z],,-[Manal-P0,-6Galpl.41,4],.~~al-P0~ [Galp 1.4Manal -Pq-],Gal Pl,4Manal -PO;-

~ a l , 2 ] , , - M a n a l -PO;-

SAPfrom L. mexicana

coon

[GalpI .4Manal-PO;-],Gal~l,4Manal -PO;-Ser

'de

I

+/-GlcP1,3

I

Manal,ZManal -PO;-Ser

+/-AraPl,Z

PPG from L. major COOH

Figure 2. Structures of the phosphoglycans from Leishmania donovani, L. mexicana and L. major. The mucin-like proteins from L. mexicuna and L. major carry large numbers of Ser-linked O-glycans, with the various structures indicated.

lian host, the parasites invade host cells, in particular they invade phagocytic cells, such as macrophages. The promastigotes bind to and enter macrophages where they lose their flagella and are transformed into amastigotes, which multiply in the host cell by binary fission. The amastigotes are released by rupture of macrophages, enter the blood stream, and complete the life-cycle when blood is taken from an infected host by another sand fly. The promastigote surface is covered with a macromolecular glycoconjugate, known as a phosphoglycan, which is occurs in a different form for all the Leishmania sp. The phosphoglycan is characterized in all species by a high mannose content (Figure 2). In L. donouani this is present in the lipophosphoglycan (LPG), which is not linked to protein. However, the phosphoglycan of L. mexicanu, such as secreted acid phosphatase (SAP), is linked to Ser residues in proteins [70]. In L. major the phosphoglycan is linked to Ser residues to form a proteophosphoglycan (PPG). The biosynthesis of the highly unusual Leishmania phosphoglycans is becoming clearer, partly due to the ability to isolate laboratory strains defective in phosphoglycan synthesis and the ability to perform complementation analysis of the altered gene [71]. The LPG of L. donovani is synthesized by sequential addition of Man-lphosphate from GDP-Man and Gal from UDP-Gal [72]. Recent biochemical and

53.3 The Major Protozoan Parasites

877

genetic evidence indicates that LPG synthesis requires at least one mannophosphoryltransferase for initiation and another that is required for elongation of the repeating units [73]. One of the mutations in Leishmania affecting mannoseincorporation into LPG was identified to be within a gene encoding a protein with predicted homologies to membrane transporters, and is predicted to encode the GDP-Man transporter [74]. Mutations in addition of the unusual galactofuranose by a galactofuranosyltransferase, neither of which are present in animal hosts, also blocks phosphoglycan assembly and adversely affects virulence of the parasite [75, 761. The Leishmania phosphoglycans are probably critical for recognition of the parasite by macrophage mannose receptors [ 771. Thus, the receptor-mediated endocytotic system may be suicidal for macrophages. But in addition to this role in adhesion, the phosphoglycans also have wide-ranging biological activities. The L. donovani LPG can inhibit phagosome-endosome fusion and prevent lysis of the endocytosed parasite. The LPG of L. donovani inhibits viral fusion with cells at low concentrations (20 pM) [78], which is associated with conformational changes in fusion peptides of the virus [79, 801. LPG blocks transendothelial transmigration of monocytes, probably by interfering with the expression of adhesion receptors, such as E-selectin, ICAM-1 and VCAM [81]. LPG suppresses induction of IL-1 in human monocytes at a unique “gene silencer” promoter sequence distinct from that of known agonists [82]. In addition, the LPG can activate HIV-1 replication in monocytoid cells and directly up-regulate HIV- 1 transcription in T cells, via activation of transcription factors that recognize NF-kappaB binding sites [83, 841. These results support the possibility that L. donovani may be a putative cofactor in HIV-1 pathogenesis. The LPG of L. donovani is generally required for the parasite’s survival; mutagenized Leishmania promastigotes that cannot synthesize LPG grow well in vitro, but they can neither grow within the sand fly nor establish infections in macrophages [72]. The purified LPG from L. mexicana efficiently activates complement and depletes hemolytic activity of normal serum, which may limit opsonization of L. mexicana amastigotes [85]. LPG inhibits protein kinase C in macrophages [86], which is normally required for oxidative burst in these cells. It has been proposed that a galactosyltransferase in L. donovani promastigotes is developmentally regulated and that up-regulation of this enzyme is associated with loss of virulence [87]. Thus, the phosphoglycans may all function to allow survival and reproduction of parasites within vacuoles of infected cells. Altogether, the phosphoglycans expressed by Leishmania sp. may be among the most potent weapons of defense presented by protozoan parasites, since these organisms are able to infect and thrive in host macrophages. Another unusual feature of leishmaniasis is the demonstrated presence of significant titers of antibodies to O-acetylacted sialic acid in sera of patients with visceral leishmaniasis, which may be useful in diagnosing the disease [ 881. In addition to exploiting the lectins of the host system for its infective mechanisms, Leishmania may also express its own types of lectins. Extracts of many species of Leishmania exhibit hemagglutination activity inhibitable by galactosamine, glucosamine and sialic acid, but not by GlcNAc or GalNAc [89]. Interestingly, the hemagglutinating activity appeared to be the highest titer in the infective compared

878

53 Glycobioloyy of Protozoan and Helminthic Parasites

to uninfective strains of Leishmania. There is also speculation that the parasites express another lectin termed the Gene B protein, which may associate with the glycocalyx of the parasite [90].

53.4 Other Protozoan Parasites 53.4.1 Entamoeba histolytica

One of the major glycoproteins expressed by E. histolytica, which is a major cause of amoebic dysentery, is a carbohydrate-binding protein. The parasite expresses a lectin [91, 921 that recognizes Gal/GalNAc residues [93, 941. The Gal/GalNAc lectin has heavy (-170 kDa) and light (31-35 kDa) subunits [95-971 and causes adhesion of trophozoites to colonic mucin, epithelium, and other target cells. Adhesion is followed by contact dependent cytolysis of host cells.

53.4.2 Acanthamoeba A. keratitis expresses a mannose-binding protein that is involved in adhesion to corneal epithelium [98] and whose recognition is highly inhibited by Manal-3Man disaccharides. Injury to the cornea may expose mannose-containing glycoproteins that act as receptors for the invading parasite [99]. This lectin-mediated adhesion of A. keratitis to host cells is a prerequisite for the amoeba-induced cytolysis of target cells.

53.4.3 Giardia lamblia

Mannose- and mannose-6-phosphate inhibit adhesion of G. lamblia trophozoites to cultured human intestinal cells [IOO]. One of the lectin-like proteins from the parasite that may bind these sugars is termed taglin, which may be in the range of 28-30 kDa [101, 1021. In addition, the parasite expresses a 148 kDa protein with a-methylmannoside binding activity [ 1031.

53.4.4 Cryotosporidiumparvum C. parvum is an opportunistic protozoan infecting individuals with compromised immunity, resulting in severe diarrhea and wasting. The sporozoites of the parasite express a lectin with hemagglutinating activity, inhibitable by Gal and GalNAc, that may be involved in parasite attachment to intestinal epithelium [ 104-1061.

53.5 Helmintlzic Parasites

819

53.4.5 Surcocystis spp The merozoites of S. nzuris express a surface antigen of -1 6 kDa that dimerizes and has lectin activity toward GalNAc and Gal [107], although it is not yet clear whether this lectin is involved in parasite adhesion to host cells. S. yigantea contains a dimeric lectin (-19 kDa subunits) with mitogenic activity that is inhibited by galactose-containing ligands [ 1081.

53.4.6 Toxoplasmu gondii T. gondii causes a debilitating disease in humans with wide ranging symptoms. T. gondii tachyzoites interact with and invade a wide variety of cells. such as fibroblasts, epithelial cells, endothelial cells, macrophages, and cells of the central nervous system. Recent evidence shows that tachyzoites express a lectin-like activity that recognizes sulfated polysaccharides [ 1091. In addition, the parasite does not bind in vitro to cells that are unable to express certain glycosaminoglycans. 53.4.7 Pneumocystis carinii

P. carinii is an opportunistic parasitic protozoan causing respiratory pneumonia. The parasite expresses a mannose-rich surface antigen termed gpA, which may be a ligand within the airspace lining for the C-type lectin surfactant protein D (SP-D) [ 1 lo]. It has also been reported that P. carinii expresses a lactose-inhibitable lectin that can be detected by hemagglutination assays [ 1 1 11.

53.5 Helminthic Parasites Schistosomiasis is a parasitic disease of humans caused by flat blood helminths of the trematode class. The disease is characterized by an enlargement of the spleen and liver together with changes in the blood circulatory system. Schistosomiasis is a major cause of morbidity and mortality in endemic areas and three main species, Schistosoma mansoni, S. japonicum and S. huematobium, are responsible for humans infections worldwide. The parasites have a complex life cycle which alternates between definitive vertebrate hosts and intermediate freshwater snail hosts, which differ for each Schistosomu species [ 112, 1131. Schistosomes are unique among trematodes in that they have separate sexes and live as male and female pairs in the blood vessels of their vertebrate host, where the females lay eggs that adhere to the endothelium. The adhered eggs have two fates. They either become lodged in host tissues to initiate the pathology associated with the disease, or they eventually pass out of the host in stool or urine, depending on the infecting species. Upon contact

880

53 Glycobiology of Protozoan and Helminthic Parasites

-

LDNF-Antigen Le-Antigen

I

Manpl-4GlcNAcp 1 -4GlcNAcP I -ASN

I

Fucal-3

Extended Le'-Antigen

I Fucal-3

Fucal-6

f

Fucal-3

'

Fuca 1-3

I

3GalP 1- 4GlcNAcp 1-6

Galp 1-4GlcNAc pl (3GaU 14GlckAcp I

\

n

Galpl-4GlcNAcpl-2Manal-

tManpl-4GlcNAcpl-4GlcNAcp Fucal-6 I

I -ASN

I GalPl-4GlcNAc!3 1-2Manal-3

Comulex-tvue N-Glvcans from Membrane and Soluble GlvcoDroteins of S. mansoni Le'-Antigen

Extended Le'-Antigen Fucal-3

'

Fucal-3

I

3GalpI- 4GlcNAcpI -6 n GalPl-3GalNAcal -SERRHR

\

FLVLVI;

0-Glvcans from Circulating Cathodic Antieen (CCA) of S.mansoni idanal-6;

Core-Xy'Antigen

I

~

Xylpl-2Man I-4GlcNA pI-4GlcNAc I-ASN

GlcNAcpl-2Manal-3

Core-Fuc Antigen

Comulex-tvue N-Glvcans from Eggs of S. mansoni and S.jammicum

Figure 3. Structures of some major N- and 0-glycans in Schistosomn sp. Denoted antigenic structures are boxed.

with fresh water, the eggs hatch into free swimming larvae, miracidia, which invade specific freshwater snails and undergo a series of multiple asexual transformations to form numerous infective larvae, cercariae. The cercariae are shed out of the snails into the surrounding water where they continue the infection cycle by seeking and penetrating prospective vertebrate hosts. Adult schistosomes generate a large and complex array of glycan structures on glycosphingolipids and 0- and N-linked glycans on membrane-bound and circulating glycoproteins [112, 1131. Many of the glycans are extensively fucosylated and are antigenic (Figures 3 and 4). Three of the notable antigenic carbohydrate structures found in adult schistosome glycans include the Lewis x (Le') antigen [ 114, 1151, LDN and LDNF [ 116, 1171. Additionally, schistosomes synthesize complextype N-glycans containing the core fucose and core xylose structures [118, 1191. Overall, fucosylation of N- and 0-glycans appears to be a common theme for most

53.5 Helrninthic PwusitPs

881

+/- Fucal-2

I +/- Fucal-2 +I-Fucal-2

I

I

+/- Fucal-3

+/- Fucal-3

I

I

{G~~NA~~~-~G~~NA~~~-~G~I~I-}~G~INA~~~-~GICNAC~I-~ n

\

GalNAcal-SEIUTHR

I {GalNAc~l-4GlcNAc~l-3Galcal}-3GlcNAcpl-4Galpl-3

I

I

n

I

GlcApI-3

I

Ga/NAcpl-6 {GalNAcPI -6) GaNAcpl-6R n>4

I

+/- Fucal-2 +/- Fucal-2

GlcAPI-3

I

+/- Fucal-3

+/- Fucal-3 +/- Fucal-3

Circulating Anodic Antigen (CAA) S. mansoni

I +/- Fucal-2

0-Glvcans in CercerialG l v w d w of S. mansoni Fucald

I Man~l-4GlcNAc~l-4GlcNAc~I-ASN +/-Fucal3 +/- Fucal-2

I

I

+/- Fucul-3 +I- Fucal-3

I

I

+/- Fucal-2

I

I

I

Fucal-3

Fucal-3

+/- Fucal-3

I

GalNAcP 1-{4GlcNAcp I -}4GlcNAcpI -3GalNAcp I -4Glcp I -CERAMIDE

Manald

\

Fucal-6

I

Xylpl-2Man~l-4GlcNAc~l-4GlcNAc~ I-ASN

n=lto3

I

GalNAcPl-4GlcPI-CERAMIDE GlcNAcpl-ZManal-3

Glvcosuhineolioids in S.mansoni

I Fucal-3

N-Glvcans in Haemonchus confoilus

Figure 4. Structures of other unusual glycans in Schistomma sp.

schistosome glycoconjugates. An interesting feature of glycans synthesized by schistosomes and other helminths studied so far is the general absence of sialic acid residues [ 120, 1211. Both al,3- and wl,2-fucosyltransferases catalyzing synthesis of fucosylated glycans in S. munsoni and the bird schistosome Tvichobilhurziu ocelluta have recently been identified [122-1241. In addition, the activity of the unusual N acetylgalactosaminyltransferase that catalyzes synthesis of the LDN antigen has been detected in these parasites [125, 1261. Schistosomes also synthesize many other interesting carbohydrate structures in their eggs, the so-called cercarial glycocalyx, and in secreted glycans [ 1 18, 127- 1291. For example, the circulating anodic antigen (CAA) has 0-glycans with an unusual repeating b1-6 linked GalNAc polysaccharide with GlcUA subsitutions in b1-3 linkage to the GalNAc [130] (Figure 4).Novel synthetic glycans with this structure hold promise in the development of new diagnostic tests for schistosomiasis [ 1311. The expression of many of these glycan structures is developmentally regulated and stage-specific, but their fundamental role in parasite development and host pathogenesis are not clear. It is also appears that the schistosome species differ in several ways in terms of their glycoconjugate structures. For example, the S. mansoni glycosphingolipids have an extended di-fucosylated oligosaccharides, but the terminal difucosylated GalNAc is absent from glycosphingolipids of S. juponicum [ 118, 1291.

882

53 Glycobiology of Protozoan and Helrninthic Parasites

Glycoproteins derived from the tegument, gut and eggs of the parasite are highly antigenic and occur in the circulation of the infected animal. LeX was among the first glycan antigens from schistosomes to be structurally characterized and shown to be antigenic [ 1141. Remarkable, among all helminths tested thus far, schistosomes are the only ones capable of synthesizing LeXglycans [132]. LeXoccurs as repeating polymers on N-linked glycans of the parasites' outer syncytial tegument membrane and on core-2 based 0-linked glycans of the secreted circulating cathodic antigen (CCA) [ 115, 1331. Individuals infected with Schistosoma species develop autoantibodies to the LeXantigen [132, 134-1361. This is surprising in many ways, since Le" antigen is a common mammalian leukocyte marker (CD15). The anti-LeXantibodies are cytolytic towards cellular targets bearing LeX determinants in the presence of complement [ 134-1361. The effect of these cytolytic autoantibodies on infected humans' leukocytes has not been evaluated. Interestingly, the intense humoral immunity induced to Le" and other carbohydrate antigens during schistosomiasis appears to be ineffective in destroying the mature worms. This may be related to the fact that the membrane of the outer syncytial tegument of the adult schistosomes is rough and highly regenerative, even in response to complement attack. However, while the anti-carbohydrate antibodies may only be weakly effective against the resilient tegumental membrane by themselves, the antibodies may be critically important in the antibody-dependent lysis of adult schistosomes by the drug praziquantel. Although there is no direct evidence as yet, antibodies to LeXand other glycans may be an important component of the concominant immunity phenomenon proposed by Smithers and Terrry [ 1371, whereby infected individuals are rendered less susceptible to new infections. Concomittant immunity is believed to be mediated by immune responses induced by existing adult worms which is effective against newly transformed invading larvae, the schistosomula, but not to the adults worms which provoked the response. Interestingly, schistosomula, the young adults, lack the resilient tegumental membrane found in adult schistosomes. LeX glycans may modulate many aspects of host immune responses and hostparasite interactions. There is now evidence that expression of the LeXantigen by the parasite may be important in compromising host cellular immunity [138, 1391. During chronic schistosome infection, Th2 immune responses (promoting humoral immunity) predominate over Thl responses (promoting cellular immunity). In response to glycans containing the LeXantigen, murine B-1 cells secrete in vitro large amounts of IL-10. Because 11-10 can depress Thl responses in animals, this may partly contribute to Th2 dominance in early stages of schistosomiasis. Animals immunized to the LeXantigen displayed an increased cellular responsiveness toward beads derivatized with soluble egg antigen (SEA), suggesting that this response may be involved in modulation of egg granuloma [140]. LeXglycans synthesized by schistosome eggs may also play a role in egg interactions with the endothelium. Recent studies show that adhesion of eggs to activated endothelium can be blocked with monoclonal antibodies to LeX[141]. Glycans derived from schistosome eggs are also ligands for selectins, such as soluble forms of L-selectin [ 1421, a member of the C-type family of carbohydrate-binding proteins expressed in the vascular bed. Surprisingly, schistosomes themselves may

53.7 Unusual Glycuns in Other Helrninthic Purusites

883

express selectin-like molecules that interact with host cells expressing fucosylated glycans. This may be important in mediating antibody-dependent cell-mediated cytotoxicity (ADCC) of schistosomes by macrophages and eosinophils [ 143, 1441. LDN and LDNF glycans are expressed by all adult and larval stages of schistosome species analyzed thus far [ 145, 1461. LDN and LDNF determinants occur on biantennary N-linked glycans on glycoproteins of the tegumental membrane [ 1 171. Both LDN and LDNF induce strong humoral immune responses in infected humans and animals [ 145, 1461. Interestingly, other helminths, such as Fusciolu hepatica, Eclzinococcus yrunulosus, Dirofiluria immitis, and Huemonchus contortus, also synthesize glycoproteins containing LDN and LDNF sequences [ 1321. The expression of these common glycan antigens is interesting because they may contribute to the immunologically cross-reactive antigenicity observed between schistosomes and other helminths and may be useful in designing specific serodiagnostic tests for helminth infections. The identification of many different antigcnic glycoconjugates from schistosomes is helping in the design of new diagnostic procedures and development of candidate carbohydrate-based vaccines for schistosomiasis. In addition, the characterization of the glycosyltransferases in schistosomes responsible for the biosynthesis of schistosome glycan and of enzymes in their intermediate snail hosts may help identify new drug targets.

53.6 Carbohydrate-Binding Proteins in Parasitic Helminths As discussed above, carbohydrate-binding proteins play an impressive role in host infection by many different parasitic protozoans [ 1471. But parasitic helminths may also express different lectins that may be involved in host-parasite interactions. For example, the cDNAs of several galectin family members (P-galactoside-binding proteins), identified initially as antigenic proteins, have been now been cloned from parasitic nematodes, such as Teladorsayia circumcincta [ 148, 1491. Although the functions of the galectins in these parasites are not known, their presence also in the free-living nematode Caenorhubditis eleyans suggests that these proteins may play a pivotal role in development, and not just in infection. Interestingly, a protein with a C-type (Ca2+-dependent)lectin motif has been identified in the nematode Toxocara canis [150]. The protein was shown to be a functional lectin and recognize mannosyl- and galactosyl-containing targets. Other genes encoding proteins with predicted C-type lectin motifs were also identified in genetic screens of T. canis [ 1511.

53.7 Unusual Glycans in Other Helminthic Parasites The major antigenic glycoconjugates found in excreted/secreted (ES) antigens synthesized by larvae of the parasitic nematodes T canis and T. cati are 0-methylated

884

53 Glycobiology of Protozoan and Helminthic Parasites Fucal-3

\

Fucal-3

Tyvelose -Antigen I

i

I

c

N

A

c

Fucald

I

Manpl-4GlcNAc pI -4GlcNAcp1-ASN

I

p 1-2Msnal-3

I Fucal-3

Comulex-tvueN-elvcans from Trichinella wirulis

GlcNAcPl-6

\

Fucal-6

I

Phosphorylchloline Antigen

Man~l-4GlcNAc~l-4GlcNAc~I-ASN

PC,

I i

GlcNAcP 1-4

ComDlex-tvueN-glvcans from Acanthocheilonemuvireae

Figure 5. Structures of complex-type N-glycans from Trypanusoma spiralis and Acanthocheilunema uiteae, demonstrating the presence of the unusual tyvelose and phosphorylchloline antigenic determinants.

trisaccharides containing 2-0-methyl fucosyl and 4-0-methyl galactosyl residues present in 0-glycans [ 1521. The structures are 2-0-methyl-Fuc(a1-2)4-O-methylGal(P1-3)GalNAc and 2-0-methyl-Fuc(a1-2)Gal(P1-3)GalNAc. Unusually, glycans in the related ascarid Ascaris mum, lack such 0-methylated sugars [ 1521. The intestinal nematode Trichinellu spiralis synthesizes several highly immunogenic glycoproteins that contain the unusual sugar tyvelose (3,6-dideoxy-~-arabinohexose; Tyv) [ 1531 (Figure 5). Tyv is found in complex-type N-glycans of T. spiralis larvae and is a critical antigenic determinant recognized by antibodies in infected animals [ 154-1 561. Strong immunity to these glycans provide protective immunity, which causes expulsion of the invading larvae from the intestine. Highly fucosylated N-glycans are also found in the ruminant parasite Haemonchus contortus [ 121, 157, 1581 including unusual core-Xyl and core-Fuc substitutions (Figure 4), which are also found in Schistosoma sp. Specific monoclonal antibodies to LDNF determinants and LeXhave been used to demonstrate that glycoproteins from adult H. contortus express LDNF, but not the LeXantigen [ 1461. As might be expected, extracts of H. contortus contain high amounts of an al,3-fucosyltransferase activity capable of synthesizing the LDNF antigen [ 1591. Consistent with these findings are structural data that another parasitic nematode, Dirofiluria immitis (dog heartworm), also synthesizes complex-type N-glycans with the LDN and LDNF antigens and fails to synthesize the LeX antigen [160]. These studies may have additional validity in the development of novel vaccines. For example, un-

53.8 Future Directions

885

defined, carbohydrate determinants on specific glycoproteins from H. contortus provide protective immunity against the parasite in goats [ 1611. Complex-type N-glycans from several filarial parasites that infect humans (Onchocercu voluulus) and rodents (Acunthocheilonema viteue) contain a conserved phosphorylcholine-substitution and also contain an unusual outer chitin oligosaccharide structure (GlcNAco-5) [ 162, 1631 (Figure 5). There is increasing evidence that the phosphorylcholine moiety may play an immunomodulatory role in the infected animal [ 1211. Immunological studies on the related parasite 0. gibsoni suggest that the ES glycoproteins may contain phosphorylcholine on 0-glycans [ 1641. The complex-type N-glycans from the cestode (tapeworm) Echinococcus yrunulosus reveal the presence of terminal sequences GlcNAc-R, Galp-GlcNAc-R and Gala-Galb-GlcNAc-R [ 1191. Surprisingly, some of these N-glycans are sialylated, but the origin of the sialic acid is unclear. The parasitic nematode Ascaris mum synthesizes several acidic glycosphingolipids that are immunogenic, including phosphoinositolglycosphingolipid (Gala1-2Ins-Plceramide) [ 1651 and the minor component A F I to be a 3-sulfogalactosylcerebroside (HS03-3GalPl-lceramide) 11661. In addition, A . mum generates a number of neutral glycosphingolipids (Manpl-4Glcp1-1ceramide, GlcNAcP 1-3ManP1-4GlcPllceramide and Gala1 -3GalNAc~1-4GlcNAc~l-3Man~1-4Glc~l-lceramide) [ 1671 and other related glycosphingolipids containing phosphorychloline in 6-linkage to GlcNAc residues [ 1681.

53.8 Future Directions Studies on parasite glycobiology are illustrating the extreme level of complexity in glycan structures among these so-called primitive organisms, and further humble us by the exceedingly cunning ways they exploit and manipulate the host defense systems. Although much remains to be learned about parasite lectins and glycans and their biological functions, the available information indicates that glycoconjugates are critically involved in the life cycle and infectivity of many protozoan and helminthic parasites. It is likely that future studies in this area will be increasingly molecular in nature and provide novel information about the enzymes of parasites that synthesize glycoconjugates and the lectins they make or exploit from the hosts for infection. It is hoped that with a more detailed chemical and molecular understanding of the how parasites enter their host and evade immune destruction, new and more powerful drugs can be found to treat or prevent the parasitemia. In addition, the new information about specific glycoconjugate antigenic structures offers hope that carbohydrate-based vaccines might be developed. Acknowledgments This work was supported by NIH Grant R 0 1 A142272 to RDC and a Grant from the Oklahoma Center for the Advancement of Science and Technology.

886

53 Glycobioloyy of Protozoan and Helminthic Parasites

References 1. Schmidt, G D and Roberts, LS (1996) Foundations of Parasitology (Fifth Edition), p 2, Wm C

Brown, NY. 2. Hoessli, DC, Davidson, EA, Schwarz; RT and Nasir-ud-Din (1996) Glycobiology of Plusmvdium falciparum: an emerging area of research. Glycoconj J 13:l-3. 3. Pancake SJ, Holt GD, Mellouk S, Hoffman SL (1992) Malaria sporozoites and circumsporozoite proteins bind specifically to sulfated glycoconjugates. J Cell Biolll7: 1351-7. 4. Pancake SJ, Holt GD, Mellouk S, Hoffman SL (1993) Malaria sporozoites and circumsporozoite protein bind sulfated glycans: carbohydrate binding properties predicted from sequence homologies with other lectins. Parassitologiu 35 Suppl:77- 80. 5. Sinnis, P, Clavijo, P, Fenyo, D, Chait, BT, Cerami, C, and Nussenzweig, V (1994) Structural and functional properties of region 11-plus of the malaria circumsporozoite protein. J Exp Med 180:297-306. 6. Cerami, C , Frevert, U, Sinnis, P, Takacs, B, Clavijo, P, Santos, MJ, and Nussenzweig, V (1992) The basolateral domain of the hepatocyte plasma membrane bears receptors for the circumsporozoite protein of Plasmodium fulcipurum sporozoites. CeN 70, 1021-1035. 7. Gantt SM, Clavijo P, Bai X, Esko JD, Sinnis P (1997) Cell adhesion to a motif shared by the malaria circumsporozoite protein and thrombospondin is mediated by its glycosaminoglycanbinding region and not by CSVTCG. J Biol Chem 272:19205-13. 8. Shakibaei M, Frevert U 1996. Dual interaction of the malaria circumsporozoite protein with the low density lipoprotein receptor-related protein (LRP) and heparan sulfate proteoglycans. J E x p Med184:1699-1711. 9. Nussenzweig V (1 997) Malaria sporozoites and chylomicron remnants compete for binding sites in the liver. Behring Inst Mitt (99):85-9. 10. Sinnis P, Willnow TE, Briones MR, Herz J, Nussenzweig V (1996) Remnant lipoproteins inhibit malaria sporozoite invasion of hepatocytes. J Exp Med 184:945--54. 11. Frevert U, Sinnis P, Esko JD, Nussenzweig V (1996) Cell surface glycosaminoglycans are not obligatory for Plasmodium herghei sporozoite invasion in vitro. Mol Biochem Purasitol 76257-66. 12. Cooke BM, Rogerson SJ, Brown GV, Coppel RL (1996) Adhesion of malaria-infected red blood cells to chondroitin sulfate A under flow conditions. Blood 88(10):4040-4. 13. Clark, DL, Su, S and Davidson, EA (1997) Saccharide anions as inhibitors of the malaria parasite. Glycoconj J 14:473-9. 14. Maubert B, Guilbert LJ, Deloron P (1997) Cytoadherence of Plasmodium falciparum to intercellular adhesion molecule 1 and chondroitin-4-sulfate expressed by the syncytiotrophoblast in the human placenta. Inject Zmmun 65(4):1251-7. 15. Evans SG, Morrison D, Kaneko Y, Havlik I (1998) The effect of curdlan sulphate on development in vitro of Plasmodiumfulciparum. Truns R Svc Trop Med Hyg 92:87-9. 16. Eda S, Eda K, Prudhomme JG, Sherman IW (1999) Inhibitory activity of human lactoferrin and its peptide on chondroitin sulfate A-, CD36-, and thrombospondin-mediated cytoadherence of Plasmodium fulcipurum-infected erythrocytes. Blood 94:326-32. 17. Beeson JG, Chai W, Rogerson SJ, Lawson AM, Brown GV (1998) Inhibition of binding of malaria-infected erythrocytes by a tetradecasaccharide fraction from chondroitin sulfate A. Infect Immun 66(7):3397-402. 18. Barragan A, Spillmann D, Kremsner PG, Wahlgren M, Carlson J (1999) Plasmodium fulciparum: molecular background to strain-specific rosette disruption by glycosaminoglycans and sulfated glycoconjugates. Exp Parasitol91: 133-43. 19. Xiao L, Yang C, Patterson PS, Udhayakumar V, La1 AA (1996) Sulfated polyanions inhibit invasion of erythrocytes by plasmodia1 merozoites and cytoadherence of endothelial cells to parasitized erythrocytes. Infect Immun 64:1373-8. 20. Chen Q, Barragan A, Fernandez V, Sundstrom A, Schlichtherle M, Sahlen A, Carlson J, Datta S. Wahlgren M (1998) Identification of Plusmodium fulcipurum erythrocyte membrane protein 1 (PfEMPI) as ‘the rosetting ligand of the malaria parasite P. falciparum. J Exp Med 187:15-23. 21. Reeder JC, Cowman AF, Davern KM, Beeson JG, Thompson JK, Rogerson SJ, Brown GV

Rrfkrences

887

(1999) The adhesion of Plasmodium .fulciparum-infected erythrocytes to chondroitin sulfate A is mediated by P. ,falcipurum erythrocyte membrane protein 1. Proc Nut1 Acud Sci USA 96(9):5 198-202. 22. Cardin, AD, and HJR Weintraub (1989) Molecular modeling of protein-glycosaminoglycan interactions. Arteriosclerosis 9:21-32. 23. Hileman RE, Fromm JR, Weiler JM, Linhardt RJ (1998) Glycosaminoglycan-protein interactions: definition of consensus sites in glycosaminoglycan binding proteins. Bioessuys 20: 15667. 24. Pouvelle B, Meyer P, Robert C, Bardel L, Gysin J (1997) Chondroitin-4-sulfate impairs in vitro and in vivo cytoadherence of Plusmodium fulciparum infected erythrocytes. Mol Med 3(8):508-18. 25. DeLuca GM, Donne11 ME, Carrigan DJ, Blackall DP (1996) Plasmodium julciparum merozoite adhesion is mediated by sialic acid. Biochem Biopkys Res Commun 225726-32. 26. Orlandi, PA, FW Klotz, and J D Haynes. 1992. A malaria invasion receptor, the 175kilodalton erythrocyte binding antigen of Plasmodium julcipurum recognizes the terminal Neu5Ac(alpha2-3)Gal-sequences of glycophorin A. J Cell B i d 116:901-909. 27. Urquiza M, Rodriguez LE, Suarez JE, Guzman F, Ocampo M, Curtidor H, Segura C, Trujillo E, Patarroyo ME (1996) Identification of Plasmodium falciparum MSP-I peptides able to bind to human red blood cells. Parasitr Immunol 18515-26. 28. Jakobsen PH, Heegaard PM, Koch C, Wasniowska K, Lemnge MM, Jensen JB, Sim BK (1998) Identification of an erythrocyte binding peptide from the erythrocyte binding antigen, EBA-175, which blocks parasite multiplication and induces peptide-blocking antibodies. Injkct h n i u n 66:4203-7. 29. El Moudni, B, Philippe, M, Monsigny, M and Schrevel, J (1993) N-acetylglucosamine-binding proteins on Plusmodium ,falciparum merozoite surface. Glycobioloyy 3:305- 12. 30. Templeton TJ, Keister DB, Muratova 0, Procter JL, Kaslow D C (1998) Adherence of erythrocytes during exflagellation of Plasrnodiuni ,falciparuni microgametes is dependent on erythrocyte surface sialic acid and glycophorins. J E,xp Med 187:1599-609. 31. Zieler H, Nawrocki JP, Shahabuddin M (1999) Plasmodiunz yallinaceum ookinetes adhere specifically to the midgut epithelium of Aedes upgypti by interaction with a carbohydrate ligand. J Exp Biol202 (Pt 5):485-95. 32. Ramasamy R , Wanniarachchi IC, Srikrishnaraj KA, Ramasamy MS (1997) Mosquito midgut glycoproteins and recognition sites for malaria parasites. Biockim Biophys Actu 1361( I ) : 1 14-22. 33. Ho M, Schollaardt T, Niu X, Looareesuwan S, Patel KD, Kubes P (1998) Characterization of Plusmodium julciparum-infected erythrocyte and P-selectin interaction under flow conditions. Blood 91( 12):4803-9. 34. Leppanen, A, Mehta, P, Ouyang, YB, Ju, T, Helin, J, Moore, KL, van Die, I, Canfield, WM, McEver, RP and Cummings, R D (1999) A Novel Glycosulfopeptide Binds to P-selectin and Inhibits Leukocyte Adhesion to P-selectin. J Biol Chrm 274 24838-24848. 35. Schmidt, A, Schwarz, R T and Gerold, P (1998) Plasmodium julciparuriz: asexual erythrocytic stages synthesize two structurally distinct free and protein-bound glycosylphosphatidylinositols in a maturation-dependent manner. ESP Purusitol88:95- 102. 36. Schofield L, Novakovic S, Gerold P, Schwarz RT, McConville MJ, Tachado SD 1996. Glycosylphosphatidylinositol toxin of Plasrwodiurn up-regulates intercellular adhesion molecule- 1, vascular cell adhesion molecule-1, and E-selectin expression in vascular endothelial cells and increases leukocyte and parasite cytoadherence via tyrosine kinase-dependent signal transduction. J Zmmunol156:1886-1896. 37. Tachado et al 1997. 38. Galili U (1991) The natural anti-Gal antibody: evolution and autoimmunity in man. Imniunol Ser 55355-73. 39. Cross, GA ( I 990) Glycolipid anchoring of plasma membrane proteins. Annu Re0 Cell Biol 6, 1-39. 40. Turner CM (1999) Antigenic variation in Trypanosoma brucei infections: an holistic view. J Cell Sci 112:3187-3192. 41. Ferguson MA (1999) The structure, biosynthesis and functions of glycosylphosphatidylinositol anchors, and the contributions of trypanosome rescarch. J Cell Sci 112:2799-809.

888

53 Glycohiology of Protozoan a n d Helminthic Parasites

42. Garg N, Postan M, Mensa-Wilmot K, Tarleton RL (1997) Glycosylphosphatidylinositols are required for the development of Trypanosoma cruzi amastigotes. Infect Zmmun 65405560. 43. Masterson, WJ, Doering, TL, Hart, GW, and Englund, PT (1989) A novel pathway for glycan assembly: biosynthesis of the glycosyl-phosphatidylinositol anchor of the trypanosome variant surface glycoprotein. Cell 56, 793-800. 44. Masterson, WJ, Raper, J, Doering, TL, Hart, GW, and Englund, PT (1990) Fatty acid remodeling: a novel reaction sequence in the biosynthesis of trypanosome glycosyl phosphatidylinositol membrane anchors. Cell 62, 73-80. 45. Menon, AK, Schwdrz, RT, Mayor, S, and Cross, GAM (1990) Cell-free synthesis of glycosylphosphatidylinositol precursors for the glycolipid membrane anchor of Trypanosoma brucei variant surface glycoproteins. Structural characterization of putative biosynthetic intermediates. J Biol Chem 265, 9033-9042. 46. Udenfriend, S, and Kodukula, K (1995) How glycosylphosphatidylinositol-anchoredmembrane proteins are made. Annu Rev Biochem 64, 563-591. 47. Mayor, S, Menon, AK, and Cross, GAM (1991) Transfer of glycosyl-phosphatidylinositol membrane anchors to polypeptide acceptors in a cell-free system. J Cell Biol114, 61-71. 48. Maxwell, SE, Ramalingam, S, Gerber, LD: Brink, L, and Udenfriend, S (1995) An active carbonyl formed during glycosylphosphatidylinositol addition to a protein is evidence of catalysis by a transamidase J Biol Chem 270, 19576-19582. 49. Sharma DK, Vidugiriene J, Bangs JD, Menon AK (1999) A cell-free assay for glycosylphosphatidylinositol anchoring in African trypanosonies. Demonstration of a transamidation reaction mechanism. J Biol Chem 274: 16479-86. 50. Schenkman, S, Eichinger, D, Pereira, ME A. and Nussenzweig, V (1994) Structural and functional properties of Trypanosoma cruzi trans-sialidase. Annu Rev Microbiol48:499-523. 51. Previato JO, Andrade AF. Pessolani MC, Mendonca-Previato L (1985) Incorporation of sialic acid into Trypanosoma cruzi macromolecules. A proposal for a new metabolic route. Mol Biochem Parasitol 16(1):85-96. 52. Zingales, B, Carniol, C, Lederkremer, RM, and Colli, W (1987) Direct sialic acid transfer from a protein donor to glycolipids of trypomastigote forms of Trypanosomu cruzi. Mol Biochem Parasitol26, 135- 144. 53. Schenkman, S, Ferguson, MAJ, Heise, N, Cardoso de Almeida, ML, Mortara, RA, and Yoshida, N (1993) Mucin-like glycoproteins linked to the membrane by glycosylphosphatidylinositol anchor are the major acceptors of sialic acid in a reaction catalyzed by transsialidase in metacyclic forms of Trypanosoma cruzi. Mol Biochem Parasitol59, 293-304. 54. Cross GA, Takle GB (1993) The surface trans-sialidase family of Trypanosoma cruzi. Annu Rev Microbial 47:385-411. 55. Vandekerckhovc F, Schenkman S, Pontes de Carvalho L, Tomlinson S, Kiso M, Yoshida M, Hasegawa A, Nussenzweig V (1992) Substrate specificity of the Trypanosoma cruzi transsialidase. Glycobiology 2541-8. 56. Scudder P, Doom JP, Chuenkova M, Manger ID, Pereira ME (1993) Enzymatic characterization of P-D-galactoside alpha 2,3-trans-sialidase from Trypanosoma cruzi. J Biol Chem 268:9886-91. 57. Egima, CM, Briones, MRS, Freitas-Junior, LHG, Schenkman, RPF, Uemura, H, and Schenkman, S (1996) Organization of trans-sialidase genes in Trypanosoma cruzi. Mol Biodiem Parasitol77, 115-125. 58. Chuenkova M, and Pereira ME (1995) Trypanosomu cruzi trans-sialidase: enhancement of virulence in a murine model of Chagas’ disease. J Exp Med 181:1693-703. 59. Weston D, Patel B, Van Voorhis WC (1999) Virulence in Trypanosoma cruzi infection correlates with the expression of a distinct family of sialidase superfamily genes. A401 Biochem Parasitol98: 105- 16. 60. Costa F, Frdnchin G, Pereira-Chioccola VL, Ribeirao M, Schenkman S, Rodrigues MM (1998) Immunization with a plasmid DNA containing the gene of trans-sialidase reduces Trypanosoma cruzi infection in mice. Vaccine 16:768-74. 61. Serrano AA, Schenkman S, Yoshida N, Mehlert A, Richardson JM, Ferguson MA (1995) The lipid structure of the glycosylphosphatidylinositol-anchoredmucin-like sialic acid accep-

Rejerences

889

tors of Trypanosoma crusi changes during parasite differentiation from epimastigotes to infective metacyclic trypomastigote forms. J Biol Chem 270(45):27244-53. 62. Previato JO, Jones C, Goncalves LP, Wait R, Travassos LR, Mendonca-Previato L (1994) 0glycosidically linked N-acetylglucosamine-bound oligosaccharides from glycoproteins of T r y panosoma cruzi. Biochem J 301: 15 1-9. 63. Previato JO, Jones C, Xavier MT, Wail R, Travassos LR, Parodi AJ, Mendonca-Previato L (1 995) Structural characterization of the major glycosylphosphatidylinositol membraneanchored glycoprotein from epimastigote forms of Trypunomna cruzi Y-strain. J B i d Chenz 270:7241-50. 64. Ortega-Barria E, Pereira ME (1991) A novel T. cruzi heparin-binding protein promotes fibroblast adhesion and penetration of engineered bacteria and trypanosomes into mammalian cells. Cell 67:411-21. 65. Herrera EM, Ming M, Ortega-Barria E, Pereira ME (1994) Mediation of Trypanosoma cruzi invasion by heparan sulfate receptors on host cells and penetrin counter-receptors on the trypanosomes. Mol Biochem Purcisitol65:73-83. 66. Kahn SJ, Wleklinski M , Ezekowitz RA, Coder D, Aruffo A, Farr A (1996) The major surface glycoprotein of Tr~punosomacruzi aniastigotes are ligands of the human serum mannosebinding protein. Infect Inirtzun 64:2649-56. 67. Parodi AJ (1993) N-glycosylation in trypanosomatid protozoa. Glvcobioloyy 3:193-9. 68. Labriola C, Cazzulo JJ, Parodi AJ (1995) Retention of glucose units added by the UDPGLC:glycoprotein glucosyltransferase delays exit of glycoproteins from the endoplasmic reticulum. J Cell Biol 130:771-9. 69. Labriola C, Cazzulo JJ, Parodi AJ (1999) Trypanosoma cruzi calreticulin is a lectin that binds monoglucosylated oligosaccharides but not protein moieties of glycoproteins. Mol Biol Cell lO(5):1381-94. 70. Lippert DN, Dwyer DW, Li F, Olafson RW (1999) Phosphoglycosylation of a secreted acid phosphatase from Leishmania donoocmi. Glycohiology 9(6):627-36. 7 I . King and Turco (1988). 72. Mengeling BJ, Beverley SM, Turco SJ (1997) Designing glycoconjugate biosynthesis for an insidious intent: phosphoglycan assembly in Leishmania parasites. Glycohiology 75373-880. 73. Descoteaux A, Mengeling BJ, Beverley SM, Turco SJ (1998) Leishmania donovani has distinct mannosylphosphoryltransferases for the initiation and elongation phases of lipophosphoglycan repeating unit biosynthesis. Mol Biochem Purasitol94:27-40. 74. Ma D, Russell DG, Beverley SM, Turco SJ (1997) Golgi GDP-mannose uptake requires Leishmania LPG2. A member of a eukaryotic family of putative nucleotide-sugar transporters. J Biol Chem 272(6):3799--805. 15. Ryan et a1 1993. 76. Huang C , Turco SJ (1993) Defective galactofuranose addition in lipophosphoglycan biosynthesis in a mutant of Leislztnania donotnni. J Biol Chem 268:24060-6. 77. Chakraborty R, Chakraborty P, Basu MK (1998) Macrophage mannosyl fucosyl receptor: its role in invasion of virulent and avirulent L. donovani promastigotes. Biosci Rep 18(3):12942. 78. Rasmusson BJ, Flanagan TD. Turco SJ, Epand RM, Petersen NO (1998) Fusion of Sendai virus and individual host cells and inhibition of fusion by lipophosphoglycan measured with image correlation spectroscopy. Biochim Biophys Acta 1404:338-52. 79. Martin I , Turco SJ, Epand RM, Ruysschaert JM (1998) Lipophosphoglycan of Leishmania donovani inhibits lipid vesicle fusion induced by the N-terminal extremity of viral fusogenic simian immunodeficiency virus protein. Eur J Biochem 258:150-6. 80. Razinkov V, Martin I, Turco SJ, Cohen FS, Ruysschaert JM, Epand RM (1999) Relationship of membrane sidedness to the effects of the lipophosphoglycan of Leishmania donooani on the fusion of influenza virus. Eur J Biochem 262:890-9. 81. Lo SK, Bovis L, Matura R , Zhu B, He S, Lum H, Turco SJ, Ho JL (1998) Leishmania lipophosphoglycan reduces monocyte transendothelial migration: modulation of cell adhesion molecules, intercellular junctional proteins, and chemoattractants. J Immunol 160:185765. 82. Hatzigeorgiou DE, Geng J, Zhu B, Zhang Y, Liu K, Rom WN, Fenton MJ, Turco SJ, Ho JL

890

53 Glycobiology of Protozoan and Helminthic Parasites

(1996) Lipophosphoglycan from Leishmania suppresses agonist-induced interleukin 1 beta gene expression in human monocytes via a unique promoter sequence. Proc Nut1 Acad Sci USA 93: 14708-1 3. 83. Bernier, R, Barbeau, B, Tremblay, MJ and Olivier, M (1998) The iipophosphoglycan of Leishmania donovani up-regulates HIV-1 transcription in T cells through the nuclear factorkappaB elements. J Immunol160:2881-2888. 84. WOlddy D, Akuffo H, Demissie A, Britton S (1999) Role of Leishmania donouani and Its Lipophosphoglycan in CD4(+) T-cell Activation-Induced Human Immunodeficiency Virus Replication. Infect Immun 67( 10):5258-5264. 85. Peters C, Kawakami M, Kaul M, Ilg T, Overath P, Aebischer T (1997) Secreted proteophosphoglycan of Leishmania mexicana amastigotes activates complement by triggering the mannan binding lectin pathway. Eur J Immunol27( 10):2666-72. 86. Giorgione JR, Turco SJ, Epand RM (1996) Transbilayer inhibition of protein kinase C by the lipophosphoglycan from Leishmania donovani. Proc Nut1 Acad Sci USA 93:11634-9. 87. De T, Roy S (1999) Infectivity and attenuation of Leishmania donovani promastigotes: association of galactosyl transferase with loss of parasite virulence. J Parasitol85( l):54-9. 88. Chatterjee M, Sharma V, Mandal C, Sundar S, Sen S (1998) Identification of antibodies directed against 0-acetylated sialic acids in visceral leishmaniasis: its diagnostic and prognostic role. Glycoconj J 15(12):1141-7. 89. Svobodova M, Bates PA, Volf P (1997) Detection of lectin activity in Leishmania promastigotes and amastigotes. Acta Trop 14;68(1):23-35. 90. Smith, D F and Rangarajan, D (1995) Cell surface components of Leishmania: identification of a novel parasite lectin? Glycobiology 5:161-6. 91. Kobiler D and Mirelman D (1980) A lectin activity in Entumoeba histolytica trophozoites. Arch Invest Med ( M e x ) 11(1 Suppl):101-8. 92. Mirelman D, Kobiler D (1981) Adhesion properties of Entamoeba histolytica. Ciba Found Symp 80:17-35. 93. Ravdin JI, Murphy CF, Salata RA, Guerrant RL, Hewlett EL (1985) N-Acetyl-D-galactosamine-inhibitable adherence lectin of Entumoeba histolytica. I. Partial purification and relation to amoebic virulence in vitro. J Inject Dis 151904-15. 94. Yi, D, Lee, RT, Longo, P, Boger, ET, Lee, YC, Petri, WA Jr and Schnaar, RL (1998) Substructural specificity and polyvalent carbohydrate recognition by the Entamoeba histolytica and rat hepatic N-acetylgalactosamine/galactose lectins. Glycobiology 8:1037-43. 95. Petri WA Jr, Smith RD, Schlesinger PH, Murphy CF, Ravdin JI (1987) Isolation of the galactose-binding lectin that mediates the in vitro adherence of Entamoeba histolytica. J Clin Invest 80:1238-44. 96. Petri, W, Jr, and Mann, BJ (1993) Molecular mechanisms of invasion by Entumoeba histolytica. Semin Cell Biol4:305-313. 97. McCoy JJ, Mann BJ, Vedvick TS, Pak Y, Heimark DB, Petri WA Jr (1993) Structural analysis of the light subunit of the Entamoeba histolytica galactose-specific adherence lectin. J Biol Chem 268:24223-3 1. 98. Cao Z , Jefferson DM, Panjwani N (1998) Role of carbohydrate-mediated adherence in cytopathogenic mechanisms of Acanthamoeba. J Biol Chem 273: 15838-45. 99. Jaison PL, Cao Z, Panjwani N (1998) Binding of Acanthamoeba to mannose-glycoproteins of corneal epithelium: effect of injury. Curr Eye Res 17:770-6. 100. Katelaris PH, Naeem A, Farthing MJ (1995) Attachment of Giardia lumblia trophozoites to a cultured human intestinal cell line. Gut 37512-8. 101. Ward HD, Keusch GT, Pereira ME (1990) Induction of a phosphomannosyl binding lectin activity in Giardia. Bioessays 12:211-5. 102. Ward HD, Lev BI, Kane AV, Keusch GT, Pereira ME (1987) Identification and characterization of taglin, a mannose 6-phosphate binding, trypsin-activated Iectin from Giardia lamblia. Biochemistry 26:8669-75. 103. Sreenivas K, Ganguly NK, Ghosh S, Sehgal R, Mahajan RC (1995) Identification of a 148kDa surface lectin from Giardia lamblia with specificity for alpha-methyl-D-mannoside. FEMS Microbiol Lett 134(1):33-7. 104. Thea, DM, MEA Pereira, D Kotler, CR Sterling, and G T Keusch (1992) Identification and

References

891

partial purification of a lectin on the surface of the sporozoite Cryptosporidium parvum. J Parasitol78:886-893. 105. Joe A, Verdon R, Tzipori S, Keusch GT, Ward HD (1998) Attachment of Cryptosporidium parvum sporozoites to human intestinal epithelial cells. Znject Zmmun 66:3429-32. 106. Joe, A, D H Hamer: MA Kelley, MEA Pereira, GT Keusch, S Tzipori, and HD Ward (1994) Role of a Gal/GalNAc-specific sporozoite surface lectin in Cryptosporidium parvum-host cell interaction. J Eukaryot Microhiol41:44S. 107. Klein H, Loschner B, Zyto N, Portner M, Montag T (1998) Expression, purification, and biochemical characterization of a recombinant lectin of Sarcocystis inuris (Apicomplexa) cyst merozoites. Glycoconj J 15147-53. 108. Macha J, Prochazkova Z, Cerva L, Gut J (1985) Isolation and characterization of a lectin from Sarcocystis giguntea. Mol Biochem Purasitol16:243-9. 109. Ortega-Barria E, Boothroyd JC (1999) A Toxoplusma lectin-like activity specific for sulfated polysaccharides is involved in host cell infection. J Biol Chem 274: 1267-76. 110. O’Riordan DM, Standing JE, Kwon KY, Chang D, Crouch EC, Limper AH (1995) Surfactant protein D interacts with Pneumocysris carinii and mediates organism adherence to alveolar macrophages. J Clin Invest 95:2699--710. 111. Ortega-Barria E, Pereira ME (1992) Identification of a lectin activity in Pneumocystis carinii. Trop Med Parasitol43: 186-90. 112. Cummings, R D and Nyame, AK (1996)Glycobiology of schistosomiasis FASEBJ 10,838-848. 113. Cummings, R D and Nyame, AK (1999) Schistosome glycoconjugates. Biochem Biophy Acta (In Press) 114. KO, AI, Drager, UC and Harn, DA (1990) A Schistosoma mansoni epitope recognized by a protective monoclonal antibody is identical to the stage-specific embryonic antigen 1. Proc Nut1 Acad Sci U S A 87, 4159-4163. 115. Srivatsan, J, Smith, DF and Cummings, R D (1992a) The human blood fluke Schisrosoma mansoni synthesizes glycoproteins containing the Lewis x antigen. J Biol Chem 267, 2019620203. 116. Nyame, K, Smith, DF, Damian, RT and Cummings, R D (1989) Complex-type asparaginelinked oligsoaccharides in glycoproteins synthesized by Schistosoma mun.soni adult males contain terminal P-linked N-acetylgalactosamine. J Biol Chem 264:3235-3243. 117. Srivatsan, J, Smith, D F and Cummings, R D (1992b) Schistosoma munsoni synthesizes novel biantennary Asn-linked oligosaccharides containing terminal p-linked N-acetylgalactosamine. Glycobiology 21445-52. 118. Khoo KH, Chatterjee D, Caulfield JP, Morris HR, Dell A (1997) Structural characterization of glycophingolipids from the eggs of Schistosoma munsoni and Schistosoma japonicum. Glycohiology 7:653-661. 119. Khoo KH, Nieto A, Morris HR, Dell A (1997) Structural characterization of the N-glycans from Echinococcus granulosus hydatid cyst membrane and protoscoleces. Mol Biochem Parasitol86(2):237-48. 120. Nyame, K, Cummings, R D and Damian, RT (1987) Schistosoma niansoni synthesizes glycoproteins containing terminal 0-linked N-acetylglucosamine residues. J Biol Chew 262, 79907995. 121. Dell, A, Haslam, SM Morris, HR and Khoo, K-H (1999) Immunogenic glycoconjugates implicated in parasitic nematode diseases. Biochem Biophys Acta (In Press) 122. DeBose-Boyd, R , Nyame, AK and Cummings, RD (1996) Schistosoma mansoni: characterization of an a l - 3 fucosyltransferase in adult parasites. Exp Parasitol82: 1-10, 123. Hokke, CH, Neeleman, AP, Koeleman, CA, and van den Eijnden, DH (1998) Identification of an a3-fucosyltransferase and a novel a2-fucosyltransferase activity in cercariae of the schistosome Trichobilharzia ocellatu: biosynthesis of thc Fucal-2Fucal-3[Gal(NAc)~l-4] GlcNAc sequence. Glycobiology 8:393-406. 124. Marques, ET, Jr, Weiss, JB and Strand, M (1998) Molecular characterization of a fucosyltransferase encoded by Schistosomu munsoni. Mol Biochem Purasitol93:237-50. 125. Srivatsan, J, Smith, DF, and Cummings, R D (1994) Demonstration of a Novel UDPGalNAc: GlcNAcpl-4N-Acetylgalactosaminlytransferasein Extracts of Schisto.roma mansoni. J Purasitol80, 884-890.

892

53 Glycobiology of Protozoan und Helminthic Parasites

126. Neeleman, AP, van der Knaap, WP and van den Eijnden, DH (1994) Identification and characterization of a UDP-Ga1NAc:GlcNAc B-Rpl+4-N-acetylgalactosaminyltransferase from cercariac of the schistosome Trichohilhurzia ocellutu. Catalysis of a key step in the synthesis of N,N’-diacetyllactosediamino (1acdiNAc)-typeglycans. Glycobiology 4:641-5 I . 127. Xu, X, Stack, RJ, Rao, N and Caulfield, JP (1994) Schistosomu mansoni: fractionation and characterization of the glycocalyx and glycogen-like material from cercariae. Exp Parasitol 79:399-409. 128. Khoo, KH, Sarda, S,Xu, X, Caulfield, JP, McNeil, MR, Homans, SW, Morris, HR and Dell, A (1995) A unique multifucosylated -3GalNAc~1-4GlcNAc!31-3Galalmotif constitutes the repeating unit of the complex 0-glycans derived from the cercarial glycocalyx of Schistosoma munsoni. J Biol Chem 270, 17114-17123. 129. Khoo, KH, Chatterjee, D, Caulfield, JP, Morris, HR and Dell, A (1997) Structural mapping of the glycans from the egg glycoproteins of Schistosoniu munsoni and Schistosoma japonicum: identification of novel core structures and terminal sequences. Glycobiology 7:663-77. 130. Bergwerff, AA, Van Dam, GJ, Rotmans, JP, Deelder, AM, Kamerling, JP and Vliegenthart, JFG (1994) The immunologically reactive part of immunopurified circulating anodic antigen from Schistosoma mansoni is a threonine-linked polysaccharide consisting of -6)-(b-D-GlcpA(1-3))-b-D-GalpNAc-(1- repeating units. J Biol Chem 269, 3 1 5 10-3 15 17. 131. Halkes, KM, Vermeer, HJ, Slaghek, TM, van Hooft, PA, Loof, A, Kamerling, JP and Vliegenthart, JF (1998) Preparation of spacer-containing di-, ti+, and tetrasaccharide fragments of the circulating anodic antigen of Schistosoma monsoni for diagnostic purposes. Carhohydr Res 309: 175-88. 132. Nyame, AK, Debose-Boyd, R, Long, TD, Tsang, VC and Cummings, RD (1998) Expression of Le” antigen in Schistosornn,juponicumand Shaemutobium and immune responses to LeXin infected animals: lack of Le” expression in other trematodes and nematodes. Glycobioloqy 81615-24. 133. Van Dam, GJ, Bergwerff, A, Thomas-Oates, JE, Rotmans, JP, Kamerling, JP, Vliegenthart, J F and Deelder, AM (1994) The immunologically reactive 0-linked polysaccharide chains derived from circulating cathodic antigen isolated from the human blood fluke Schistosoma nzansoni have Lewis x as repeating unit. Eur J Biochem 225467-82. 134. Nyame, AK, Pilcher, JB, Tsang, VC and Cummings, R D (1 996) Schistosonm munsoni infection in humans and primates induces cytolytic antibodies to surface LeX determinants on myeloid cells. Exp Parusitol82: 191-200. 135. Nyame, AK, Pilcher, JB, Tsang, VC and Cummings, RD (1997) Rodents infected with Schistosomu munsoni produce cytolytic IgG and IgM antibodies to the Lewis x antigen. GI-vcobidoqy 7:207-15. 136. van Dam, GJ, Claas, FH, Yazdanbakhsh, M, Kruize, YC, van Keulen, AC, Ferreird, ST, Rotmans, JP and Deelder, AM (1996) Schistosorna rnansoni excretory circulating cathodic antigen shares Lewis-x epitopes with a human granulocyte surface antigen and evokes host antibodies mediating complement-dependent lysis of granulocytes. Blood 88:4246-5 1. 137. Smithers, SR and Terry, RJ (1969) Immunity in schistosomiasis. Ann N Y Acad Sci 160:826840. 138. Velupillai, P and Harn, DA (1994) Oligosaccharide-specific induction in interleukin 10 production by B22f cells from schistosome-infected mice: A mechanism for regulation of CD4+ T-cell subsets. Proc Nut1 Acud Sci U S A 91, 18-22. 139. Velupillai, P, Secor, WE, Horauf, AM and Harn, DA (1997) B-1 cell (CD5+B220+) outgrowth in murine schistosomiasis is genetically restricted and is largely due to activation by polylactosamine sugars. J Immunol 158:338-44. 140. Jacobs W, Deelder A, Van Marck E (1999) Schistosomal granuloma modulation. 11. Specific immunogenic carbohydrates can modulate schistosome-egg-antigen-induced hepatic granuloma formation. Parasitol Res 85( 1):14-8. 141. Lejoly-Boisseau H, Appriou M, Seigneur M, Pruvost A, Tribouley-Duret J, Tribouley J (1999). Schistosoma munsoni: in vitro adhesion of parasite eggs to the vascularendothelium. Subsequent inhibition by a monoclonal antibody directed to acarbohydrate epitope. Exp Parasitol91:20-9. 142. El Ridi, R, Velupillai, P and Harn, DA (1996) Regulation of schistosome egg granuloma

References

893

formation: host-soluble L-selectin enters tissue-trapped eggs and binds to carbohydrate antigens on surface membranes of miracidia. Infect Irnrnun 64:4700-5. 143. Trottein, F, Nutten, S, Papin, JP, Leportier, C, Poulain-Godefroy, 0, Capron, A and Capron, M (1997) Role of adhesion molecules of the selectin-carbohydrate families in antibodydependent cell-mediated cytoxicity to schistosome targets. J Irnmunol159:804-11. 144. Nutten S, Papin JP, Woerly G, Dunne DW, MacGregor J, Trottein F, Capron M (1999) Selectin and Lewis(x) are required as co-receptors in antibody-dependent cell-mediated cytotoxicity of human eosinophils to Schistosorna rnansoni schistosomula. Eur J Irnmunol29:799808. 145. Nyame, AK, Leppanen, AM, Debose-Boyd, R and Cummings. R D (1999). Mice infected with Schistosorna rnansoni generate antibodies to LacdiNAc (GalNAcPI 44GlcNAc) determinants. Glycobiology 9:1029-1035. 146. Nyame, K and Cummings, RD; manuscript in preparation. 147. Ward, H D (1997) Glycobiology of parasites: role of carbohydrate-binding proteins and their ligands in the host-parasite interaction. In Glycosciences: Stutus and Perspecticrs (HJ Gabius and S Gabius, Eds), pp. 399-413, Chapman and Hall, Weinheim, Germany. 148. Newton, SE, Monti, JR, Greenhalgh, CJ. Ashman, K and Meeusen, EN (1997) cDNA cloning of galectins from third stage larvae of the parasitic nematode Teladorsagia circurncincta. hilo1 Biochern Parasitol86:143-53. 149. Greenhalgh CJ, Beckham SA, Newton SE (1999) Galectins from sheep gastrointestinal nematode parasites are highly conserved. Mol Biochern Purusitol98(2):285-9. 150. Loukas A, Mullin NP, Tetteh KK, Moens L, Maizels RM (1999) A novel C-type lectin secreted by a tissue-dwelling parasitic nematode. Curr Biol 12;9(15):825-828. 151. Tetteh KK, Loukas A: Tripp C. Maizels RM (1999) Identification of abundantly expressed novel and conserved genes from the infective larval stage of Toxocaru canis by an expressed sequence tag strategy. Infect Irnrnun 67(9):4771-9. 152. Khoo KH, Maizels RM, Page AP, Taylor GW, Rendell NB, Dell A (1991) Charactcrization of nematode glycoproteins: the major 0-glycans of Toxocara excretory-secretory antigens are 0-methylated trisaccharides. Glycobiology I: 163-7 1. 153. Reason AJ, Ellis LA, Appleton JA, Wisnewski N , Grieve RB, McNeil M, Wassom DL, Morris HR, Dell A (1994) Novel tyvelose-containing tri- and tetra-antennary N-glycans in the immunodominant antigens of the intracellular parasite Trichinella spiralis. Glycobioloqy 4593-603. 154. Ellis LA, McVay CS, Probert MA, Zhang J, Bundle DR, Appleton JA (1997) Terminal 0linked tyvelose creates unique epitopes in Trichinella spiralis glycan antigens. Glycobioloqy 7:383-90. 155. Ellis LA, Reason AJ, Morris HR, Dell A, Iglesias R, Ubeira FM, Appleton JA (1994) Glycans as targets for monoclonal antibodies that protect rats against Trichinella spirulis. Glycobiologj) 4:585-92. 156. Peters PJ, Gagliardo LF, Sabin EA, Betchen AB, Ghosh K, Oblak JB, Appleton JA (1999) Dominance of immunoglobulin G2c in the antiphosphorylcholine response of rats infected with Trickinella spiralis. Infiw Irnnuiti 67:466 1-7. 157. Haslam, SM, Coles, GC, Munn, EA, Smith, TS. Smith, HF, Morris, HR and Dell, A (1996) Huernonchus contortus glycoproteins contain N-linked oligosaccharides with novcl highly fucosylated core structures. J Biol Clicwi 271:30561-70. 158. Haslam, SM, Coles, GC, Reason, AJ, Morris, HR and Dell, A (1998) The novel core fucosylation of Huemonckus contorfirs N-glycans is stage specific. Mo/ Biochun Purusirol 93:143-7. 159. DeBose-Boyd RA, Nyame A K , Jasmer DP, Cummings R D (1998) The ruminant parasite Huemonchus contortus expresses an u I ,3-fucosyltransferase capable of synthesizing the Lewis x and sialyl Lewis x antigens. Glycoconj J 15:789-98. 160. Kang et al 1993. 161. Jasmer, DP, Perryman, LE, Conder, GA, Crow, S and McGuire, T (1993) Protective immunity to Huernonchus contortus induced by immunoaffinity isolated antigens that share a phylogenetically conserved carbohydrate gut surface epitope. J Immunol 151:5450-60. 162. Haslam SM, Houston KM, Harnett W, Reason AJ, Morris HR, Dell A (1999) Structural

894

53 Glycobiology of’ Protozoan and Helminthic Parasites

studies of N-glycans of filarial parasites. Conservation of phosphorylcholine-substitutedglycans among species and discovery of novel chito-oligomers. J Biol Chem 274(30):20953-60. 163. Haslam SM, Khoo KH, Houston KM, Harnett W, Morris HR, Dell A (1997) Characterisation of the phosphorylcholine-containingN-linked oligosaccharides in the excretory-secretory 62 kDa glycoprotein of Acunthocheilonvma viteae. Mol Biochem Parasitol85( 1):53-66. 164. MacDonald M, Copeman DB, Harnett W (1996) Do excretory-secretory products of Onchocerca gibsoni contain phosphorylcholine attached to 0-type glycans? Int J Parusitol 26(10):1075-80. 165. Sugita M, Mizunoma T, Aoki K, Dulaney JT, Inagaki F, Suzuki M, Suzuki A, Ichikawa S, Kushida K, Ohta S, Kurimoto A (1996) Structural characterization of a novel glycoinositolphospholipid from the parasitic nematode, Ascaris suum. Biochim Biophys Acta 1302(3):18592. 166. Lochnit G, Nispel S, Dennis RD, Geyer R (1998) Structural analysis and immunohistochemical localization of two acidic glycosphingolipids from the porcine, parasitic nematode, Ascaris suum. Glycobiology 8:891-9. 167. Lochnit G, Dennis RD, Zahringer U, Geyer R (1997) Structural analysis of neutral glycosphingolipids from Ascaris mum adults (Nemat0da:Ascaridida). Glycoconj J 14:389-99. 168. Lochnit G, Dennis RD, Ulmer AJ, Geyer R (1998) Structural elucidation and monokineinducing activity of two biologically active zwitterionic glycosphingolipids derived from the porcine parasitic nematode Ascaris suum. J Biol Chem 273:466-74.

Carbohydrates in Chemistry and Biology Beat Emst, Gerald W. Hart, Pierre Sinay copyright Q WILEY-VCH Verlag GmbH. 2000

54 The Involvement of the Oligosaccharide Chains of Glycoproteins in Gamete Interactions at Fertilization Noritaka Hirohashi and William J. Lennarz

54.1 Introduction Fertilization is the result of a series of temporally and specially programmed interactions between molecules on the surfaces of the egg and the sperm. Although it has long been proposed that glycoconjugates play a major role in these molecular interactions, only recently has direct evidence been provided for their role in gamete recognition and binding. A diversity of systems, for example the fresh water clam [ 11, have been studied. However, coverage in this Chapter will be limited to the roles of glycoconjugates in sperm-egg interactions in marine invertebrates and amphibians. The mammalian system has recently been reviewed [2-41, and will only be discussed in relation to the other systems to be described.

54.2 Advantages of Marine Invertebrates as an Experimental System The major advantages of marine invertebrates in studying sperm-egg interaction are:

1) ease in collecting gametes and the ability to conduct in vitro assays under conditions that are similar to those that occur in vivo; and 2) the ability to collect a large amount of gametes, which facilitates biochemical approaches to isolating the biologically active components. This is important because structure analysis of carbohydrates requires significant quantities of glycoconjugates. The availability of material adequate for quantitative analysis of biological activities is also very important for studying structurefunction relationships. The specific activities of the putative interacting molecules

896

54 The Involvement o j the Oligosacchavide Chains of Glycoproteins

should be determined, because in many systems multiple components appear to be involved in the binding process. For instance, in the mouse a number of different adhesion molecules on the sperm surface have been reported to be candidates for binding to the zona pellucida (ZP). In starfish, the jelly coat contains sperm activating peptides of different structures that share common structural features for biological activity. In sea urchin egg jelly, different sulfation patterns on the same polysaccharide backbone occurs in different species that result in species specific induction of the acrosome reaction. Also, in sea urchins the macromolecules both on egg and on sperm that are involved in sperm-egg binding consist of multiple domains with different adhesive characteristics. In the case of mammals, state-of-the-art structure analysis of oligosaccharides of the extracellular matrix of the egg has been applied and the structure of the N linked chains of a 55 kDa family of glycoproteins in the porcine ZP has been determined [5, 61. In addition, an 0-linked trisaccharide with the structure GlcNAcGalpl-3GalNAc has been identified in mouse ZP2 and ZP3 [7]. Numerous partial structures of the oligosaccharides of mammalian ZPs are known. However, the complete structure of an oligosaccharide that is known to serve a ligand for sperm binding has not been determined in any mammal, or in fact, in any animal. In contrast recent studies in marine invertebrates have resulted in elucidation of the detailed structures of the molecules that function in inducing the acrosome reaction in sperm. In fact, because of the advantages mentioned above, the role of glycoconjugates in gamete interactions is best understood in marine invertebrates.

54.3 Induction of the Acrosome Reaction Induction of the acrosome reaction is a key event that occurs in sea urchins, starfish, ascidians and mammals. However, it is not clear that in the amphibian Xenopus sperm undergo the same type of acrosome reaction, because thus far it has not been possible to detect a morphologically identifiable acrosome vesicle in these sperm. The acrosome reaction in echinoderms and in mammals has been well studied. During the acrosome reaction the acrosomal vesicle fuses with the plasma membrane and the acrosomal components are externalized. Several proteolytic activities thought to be involved in penetration through the extracellular matrix of the egg are detected in the acrosomal contents. In addition, exposure of egg coat binding molecules on the surface of the sperm occurs (see below).

54.3.1 Studies in Sea Urchins In sea urchins, the sperm, once spawned into sea, initiate swimming. The motile sperm, with an aid of a chemoattractant released from the egg, contact the jelly coat, a thick gelatinous matrix surrounding the egg. Upon contact with jelly coat, sperm undergo the acrosome reaction (AR), a regulated exocytic event that is

54.3 Induction of the Acrosorne Reuctioii

897

essential for subsequent steps. Acrosome-reacted sperm bind to the vitelline layer, an extracellular matrix tightly overlying the egg plasma membrane. This binding occurs via the tip of the acrosomal process at the sperm head. Presumably the sperm then penetrates the vitelline layer and then binds to and fuses with the egg plasma membrane. In the sea urchin the jelly coat is composed of a fucose sulfate polysaccharide (FSP), several glycoproteins, and peptides. It has been argued for many years whether the AR-inducing activity resides in the carbohydrate chains or in the polypeptide backbone. Evidence suggesting that the protein moiety has an ARinducing activity was obtained in an experiment that indicated that Pronase digestion of jelly coat decreased the AR-inducing activity by 50% [8, 91. Fractionation of jelly components showed that the AR-inducing activity co-migrated with the glycoprotein fraction, but not the FSP [ 101. In contrast, much earlier studies reported that FSP had AR-inducing activity; the purified FSP obtained by the treatment of p-elimination and followed by successive chromatographic steps exhibited ARinducing activity [ 11, 121. FSP isolated from two species, despite their similarity in gross chemical composition, induced the AR only in homologous sperm, suggesting that structure differences in the carbohydrate linkages and/or location or degree of sulfation is crucial for a species-specific induction of AR. More recently the sulfated polysaccharides from the jelly coat of three species were isolated, and purified [13, 141. In confirmation of the early studies these purified sulfated polysaccharides were shown to exhibit AR inducing activity in a species-specific manner. The structures of these sulfated polysaccharides were characterized and it was established that they are linear homopolymers with 0-sulfation in a regular manner, but at different positions among the different species (Figure 1). Interestingly, two sulfated fucan isotypes were found in Strongylocentrotus purpuratus jelly coat; one isotype is a 1-3 linked linear fucose polysaccharide with a regular trisaccharide repeat unit defined by the pattern of 0-sulfation. Another isotype is also a 1-3 linked linear fueose polysaccharide, entirely sulfated at the 0-2 position, but with a heterogeneous sulfation pattern at the 0-4 position. The eggs of most (87%) female sea urchins had only one isotype and 13% had both. The two isotypes had virtually equal potency in inducing the AR [ 151. It will be of interest to examine the sulfated polysaccharide structures among the various echinoderms species, because it is well known that cross-reactivity of AR-inducing activity by heterologous jelly coat can occur, depending on the particular species tested. Further comparison should better define the relationship between polysaccharide structure (especially the location of sulfation) and biological activity. In parallel with efforts to identify the AR-inducer in jelly coat, studies have identified a sperm cell surface molecule that binds jelly coat. In preliminary experiments, it was concluded there must be a receptor-like protein on the surface of the acrosome-intact sea urchin sperm, because trypsin treatment of intact sperm blocked induction of the AR by jelly coat. Subsequently the receptor was isolated and shown to be a 210 kDa glycoprotein with the ability to bind jelly coat [16]. A monoclonal antibody to the 210 kDa glycoprotein was shown to induce the AR. Immuno-localization studies showed the 2 I0 kDa glycoprotein was present on the surface of the sperm flagellum, as well as on a thin belt of the membrane surface

898

54 The Involvement of the Oligosucchuride Chains of Glycoproteins

H

R= OSO-3 or OH

o

e

o

-

E. lucunter

S. purpuratus

S. purpuratus

L. variegatus

A. lixula

L. grisea (sea cucumber)

A. amurensis (sea star)

I

OH

S. purpuratus H . pulcherrimus

Figure 1. Structures of the jelly coat polysaccharides found in a variety of echinoderm species.

54.4 Sperm-Egg Coat Binding

899

directly over the acrosome granule at the anterior apex of the sperm head. Molecular cloning revealed that the receptor of egg jelly (REJ) is modular in design, containing one EGF and two C-type lectin carbohydrate-recognition modules. In addition, it contains termed a novel domain (the so-called REJ module) that exhibits extensive homology with human polycystic kidney disease protein [17]. It is not clear how this homology relates to the functions of these two proteins. 54.3.2 Studies in Starfish

Unlike the sea urchin, in starfish induction of the AR requires three major components of jelly coat: ARIS (acrosome reaction-inducing substance), a steroid saponin called Co-ARIS (cofactor for ARIS), and Asterosap (asterodical sperm activating peptide). Three Co-ARIS species consisting of a sulfated steroid and a pentasaccharide chain have been structurally and biologically characterized. Asterosap has structural diversity, but in all instances the Astersap has been shown to consist of glutamine-rich tetratriacontapeptides with an intramolecular disulfide linkage [ 181. All of the Astersap species stimulate sperm respiration and motility, and increase the intracellular pH of the sperm. The intramolecular disulphide linkage is essential for their activity. ARIS is a sulfated fucose-rich glycoprotein with an apparent molecular weight of over lo4 kDa. Although the AR-inducing activity of ARIS was retained after Pronase treatment, it was abolished by periodate oxidation or solvolytic desulfation, suggesting that biological activity depends on the saccharide and sulfate moieties. ARIS treated with Pronase, followed by sonication and fractionation, yielded a saccharide fragment whose structure was determined to be [4-p-~-Xylp143-a-D-Galp-1+3-a-l-Fucp-4(OS03)- 143-a-L-Fucp-4 (Oso3)1+4-a-~-Fucp-l],,.A polymer consisting of 10-1 1 repeating units of this pentasaccharide contained the AR-inducing activity. Removal of 1 mole of sulfate per repeating unit destroyed the AR-inducing activity, suggesting the importance of the sulfate group for activity [ 191. To estimate the minimum size required for its biological activity ARIS was irradiated with high-energy electrons. It was estimated to be 12.7-15.4 kDa (mean 14 kDa). which is good agreement with the molecular mass of the carbohydrate fragment (about 10.4 kDa) [20].

54.4 Sperm-Egg Coat Binding Following the acrosome reaction that sperm encounters a species specific coat. This coat is called the vitelline layer (VL) in sea urchins, the zona pellucida (ZP) in mammals, the vitelline envelope (VE) in frogs and the vitelline coat (VC) in ascidians. In these systems the general conclusion is that initial sperm-egg coat binding is mediated by the interaction between the oligosaccharide chains on an egg coat protein and a carbohydrate binding protein on the sperm. In some cases it is known that the sperm protein interacts with the non-reducing end of oligosaccharide chain.

900

54 The Involvement of the Oligosucchuride Chains of Glycoproteins

Although multiple approaches have been developed to identify the specific molecules functionally involved in sperm-egg coat interaction, in not all cases have the following criteria been met in testing candidate molecules. Carbohydrate residues should be present at the site on the egg where sperm bind. Of course, in addition carbohydrate-binding proteins should be present at the proper site on the sperm surface membrane. Competitors for the carbohydrate-binding protein, such as synthetic oligosaccharides or glycosides derived from the egg coat or from other sources should inhibit sperm-egg binding. It should be established that the competitors do not affect sperm motility and induction of the acrosome reaction. Specific modifications of carbohydrate moieties by chemical reagents, or by enzymatic reactions such as glycosidases or glycosyltransferases, should result in loss of spermegg binding. Masking of carbohydrate moieties with monovalent lectins, or with Fab fragments of antibodies, should inhibit sperm-egg binding. In some cases, lectins promote sperm-egg binding, which can be accounted for their crosslinking to sugar moieties present on both gametes. In addition to these criteria, it would be expected that the biochemcally purified carbohydrate-binding protein isolated from the sperm would interact with the carbohydrate chains of the egg protein under physiological conditions. Finally, an obviously complementary approach is to utilize genetic knockouts to define the function of the molecules. Recently this genetic approach has been taken using transgenic mice that have had genes knocked out for putative carbohydrate-binding proteins in sperm, or glycosyltransferase genes responsible for assembly of the complete oligosaccharide chains of the ZP glycoproteins. 54.4.1 Studies in Mammals To date in mammals many sperm surface molecules have been proposed as candidates for ZP binding. Among these molecules, two major candidates have emerged based on several lines of evidence that basically fulfill the criteria described above. Shur and his colleagues have been presented the hypothesis that sperm plasma membrane p- 1,4-galactosyItransferase (GalTase) recognizes terminal N acetylglucosamine residues on O-linked oligosaccharides in ZP3. The fact that GalTase forms a complex with its substrate(s) in the absence of UDP-Gal provides support for the idea that a stable sperm attachment to zona pellucia is mediated by this molecule. Another hypothesis, proposed by Wassarman and his colleagues, is that a lectin-like molecule, called sp56, recognizes a-galactosyl residues of 0-linked oligosaccharides in ZP3. For more detail the reader is referred to excellent reviews focused on mammalian gamete interactions 14, 21, 221. 54.4.2 Studies in Frog Although a great deal of work has been done to identify the components of the jelly coat and the vitelline envelope of the frog (mostly in Xenopus) until recently was little or no information on components of the egg and/or sperm that mediate gamete binding. Recently, a X . laevis sperm binding glycoprotein, gp69/64, present

54.4 Sperm-Egg Coat Binding

901

as a minor component of the egg vitelline envelope was identified as a sperm binding protein [23, 241. When the isolated glycoprotein was added to sperm it blocked their binding to the VE of eggs. In addition, antibody prepared to either gp69 or gp64 blocked sperm-egg binding and fertilization. Furthermore, when gp69/64 was coupled to beads, it could be shown that the beads bound sperm. Binding did not occur if the coupled beads were treated with antibody to gp69/64 prior to addition to sperm. Further, it was shown that both isolated gp69 and gp64 were active in inhibiting sperm binding, and were glycoforms of the same protein, because after deglycosylation a single polypeptide of 55 kDa was produced. As a prelude to cloning and sequencing the gp69/64, N-terminal sequencing of gp69 and gp64 was carried out. As expected both sequences were found to be the same. It has been know for some time [25] that during fertilization, gp69/64 undergoes limited proteolysis and that thereafter sperm no longer bind to the VE. Subsequently, isolation and N-terminal sequencing of the form of this glycoprotein present in the coat of the fertilized egg revealed that it had been proteolytically truncated by 27 amino acids. Similarly, isolation of the glycoprotein after treatment of intact eggs with crude collagenase indicated a truncation of 31 amino acids. Based on the identification of these three N-terminal sequences the corresponding oligonucleotides were prepared and PCR was carried out. Ultimately this approach yielded an ORF encoding a polypeptide chain of 699 amino acid residues with a calculated mass of 77,867 Daltons. All three chemically determined peptide sequences (two of which overlapped each other) were found at the predicted positions in the deduced sequence of the clone. The calculated mass of 77 kDa is significantly larger than the apparent mass (-54 kDa) of the mature, deglycosylated gp69/64 protein, suggesting that posttranslational processing of the nascent polypeptide chain may occur. Four potential N-glycosylation sites were found in the proposed mature form of the receptor. Of course, numerous potential 0-linked glycosylation sites (Ser or Thr) also were found in the sequence. The 27 amino acid residue fragment that is cleaved upon fertilization lacks any potential N-linked glycosylation sites, but does contain potential 0-glycosylation sites. Earlier it was found that loss of the ability of gp69/64 to inhibit sperm binding did not occur after Pronase digestion of gp69/64, but that all activity was lost upon Pronase digestion followed by periodate treatment. This observation, coupled with the finding that during N-terminal sequencing of the mature protein the Ser deduced to be present at residue 6 is not detected, suggest that an 0-linked oligosaccharide functional in sperm binding may be present at the N-terminus at Ser6 (Figure 2). Both in vivo and in vitro data indicated that the N-terminus of gp69/64 is essential for its binding to sperm. Cleavage of the N-terminal 27 residues during fertilization or the 31 residues by type-I collagenase, abolishes sperm binding to the egg surface. Moreover, the isolated purified, N-terminally truncated receptor no longer inhibits sperm-egg binding in competition assays. However, as mentioned above it remains to be determined whether the epitope required for sperm binding includes polypeptide sequences, as well as the putative oligosaccharide chain. Based on the above findings, it is proposed that the function of cleavage of the Nterminus of the sperm receptor is to abolish sperm-egg interaction after egg activation (see [23], for additional evidence). Analysis of the deduced amino acid se-

902

54 The Involvement of the Oligosaccharide Chuins of Glycoproteins

Figure 2. Sequence of the N-terminal region of gp69/64. Although many potential 0-glycosylation sites were found in the N-terminus, current data suggest that Sers is a likely site for 0-glycosylation site. Upon fertilization, the bond between Asp27-Aspzg is cleaved and causes the 27 amino acid fragment to be released. The cleavage site for crude collagenase is indicated to be localized at residues Phe31-Va13z.

quence of gp69/64 revealed a domain towards the C-terminus with high sequence identifies with a similar domain in the ZP2 family of mammalian zona pellucida glycoproteins. The mouse homolog, mZP2, is also proteolytically modified during fertilization and its apparent MW is reduced from 120,000 to 90,000 (ZP2f) [28, 291. However, the molecular details of this modification are not clear. In view of the close structural similarities found between the mouse ZP2 and the Xenopus gp69/64, it is possible that the mouse ZP2 is also cleaved at a site close to its N-terminus and that this cleavage blocks mouse sperm-egg binding. Indeed, there is evidence that cleavage post-fertilization occurs at or near this same site in the pig ZP2 homolog of gp69/64 [30]. These studies have provided strong evidence for participation of the egg VE glycoprotein gp69/64 in sperm binding. Nothing is yet known about the complementary binding protein in the sperm. As noted earlier, we speculate that an 0-linked oligosaccharide chain possibly attached to Ser residue 6 is involved in the binding to sperm. If this model proves to be correct it is possible that the protein on the sperm that binds to gp69/64 is a lectin-like molecule.

54.4.3 Studies in Ascidians

In the ascidians, which are marine invertebrates, it appears that sperm utilize a glycosidase for sperm-vitelline coat binding. In Ciona intestinalis it was established that a sperm surface a-L-fucosidase interacts with fucosyl residues present in vitelline coat oligosaccharide chains and results in formation of an enzyme-substrate complex. Substrates for u-L-fucosidase competitively inhibited sperm-VC binding in a dose dependent fashion, and in an anomer/stereo specific manner. Sperm u-Lfucosidase has an optimal pH at around 4; it has very low enzyme activity at physiological conditions (pH 8-8.2), leading to the assumption that it may function by binding to a fucose-containing oligosaccharide rather than by its hydrolytic

54.4 Sperm-Egg Cout Binding

903

Figure 3. Sperm-vitelline coat (VC) binding in the ascidian. Binding of sperm was localized in the vicinity of hexagonal structures on the surface of the VC in thc ascidian, Hulocynthiu roretzi. These structures stained with several lectins, suggesting that they consist of glycosylated proteins.

activity [31]. However, it is not clear if the fertilizing sperm also can hydrolyze fucosyl residues after sperm-VC binding has occurred. A monoclonal antibody to a-L-fucosidase inhibited sperm-vitelline coat binding and an immunofluoresence study showed that w-L-fucosidase is localized at the tip of the head of the sperm. In another ascidian, Hulocynthiu roretzi, the sperm w-L-fucosidase has been implicated in sperm-VC binding based on the same types of experiments. The distribution of the sperm bound on the surface of the VC is found near the hexagonalrepeated filamentous structure that forms a network on the vitelline coat (Figure 3). This structure was found to stain with several lectins, suggesting that oligosaccharides with terminal fucosyl, N-acetylglucosaminyl and N-acetylgalactosaminyl residues are present. Both 0-linked and N-linked oligosaccharides were found in the vitelline coat glycoproteins. Competitive bioassays showed that only 0-linked oligosaccharide chains had potential sperm ligand activity. Removal of fucosyl residues from 0-linked oligosaccharides decreased their inhibitory activity by 50% in a competitive sperm-VC-binding assay and an in vitro fertilization assay. The compositional analysis of 0-linked oligosaccharides revealed that they were heavily sulfated, and their desulfation completely abolished competitive inhibitory activity.

904

54 The Involvement of the Oliyosaccharide Chains of Glycoproteins

This finding strongly implies that the sulfate groups on 0-linked oligosaccharides are a recognition determinant [32, 331. In Phallusia mummillutn, another ascidian, sperm bind to N-acetylglucosamine groups on the vitelline coat via a sperm surface N-acetylglucosaminidase [34]. Two egg N-acetylglucosaminidases also have been identified; one is GPI-anchored on the egg plasma membrane and another is present in the follicle cells, which are accessory cells on the outside of the egg vitelline coat. The latter enzyme, when released, modified N-acetylglucosamine residues on the vitelline coat and the bound sperm detached. This enzymatic modification of the sperm binding site on the vitelline coat may be an early first block to polyspermy that precedes the electrical block operating at the egg plasma membrane level and requiring sperm-egg fusion [35]. 54.4.4 Studies in Sea Urchins

In sea urchins sperm that have undergone the acrosome reaction bind to the vitelline layer of the egg via their acrosomal process. In earlier studies evidence was presented that bindin, a 30.5 kDa protein, stored in the acrosomal vesicle and externalized on the surface of the acrosomal process following the acrosome reaction, is involved in sperm-egg binding. Subsequently, studies were undertaken to identify the sperm binding macromolecule on the surface of the egg. Several groups have reported the isolation of high molecular weight glycoconjugates from the egg surface that inhibits sperm-egg binding. However, further characterization of these molecules proved difficult because of their high molecular weight (over 300 kDa) and extensive glycosylation. An alternative approach that was taken was to identify a proteolytic fragment of the putative receptor by releasing a biologically active domain of the binding protein from the egg surface. Thus, when S. purpuratus eggs were treated with lysylendoprotease C (LysC) they lost their ability to bind sperm. Furthermore a released fragment could be recovered that inhibited sperm-egg binding and fertilization in a dose-dependent, and species-specific manner [ 361. This fragment was purified and found to be a 70 kDa sulfated glycoprotein fragment. A rabbit antiserum specific for the 70 kDa glycoprotein recognized a -350 kDa glycoprotein. It was proposed that this was the putative intact receptor present in the preparation of egg plasma membrane and vitelline layer (PMVL). The anti-70 kDa glycopeptide IgG and Fab fragments were found to bind to the egg surface and inhibited sperm binding [37]. A cDNA encoding a portion of the receptor (45A) was identified by using this antibody to screen a a ZAP cDNA expression library made from polyadenylated RNA of immature ovaries. The 45A cDNA was found to contain segments of a deduced amino acid sequence that corresponded to two proteolytic fragments isolated from 70 kDa glycoprotein. The 45A cDNA clone was bacterially expressed as a glutathione S-transferase (GST) fusion protein. This protein exhibited the ability to bind sperm, suggesting that the polypeptide chain per se serve as a ligand for two process. When GST45A was coupled to polystyrene beads, interaction of GST45Abeads with sperm found to occur via the acrosomal process of the sperm [38]. In support with this observation, GST45A was shown to interact with recombinant

54.4 Sperm-Egg Coat Binding

905

bindin protein with simple bimolecular kinetics [39]. After initial sequencing of the full length of receptor several sequencing errors were corrected and the revised structure indicated that the “receptor” lacked a transmembrane domain and, therefore really is a binding protein rather than a receptor. The revised sequence encoded an ORF of 889 amino acids with high sequence similarity to a subfamily of heat-shock proteins called hspl10. The sequence identity in the N-terminal 500 amino acid residues (which contains 45A) of the receptor was found to be 63% when compared with Hsp 110 from CHO cells. In contrast, in the C-terminal half there were two blocks of polypeptide, encompassing residues 500-625 and 825-889, that had no sequence similarity to Hsp 110 1401. To examine the sperm binding domain of recombinant 45A protein, a series of truncated constructs of GST-45A were generated and tested in a sperm binding assay. It was found that a 32 amino acid segment encompassing residues 380-411 exhibited sperm binding activity. Initially, this binding was thought to be speciesspecific since L. pictus sperm did not bind to it. However, cross-species binding was observed when sperm of S. franciscanus, a species closely related to S. purpurutus, was used. Therefore the binding of sperm to the recombinant protein is more correctly termed genus specific [41]. In spite of a previous report that S. fianciscanus sperm bind more efficiently to S. purpurutus eggs than S. purparatus sperm binding to S. franciscunus eggs [42], a significant level of sperm binding to heterologous eggs with either combination of these two species could not be observed 1.561. Thus it appears that control of species specificity cannot be ascribed to the polypeptide chain alone. As mentioned earlier, the deduced sequence of the sperm binding protein encodes 889 amino acids, which corresponds to a calculated mass of approximately 100 kDa. The molecular weight of the intact binding protein isolated from a egg plasma membrane/vitelline layer preparation is estimated to be 350 kDa by SDS-PAGE under reduced conditions, which is markedly larger than the mass of the recombinant receptor protein. Presumably co- or post-translational modifications are responsible for the difference in mass of the intact binding protein. Indeed, chemical analysis showed that the 350 kDa sperm binding protein contains 70% carbohydrate by mass [43, 441, which would indicate a polypeptide chain of approximately 105 kDa. The sequence of the binding protein indicates at least 22 potential 0glycosylation sites and five potential N-glycosylation sites. In parallel with the effort to clone the binding protein, the role of carbohydrates in the receptor was investigated. Earlier studies showed that carbohydrate moieties could function as a ligand for sperm. A hypothesis on how the species-specificity is controlled is discussed below. Compositional analysis of monosaccharides of the binding protein suggested the presence of N-linked oligosaccharides in the protein [43]. No significant change in amount of N-acetylgalactosamine, galactose and N-glycolylneuramic acid were found in the preparations before and after de-Nglycosylation of the protein by PNGase, suggesting that it also had O-linked oligosaccharides. In fact, subsequent structural studies demonstrated that the binding protein has sulfated oligosialic acid chains, (S0~-)-9NeuGca2(-5-O,1,,,1,1NeuGca2-),, that are O-linked to the polypeptide chain via GalNAc 1441. To examine the biological role of carbohydrates of the binding protein, it was

906

54 The Involvement of the Oliyosaccharide Chains of Glycoproteins

subjected to hydrazinolysis and fractionated on a concanavalin A column. These fractions were shown by compositional analysis to consist of N-glycans (bound) and 0-glycans (not bound). The biological activity resided in the fraction containing 0glycans judged by a competitive fertilization assay. Further fractionation by ion exchange showed that the most heavily sulfated 0-glycans had the highest inhibitory activity in a competitive bioassay, suggesting that the sulfated 0-glycans serve as a ligand for sperm 1451. Isolated 0-glycans were covalently linked to either the recombinant binding protein or to bovine serum albumin. The resulting neoglycoproteins were coated on the beads to conduct sperm-bead-binding assay. The assays revealed that sperm bound more stably to the neoglycoproteins than to recombinant binding protein [46]. Furthermore, the number of bound sperm were increased by increasing number of 0-glycans attached to protein, suggesting that the density of the carbohydrate chains is also key factor to control sperm binding to the receptor.

54.5 Carbohydrate as a Species-Specific Determinant Since in many animals sperm interact with the carbohydrate moieties on the egg coat and sperm-egg coat binding takes place in a species-specific manner, it is assumed that carbohydrate can act as a species-specific determinant. In mammals cross-fertilization between species does not occur if the zona pellucida is present on the eggs. However when the zona pellucida is eliminated from eggs, heterologus sperm can fuse to the egg plasma membrane and produce hybrid zygotes 147, 481. This result suggests that the mechanism of heterospecific incompatibility resides in the zona pellucida. It also suggests the possibility that sperm-egg fusion is not species specific. Recombinant hamster ZP3 expressed in mouse embryonal carcinoma cells and Chinese hamster ovary cells exhibited species-specific binding of hamster sperm, but not mouse sperm [49]. Site-specific mutagenesis of the mouse zona protein, mZP3, and its expression in cells in culture has established that two glycosylation sites are essential for sperm binding activity 1501. These observations open a number of possible experimental approaches to address how species-specificity is realized. However, it remains to be determined if the glycoforms synthesized in heterologous systems are identical to those in a homologous system, since the activity and the specificity of glycosyltransferases varies in different cell types. The species-specificity of sea urchin fertilization has been studied since the late 1800s. Blocks to heterospecific fertilization may occur at two steps: induction of the acrosome reaction by jelly coat and gamete binding. Species-specificcontrol of the induction of the acrosome reaction by homologous jelly coat would be expected to prevent heterospecific sperm-vitelline layer interactions. However, species specificity at this first stage often is not observed. A earlier study showed that out of 11 heterospecific combinations of gametes from four echinoids, the acrosome reaction was induced in nine combinations [51]. In good agreement of this, current studies clearly demonstrated that isolated FSPs from S. purpuratus and S. $ranciscanus

References

907

have this ability to induce the acrosome reaction in both species [52].However, the binding of these acrosome reacted sperm to the vitelline layer was never observed in 1 1 heterospecific combinations, suggesting that in the sea urchin species-specific fertilization is primarily regulated by species-specific sperm binding to the vitelline layer. Similar to mammalian eggs, when the vitelline layer of the sea urchin eggs was removed cross-fertilization readily took place. Furthermore, this species-specificity in sperm binding was retained in aldehyde-fixed eggs, suggesting the possible involvement of carbohydrates in the species specificity of binding [53]. Experiments involving proteolysis of the sperm binding protein supports the idea of involvement of its carbohydrate chains in sperm binding. Extensive Pronase digestion of crude preparations of the S. purpuratus and the A . punctulata sperm binding proteins yielded glycopeptides that had very little protein remaining, but still exhibited inhibition activity in the sperm-egg binding assay. Moreover, in the case of S. purpuratus the purified intact binding protein or the binding protein fragment obtained by limited proteolysis exhibited species specific inhibitory activity in sperm-egg binding. However, as with the crude binding protein, the small glycopeptides generated by the treatment of the purified sperm binding protein with Pronase, although retaining inhibitory activity, lost species specificity [ 36, 541. When a tryptic peptide was prepared the longer glycopeptides produced retained species specificity, whereas the shorter glycopeptides although active, lacked species specificity [55].As discussed earlier in the case of S. purpuratus, by use of bacterially expressed recombinant sperm binding protein it became clear that the polypeptide has two sperm binding domains: a non-specific binding domain and a genus-specific binding domain [41]. These findings, combined with the fact that the sperm binding protein is highly glycosylated, raise the possibility that the carbohydrate chains attached to the appropriate sites on the protein backbone may impart the observed species specificity. Clearly a good deal is yet to be learned about the relative roles of the oligosacchdride chains and the polypeptide backbone of sperm binding proteins in gamete interactions and how species specificity is achieved. References 1. 2. 3. 4. 5. 6. 7.

8. 9. 10. 11. 12. 13. 14.

R. Focarelli and F. Rosati, (1995) Develop B i d , 1995, 171, 606-614. R. Shalgi and T. Raz (1997) Histol Histoputhol 12, 813-822. F. Sinowatz, J. Plendl and S. Kolle, Actu Anat (Basel), 1998, 161, 196-205. P. M. Wassarman, Cell, 1999, 96, 175-183. S. Noguchi, Y. Hatanaka, T. Tobita and M . Nakano, Eur J Biochem, 1992,204, 1089-1 100. S. Noguchi and M. Nakano, Eur J Biorhem, 1992,209, 883-894. S. K. Nagdas, Y. Araki, C. A. Chayko, M. C. Orgebin-Crist and D. R. Tulsiani, Biol Reprod, 1994, 51, 262-272. K. Ishihara and J. C. Dan, Dev Growth DiSfer, 1970, 12, 179-187. M. Yamaguchi, M. Kurita and N. Suzuki, net; Growth Differ, 1989, 31, 233-239. T. Shimizu, H. kinoh, M. Yamaguchi and N. Suzuki, Dev Growth Differ, 1990, 32, 473-487. G. K. SeGall and W. J. Lennarz, Der Biol, 1979, 71, 33-48. G . K. SeGall and W. J. Lennarz, Dev Biol, 1981,86, 87-93. A. P. Alves, B. Mulloy, J. A. Diniz and P. A. Mourao, J Biol Chem, 1997,272, 6965-6971. A. P. Alves, B. Mulloy, G. W. Moy, V. D. Vacquier and P. A. Mourao, Glycobiology, 1998, 8, 939-946.

908

54 The Involvement of the Oligosacchuride Chains of Glycoproteins

15. V. D. Vacquier and G. W. Moy, Dev B i d , 1997, IY2, 125-135.

16. S. B. Podell and V. D. Vacquier, J Biol Cheni, 1985,260, 2715-2718. 17. G. W. Moy, L. M. Mendoza, J. R. Schulz, W. J. Swanson, C. G. Glabe and V. D. Vacquier, J Cell B i d , 1996, 133, 809-817. 18. T. Nishigaki, K. Chiba, W. Miki and M. Hoshi, Zygote, 1996, 4, 237-245. 19. S. Koyota, K. M. Wimalasiri and M. Hoshi, J Biol Chem, 1997,272, 10372-10376. 20. A. Ushiyama, K. Chiba, A. Shima and M. Hoshi, Zygote, 1995, 3, 351-355. 21. B. D. Shur, S. Evans and Q. Lu, Glycoconj J , 1998, 15, 537-548. 22. B. D. Shur, Biorhem Biophys Res Commuiz, 1998,250, 537-543. 23. J. Tian, H. Gong, G. H. Thomsen and W. J. Lennarz, Dev Biol, 1997, 187, 143-153. 24. J. Tian, H. Gong, G. H. Thomsen and W. J. Lennarz, J Cell B i d , 1997, 136, 1099-1 108. 25. G. L. Gerton and J. L. Hedrick: Dev B i d , 1986, 116: 1-7. 26. D. J. McGeoch, Virus Res, 1985, 3, 271-286. 21. G. von Heijne, Nucleic Acids Res, 1986, 14, 4683-4690. 28. J. D. Bleil, C. F. Beall and P. M. Wassarman, Dez: Biol, 1981, 86, 189-197. 29. C. C. Moller and P. M. Wassarman, Deu B i d , 1989, 132, 103-112. 30. A. Hasegawa, K. Koyama, Y. Okazaki, M. Sugimoto and S. Isojima, J Reprod Fertil, 1994, 100, 245-255. 31. M. Hoshi, Ado Exp Med Biol, 1986,207, 251-260. 32. M. Hoshi, S. Takizawa and N. Hirohashi, Semin Develop Biol, 1994, 5, 201-208. 33. T. Baginski, N. Hirohashi and M. Hoshi, Develop Growth Differ, 1999, 41, 357-364. 34. A. Godknecht and T. G. Honegger, Dev Biol, 1991, 143, 398-407. 35. C. Lambert, H. Goudeau, C. Franchet, G. Lambert and M. Goudeau, Mol Reprod Dev, 1997, 48, 137-143. 36. K. R. Foltz and W. J. Lennarz, J Cell Biol, 1990, 111, 2951-2959. 37. K. R. Foltz and W. J. Lennarz, J Cell B i d , 1992, 116, 647-658. 38. K. R. Foltz, J. S. Partin and W. J. Lennarz, Science, 1993, 259, 1421-1425. 39. R. A. Cameron, T. S. Walkup. K. Rood, J. G. Moore and E. H. Davidson, Dev Biol, 1996, 180, 348-352. 40. M. L. Just and W. J. Lennarz, Dev B i d , 1997, 184, 25-30. 41. R. L. Stears and W. J. Lennarz, Dev Bid, 1997, 187, 200-208. 42. A. Lopez, S. J. Miraglia and C. G. Glabe, Dev B i d , 1993; 156,24433. 43. K. Ohlendieck, S. T. Dhume, J. S. Partin and W. J. Lennarz, J Cell B i d , 1993, 122, 887-895. 44. S. Kitazume-Kawaguchi, S. Inoue, Y. Inoue and W. J. Lennarz, Proc Nut1 Acad Sri USA, 1997, 94, 3650-3655. 45. S. T. Dhume and W. J. Lennarz, Glycobidogy, 1995, 5, 11-17. 46. S. T. Dhume, R. L. Stears and W. J. Lennarz, Glycohiology, 1996, 6, 59-64. 47. R. Yanagimachi, In: The Physiology of’ Reproduction, ed. E. Knobil & J. D. Neil/, New York: Raven Press, 1994, 189-3 17. 48. P. M. Wassarman and E. S. Litscher, Curr Top Dev Biol, 1995, 30, 1-19. 49. E. S. Litscher and P. M. Wassarman, Zygote, 1996, 4, 229-236. 50. J. Chen, E. S. Litscher and P. M. Wassarman, Proc Natl Acud Sci USA, 1998, 95, 6193-6197. 51. R. G. Summers and B. L. Hylander, Exp Cell Res, 1975, 96, 63-68. 52. A. C. Vilela-Silva, A. P. Alves, A. P. Valente, V. D. Vacquier and P. A. Mourao, Glycobiology, 1999, 9, 927-933. 53. K. H. Kato and M. Sugiyama, Dev Growth Differ, 1978,20, 337-347. 54. D. P. Rossignol, B. J. Earles, G. L. Decker and W. J. Lennarz, Dev Biol, 1984, 104, 308-321. 55. N. Ruiz-Bravo and W. J. Lennarz, Dev Biol, 1986, 118, 202-208. 56. N. Hirohashi, unpublished observations.

Carbohydrates in Chemistry and Biology Beat Emst, Gerald W. Hart, Pierre Sinay copyright Q WILEY-VCH Verlag GmbH. 2000

55 Glycosylation and Development MichPle Aubery and Christian Deruppe

55.1 Summary Carbohydrates and carbohydrate-binding proteins, or lectins, are present on the cell surface and within the extracellular matrix, and their expressions are controlled both spatially and temporally during embryo development [ 1-31. The cross-talk between glycosylated molecules and lectins is important for harmonious embryo development, not only in cellular recognition phenomenon but also as an intermediate in the transduction of various signals depending on the developmental stage. The contact between cells and their immediate environment is primarily mediated by cell-adhesion molecules which influence the earliest developmental stages, such as preimplantation and compaction at the morula stage, as well as the processes of gastrulation, neurulation, and subsequent organogenesis. Herein, we described the developmental changes of the carbohydrate structures borne by glycoproteins, especially certain cell-surface glycoproteins implicated in cellular recognition processes that might be involved with regulation of the latter at specific stages of embryogenesis, together with the modified expressions and activities of glycosyltransferases. The developmental alterations of endogenous lectins are also considered as are their subsequent consequences in embryo development.

1) Lectins as tools to analyze changes of cell-surface glycoconjugates during embryo development. Different plant lectins and neolectins have been used as markers of glycosides which are specifically expressed at successive stages of embryogenesis in different species, ranging from chicken to man. Table 1 gives the most commonly used lectins with their usual abbreviations and their specific carbohydrate ligands. 2) Cell-adhesion molecules: glycosylation changes during embryo development. The functional specificity of neural cell adhesion molecule (NCAM) can be modu-

910

55 Glycosylution and Development

Table 1. Plant lectins commonly used for structural investigations on cell-surface glycoconjugates. Lectin source

Lectin abbreviation

Ligand

Canavalia ensifarmis Vicia villosa Triticum vulgaris Helix pomatia Clycinc max Arachis hypogaea Ricinus communis I Dolichos bifiorus Sambucus niqru Ulex europaeus I Limulus polyphenzus

Con A VVA WGA HPA SBA PNA RCA I DBA SNA UEA I LPA

u / P-Mannose

~

P-Galactose P-N-Acetylglucosamine a-N-Acetylgalactosamine a-N-Acetylgalactosamine P-Galactose P-Galactose a-N-Acetylgalactosamine Sialic acid a-Fucose Sialic acid

lated either by alternative splicing or by posttranslational modifications, such as modified polysialylation of carbohydrate chains. These changes occur during rat and human embryonic development. Glycosylation of another cell adhesion molecule, L1, was also altered during nervous system development. 3) Glycosyltransferase changes during embryo development and their consequences. The expressions of several glycosyltransferases, polysialyltransferases, P-1,4-galactosyltransferaseand p-N-acetylglucosaminyltransferase IV during embryonic development have been studied, and relationships between polysialyltransferase activities and sialylation of NCAM, and p-galactosyltransferase activities during the mouse brain development were observed. Different activities of N-acetylglucosaminyltransferase IV have also been noted at two stages of chick embryo development. 4) Altered expression of endogenous lectins during embryo development. Two main families of lectins, galectins and selectins, are involved. Altered expressions of galectin 1 and galectin 3 are associated with the development of the olfactory system and the kidney during rat embryogenesis. Mice carrying a null mutation in the galectin-1 or the galectin-3 gene were produced to analyze the roles of these galectins during embryogenesis; however, redundant properties and multiplicity of these galectins (three or more galectins can be present in the same tissue) limit the usefulness of such analyses. Various changes of the expression of E-, L- or P-selectins and also of another endogenous lectin (R1 from rat cerebellum) have also been observed during embryo development. Taken together, these results suggest that carbohydrates are involved in various embryogenetic processes, probably by modulating of several enzyme activities and/ or signalling pathways. Nevertheless, their exact biological roles remain to be demonstrated and further investigations are needed. This review cites 72 references, 30 among the 44 published during the last five years appeared between 1996 and 1998.

55.3 Lectins as Tools to Analyze Clzanyes in Cell-surjucr Gl.vcoconjugates

9 11

55.2 Introduction Carbohydrates and carbohydrate-binding proteins, or lectins, are present on the cell surface and within the extracellular matrix, and their expressions are controlled both spatially and temporally during embryo development [ 1-31. The cross-talk between glycosylated molecules and lectins is important for the harmonious embryo development, not only in cellular recognition phenomena but also as an intermediate in transduction of various signals depending on the developmental stage. Contact between cells and their immediate environment is primarily mediated by membrane macromolecules known as cell-adhesion molecules involved in cell-cell and cell-extracellular matrix adhesion. Adhesion molecules influence the earliest developmental stages, such as preimplantation, during which the cell proliferation and morphological changes necessary for further development occur, the first of them being compaction at the morula stage. Then, gastrulation leads to the formation of the three primitive embryonic layers: ectoderm, mesoderm and endoderm. During organogenesis, these three cell layers interact and rearrange themselves to build the complex structures of adult organs. For example, neurulation, the formation of the neural tube by the folding of the ectoderm, is the start of these processes in vertebrates. Herein, we describe the developmental alterations of the carbohydrate structures borne by glycoproteins, especially certain cell-surface glycoproteins implicated in cellular recognition processes that might be involved with regulation of the latter at specific stages of embryogenesis, together with the modified expressions and activities of glycosyltransferases. The developmental changes of endogenous lectins are also considered as are their subsequent consequences in embryo development.

55.3 Lectins as Tools to Analyze Changes in Cell-surface Glycoconjugates During Development Table 1 gives the lectins most commonly used to analyze structural changes of glycoconjugates present at cell surface with their usual abbreviations and their specific carbohydrate ligands. Using biotinylated lectins, Chisholm and Adi [4] analyzed changes of glycoconjugate expression in human sublingual salivary glands during fetal growth and development (10-38 weeks of gestation). They observed that N - and 0-linked oligosaccharides were weakly expressed over 10- 19 weeks, and that their presence increased with maturity through the middle and the late stages of development. Recently, Paul and Ulfig [ 51 demonstrated a relationship between glycoconjugate expression and the development of human prosencephalon over weeks 16-36 of gestation. Different expression patterns were observed: PNA and VVA have affinity only for the basal nucleus (presence of P-galactosides), whereas Con A has affinity

9 12

55 Glycosylution and Development

for both basal nucleus and hypothalamic tuberomamillary nucleus (presence of mannosides). Miosge et al. [6] analyzed carbohydrate patterns in malformed and normal human embryos and fetuses using following lectins: RCA I, PNA, WGA, SBA, SNA, Con A and LPA. During development, malformed tissues exhibited carbohydrate patterns different from those of normally developed specimens. In addition, histologically normal tissues from secondary malformations accompagning the primary defect, also had modified carbohydrate patterns, suggesting a possible general alteration of these patterns in malformed embryos and fetuses. In their study on glycoconjugate changes during the development of brain microvasculature in 9, 14, 20 day-old chicken embryos and 30-day-old chickens, Nico et al. [7] observed that the endothelial cells of the optic tectum exhibited altered spatial distributions of RCA-binding (p-galactoside sites) and WGA-binding (B-Nacetylglucosaminoside sites) during embryo development. Zhou et al. [8] studied mouse Leydig cell glycoconjugate expressions during fetal and postnatal development using Con A, WGA, RCA I, UEA I, PNA and SBA. They observed that certain glycoconjugates bearing galactose, N-acetylgalactosamine and N-acetylglucosamine residues were expressed on the cell surface and in the cytoplasm of Leydig cells from the 13th day postconception to around the 20th day postpartum, suggesting a role for glycoconjugates in the modulation of hormone-receptor interactions in Leydig cells before day 20. Several artificial neolectins obtained by covalent linkage of a sugar residue to a protein (generally bovine serum albumin, BSA) were also used to analyze the carbohydrate expression pattern. Thus, p-D-galactose-BSA and lactose-BSA were synthesized 191 and used to detect the presence of P-galactoside-binding sites in rabbit fallopian tubes, uterus and blastocysts during the preimplantation phase and implantation. These carbohydrate-binding sites could be considered endogenous lectins, and their temporal expression pattern suggests their involvement in embryogenesis.

55.4 Cell-adhesion Molecules 55.4.1 Neural Cell-adhesion Molecule The neural cell-adhesion molecule, NCAM, is a member of the immunoglobulin family that was originally isolated from chicken retina cells. NCAM is implicated in intercellular adhesion via NCAM-NCAM homophilic binding, and is expressed in various cell types throughout embryonic development. During embryo development, NCAM modulates various cellular interactions [ 101 implicated in guidance and targeting of axons [ 11, 121, migration of neuronal and glial precursor cells [ 13, 141, in vitro differentiation of astrocytes [ 151 and development of muscle myotubes [16]. Moreover, it was reported that NCAM is also expressed in adult tissues associated with sensory systems and neuronal plasticity [ 171. The functional specificity of NCAM is modulated by alternatively spliced tran-

55.4 Cell-adhesion Molecules

913

scription of a single gene [18], and also by post-translational modifications, such as glycosylation [2]. It has been shown that the NCAM glycosylation pattern varied according to the embryonic development stage, and influenced the function(s) of the adhesion molecule. In embryonic cells, the elevated long-chain polysialic acid (PSA) binding to NCAM results from homopolymerization of 2,8-linked sialic acid units, whereas, in adult tissues, NCAM is poorly or not sialylated. PSA-NCAM is essential for various developmental processes at specific embryonic stages because it reduces NCAM-mediated homophilic binding in cell-cell adhesion [ 171. This observation was reinforced by the transfection of ~-galactoside-w2,6-sialyltransferase into Xenopus oocytes which inhibited PSA-NCAM formation and, consequently led to neural development abnormalities [ 191. Based on their studies on the F11 neuronal hybrid cell line, produced by fusion of rat dorsal root ganglion cells with mouse neuroblastoma cells, [20] concluded that, because PSA is highly charged, the presence of PSA-NCAM could be a source of repulsive forces causing an increase of the intercellular space. The presence of PSA on NCAM is regulated during embryonic development. Immunocytochemistry and immunoblotting experiments using the mouse IgG isotype-2a monoclonal antibody 735 (mAb 735) and bacteriophage endoneuraminidase H, which reduces the length of the PSA chain and abolishes immunoreactivity with mAb 735, demonstrated the expression and function of PSA. It has been shown that during rat brain development the elevated PSA-NCAM level decreased, and became undetectable in adult rat brain. This developmental change results in an enhanced NCAM level without PSA and/or a shortened PSA chain on NCAM [21]. Dowsing et al. [22] identified two novel carbohydrates bound to rat NCAM, NOC-3 and NOC-4. The NOC-4-NCAM glycoform contains a human blood group B-like antigen, and the NOC-3-NCAM glycoform contains a carbohydrate epitope that remains uncharacterized at this time. The rat NCAM bearing these novel carbohydrates are expressed in subpopulations of primary rat sensory olfactory neurons during axogenesis and synaptogenesis within the olfactory bulb. NOC-3 and NOC-4, as well as PSA, may play active role(s) in mediating axon growth by affecting the NCAM generating transmembrane signaling through a second-messenger activation pathway. Kajikawa et al. [23] investigated the expression and role of PSA-NCAM in the auditory organ at different embryonic stages of the chicken by means of immunohistochemical labeling with a mAb specific to the PSA portion of the PSANCAM. PSA-NCAM was detected at stage 24 (embryonic day E4), when peripheral nerve fibers begin to emerge from the acoustic ganglion. The nerve fibers remained positive during neurogenesis, stages 24-40 (E4-E 14), and disappeared at stage 42 (E16) when nerve fibers were mature. Auditory epithelial cells also expressed PSA-NCAM during stages 29-34 (E6-ES) in association with the peripheral fibers that penetrate and extend into the epithelium. At stage 38 (E12), PSANCAM was located at the base of the hair cells. These data suggest an important role of PSA-NCAM in the developing inner ear of the chicken. PSA-NCAM expression is modified during motoneuron and myotube development in the fetal rat (embryonic days Ell-E19). Using a model of the mammalian neuromuscular system, the phrenic nerve-diaphragm axis in the fetal rat, Allan and Greer [24] established the first clear immunohistochemical demonstration of the

9 14

55 Glycosylution and Development

spatiotemporal modulation of PSA-NCAM expression on myotubes during each stage of myogenesis. PSA-NCAM was identified on primary myotubes, early after initial myoblast fusion. PSA-NCAM plays a major role in phrenic axon guidance at the brachial plexus and during intramuscular branching. It is strongly expressed by the phrenic nerve during nerve outgrowth, and is down-regulated once intramuscular branching has been completed and is functional. Moreover, PSA-NCAM has also been detected and implicated in heart, kidney and hair-follicule development [25]. 55.4.2 The Adhesion Molecule L1

The mammalian cell-adhesion molecule L1 is abundant in both the nervous and immune systems. L1 plays an important role in human ontogeny, as supported by neurological disorders associated with mutations in the Ll gene (hydrocephalus, mental retardation). Ll is highly conserved in rodents and man. It is implicated in cell adhesion and migration, axonal outgrowth and fasciculation, Schwann cellaxon recognition and leukocyte interactions [26]. In neurons, homophilic L1 binding appears to be modulated by co-expression with NCAM, and is down-regulated in the presence of PSA-NCAM [27]. In rat neurons, homophilic L1 binding mediates adhesion and triggers neurite extension via fibroblast-growth-factor receptormediated signal transduction [ 12, 281. Heterophilic interactions involving glycans have also been described [29, 301. NCAM is normally tightly associated with L1, and carbohydrates on NCAM mediate L1 -L1 interactions. Carbohydrate-mediated binding of NCAM to Ll is a key requirement for the outgrowth of neurons derived from early-postnatal mouse cerebellum [27, 3 I]. Allan and Greer [24] concluded that PSA-NCAM limited the L1 -mediated adhesion between phrenic and brachial axonal nerve fibers. Lis and Sharon [2] reported that L1 and NCAM have similar expression patterns and share a common carbohydrate epitope. When, released Ll becomes embedded in the extracellular matrix, where it participates in cell-extracellular matrix interactions, and might mediate leukocyte interactions [32, 331. In addition, taking into account the abundance of L1 in the immune system, and the hypothesis developed [34] that implicates specific glycoconjugate interactions in fetoembryonic defense mechanisms, it can be postulated that L1 might directly or indirectly intervene to protect the embryo/fetus from the maternal immune response. Indeed, Clark et al. [34] observed that the placental glycoprotein designated glycodolin A might suppress a lymphocyte reaction; this glycoprotein expresses N-linked oligosaccharides with highly fucosylated or sialylated carbohydrates, and is present at very high levels at the time of embryo implantation and in the amniotic fluid.

55.5 Glycosyltransferases Polysiulyltrunsferases associated with the Golgi apparatus are implicated in the developmental changes noted in PSA-NCAM levels. The various polysialyltransfer-

55.6 Altered Expression ef Endogenous Lectins During Development

9 15

ases that have been cloned from different animal species (human, hamster, mouse, rat) exhibit strong homology and share common enzymatic properties. Both polysialyltransferase (PST) and sialyltransferase X (STX) participate in the in vivo and in vitro polysialylation of NCAM, and their expression levels are dynamically controlled during embryonic development [ 35-38]. In embryonic chick brain, w2,S-PST catalyzes 1x2,s-specificpolysialylation of endogenous NCAM [38] and this enzyme’s specific activity is maximal about 12 days after fertilization. Ong et al. [37] demonstrated that PST and STX expressions in mouse embryo varied during development, and were strongly present at E l l-El5, when maximum PSA synthesis is required. Ten days after birth, the STX-transcript level decreased sharply, whereas the PSTtranscript levels declined gradually but remained present in adult brain. Bruses and Rutishauser 1391 investigated the cellular mechanisms that regulate the PSA-NCAM level during the development of ciliary ganglion motoneurons in chicken. Using pharmacological and biochemical methods to analyze PSA synthesis, they observed the calcium dependence of both PSA biosynthesis and transferase activity, suggesting that the intracellular calcium pool might rapidly intervene and modulate PSA synthesis during embryonic development. Several observations indicate that galactose and poly-N-acetyllactosamine structures are particularly important in embryogenesis and that their biosyntheses require developmentally regulated glycosyltransferase activities [40, 4 11. The detection of galactosyltransferase at the cell surface [42] supports its role in cellular recognition phenomena. Zhou et al. [43] demonstrated the developmentally regulated expression of p-1,4-galactosyltransferasein the mouse brain; this enzyme is strongly expressed on day 16 of development, gradually declines thereafter and drops sharply after birth but remains detectable. During chick embryo development, p-NL1cetylglucosarninyltvansferase ZV activity increases, leading to the expression of more complex carbohydrate structures in 16-day-old embryos [44].

55.6 Altered Expression of Endogenous Lectins During Development During the past few years, numerous proteins (or glycoproteins) exhibiting affinity for one or several carbohydrate ligands in different animal tissues and various cell lines have been characterized which has resulted in multiple names for the same molecule (for example, more than 15 names for galectin-1). To facilitate discussions concerning these lectins, it was proposed to classify endogenous lectins into four categories [45]: C-type lectins, P-type lectins, pentraxins and galectins (previously referred to as S-type lectins). C-type lectins need calcium to bind to their carbohydrate ligands; selectins fall into this class. 55.6.1 Galectins Galectins constitute a class of calcium-independent endogenous lectins which bind only to P-galactosides. Several reviews have been devoted to galectins: [46-481. To

9 16

55 Glycosylution and Development

date, nine mammalian galectins, termed galectin- 1 to galectin-9, have been identified. Two of them have been investigated in greater depth: galectin-1 and galectin-3 [49]. Galectin-1 has a single carbohydrate recognition domain (CRD) and is present as a dimer, whereas monomeric galectin-3 has two CRD. All galectins share extensive structure similarities, for example, compared to galectin-4, galectin-6 lacks a 24 amino acid fragment in the link region between the C R D 1501. Several changes of galectin expression during embryonic development have been observed. Galectin-1 appears to be involved in the development of the mouse 01factory system [ 5 1, 521: modifications of its expression were noted at different stages of embryonic development. At El 5.5, galectin-1 was expressed in the cartilage and mesenchyme surrounding the nasal cavity, but was absent from the olfactory neuroepithelium nerve and bulb. Between E16.5 and birth, galectin-1 began to be expressed by olfactory nerve-ensheathing cells in the lamina propria of the neuroepithelium and nerve fiber layer. Neither primary sensory neurons in the olfactory neuroepithelium nor their axons in the olfactory nerve expressed galectin-1. A similar expression pattern was observed for laminin, one of the known galectin-1 ligands. It was postulated that expression in the embryonic nerve fiber layer of galectin-1 and its ligand facilitated the adhesion and fasciculation of axons en route to their glomerular targets. However, detailed analysis of galectin- 1 mutants has revealed that a subset of primary olfactory neurons fails to project into their target sites in the caudal olfactory bulb, suggesting galectin-1 involvement in the growth or guidance of these axons [53]. Using high-resolution, two-dimensional gel electrophoresis, Choi et al. [ 541 analyzed nuclear matrix preparations from cultured rat calvarial osteoblasts, at two stages of development: proliferating osteoblasts (day 3) and more mature differentiated cells with mineralization (day 18); galectin-1 was mainly expressed on day 18. Although, galectin-1 could be detected in the cytoplasmic and nuclear fractions of both proliferating and differentiated osteoblasts, with various immunological methods, it was exclusively associated with the nuclear matrix of differentiated osteoblasts. Using Western blots of kidney extracts, Foddy et al. [55] showed that day 16 hamster embryos strongly express galectin-3, which becomes undetectable a few days after birth. This pattern is tissue specific, since galectin-3 expression remains constant in lung and colon throughout development. The period of highest galectin3 expression corresponds to the medullary phase of kidney development during which there is extensive differentiation and elongation of tubular epithelia. From these observations and others, a model was proposed in which galectin-3 would link to a basement-membrane component (laminin) and thereby prevent the interaction between laminin and its integrin hgand (a6B1). Such an anti-adhesion role has been advanced for galectin-3 because of its presence in S-shaped bodies, which are the early form of kidney tubules 1.561. A similar activity has been reported for galectin-1 during muscle differentiation 1571. These galectins could also be involved in the developing nervous system in a manner that is consistent with roles in cell migration, axonal growth and synaptogenesis [58, 591. A relationship was found between the presence of galectin-1, but not of galectin3, and the surface expression of a ganglioside sialidase mediated GM(1) during cell

55.6 Altered Expression of’ Endogenous Lectins During Decelopmeiit

917

growth and neural differentiation, in human neuroblastoma cell line SK-N-MC [601. Another approach to explore the role of galectins in development consists of the generation of mice carrying a null mutation in the gene encoding galectin-1 or other galectins. Homozygous mutant animals that lack galectin- 1 develop normally and are viable and fertile. Similarly, the generation of mice from which either the galectin-1 or galectin-3 gene had been deleted has shown that both of these mutations are viable and the animals are fertile [61, 621. Moreover, double mutants lacking both galectin-1 and galectin-3 could implant successfully [62]. It should be noted that galectin-5 was detected in the blastocyst, trophoectoderm and inner cell mass of these double mutants. These observations suggest that these galectins have redundant functions in the trophoectoderm at implantation and that one galectin can potentially compensate for the absence of another galectin. Altered expression pattern and the presence of galectins in different tissues during development [48, 631 lead to the question of their biological role(s). No evident function can be attributed to these proteins, except for galectin-3 which has been proposed as a factor involved in pre-mRNA splicing [64]. Independently of a possible role during development, it was demonstrated that galectins are directly involved in apoptosis and part of a recent review was devoted to this aspect [48]. 55.6.2 Selectins

The roles of selectins, another class of endogenous lectins, have also been investigated. Nguyen et al. [65] showed that E-selectin influences capillary morphogenesis by binding to bovine capillary endothelial cell sialyl Lewis-X- and/or sialyl Lewis-A-containing ligands. Tube formation can be inhibited by blocking the interaction between E-selectin and its ligand with antibodies directed against sialyl Lewis-A or -X, or E-selectin. Modifications of L-selectin expression were studied on T cells, immediately after their exit from the thymus of fetal and newborn animals [66]. It was observed that the percentage of L-selectin-expressing thymocytes exported per day decreased by half after birth, with a remarkable heterogeneity of L-selectin expression (number of cells and intensity of expression) on exported and mature T cells. Furthermore, emigrant immature T cells weakly express L-selectin and during maturation Lselect in is up-regulated independently of extrinsic antigen. P-selectin is expressed in human preimplantation embryos [67]. As observed for galectins, homozygous mutant mice null for all three selectins undergo normal embryonic development and implantation [68-70]. Similar conclusions to those advanced above can be drawn concerning possibility for glycoproteins with redundant biological functions to originate from different genes. 55.6.3 Other Endogenous Lectins

Another endogenous lectin, called R1, was first isolated from rat cerebellum by Zanetta et al. [71]. R1 is involved in the formation of myotubes during muscle de-

9 18

55 Glycosylation a n d Development

velopment and regeneration [72], and might also to be involved in a recognition process between axons and muscle cells during neuromuscular junction formation.

55.7 Conclusion The elucidation of the molecular mechanisms involved in the cellular recognition functions of carbohydrates during embryo development remains a challenge to be met in order to understand the regulation of specific key developmental stages. Moreover, since glycosylated molecules and endogenous lectins have also been detected intracellularly, in the cytoplasm and in the nucleus, their potential roles in embryo development need to be more intensively explored. References 1. Bourrillon, R., Aubery, M. Cell surface glycoproteins in embryonic development. Intl Reo. Cytol. 1989, 116, 257-338. 2. Lis, H., Sharon, N. Protein glycosylation structural and functional aspects. Eur. J. Biochenz., 1993, 18, 1-27. 3. Poirier, F., Kimber, S. Cell surface carbohydrates and lectins in early development. Mol. Hum. Reprod., 1997, 3, 907-918. 4. Chisholm, D.M., Adi, M.M. A histological, lectin and S-100 histochemical study of the developing prenatal human sublingual salivary gland. Arch. Oral B i d . , 1995, 40, 1073-1076. 5. Paul, A,, Ulfig, N. Lectin staining in the basal nucleus (Meynert) and the hypothalamic tuberomamillary nucleus of the developing human prosencephalon. Anat. Rec., 1998,252, 149-158. 6. Miosge, N., Gotz, W., Quondamatteo, F., Herken, R. Comparison of lectin binding patterns in malformed and normal human embryos and fetuses. Terutoloyy, 1998, 57, 85-92. 7. Nico, B., Quondamatteo, F., Ribatti, D., Bertossi, M., Russo, G., Herken, R., Roncali, L. Ultrastructural localization of lectin binding sites in the developing brain microvasculature. Anat. Embryol., 1998, 197, 305-3 15. 8. Zhou, X.H., Kawakami, H., Hirano, H. Changes in lectin binding patterns of Leydig cells during fetal and postnatal development in mice. Histochem. J . , 1992, 24, 354-360. 9. Biermann, L., Gabius, H.J., Denker, H.W. Neoglycoprotein-binding sites (endogenous lectins) in the fallopian tube, uterus and blastocyst of the rabbit during the preimplantation phase and implantation. Acta Anat., 1997, 160, 159-17 1. 10. Rutishauser, U., Landmesser, L. Polysialic acid in the vertebrate nervous system: a promoter of plasticity in cell-cell interactions. TINS, 1996, 19, 422-427. 1 1 . Tang, J., Landmesser, L., Rutishauser, U. Polysialic acid influcnces specific pathfinding by avian motoneurons. Neuron, 1992, 8, 1031-1 044. 12. Tang, J., Rutishauser. U., Landmesser, L. Polysialic acid regulates growth cone behavior during sorting of motor axons in the plexus region. Neuron, 1994, 13, 405-414. 13. Hu, H., Tomasiewicz, H., Magnuson, T., Rutishauser, U. The role of polysialic acid in the migration of olfactory interneuron precursors in subventricular zone. Neuron, 1996, 16, 735743. 14. Wang, C., Pralong, W.F., Schulz, M.F., Pougon, G., Aubry, J.M., Pagliusi, S., Robert, A,, Kiss, J.Z. Functional N-methyl-D-aspartate receptors in 0-2A glial precursor cell: a critical role in regulating polysialic acid-neural cell adhesion molecule expression and cell migration. J. Cell Biol., 1996, 135, 1565-1581. 15. Minana, R., Sancho-Tello, M., Climent, E., Segui, J.M., Renau-Piqueras, J., Guerri, C. Intra-

References

9 19

cellular location, temporal expression, and polysialylation of neural cell adhesion molecule in astrocytes in primary culture. Gliu, 1998, 24, 41 5-427. 16. Fredette, B., Rutishauser, U., Landmesser, L. Regulation and activity-dependence of Ncadherin, N-CAM isoforms, and polysialic acid on chick myotubes during development. J. Cell Biol., 1993, 123, 1867-1888. 17. Rutishauser, U. Polysialic acid at the cell surface: biophysics in service of cell interactions and tissue plasticity. J. Cell. Biochem., 1998, 70, 304-3 12. 18. Reyes, A.A., Small, S.J., Akeson, R.A. At least 27 alternatively spliced forms of the neural cell adhesion molecule mRNA are expressed during rat heart development. Mol. Cell. Biol., 1991, 11, 1654-1661. 19. Livingstone, B.D., De Robertis, E.M., Paulson, J.C. Expresion of p-galactoside a2,6 sialyltransferase blocks synthesis of polysialic acid in xenopus embryos. G[j,cohio/ogy,1990, I , 3944. 20. Yang, P., Yin, X., Rutishauser, U. Intercellular space is affected by the polysialic acid content ofN-CAM. J. Cell Biol., 1992, 116, 1487-1496. 21. Doherty, P., Walsh, F.S. Polysialic acid as a specific and positive modulator of NCAM dependent axonal growth. In Polysialic uc~irlfrommicrobes to man, (J. Roth, U. Rutishauser & F.A. Troy-11, eds-Birkhaiiser Verlag, Boston, Berlin), 1993, 241-255. 22. Dowsing, B., Puche, A,, Hearn, C., Key, B. Presence of novel N-CAM glycofornis in the rat olfactory system. J. Neurobiol., 1997, 32, 659-670. 23. Kajikawa, H., Umemoto, M., Mishiro, Y., Sakagami, M., Kubo, T., Yoneda, Y. Expression of highly polysialylated NCAM (NCAM-H) in developing and adult chicken auditory organ. Hear. R e x , 1997, 103, 123-130. 24. Allan, D.W., Greer, J.J. Polysialylated NCAM expression during motor axon outgrowth and myogenesis in the fetal rat. J. Comp. Neurol., 1998, 391, 275-292. 25. Lackie, P., Zuber, C., Roth, J. Expression patterns of polysialic acid during vertebrate organogenesis. In Polysialic acidfrom microbes to man (J. Roth, U. Rutishauser & F.A. Trug IT, eds-Birkhaiiser Verlag, Boston, Berlin), 1993, 263-278. 26. Kadmon, G., Altevogt, P. The cell adhesion molecule L1: species- and cell-type-dependent multiple binding mechanisms. Differentiation, 1997, 61, 143-150. 27. Kadmon, G., Kowitz, A., Altevogt, P., Schachner, M. Functional cooperation between the neural adhesion molecules LI and N-CAM is carbohydrate dependent. J . Cell Biol., 1990, 110, 209-21 8. 28. Doherty, P., Williams, E., Walsh, F.S. A soluble chimeric form of the LI glycoprotein stimulates neurite outgrowth. Neuron, 1995, 14, 57-66. 29. Mauro, V.P., Krushel, L.A., Cunningham, B.A., Edelman, G.M. Homophilic and heterophilic binding activities of N-CAM a nervous system cell adhesion molecule. J. Cell B i d , 1992, 119, 191-202. 30. Murray, B.A., Jensen, J.J. Evidence for heterophilic adhesion of embryonic retinal cells and neuroblastoma cells to substratum adsorbed NCAM. J. Cell Biol., 1992, 117, 131 1-1320. 3 1. Feizi, T. Evidence for carbohydrate-mediated interactions between the neural-cell-adhesion molecules NCAM and LI. TZBS, 1994, 19, 233-234. 32. Hubbe, M., Kowitz, A.. Schirrmacher, V., Schachner, M., Altevogt, P. L1 adhesion molecule on mouse leukocytes: regulation and involvement in endothelial cell binding. Eur. J. Immunol.. 1993,23, 2927-2931. 33. Ebeling, O., Duczmal, A,, Aigner, S., Griger, C., Scholhammer, S., Kemshead, J.T., Miiller, P., Schwartz-Albiez, R., Altevogt, P. LI adhesion molecule on human lymphocytes and monocytes: expression and involvement in binding to avP3 integrin. Eur. J. Zmmunol., 1996, 26, 2508-25 16. Oehninger, S., Patankar, M.S., Koistinen, R.; Dell. A,, Morris, H.R., Koistinen, M. A role for glycoconjugates in human development: the human feto-embryonic defence system hypothesis. Hum. Repod., 1996, 11, 467-473. 35. Phillips, G.R., Kenshel, L.A., Crossin, K.L. Developmental expression of two rat polysialyltransferases that modify the neural cell adhesion molecule N-CAM. Bruin Res. Dell. Bruin Rex, 1997, 102, 143-155. 36. Nakayama, J., Angata, K., Ong, E., Katsuyama, T., Fukuda, M. Polysialic acid, a unique

920

55 Glycosylation and Development

glycan that is developmentally regulated by two polysialyltransferases, PST and STX, in the central nervous system: from biosynthesis to function. Pathol. Int., 1998, 48, 665-677. 37. Ong, E., Nakayama, J., Angata, K., Reyes, L., Katsuyama, T., Arai, Y., Fukuda, M. Developmental regulation of polysialic acid synthesis in mouse directed by two polysialyltransferases PST and STX. Glycohiology, 1998. 8; 415-424. 38. Sevigny, M.B., Yi, J., Kitamore-Kawaguchi, S., Troy-11, S.A. Developmental expression and characterization of the alpha-2,8-polysialyltransferaseactivity in embryonic chick brain. Glycobiology, 1998, 8, 857-867. 39. Bruses, L.J., Rutishauser, U. Regulation of neural cell adhesion molecule polysialylation: evidence for nontranscriptional control and sensitivity to an intracellular pool of calcium. J. Cell Biol. 1998, 140, 1177-1 186. 40. Shur, B.D. Cell surface p-1,4 galactosyltransferase: twenty years later. Glyrobiology, 1991, I , 563-515. 41. Varki, A . Biological roles of oligosaccharides: all the theories are correct. Glycobiology, 1993,3, 97-130. 42. Shur, B.D. Glycosyltransferases as cell adhesion molecules. Curr. Opin. Cell B i d , 1993, 5, 854-863. 43. Zhou, D., Chen, C., Jiang, S., Shen, Z., Chi, Z . , Gu, J. Expression of pl-4 galactosyltransferase in the development of mouse brain. Biochim. Biophys. Acta, 1998, 1425, 204-208. 44. Ogier-Denis, E., Bauvy, C., Moutsita, R., Aubery, M., Codogno, P. Increased UDP-GlcNAc: a-mannoside P( 1-4) N-acetylglucosaminyltransferaseactivity during chick embryo development. Biochim. Biophys. Acta, 1990, 1054, 149-1 53. 45. Barondes, S.H., Castronovo, V., Cooper, D.N.W., Cummings, R.D.: Drickamer, K.: Feizi, T., Gitt, M.A., Hirabayashi, J., Hughes, C., Kasai, K.I., Leffler, H., Liu, F.T., Lotan, R., Mercurio, A.M., Monsigny, M., Pillai, S., Poirier, F., Raz, A., Rigby, P.W.J. Galectins: a family of animal P-galactoside-binding Iectins. Cell, 1994, 76, 597-598. 46. Barondes, S.H., Cooper, D.N.W., Gitt, M.A., Leffler, H. Structure and function of a large family of animal lectins. J. Biol. Chem., 1994, 265, 20807-20810. 47. Hirabayashi, J., Kasai, K.I. The family of metazoan metal-independent P-galactoside-binding lectins: structure, function and molecular evolution. Glycohiology, 1993, 3, 297-304. 48. Perillo, N.L., Marcus, M.E., Baum, L.G. Galectins: versatile modulators of cell adhesion, cell proliferation, and cell death. J. Mol. Med., 1998, 76, 402-412. 49. Wang, J.L., Laing, J.G., Anderson, R.L. Lectins in the cell nucleus. Glycobiology, 1991, 1, 243-252. 50. Gitt, M.A., Colnot, C., Poirier, F., Nani, K.J., Barondes, S.H., Leffler, H. Galectin-4 and galectin-6 are two closely related lectins expressed in mouse gastrointestinal tract. J. Biol. Chem., 1998,273,2954-2960. 51. Tenne-Brown, J., Puche, A.C., Key, B. Expression of galectin-1 in the mouse olfactory system. Intl J. Deo. Bid. 1998, 42, 791-799. 52. Plendl, J., Sinowatz, F. Glycobiology of the olfactory system. Acta Anaf., 1998, 161, 234.253. 53. Puche, A.C., Poirier, F., Hair, M., Bartlett, P.F., Key, B. Role of galectin-1 in the developing olfactory system. Dev.Biol., 1996, 179, 274-287. 54. Choi, J.Y., Van Wijnen, A.J., Aslam, F., Leszyk, J.D., Stein, J.L., Lian, J.B., Penman, S. Developmental association of the beta-galactoside-binding protein galectin-1 with the nuclear matrix of rat calvarial osteoblasts. J. Cell Sci., 1998, 11I , 3035-3043. 55. Foddy, L., Stamatoglou, S.C., Hughes, R.C. An endogenous carbohydrate-binding protein of baby hamster kidney (BHK21 C13) cells. J. Cell Sci., 1990, 57, 139-148. 56. Sorokin, L., Klein, G., Mugrauer, G., Fecker, L., Ekblom, W.M., Ekblom, P. Development of kidney epithelial cells. In Epithelial organization and developnzent. (T.P. Fleming, ed.Chapman & Hall, London) 1992, 163-190. 57. Cooper, D.N.W., Massa, S.M., Barondes, S.H. Endogenous muscle lectin inhibits myoblast adhesion to laminin. J. Cell Biol., 1991, 115, 1437-1448. 58. Li, W.X., Joubert-Caron, R., El Oumami, H., Bladier, D., Caron, M., Baumann, N. Regulation of a beta-galactoside-binding lectin and potential ligands during post-natal maturation of rat brain. Dev. Neurosci., 1992, 14, 290-295.

References

921

59. Poirier, F., Tirnmons, P.M., Chan, C.-T.J., Guenet, J.-L., Rigby. P.W.J. Expression of the L14 lectin during mouse embryogenesis suggests multiple roles during pre- and post-implantation development. Development, 1992, 115, 143.- 155. 60. Kopitz, J., von Reitzenstein, C., Burchert, M., Cantz, M., Gabius. H.J. Galectin-1 is a major receptor for ganglioside GM( I ) , a product of the growth-controlling activity of a cell surfxe ganglioside sialidase, on human neuroblastoma cells in culture. J. Bid. Chem., 1998, 273, 11205-11211. 61. Poirier, F.: Robertson, E.J. Normal development of mice carrying a null mutation in the gene encoding the L14 S-type lectin. Developnient, 1993, 119, 1229- 1236. 62. Colnot, C., Fowlis, D. Ripoche, M.A.. Bouchaert, I., Poirier, F. Embryonic implantation in galectin l/galectin 3 double mutant. Der. Dynrmz., 1998,211, 306-313. 63. Catt, J.W., Harrison, F.L., Carleton, J.S. Distribution of an endogenous b-galactoside-specific lectin during foetal and neonatal rabbit development. J. Cell Sci.. 1987, 87, 623-633. 64. Dagher, S.F., Wang, J.L., Patterson, R.J. ldentification of galectin-3 as a factor in pre-mRNA splicing. Proc. Natl Acad. Sci. USA, 1995, 92, 1213-~1217. 65. Nguyen, M., Strubel, N.A., Bischoff, J. A role for sialyl Lewis-X/A glycoconjugates in capillary morphogenesis. Nature, 1993, 365, 267- 269. 66. Witherden, D.A., Abernethy, N.J., Kimpton, W.G., Cahill, R.N. Changes in thymic export of L-selectin+ gamma delta and alpha beta T cells during fetal and postnatal development. J. Nruropathol. Exp. Neural., 1994, 53, 144 149. 67. Campbell, S., Swann, H.R., Aplin, J.D. Cell adhesion molecules on the oocyte and preimplantation human embryo. Hunt. Reprod., 1995, 10, 1571Ll578. 68. Mayadas, T.N., Johnson, R.C., Rayburn, H. Leukocyte rolling and extravasation are severely compromised in P selectin deficient mice. Cell, 1993, 74, 541-554. 69. Arbones, M.L., Ord, D.C., Key, K . Lymphocyte homing and leukocyte rolling and migration are impaired in L-selectin-deficient mice. Irninunity, 1994, 1, 247--260. 70. Labow, MA., Norton, C.R., Rumberger, J.M. Characterization of E-selectin-deficient mice: demonstration of overlapping function of the endothelial selectins. Immunity, 1994, I , 709--720. 71. Zanetta, J.P., Dontenwill, M., Meyer, A,, Roussel, G. Isolation and immunohistochemical localization of a lectin like molecule from the rat cerebellum. Dec. Brain Rex, 1985, 17. 233--243. 72. Thomas, D., Lehmann, S., Kuchler, S . , Marschal, P., Zanetta, J.P. Differential expression of an endogenous mannose-binding protein R I during muscle development and regeneration delineating its role in myoblast fusion. Glycohiology, 1994, 4 , 23-38.

Carbohydrates in Chemistry and Biology Beat Emst, Gerald W. Hart, Pierre Sinay copyright Q WILEY-VCH Verlag GmbH. 2000

56 Protein Glycosylation and Cancer James W. Dennis and Maria Granovsky

56.1 Introduction Molecular changes that accompany malignant transformation include altered glycosylation of glycoproteins and glycolipids. Interest in glycosylation changes in cancers stems from the observations that expression of certain glycan structures in tumors correlates with clinical prognosis (reviewed in [ l , 21). Some of the more common alterations in N- and 0-glycan structures are an increase in size due to GlcNAc-branching of core sequences proximal to the protein, and variation in the terminal sequences (Table 1). Protein glycosylation modifies extracellular domains of secreted and cell surface glycoproteins. As a class of macromolecules, glycoproteins mediate many cell interactions with the extracellular environment. Cell adhesion receptors and cytokine receptors transmit intracellular signals controlling cell growth, and are of prime importance to our understanding of cancer development. In this Chapter, we consider functions of N- and 0-glycan structures that may contribute to cancer growth and metastasis. We begin by briefly considering roles for glycosylation in normal developmental processes, and how modification of proteins with glycans of diverse structure can affect their functions.

56.2 Protein Glycosylation Generates Molecular Diversity Variation in glycan structures is regulated in a tissue-specific manner [3]. The GlcNAc-branching of core N- and 0-glycans, the extension of these branches, and capping with terminal sequences vary between tissues and between normal and cancer cells. Structural diversity is dictated in part by tissue-specific regulation of genes encoding glycosyltransferases of the medial- and trans-Golgi compartments 141. The transcriptional regulation of glycosyltransferase genes can be complex, as

924

56 Protein Glycosylation und Cuncer

Table 1. Summary of N- and 0-glycan changes in cancer cells. Cancers

Gain or loss of glycans

Enzymes

References

Rodent tumors; human melanoma, carcinomas of breast, colon

Gain of fi1,6GlcNAcbranched N-glycans. Preferred substrate for extension with polylactosamine. Gain of fi1,4GlcNAcbranched in N-glycans Gain of Bisecting fi1,3GlcNAc in N-glycans Gain of extended chain Lewis antigens on 0- N - and glycolipids (Le', SLe". Ley and SLe") Truncated 0-glycans, Tn, T antigen Gain of pl,6GlcNAc-branched 0-glycans. Substrate addition of extended chain Lewis antigens Loss of 0-glycans on MUCl mucin exposes protein and carbohydrate epitopes

GIcNAc-TV

[32, 36, 52, 53, 581

GICNAc-TIV

[1151

GICNAc-TI11

[57, 581

FUC-TIV FUC-TVII

[93, 117, 1181

ND

[I 191

core 2 GlcNAc-T

[87, 981

Core 2 GlcNAc-T

[120, 1211

Human choriocarcinoma Liver chirosis, hepatomocarinoma Human carcinomas of colon and baldder

Human breast cancer Rodent and human carcinoma

Human breast carcinomas

observed for pl,4Gal-T1 [5] and the ST6Gal [6]; both genes have multiple tissuesspecific promoters and transcription start-sites. Current information on the regulation of glycosyltransferase genes in normal and cancer tissues is incomplete, but clearly important to our understanding of glycan functions in cancer. Glycan processing, and particularly the terminal sequences, are variable, resulting in a range of related glycoforms at each glycosylation site [7]. Golgi glycosyltransferases generally show specificity for oligosaccharide intermediates, and transfer a monsaccharide from a sugar-nucleotide donor to a single hydroxyl position of the acceptor. Glycan products are also regulated by competition between certain enzymes for common acceptor intermediates [8]. The relative levels of competing enzymes affect the distribution of glycan structures on mature glycoproteins (Figure 1). For some glycoproteins, glycan structural diversity imparts functional diversity, and effectively increases the range biological activity for a single gene product. For example, certain glycan sequences on serum glycoproteins bind to hepatic lectins, which mediate their clearance. The glycopeptide hormones are produced in the pituitary with glycans bearing either terminal N-acetylgalactosamine-4 sulphate or sialic acid, and these glycoforms differ in liver clearance rates and biological activity [ 91. Similarly, relative levels of sialylated and branched N-glycans on erythropoietin

56.2 Protein Glycosylation Generates Moleculur Diversity

Hybrid-type

t

925

Complex-type

t

I

GlcNAc Man

Figure 1. Schematic of N-linked glycan biosynthesis showing the Golgi compartments. Abbreviations used are oligosaccharyltransferase, OT; the a-glucosidases, GI, CII; the P-N-acetylglycosaminyltransferases, TI, TII, TIII, TIV, TV, T(i); the al,2mannosidases, MI, al,3/6mannosidases MII, MIII; P I .4-galactosyltransferases (Gal-T), a-fucosyltransferases (Fuc-T), a-sialyltransferases (ST), swainsonine (SW), castinospermine (Cast). The terminal sequences are added to both N- and 0-linked glycans by P1,4Gal-T, P I ,3ClcNAc-T(i), P1,6GlcNAc-T(I), u2,3ST, al ,3Fuc-T. The lactosamine antenna initiated by GlcNAc-TV is preferentially elongated with polylactosamine and Lewis antigens as indicated by the gray box. Note that GlcNAc-TI11 re-directs the pathway into “bisected glycans” rather than “complex-type”. The GlcNAc-TI product is required for substrate recognition by a-mannosidase 11 and GlcNAc-TII. and the subsequent biosynthesis of the complex-type N-glycans. GlcNAc-TI11 substitutes the core p-mannose residue and redirects the pathway into “bisected N-glycans”. blocking subsequent action of GlcNAc-branching transferases { {254}}. GlcNAc-TV is one of several mcditrl-Golgi enzymes that initiate a specific branch that is variably elongation in the trtrns-Golgi generating structural diversity in the mature N-glycans. The GlcNAc-TV product is the preferred intermediate for extension with polylactosamine chains (i.e. G a l ~ 1 , 4 G l c N A c ~ lrepeating ,3 units of 2 to >I0 in length) (1221. Polylactosamine synthesis is also regulated by fi1,3GlcNAc-T(i) activity [ 1231, glycoprotein transit time in the truns-Golgi 11241, and competition by chain-terminating enzymes including ul,2Fuc-T and a2,6SA-T [ 1251.

affect the stability and activity of the glycoprotein in vivo [lo]. Glycans are also structural components of glycoprotein receptors and cytokines that transmit information between cells and the environment to control fundamental aspects of cell behavior. Glycoform heterogeneity observed for CD44 [ 111 ICAM-1 [ 121 and CD43 [ 131 cell surface receptors is suggested to modify their cell adhesion activities. As discussed below, tissue-specific and disease-associated programs of glycosylation can regulate the activity of intracellular signaling pathways, and thereby cell behavior. The Golgi biosynthetic pathways appear to have evolved to generate structural diversity on glycoproteins. However, specific functions in metazoans, and therefore

926

56 Protein Glycosylution and Cuncer

selection pressures favoring the observed conservation of glycosylation, remain unclear. Glycan binding to lectins is a common means of adhesion between multicellular organisms and parasitic or symbiotic organisms [ 141. The influenza virus surface haemagglutinin-neuraminidase recognizes sialic acids present in bronchial epithelium, and strain variation affects binding specificity, cell surface attachment and virulence [ 151. Different terminal glycan sequences on glycoproteins can also mask lectin-binding sites for parasite infection. Thus, on an evolutionary time scale, continuous generation of molecular diversity in metazoans may enhance population fitness in the ongoing battle to evade parasitic organisms. However, glycan structures have been co-opted into developmental functions unrelated to pathogen evasion. Conserved glycan structures become acceptor intermediates for newly evolved glycosyltransferases, as the pathway and glycan diversity expands. The earlier the biosynthetic capacity for an oligosaccharide structure arose, the more likely it is that the glycans have acquired functional niches within the biology of the organism. Thus, mutations in upstream, and evolutionarily ancient, parts of the glycosylation pathway will affect many downstream oligosaccharide structures, which are likely to have been co-opted into developmental functions. Indeed, the phenotypes of null mutations in glycosyltransferase genes in mice appear to be progressively milder for enzymes operating later in the biosynthetic pathway [ 161. GlcNAc-TI deficient mouse embryos (MgatZ-’-) lack all complex-type N-glycans, and they die at around E9.5 due to a failure of multiple organ systems [17, 181, whereas mutations affecting subsets of complex-type N-glycans are less severe. GalTI-’- [ 191, Mgat2-1- [20] and MqatS-1- mice are viable [126]. Characterization of these mutant embryos and mice suggests that specific subsets of glycan structures have celltype specific functions that can affect cell differentiation, growth and migration; processes of central importance to cancer development and metastasis.

56.3 Cancer Initiation and Progression Cancer genetics is the basis of our understanding of the disease, and provides the context to considering protein glycosylation as a modifier of cancer progression [21, 221. Multiple genetic changes, combined with selection pressure in the host environment, provide conditions for tumor formation and progression. Mutations that promote growth and survival are maintained in the cell population as tumors expand and progress to become metastatic [23]. For example, the pre-neoplastic cells of aberrant crypts in colon display increased proliferation index; a condition that provides the cell substrate for subsequent rare mutations. In this manner, disease progression leads to pre-malignant polyps, then to adenomas and finally to metastatic colon carcinoma [24]. Tumor evolution or progression often produces wide phenotypic and molecular heterogeneity in the cell population, and this has complicated the analysis of the malignant phenotypes at the level of biochemistry and cell biology [25]. Cancer mutations result in either loss-of-function in a gene product, designated

56.4 Tumor Cell Proliferation

927

“tumor suppressor proteins” (e.g. p53, APC, WTl), or missense mutations which activate a “proto-oncogenes” (e.g. H-RAS). Cancer genes have also been described as “caretaker” and “gatekeeper” genes, the latter regulate aspects of cellular proliferation, and the former control genomic integrity (reviewed in [26]. The retinoblastoma gene Rb-1 is a gatekeeper gene, where loss-of function by mutation promotes entry into the cell cycle and tumorigenesis in retinal epithelial cells. The mismatch DNA repair enzymes MSH2 and MLHl are caretaker genes, their inactivation in colon cancers leads to mutation in other genes that can enhance growth, such as the TGF-P receptor type I1 gene [27]. P53 and other genes that ensure faithful segregation of chromosomes during mitotic also serve a caretaker function. Mutations in genes that block cell death by apoptosis (e.g. Bax) contribute to cell survival in the hypoxic environment of tumors [28]. Many cancer mutations have been engineered in mice, and shown to increase the incidence of pre-malignant lesion and tumors, thereby providing evidence of cause and effect (reviewed in [29, 301. In addition, familial or heritable mutations in tumor suppressor genes such as P53, Rb-1, and APC greatly increase the risk of cancer in certain tissues, and this usually requires loss of the wild type allele in cells of the target organ. The search for somatic and heritable mutations responsible for cancer initiation or progression does not yet include genes in the protein glycosylation pathways. However, changes in glycosylation are commonly observed in human carcinomas, and in cell lines transformed with activated RAS and v-Src [31-341. In earlier studies of tumor cells glycosylation mutants, a2-3sialylation, increased polylactosamine and P1,6GlcNAc-branching of N-glycan were suggested to enhance tumor growth and metastatic in mice [35-381. More recently, topical expression of glycosyltransferase genes in tumor cell lines, and studies in gene-knockout mice confirm that cancer-associated changes in glycosylation play a causal role in cancer progression and metastasis.

56.4 Tumor Cell Proliferation RAS proto-oncogenes sustain activating mutations in approximately 20% of all human tumors. In addition, RAS signaling is induced by other common mutations such as amplification of Neu/ErbB-2 in breast cancer. RAS is a GTPase and activator of Raf kinase leading to the activation of AP1 (i.e. c-Fos/c-Jun dimers) and Ets transcription factors [ 391. These transcription factors regulate expression of multiple genes involved in cell cycle progression (e.g. cyclin D), cell motility (Rho/ Ccd42/Rac-l/Taiml), as well as metalloproteases (e.g. MMP-2, -3, -9), growth factors (e.g. VEGF and bFGF) and glycosyltransferases (e.g. GlcNAc-TV). Genetic analysis in mice has validated the relationship between RAS activation and downstream gene transcription required for tumor growth. Expression of activated RAS in transgenic mice results in invasive skin tumors following application of skin carcinogens, but on a c-Fos deficient genetic background, matrix metalloproteases (MMPs) and vascular endothelial growth factor (VEGF) transcripts are suppressed,

928

56 Protein Glycosylation and Cancer

and only hyperkeratinized benign tumors are observed [40]. VEGF induces host endothelial cells to develop microvasculature, thereby providing the necessary oxygen and nutrients to the expanding tumor [41]. MMPs secreted by tumor cells digest extracellular matrix and facilitate tumor cell invasion through extracellular matrix that separates tissue compartments. The RAS signaling pathway is activated by growth factors including members of the EGF, PDGF and F G F families. EGF receptor levels and signaling were observed to decreased in GlcNAc-TI11 transfected U373 glioma cells, and both the EGF receptor and the N G F receptor Trk are substrates of GlcNAc-TI11 [42]. Dimerization and phosporylation of Trk receptor was also reduced in GlcNAc-TI11 transfected PC12 cells [43]. In PC12 cells, NGF/Trk and EGFiEGFR both activate RAS/MAPK, but with different kinetics and thereby quite different outcomes; differentiation or proliferation, respectively [44]. As suggested by the studies discussed above, the kinetics of receptor aggregation might be regulated by glycosylation, and thereby influence intracellular signaling. Indeed, Mgat.5-dependent glycosylation in T cells affects the kinetics of agonistinduced T cell receptor (TCR) aggregation and signaling [126]. Rates of TCR internalization following addition of agonist were increased for M g a t Y - T cells compared to wild type cells. This results in a marked increase in sensitivity to low agonist concentrations, and in addition, the cooperativity of Mgat.5-/- TCR signaling is enhanced as indicated by the Hill coefficient for entry of cells into S phase. Therefore, in the absence of cell surface p176GlcNAc-branched N-glycans, TCR oligomerize more efficiently, resulting in enhanced sensitivity and cooperativity of receptor signaling. GlcNAc-TV activity has been shown to increase following activation of CD4+ and CD8+ T cells [45]. This is due to activation of Mgat.5 transcription probably through the RAS/MAPK/ETS pathway, which is activated in T cells following antigen stimulation. Therefore, the program of gene expression associated with normal T cell activation includes the regulation of 01 ,6GlcNAcbranched N-glycans, which in turn regulates T cell responses at the level of receptor aggregation [46]. B cell responses are unaffected in MgatS-’- mice, demonstrating specificity of the GlcNAc-TV dependent effects on receptor signaling. T cell receptor aggregation may be impeded by the increased mass of the p1,6GlcNAcbranched glycans. Alternatively, p 1,6GlcNA~-branchedglycans may form lattices with multivalent lectins that enhance the stability of T cell receptors in the resting state. Indeed, exogenous galectin-1 added to cultured T cells antagonizes TCR responses [47]. CD22 provides another example of an endogenous lectin acting in cis. The extracellular domain of CD22 on the B cell surface has affinity for the product of ST6Gal (i.e. SAa2,3Galpl,4GlcNAc). Following antigen stimulation, the cytosolic domain of CD22 is phosphorylated, causing recruitment of Shpl tyrosine phosphatase, which dampens intracellular signaling [47]. B cells of CD22 deficient mice are hypersensitive to antigen stimulation. Conversely, ST6Gal knocked-out mice show impaired B cell maturation and IgM production [48]. The RAS protein is activated in response to ligand-induced receptor aggregation at the inner surface of the plasma membrane where complexes of signaling molecules accumulate to attain a critical level for signal transduction (Figure 2). Activating mutations in RAS, of the type found in human tumors, bypass the need for

56.4 Tumor Cell Proliferation

MMPs - invasion VEGF- angiogenesis

929

I’

Figure 2. Model depicting up-regulation of GlcNAc-TV gene expression by activation of the RAS pathway, leads to increased pl,6GlcNAc-branching of N-glycans on cell surface receptors including integrin u1 ps (others are yet to be catalogued). This enhances focal adhesions and cell motility, with positive feedback amplifying the intracellular signaling pathways. The red barbells represent galectins, and the fork-like symbols represent N-glycans. E-cadherin homotypic cell-cell adhesion is a negative regulator of cells motility and invasion. lntegrins are substratum adhesion receptors; RTK represents receptor tyrosine kinases; Fak and c-Src are intracellular tyrosine protein kinases. RAF, MEK and MAPK are S/T protein kinases; Grb-2 is an adapter protein; Ets, c-Jun, c-Fos are transcription factors. PyMT is also an adaptor protein with phosphopeptide domains for binding the proteins c-Src, Shc and PI3K. Note that signaling pathways interact at multiple levels, and other interactions not shown here also contribute tumor progression.

growth factors and cell surface glycoprotein receptors. Therefore, the glycosylation status of glycoprotein receptors and cytokines upstream of RAS may be less significant in cells with a mutated RAS gene, as the cells acquire a large measure of growth-factor independence. A similar phenomenon may apply to other cytokine pathways. Recent genetic studies in Drosoplzilu demonstrate that Wnt receptor (Frz) signaling requires the gene encoding UDP-glucose dehydrogenase, which produces glycosaminoglycans, required as co-receptors [49]. Inactivating mutations in pcatenin and APC, which are negative regulators of growth stimulation by the Wnt pathway, circumvent the requirement for Wnt/Frz, and for UDP-glucose dehydrogenase. As discussed above, RAS activation induces expression of genes that act as downstream effectors of tumor cell invasion and metastasis, including VEGF and MMPs. RAS activation also induces Mgat5 gene expression. The promoter region of the Myat.5 has AP1, AP2 and PEA3Jets binding sites [50],and can be induced by the v-src oncogene [46]. Activation of RAS-dependent kinase Raf-1 and the Ets-2 transcription factor induces M g a d expression [42, 511. GlcNAc-TV activity is in-

930

56 Protein Glycosylution and Cancer

duced in cells transfected with u-Src [32, 521, H-RAS and the activated tyrosine kinase receptor v-fps/fes [53]. More importantly, GlcNAc-TV appears to be a downstream effector of RAS that is required for efficient metastasis. Mammary carcinoma cells transfected with GlcNAc-TV expression vector produced 4-40 times more lung metastases [54]. Human carcinomas of breast, colon, and melanomas commonly show increased levels p 1,6GlcNA~-branchedN-glycan measured by L-PHA immunohistochemistry [55].L-PHA staining in human colorectal carcinoma sections provides an independent prognostic indicator for tumor recurrence and patient survival and is associated with the presence of lymph node metastases [56]. GlcNAc-TI11 and GlcNAc-TV activities increase in chemically induced rat hepatocarcinomas [57],as do transcript levels in pre-malignant hepatitis and tumors of LEC rats, a strain, which has a hereditary predisposition to hepatitis and hepatocarcinomas [ 581. GlcNAc-TI11 and GlcNAc-TV compete for glycan acceptor intermediates based on in vitro biochemical studies [8]. The action of GlcNAc-TI11 renders glycan intermediates poor substrate for GlcNAc-TIV and GlcNAc-TV, and thereby routes the pathway into the less branched hybrid glycans. Indeed, B16 melanoma cells transfected with a GlcNAc-TI11 expression vector showed an increase in hybrid-type glycans and a reduction in GlcNAc-TV products [59]. The transfection of B16 melanoma cells showed enhanced cell-cell adhesion in vitro, and reduced lung colonization when injected intravenously into mice. GlcNAc-TI11 transfected B 16 mouse melanoma were also less invasive through extracellular matrix in vitro. Unlike rat and human liver, mouse hepatocarinomas do not express GlcNAcTI11 transcripts. However, in Mgat3-1- mice treated with the carcinogen diethylnitrosamine, hepatocarinomas progress more slowly compared to wild type littermates. This suggests that a host paracrine effect is dependent upon GlcNAc-TI11 to promote tumor growth. For example, it is possible that host-derived growth factor acting on the tumor may require glycosylation by GlcNAc-TI11 [60]. In any case, these results suggest that glycoslyation can regulated host paracrine as well as tumor cell autonomous phenotypes.

56.5 Cell Migration Metastatic carcinoma cells are not restricted by the tight junctions characteristic of normal epithelial cells. With the transition from normal to malignant status, cellcell adhesion molecules are switched off, and substratum adhesion is modified to optimize focal adhesion-turnover and cell motility. In normal epithelial cells, tight junctions and desmosomes associate with cytoskeleton intermediate filaments to maintain cell shape and polarity. Critical to maintenance of these contacts is Ca2+dependent homophilic binding of the cell-adhesion molecule E-cadherin in the zones of adhesion [61].Germline mutations in the E-cadherin gene have been found in cases of familial gastric cancer. In sporadic carcinomas of most tissues, either Ecadherin expression or its function is suppressed. Forced over-expression of E-

56.5 Cell Migration

931

cadherin in tumor cells suppresses tumor growth in mice [62, 631. The cell adhesion receptor CD44 binds hyluronic acid, and its over-expression also suppresses tumor progression [64]. E-cadherin and CD44 were subject to GlcNAc-TIII-dependent Nglycosylation in transfected B16 cells [65, 661 and levels of E-cadherin increased. Therefore, an inactivating mutation, reduced gene expression or glycosylation of the gene product can potentially suppress E-cadherin activity, and thereby enhance tumor cell growth autonomy. CD44 activity is glycosylation-dependent, as enzymatic removal of sialic acid and galactose from complex-type N-glycans o f CD44 enhanced binding to hyaluronate [ 111. GlcNAc-TI11 transfection of B16 melanoma cells enhanced CD44-mediated cell adhesion [66]. These studies suggest that GlcNAc-TI11 and GlcNAc-TV may regulate several cell adhesion receptors and thereby coordinate cell motility. Metastatic tumor cells must migrate over extracellular matrix (ECM) to make their escape from the primary tumor, then gain access to the blood stream, attach to the distal vascular bed, migrate over ECM and grow in the secondary organ [67]. Tumors of various origins differ greatly in phenotypes that affect metastasis, and various steps in the metastatic process can be rate limiting for metastasis depending on the tumor phneotype. Cell migration on ECM may be rate limiting in both escape from the primary tumor and seeding at the secondary site. The integrin receptors mediate attachment to substratum, aggregating into focal adhesions when bound to the ECM glycoproteins fibronectin, laminin and collagen [68]. In growtharrested non-transformed cells, ligand-engaged integrins form stable adhesion plaques with links to actin stress fibers inside the cell. In either growth factor stimulated cells or transformed cells, focal adhesions induce continuous recruitment of signaling complexes on the cytosolic side of the plasma membrane (Figure 2). This results in turnover of the actin microfilaments, and activation of the RAS/MAPK [69] and PI3K/PKB [70] signaling pathways. Cell migration rates depend upon optimal turnover of focal adhesion complexes. Integrin receptor levels and activity are balanced against ECM adhesion domains to regulate focal adhesion turnover and thereby migration rates. Substratum density changes can vary reciprocally to integrin levels or their affinities [71]. As such, rates of focal-adhesion turnover and cell motility exhibit bell-shaped responses to changing ligand or receptor levels. Focal adhesion turnover also stimulates intracellular signaling and cell proliferation via PI3K and c-Src kinases, creating a positive feedback loop (Figure 2). Transfection of Mvl Lu epithelial cells with a GlcNAc-TV expression vector resulted in loss of contact inhibition of growth, enhanced cell motility, and morphological transformation. More importantly, the GlcNAc-TV expressing cells formed tumors when injected into nude mice, while none were observed for the control cell lines [72]. Breast tumor progression and metastasis have recently been examined in GlcNAc-TV deficient mice. MyatSPl- mice were crossed with transgenic mice expressing the polyomavirus middle T oncogene under the control of the mouse mammary tumor virus long terminal repeat (MMTV-PyMT) in breast epithelium. PyMT transgenic mice develop multifocal breast tumors, which metastasize to the lung [73]. The PyMT viral oncogene activates c-Src, as well as the Shc/ RAS and PI3K/PKB pathways, each of which contributes to rapid carcinoma formation [74]. Breast carcinomas develop in MqutS-1- mice with a longer latency and

56 Protein Glycosylation und Cancer

932

E

F

h

F5

Mgat5-/-

c I 0: o.oo1

. -. ......

I

0.010

.

*

..

Detached I...,

0.700

a-CD3 (Fglml)

1

*

-1

0

Substratum

Figure 3. Focal adhesions in (A,B) embryonic fibroblasts and (C,D) tumor cells removed from MMTV-PyMT mice. Cells were plated on fibronectin in serum-free medium. Fibroblasts were stained with rhodamine-phalloidin, FITC conjugated anti-paxillin antibodies, and Hoechst 33258 stain. Tumor cells were stained with rhodamine-phalloidin and FITC conjugated anti-vinculin antibodies. Fluorescence images of the cells were obtained using a decovolution microscope. Paxillin and vinculin (green) are localized to ends of microfilaments (red) in focal adhesions. MgatS+'- (A) fibroblasts and ( C ) tumor cell show focal adhesions but these structures are absent in Mgat5-i- (B) fibroblasts and (D) tumor cells. (E), T cell receptor dependent stimulation measured by 3H-thymidine incorporation in response to anti-CD3 antibodies at 48h (F) Model of responses to variable substratum adhesions for MgatS+/- and Mgat5-l- cells.

tumors grow more slowly. In addition, the incidence of lung metastases are reduced ten-fold [ 1271. A similar 2-3-fold delay in tumor progression is observed in MMTVPyMT mice on a Grb2'i- [75] genetic background, which causes reduced levels of signaling through the RAS pathway (Figure 2). Transgenic mice expressing mutated forms of the PyMT gene, lacking either the Shc or the PI3K binding domains show a delay in tumor progression similar to that of wild-type PyMT on the MgatS-i- genetic background [76]. Lymphocyte infiltration into the tumor was similar in both MgatSili and MgatS-1- mice, suggesting a similar host immune response. In addition, MgatS+i+ embryonic stem cells (ES) grew as teratocarcinomas when the cell were injected subcutaneously into either MgatS'i+ or MgatS-1mice. Therefore, slower tumor growth in Mgat5-i- mice does not appear to be due to superior host immunity in the mutant mice. Mgat5-i- breast tumor cells removed from the mice and placed in tissue culture showed impaired focal adhesion formation in low serum conditions (Figure 3). By activating c-Src/FAK, Shc/RAS and P13K, the PyMT protein contributes to the reorganization of cytoskeletal structures and turnover of focal-adhesion contacts

56.6 Sialylution arid Metastasis

933

(Figure 3C,D). The same impairment of focal adhesions was observed for the nontransformed A4gat.Y- embryonic fibroblasts when cultured in low serum conditions (Figure 3A,B). Fibroblasts deficient in focal adhesion kinase (FAK) are similarly impaired in focal-adhesion turnover [77]. Wild type cells spread more extensively and pseudopodia showed fine actin microfilaments, while A4gat.Y- cells had coarse microfilament stress fibers characteristic of non-motile cells. In addition, phosphorylation of PKB, a kinase in the PI3K pathway was reduced in A 4 g ~ z t S - l ~ fibroblasts relative to wild type cells when cultured in low serum conditions. The cytosolic protein paxillin binds pl integrin, as well as signaling molecules FAK, CSK and c-Src at discrete sites of cell attachment in spreading and motile cells 1781. In wild type cells, paxillin localized to the distal ends of microfilaments characteristic of focal adhesions, but in MgatSplp fibroblasts, the protein was dispersed in the cytoplasm. Consistently with these histochemical observations, leukocyte migration into the peritoneum in vivo was delayed in MgatS-1- mice following injection of the adjuvant thioglycolate [ 1261. Leukocytes from Mgat5-lp mice attached more avidly to fibronectin coated plastic than wild type cells. These observations show that GlcNAc-TV-dependent glycosylation regulates cell motility of both normal leukocytes and metastasic tumor cells in mice. The Nglycan may reduce the stability of integrin receptor aggregates that maintain firm and stable cell-substratum attachment in resting cells, and thereby facilitate cell motility. For the T cell receptor, and possibly other receptor systems where the resting state is free of ligand, the pl-6GlcNAc-branched N-glycans may act in a similar manner to attenuate receptor aggregation, thereby preventing or reducing spurious activation in the absence of high-affinity ligands (Figure 3E,F). The glycans may impede protein-protein interactions by steric hindrance. Alternatively, multivalent lectins, such as the galectins, bind p1-6GlcNAc-branched N-glycans and form lattices that impede receptor aggregation. Mvl Lu epithelial cells transfected with a GlcNAc-TV expression vector were also observed to be less adhesive on fibronectin and to be more motile. The GlcNAc-TV substrates in the MvlLu cells included integrins u5, u, and p,. p l integrins in S115 mammary carinoma cells [79] and Sezary syndrome T cells are also glycosylated with p1,6GlcNAc-branched N-glycans [80].

56.6 Sialylation and Metastasis Terminal N- and O-glycan sequences added in the truns-Golgi compartment contribute to malignancy, as suggested by studies on tumor cell glycosylation mutations. MDAY-D2 mutants over-expressing a2,6SA-T [81] due to a retroviral insertion into the gene promoter, showed 3-10-fold fewer metastases and 60% slower tumor growth [128]. The mutant cells had predominantly u2,6SA rather than the wild-type a2,3SA on the cell surface 1811. Transfection of a glioma cell line, U373 MG cells with u2,6SA-T reduced invasion [82]. However, a2,6-sialylation in HRAS-transformed rat fibroblasts correlates positively with invasive potential [ 831,

934

56 Protein Glycosylution and Cancer

and loss of sialylation in B16 melanoma mutants due to over-expression of al,3Fuc-T results in loss of metastatic potential [38]. Similarly, metastatic properties of B16 melanoma cells were reduced in cells transfected with either al,2-Fuc-T or al,3Gal-T [ 841. CMP-NeuNAc-hydroxylase expressing mutants of MDAY-D2 also grew more slowly as solid tumors [35, 851. These results imply that capping of N-acetyllactosamine with N-acetylneuraminic acid in a a2-3 linkage may be conducive to efficient metastasis.

56.7 Endogenous Lectins and Tumor Cell Adhesion The GlcNAc-TV enzyme product in the N-glycan pathway 1861 and core 2 GlcNAcT product of the 0-glycan pathway [87] are preferred intermediate for extension with polylactosamine (i.e. Galpl,4GlcNAc~l,3repeating units of 2 to >10 in length). Polylactosamine adds heterogeneity in the form of polymer length and also capping with various sequences including the Lewis antigens. Galectins binding to N-acetyl-lactosamine, are widely expressed 1881, and have been implicated in tumor cell adhesion during metastatic spread. Galectins 1 and 3 are expressed on the surface of B16 melanoma, the UV-2237 fibrosarcoma and the K-1735 melanoma cells, and have previously been shown to facilitate organ colonization and metastasis by blood-borne tumor cells [89]. Intravenous infusion of Gal or arabinogalactan inhibited liver colonization by murine tumor cells, presumably by blocking their retention to the microvasculature [90]. Restoring Gal to the surface of UDP-Gal transporter mutants, using bovine Dl ,4Gal-T, increased tumor cell adhesion to nonactivated endothelial cells and enhanced metastasis of the mutants in mice [91]. Genetic revertants of the UDP-Gal transporter mutation also regain the malignant phenotype [92]. UDP-Gal transporter mutants lack polylactosamine in both 0-.and N-glycans, and show the most severe attenuation of tumor growth and metastasis. The Lewis carbohydrate antigens Le", Sialyl-Le", Ley and SLe" are often overexpressed in human carcinomas [93] and have been shown to mediate attachment of colon tumor cells to selectins in vitro [94]. Polylactosamine forms the backbone of the dimeric Le" and Ley antigens, and dimeric Le sequences correlate with a poor prognosis in colon cancers 1951. E- and P- selectins on endothelial cells bind SLe" and related sequences found on PSGL-1, ESL-1, GlyCAM-1, CD43 and MadCam in leukocytes (reviewed in [96]. The SLe" sequences are found on core 2 GlcNAcbranched 0-glycan [97]. Core 2 GlcNAc-T transcripts increase in human colon carcinomas 1981, as do Fuc-TIV and ST3Gal I1 [99]. These changes in gene expression may contribute to SLeX production on tumor cells. Indeed, inhibition of GalNAc-Ser/Thr elongation by treating tumor cells with benzyl-a-GalNAc reduces organ colonization [ 1001. Mice lacking al,3Fuc-TVII are deficient in leukocyte extravasation into areas of inflammation, a phenotype similar to P- and E- selectin deficient mice and to LADII pathology in humans [ 1011. Unlike neutrophils, carcinoma cells may not express leukocyte cell surface glycoproteins (e.g. PSGL-l), however selectin ligands

56.8 Carbohydrate Processing Inhibitors us Anti-Cancer Agents

935

on other glycoproteins may contribute to metastasis in vivo. Blood-borne tumor cells can aggregate with platelets and leukocytes, which may express L-selectins and can mediate attachment indirectly. Forced expression of E-selectin in the liver of transgenic mice enhanced metastasis of SLeX-expressing B 16F10 melanoma cells to the liver, rather than to their usual destination of the lung [102]. However, it is unclear that the attachment of blood-borne tumor cells to endothelium is a ratelimiting step in clinical metastasis for most tumors. Clinical studies on patients with ascites tumors treated with peritoneal shunts to maintain their salt balance push millions of peritoneal tumor cells into the circulation, but this does not significantly increase the number of metastases observed at autopsy [ 1031. Carcinomas do not induce clinically meaningful anti-tumor immune responses, even though “tumor antigens” may be present. Tumors also produce cytokines such as TGF-P and IL-10 that act to suppress cell-mediated immunity or T helper cells (i.e. Thl). LeX and polylactosamine found in Schistosoma mansoni eggs has been shown to suppress T cell helper (Thl) response required to clear the infection. Susceptible mice mount the ineffective humoral Th2 response: which leads to tissue damage and continued infection [ 1041. Thus, it is suggested that Lewis sequences and polylactosamine on tumor cells may also suppress Thl cellular immunity and T cell production of INF-)I and IL-2 cytokines that can reduce the growth of some tumors. Ironically, an intron of the GlcNAc-TV gene encodes a transcript detected in tumor cells but not normal tissues, producing a widely occurring tumorassociated antigen. This peptide sequence was recognized by tumor-infiltrating T lymphocytes in an HLA dependent manner and tumor cells displaying it were lysed by cytotoxic T cells. The antigen was expressed in 50%) of human melanomas, and also observed in tumors of other tissues [105].

56.8 Carbohydrate Processing Inhibitors as Anti-Cancer Agents The alkaloids swainsonine and castanospermine block tumor cell metastasis and invasion through extracellular matrix in vitro [ 106). Swainsonine is a competitive inhibitor of Golgi w-mannosidaseII, which blocks the N-glycan biosynthetic pathway prior to PI ,6GlcNAc-branching, and results in production of hybrid-type glycans. Swainsonine-treated cells, as well as GlcNAc-TV deficient mutant cells, showed increased transcription rates for tissue inhibitor of metalloproteinases (TIMP-1) [ 1071. Swainsonine also suppresses MMP-2 expression in human tumor cells [ 1081, a metalloproteinase associated with cancer progression in humans [ 1091. Swainsonine has been particularly useful for anti-cancer studies due to its apparent lack of toxicity (reviewed in [IlO]. Swainsonine has been tested in two phase I clinical trials with encouraging results that show both low toxicity and evidence of clinical responses. In the first trial, patients were given swainsonine by continuous intravenous infusion over 5 days [ 1 1 1, 1 121. Dose levels from 50-550 pg/kg per day were administered to 19 patients with advanced cancers for a total of 31 courses of treatment. Common side effects included peripheral edema ( n = 11/19), mild liver

936

56 Protein Glycosylation and Cancer

dysfunction in all patients (AST up to four-fold normal) and a rise in serum amylase ( n = 8/ 19). One patient with head and neck cancer showed >50% tumor shrinkage, and two patients with lymphangitis carcinomatosis on chest X-ray showed symptomatic improvement during the infusion of swainsonine and for a week thereafter. Clearance and serum half-life for swainsonine were determined to be approximately 2 ml/h.Kg, and 0.5 days, respectively. Chronic oral administration of swainsonine in 16 cancer patients was also well tolerated at 150 pg/kg per day given twice weekly, and similar rises in liver enzymes as well as fatigue were also reported as toxicities. Phase I1 trials designed to measure efficacy of oral swainsonine treatment in renal cell carcinoma are currently being done. Swainsonine should be considered a first generation compound, with room for improvement. Swainsonine inhibits lysosomal a-mannosidases with potency equal to that of the Golgi a-mannosidaseI1, and therefore induces lysosomal storage. Complex-type N-glycan biosynthesis can proceed via alternate pathways that circumvent a-mannosidase I1 and the swainsonine block [ 1131. Somatic tumor cell mutants with a deficiency in UDP-Gal transport activity show the most severe attenuation of malignancy, suggesting that a blocker of lactosamine extension in N- and 0-glycans may be effective. Inhibitors of polylactosamine extension, p 1,6GlcNA~-branching in the N - and 0-glycosylation pathways, or some combinations thereof may be potent anti-cancer agents. Further cancer studies with mutant mice lacking specific glycosyltransferase gene will increase our understanding of the relative importance of these glycan structures to cancer growth and metastasis, and should provide direction to drug development efforts.

56.9 Other Considerations pl,6GlcNAc-branched carbohydrates are added to multiple glycoproteins, but those that affect cell migration and proliferation are not well defined. Once these glycoproteins are identified, it will be important to determine how the branched glycans affect receptor function and intracellular signaling. However, current information suggests that regulated changes in glycosylation may integrate signaling via multiple receptor systems to co-ordinate cellular phenotypes. Leukocyte migration into sites of inflammation in the MaytS-1- mice was reduced but the cells were more adhesive on fibronectin. Similarly, MagtS-l- fibroblasts and tumor cells attached firmly to substratum, but showed impaired focal-adhesion formation in cell culture. In addition to regulation of cell adhesion, studies on T cells show that b1,6GlcNAc-branched glycans dampen agonist-dependent T cell receptor aggregation and signaling. Regulated expression of b 1,6GlcNAc-branched glycan structures may accelerate dissociation and turnover of certain glycoprotein receptors and thereby alter the activity of multiple receptor systems to produce an integrated cellular response (Figure 3). Certainly, glycosylation as an integrator of signaling via multiple ligand-receptor systems may also involve other tissue-specific patterns of glycosylation. At a molecular level, bulky fi 1,6GlcNAc-branched glycan chains

with polylactosamine extensions may inhibit glycoprotein receptors (e.g. T cell receptor), slow their movement in the plane of the membrane, and impede liganddependent aggregation. The distal portions of N - and 0-glycan structures are flexible about the glycosidic linkages, more so for 1,6 linkages. Therefore, N - and 0-glycans are not well resolved in most X-ray crystal structures of glycoproteins, and their flexible nature is suggested to often impede glycoprotein crystallization [ 1141. The relative mobility of glycan, compared to surrounding protein surface, may impede protein-protein interactions, either alone or through cis-mediated interactions with endogenous lectins on the cell surface. 1,4GlcNA~-branchingof the trimannosyl core is increased in choriocarcinoma glycoproteins, and might also contribute this phenotypes [ 1 151. Finally, it is possible that tissue-specific glycosylation and glycan functions may present certain common themes in metazoans. The recent completion of the Cuenorhabditis eleyans genome [116], suggests genes involved in N- and 0glycoconjugate biosynthesis are conservation with that of mammals. Mammalian and C. eleguns GlcNAc-TV genes share catalytic specificity and the Golgi localization signal. More importantly, the C. eleyans gene is a functional homolog of the mammalian gene, as it complements the Lec4 mutation in CHO cells [129]. The phylogenetic conservation of medial- and trans-Golgi processing enzymes suggests that complex-type glycans may regulate signaling pathways with equally early origins (e.g. RAS). Both GlcNAc-TV catalytic specificity and the essential elements of the RAS signaling pathway are conserved in worms and mammals. Furthermore, RAS signaling regulates Myat5 gene transcription. Therefore, it will be instructive to determine whether specific glycan structures function in signaling pathways conserved between worms and mammals. This is a testable hypothesis worth pursuing as a means determining glycan functions in other cellular contexts. Acknowledgments The authors thank NCI of Canada, the Mizutani Foundation, The National Science and Engineering Research Council of Canada, and GlycoDesign Inc., Toronto for research grants awarded to JWD. References 1. !%-I. Hakomori, Tumor malignancy defined by aberrant glycosylation and sphingo(g1yco)lipid metabolism. Cancer R e x , 56 (1996) 5309-5318. 2. A. Kobata, A retrospective and prospective view of glycopathology. Glycoconj. J., 15 (1998) 323-33 1. 3. T. Feizi, Demonstration by monoclonal antibodies that carbohydrate structures of glycoproteins and glycolipids are onco-developmental antigens. Nature, 314 (1985) 53-57. 4. H. Kitagawa and J.C. Paulson, Differential expression of five sialyltansferase genes in human tissues. J. Biol. Chem., 269 (1 994) 17872- 17878. 5. B. Rajput, N.L. Shaper and J.H. Shaper, Transcriptional regulation of murine betal,4galactosyltransferase in somatic cells. Analysis of a gene that serves both a housekeeping and a mammary gland-specific function. J. Bid. Chem., 271 (1996) 131-142.

938

56 Protein Glycosylation and Cancer

6. X. Wang, T.P. O’Hanlon, R.F. Young and J.T.Y. Lau, Rat P-galactoside a2,6-sialyltransferase genomic organization: alternative promoters direct the synthesis of liver and kidney transcripts. Glycobiology, 1 (1990) 25-31. 7. T.W. Rademacher, B. Parekh and R.A. Dwek, Glycobiology. Annu. Reu Biochem., 57 (1988) 785-838. 8. H. Schachter, Biosynthetic controls that determine the branching and microheterogeneity of protein-bound oligosaccharides. Biochem. Cell Biol., 64 (1986) 163-1 8 1. 9. D. Fiete, V. Srivastava, 0. Hindsgaul and J.U. Baenziger, A hepatic reticuloendothelkal cell receptor specific for S04-4GalNAcPI ,4GlcNAcP1,Manu that mediates rapid clearance of lutropin. Cell, 67 (1991) 1103-1110. 10. M. Takeuchi and A. Kobata, Structures and functional roles of the sugar chains of human erythropoietins. Glycobiology, 1 (1991) 337-346. 11. T.P. Skelton, C. Zeng, A. Nocks and I. Stamenkovic, Glycosylation provides both stimulatory and inhibitory effects on cell surface and soluble CD44 binding to hyaluronan. J. Cell Biol., 140 (1998) 431-446. 12. M.S. Diamond, D.E. Staunton, S.D. Martin and T.A. Springer, Binding of the integrin Mac-I (CDI lb/CD18) to the third immunoglobulin-like domain of ICAM-I (CD54) and its regulation by glycosylation. Cell, 65 (1991) 961-971. 13. L.M. McEvoy, H. Sun, J.G. Frelinger and E.C. Butcher, Anti-CD43 inhibition of T cell homing. J. Exp. Med., 185 (1997) 1493-1498. 14. J. Sandros, E. Rozdzinski, J. Zheng, D. Cowburn and E. Tuomanen, Lectin domains in the toxin of Bordetella pertussis: selectin mimicry linked to microbial pathogenesis. Glycoconj. J., 11 (1994) 501-506. 15. T.J. Pritchett, R. Brossmer, U. Rose and J.C. Paulson, Recognition of monovalent sialosides by influenza virus H3 hemagglutinin. Virology, 160 (1987) 502-506. 16. J.W. Dennis, M. Granovsky and C.E. Warren, Protein Glycosylation in Development and Disease. BioEssays, 21 (1999) 412-421. 17. M. Metzler, A. Gertz, M. Sarkar, H. Schachter, J.W. Schrader and J.D. Marth, Complex asparagine-linked oligosaccharides are required for morphogenic events during postimplantation development. EMBO J., 13 (1994) 2056-2065. 18. E. Ioffe and P. Stanley, Mice lacking N-acetylglucosaminyltransferase I activity die at midgestation, revealing an essential role for complex or hybrid N-linked carbohydrates. Proc. Natl Acad. Sci. USA, 91 (1994) 728-732. 19. M. Asano, K. Furukawa, M. Kido, S. Matsumoto, Y. Umesaki, N. Kochibe and Y. Iwakura, Growth retardation and early death of beta-l,4-galactosyltransferaseknockout mice with augmented proliferation and abnormal differentiation of epithelial cells. EMBO J., 16 (1997) 1850-1857. 20. J.J. Priatel, M. Sarkar, H. Schachter and J.D. Marth, Isolation, characterization and inactivation of the mouse Mgat3 gene: the bisecting N-acetylglucosamine in asparagine-linked oligosaccharides appears dispensable for viability and reproduction. Glycohiology, 7 (1997) 4556. 21. S.P. Hussain and C.C. Harris, Molecular epidemiology of human cancer: contribution of mutation spectra studies of tumor suppressor genes. Cancer Rex, 58 (1998) 4023-4037. 22. C.J. Sherr, Cancer cell cycles. Science, 274 (1996) 1672Z1677. 23. J. Cairns, The origin of human cancers. Nature, 289 (1981) 353-358. 24. C. Lengauer, K.W. Kinzler and B. Vogelstein, Genetic instability in colorectal cancers. Nature, 386 (1997) 623-627. 25. G. Poste and I.J. Fidler, The pathogenesis of cancer metastasis. Nature, 283 (1980) 139-146. 26. K.W. Kinzler and B. Vogelstein, Cancer-susceptibilty genes. Gatekeepers and caretakers. Nature, 386 (1997) 761-763. 27. W.M. Grady, L.L. Myeroff, S.E. Swinler, A. Rajput, S. Thiagalingam, J.D. Lutterbaugh, A. Neumann, M.G. Brattain, J.k.S.J. Chang, K.W. Kinzler, B. Vogelstein, J.K. Wilson and S. Markowitz, Mutational inactivation of transforming growth factor beta receptor type 11 in microsatellite stable colon cancers. Cancer Res., 59 (1999) 320-324. 28. C.V. Dang and G.L. Semenza, Oncogenic alterations of metabolism. Tends Biochem. Sci., 24 (1999) 68-72.

References

939

29. T. Jacks, Tumor suppressor gene mutations in mice. Annu. Rev. Genet., 30 (1996) 603-636. 30. M.A. Webster and W.J. Muller, Mammary tumorigenesis and metastasis in transgenic mice. Semin. Cuncer B i d , 5 (1994) 69-16. 31. K . Yamashita, Y. Tachibana, T. Ohkura and A. Kobata, Enzymatic basis for the structural changes of asparagine-linked sugar chains of membrane glycoproteins of baby hamster kidney cells induced by polyoma transformation. J. B i d Chetn., 260 ( 1 985) 3963-3969. 32. M. Pierce and J. Arango, Rous sarcoma virus-transformed baby hamster kidney cells express higher levels of asparagine-linked tri- and tetraantennary glycopeptides containing [GlcNAc-p (1,6)Man-a(1,6)Man] and poly-N-acetyllactosamine sequences than baby hamster kidney cells. J. Biol. Chem., 261 (1986) 10772-10777. 33. E.W. Easton, J.G.M. Bolscher and D.H. van den Eijnden, Enzymatic amplification involving glycosyltransferases forms the basis for the increased size of asparagine-linked glycans at the surface of NIH 3T3 cells expressing the N-rus proto-oncogene. J. Biol. Chem., 266 (1991) 21 674-21 680. 34. Y. Lu and W. Chaney, Induction of N-acetylglucosaminyltransferase V by elevated expression of activated or proto- Ha-ras oncogenes. Mol. Cell. Biochem., 122 (1993) 85-92. 35. J.W. Dennis, Different metastatic phenotypes in two genetic classes of WGA-resistant tumor cell mutants. Cuncer Rex, 46 (1986) 4594-4600. 36. J.W. Dennis, S. Laferte, C. Waghorne, M.L. Breitman and R.S. Kerbel, fi 1-6 branching of Asn-linked oligosaccharides is directly associated with metastasis. Science, 236 (1987) 582585. 37. Y. Lu, J.C. Pelling and W.G. Chaney, Tumor cell surface beta 1-6 branched oligosaccharides and lung metastasis. Clin. Exp. Metustusis, 12 (1994) 47-54. 38. J. Finne, M.M. Burger and J.P. Prieels, Enzymatic basis for a lectin resistant phenotype: increase in a fucosyltransferase in mouse melanoma cells. J. Cell Biol., 92 (1982) 277-282. 39. B. Wasylyk, J. Hagman and A. Gutierrez-Hartmann, Ets transcription factors: nuclear effectors of the Ras-MAP-kinase signaling pathways. Trends Biochem. Sci, 23 (1998) 213-216. 40. E. Saez, S.E. Rutberg, E. Mueller, H. Oppenheim, J. Smoluk, S.H. Yuspa and B.M. Spiegelman, c-fos is required for malignant progression of skin tumors. Cell, 82 (1995) 721732. 41. W. Risau, Mechanisms of angiogenesis. Nature, 386 (1 997) 67 1-674. 42. A. Rebbaa, H. Yamamoto, T. Saito, E. Meuillet, P. Kim, D.S. Kersey, E.G. Bremer, N. Taniguchi and J.R. Moskal, Gene transfection-mediated overexpression of beta 1,4-Nacetylglucosamine bisecting oligosaccharides in glioma cell line U373 MG inhibits epidermal growth factor receptor function. J. Bid. Chem., 272 (1997) 9275- 9279. 43. Y. Ihara, Y. Sakamoto, M. Mihara, K. Shimizu and N . Taniguchi, Overexpression of Nacetylglucosaminyltransferase TIT disrupts the tyrosine phosphorylation of Trk with resultant signaling dysfunction in PC12 cells treated with nerve growth factor. J. Biol. Chem., 272 ( 1 997) 9629-9634. 44. P.B.O. Tan and S.K. Kim, Signaling specificity: the RTK/RAS/MAP kinase pathway in metazoans. Trends Genet., 15 (1999) 145-149. 45. S. Lemaire, C. Derappe, J.C. Michalski, M. Aubery and D. Neel, Expression of pl-6branched N-linked oligosaccharides is associated with activation in human T4 and T8 cell populations. J. Bid. Chem., 269 (1 994) 8069-8074. 46. P. Buckhaults, L. Chen, N. Fregien and M. Pierce, Transcriptional regulation of Nacetylglucosaminyltransferase V by the .src Oncogene. J. Bid. Chem.. 272 (1997) 1957519581. 47. G.N. Vespa, L.A. Lewis, K.R. Kozak, M. Moran, J.T. Nguyen, L.G. Baum and M.C. Miceli, Galectin-1 specifically modulates TCR signals to enhance TCR apoptosis but inhibit IL-2 production and proliferation. J Immunology, 162 (1999) 799-806. 48. T. Hennet, D. Chui, J.C. Paulson and J.D. Marth, Immune regulation by the ST6Gal sialytransferase. Proc. Nut1 Acad. Sci. USA, 95 (1998) 4504-4509. 49. S. Cumberledge and F. Reichsman, Glycosaminoglycans and WNTs: just a spoonful of sugar helps the signal go down. Trends Genet., 13 (1997) 421-423. 50. H. Saito, J. Gu, A. Nishikawa, Y. Ihara, J. Fujii, Y. Kohgo and N. Taniguchi, Organization of the human N-acetylglucosaminyltransferase V gene. Eur. J. Biochem., 233 (1995) 18-26.

940

56 Protein Glycosylution and Cuncer

51. R. Kang, H. Saito, Y. Ihard, E. Miyoshi, N . Koyama, Y. Sheng and N . Taniguchi, Tran-

52. 53.

54. 55. 56. 57. 58. 59. 60.

61. 62. 63. 64.

65. 66.

67. 68. 69. 70.

scriptional regulation of the N-acetylghxosaminy~transerase V gene in human bile duct carcinoma cells (HuCC-TI) is mediated by Ets-1. J. Biol. Chern., 271 (1996) 26706-26712. K. Yamashita, T. Ohkura, Y. Tachibana, S. Takasaki and A. Kobata, Comparative study of the oligosaccharides released from baby hamster kidney cells and their polyoma transformant by hydrazinolysis. J. Biol. Ckem., 259 (1984) 10834-10840. J.W. Dennis, K. Kosh, D.-M. Bryce and M.L. Breitman, Oncogenes conferring metastatic potential induce increased branching of Asn-linked oligosaccharides in rat2 fibroblasts. Oncogene, 4 (1989) 853-860. P.J. Seberger and W.G. Chaney, Control of metastasis by asn-linked, PI-6 branched oligosaccharides in mouse mammary cancer cells. Glycohiology, 9 (1999) 235-241. B. Fernandes, U. Sagman, M. Auger, M. Demetriou and J.W. Dennis, pl-6 branched oligosaccharides as a marker of tumor progression in human breast and colon neoplasia. Cancer Res., 51 (1991) 718-723. W.K. Seelentag, W.P. Li, S.F. Schmitz, U. Metzger, P. Aeberhard, P.U. Heitz and J. Roth, Prognostic value of beta 1,6-branched oligosaccharides in human colorectal carcinoma. Cancer Res., 58 (1998) 5559-5564. S. Narasimhan, H. Schachter and S. Rajalakshmi, Expression of N-acetylglucosaminyltransferase I11 in hepatic nodules during rat liver carcinogenesis promoted by orotic acid. J. Cell B i d , 263 (1988) 1273-1281. E. Miyoshi, A. Nishikawa, Y. Ihara, J. Gu, T. Sugiyama, N. Hayashi, H. Fusamoto, T. Kamada and N. Taniguchi, N-acetylglucosaminyltransferase 111 and V messenger RNA levels in LEC rats during hepatocarcinogenesis. Cancer Rex, 53 (1993) 3899-3902. M. Yoshimura, A. Nishikawa, Y. Ihara, S. Taniguchi and N. Taniguchi, Suppression of Iund metastasis of B16 mouse melanoma by N-acetylglucosaminyltransferase I1 gene transfection. Proc. Nut1 Acad. Sci. USA, 92 (1995) 8753-8758. M. Bhaumik, T. Harris, S. Sundaram, L. Johnson, J. Guttenplan, C. Rogler and P. Stanley, Progresson of hapatic neoplasms in severely retarded in mice lacking the bisecting Nacetylglucosamine on N-glycans: evidence for a glycoprotein factor that facilitates hepatic tumor progression. Cancer Rex, 58 (1998) 2881L2887. G. Christofori and H. Semb, The role of the cell-adhesion molecule E-cadherin as a tumoursuppressor gene. Trends Biochem. Sci., 24 (1999) 73-76. U.H. Frixen, J. Behrens, M. Sachs, G. Eberle, B. Voss, A. Warda, D. Lochner and W. Birchmeier, E-Cadherin-mediated Cell-Cell Adhesion Prevents Invasiveness of Human Carcinoma Cells. J. Cell Biol., 113 (1991) 173-185. W. Birchmeier, E-cadherin as a tumor (invasion) suppressor gene. BioEssays, 17 (1995) 97-99. R. Schmits, J.G.N. Filmus, G. Senaldi, F. Kiefer, T. Kundig, A. Wakeham, A. Shahinian, C. Catzavelos, J. Rak, C. Furlonger, A. Zakarian, J.J.L. Simard, P.S. Ohashi, C.J. Paige, J.C. Guiterrez-Ramos and T.W. Mak, CD44 regulates hematopoietic progenitor distribution, granuloma formation, and tumorigenicity. Blood, 90 (1997) 221 7-2233. M. Yoshimura, Y. Ihara, Y. Matsuzawa and N. Taniguchi, Aberrant glycosylation of Ecadherin enhances cell-cell binding to suppress metastasis. J. Biol. Ckem., 271 (1996) 1381l13815. Y. Sheng, M. Yoshimura, S. Inoue, K. Oritani, T. Nishiura, H. Yoshida, M. Ogawa, Y. Okajima, Y. Matsuzawa and N. Taniguchi, Remodeling of glycoconjugates on CD44 enhances cell adhesion to hyaluronate, tumor growth and metastasis in B16 melanoma cells expressing Dl ,4-N-acetylglucosaminyltransferase 111. Znt. J. Cancer, 73 (1997) 850-858. G. Nicolson, Organ specificity of tumor metastasis: role of preferential adhesion, invasion and growth of malignant cells at specific secondary sites. Cancer Metastasis Rev., 7 (1988) 143188. K. Burridge and M. Chrzanowska-Wodnicka, Focal adhesions, contractility, and signaling. Annu Rev Cell Dev Biol, 12 (1996) 463-518. D.D. Schlaepfer, S.K. Hanks, T. Hunter and P. van der Geer, Integrin-mediated signal transduction linked to Ras pathway by GRB2 binding to focal adhesion kinase. Nature, 372 (1995) 786-791. Shaw LM, Rabinovitz I, Wang HH, Toker A and Mercurio AM, Activation of phosphoino-

References

941

sitide 3-OH kinase by the alpha6beta4 integrin promotes carcinoma invasion. Cell, 91 (1997) 947 -960. 71. S.P. Palecek, J.C. Loftus, M.H. Ginsberg, D.A. Lauffenburger and A.F. Horwitz, Integrinligand binding properties govern cell migration speed through cell-substratum adhesiveness. Nature, 385 (1997) 537-540. 72. M. Demetriou, I.R. Nabi, M. Coppolino, S. Dedhar and J.W. Dennis, Reduced contactinhibition and substratum adhesion in epithelial cells expressing GlcNAc-transferase V. J. Cell Biol., 130 (1995) 383-392. 73. C.T. Guy, R.D. Cardiff and W.J. Muller, Induction of mammary tumors by expression of polyomavirus middle T oncogene: a transgenic mouse model for metastatic disease. Mol. Cell. Biol., 12 (1992) 954-961. 74. C.T. Guy, S.K. Muthuswamy, R.D. Cardiff, P. Soriano and W.J. Muller, Activation of the c-Src tyrosine kinase is required for the induction of mammary tumors in transgenic mice. Genes Dev., 8 (1994) 23-32. 75. A.M. Cheng, T.M. Saxton, R. Sakai, S. Kulkarni, G. Mbamalu, W. Vogel, C.G. Tortorice, R.D. Cardiff, J.C. Cross, W.J. Muller and T. Pawson, Mammalian Grb2 regulates multiple steps in embryonic development and malignant transformation. Cell, 95 (1999) 793-803. 76. M.A. Webster, J.N. Hutchinson, M.J. Rauh, S.K. Muthuswamy, M. Anton, C.G. Tortorice, R.D. Cardiff, F.L. Graham, J.A. Hassell and W.J. Muller, Requirement for both Shc and phosphatidylinositol 3’ kinase signaling pathways in polyomavirus middle T-mediated mammary tumorigenesis. Mol. Cell. Biol., 18 (1998) 2344-2359. 77. AnonymousProtein-tyrosine phosphatase Shp-2 regulates cell spreading, migration, and focal adhesion. J. Biol. Chem., 273 (1998) 21125-21131. 78. M.D. Schaller, C.A. Otey, J.D. Hildebrand and J.T. Parsons, Focal adhesion kinasc and paxillin bind to peptides mimicking beta integrin cytoplasmic domains. J Cell Biol, 130 (1995) 1181-1 187. 79. S. Leppa, J. Heino and M. Jalkanen, Increased glycosylation of beta 1 integrins affects the interaction of transformed SI 15 mammary epithelial cells with laminin-1. Cell Growth Dzffer., 6 (1995) 853-861. 80. F. Braut-Boucher, J. Font, J. Pichon, Y. Paulin, M. Boukhelifa, M. Aubery and C. Derappe, T lymphocytes from sezary syndrome patients express beta1 integrins whose beta( 1-6)branched N-linked oligosaccharides reflect their adhesive capacity. Leuk. Res., 22 (1998) 947952. 81. R. Takano, E.A. Muchmore and J.W. Dennis, Sialylation and malignant potential in tumor cell glycosylation mutants. Glycobiology, 4 (1994) 665-674. 82. H. Yamamoto, Y. Kaneko, A. Rebbaa, E.G. Bremer and J.R. Moskal, alpha2,6-sialyltransferase gene transfection into a human glioma cell line (U373 MG) results in decreased invasivity. J. Neurochem., 68 (1997) 2566-2576. 83. N. Le Marer and D. Stehelin, High alpha-2,6-sialylation of N-acetyllactosamine sequences in ras-transformed rat fibroblasts correlates with high invasive potential. Glycobiology, 5 (1995) 219-226. 84. E. Gorelik, F. Xu, T. Henion, F. Anaraki and U. Galili, Reduction of metastatic properties of BL6 melanoma cells expressing terminal fucose(alpha)l-2-galactose after alphal,2-fucosyltransferase cDNA transfection. Cancer Res., 57 (1997) 332-336. 85. L. Shaw, S. Yousefi, J.W. Dennis and R. Schauer, CMP-N-acetylneuraminic acid hydroxylase activity determines the wheat germ agglutinin-binding phenotype in two mutants of the lymphoma cell line MDAY-D2. Glycoconj. J. 8 (1991) 434-441. 86. R.D. Cummings and S. Kornfeld, The distribution of repeating Gal PI-4GlcNAc PI-3 sequences in asparagine-linked oligosaccharides of the mouse lymphoma cell line BW5147 and PHAR 2.1. J. Biol. Chem., 259 (1984) 6253-6260. 87. S. Yousefi, E. Higgins, Z. Doaling, 0. Hindsgaul, A. Pollex-Kruger and J.W. Dennis, Increased UDP-G1cNAc:Gal 0 1-3GalNAc-R (GlcNAc to GalNAc) P 1-6 N-acetylglucosaminyltransferase activity in transformed and metastatic murine tumor cell lines: control of polylactosamine-synthesis. J. Bid. Chem., 266 (1991) 1772-1783. 88. N.L. Perillo, M.E. Marcus and L.G. Baum, Galectins versatile modulators of cell adhesion, cell proliferation, and cell death. J. Mol. Med., 76 (1998) 402-412. ~

942

56 Protein Glycosylution and Cancer

89. D.W. Ohannesian, D. Lotan, P. Thomas, J.M. Jessup, M. Fukuda, H.-J. Gabius and R. Lotan, Carcioembryonic antigen and other glycoconjugates act as ligands for galectin-3 in human colon carcinoma cells. Cancer Rex, 55 (1995) 2191-2199. 90. J. Beuth, H.L. KO, K. Oette, G. Pulverer, K. Roszkowski and G . Uhlenbruck, Inhibition of liver metastasis in mice by blocking hepatocyte lectins with arabinogalactan infusions and Dgalactose. J. Cancer Rex Clin. Oncol., 113 (1987) 51-55. 91. I. Cornil, R.S. Kerbel and J.W. Dennis, Tumor cell surface pl-4 linked galactose binds to lectin(s) on microvascular endothelial cells and contributes to organ colonization. J. Cell Biol., 111 (1990) 773-782. 92. J.W. Dennis and S. Laferte, Co-reversion of a lectin-resistant mutation and non-metastatic phenotype in murine tumor cells. Znt. J. Cancer, 38 (1986) 445-450. 93. S.H. Itzkowitz, M. Yuan, Y. Fukushi, A. Palekar, P.C. Phelphs, A.M. Shamsuddin, B.F. Trump, S. Hakomori and Y.S. Kim, Lewis' and sialylated Lewisx-related antigen expression in human malignant and non-malignant colonic tissue. Cancer Rex, 46 (1986) 2627-2632. 94. G. Mannori, P. Crottet, 0. Cecconi, K. Hanasaki, A. Amffo, R.M. Nelson, A. Varki and M.P. Bevilacqua, Differential coloncancer cell adhesion to E-, P-, and L-selectin: role of mucin-type glycoproteins. Cancer Res., 55 (1995) 4425-4431. 95. S.D. Hoff, Y. Matsushita, D.M. Ota, K.R. Cleary, T. Yamori, S.-I. Hakomori and T. Irimura, Increased expression of sialyl-dimeric LeX antigen in liver metastases of human colorectal carcinoma. Cancer Rex, 49 (1989) 6883-6888. 96. T.A. Springer, Traffic signals for lymphocyte recirculation and leukocyte migration: the multistep paradigm. Cell, 76 (1994) 301-314. 97. F. Li, P.P. Wilkins, S. Crawley, J. Weinstein, R.D. Cummings and R.P. McEver, Posttranslational modifications of recombinant P-selectin glycoprotein ligand- 1 required for binding to P- and E-selectin. J. Biol. Chem., 271 (1996) 3255-3264. 98. K. Shimodaira, J. Nakayama, N. Nakamura, 0. Hasebe, T. Katsuyama and M. Fukuda, gene in Carcinoma-associated expression of core 2 0- 1,6-N-Acetylglucosaminyltransferase human colorectal cancer: Role of 0-glycans in tumor progression. Cancer Rex, 57 (1997) 520 1-5206. 99. H. Ito, N. Hiraiwa, M. Sawada-Kasugai, S. Akamatsu, T. Tachikawa, Y. Kasai, S. Akiyama, K. Ito, H. Takagi and R. Kannagi, Altered mRNA expression of specific molecular species of fucosyl- and sialyl-transferases in human colorectal cancer tissues. Int. J. Cancer, 71 (1999) 556-564. 100. R.S. Bresalier, Y. Niv, J.C. Byrd, Q.Y. Duh, N.W. Toribara, R.W. Rockwell, R. Dahiya and Y.S. Kim, Mucin production by human colonic carcinoma cells correlates with their metastatic potential in animal models of colon cancer metastasis. J. Clin.Invest., 87 (1991) 1037-1045. 101. P. Maly, A.D. Thall, B. Petryniak, C.E. Rogers, P.L. Smith, R.M. Marks, R.J. Kelly, K.M. Gersten, G. Cheng, T.L. Saunders, S.A. Camper, R.T. Camphausen, F.X. Sullivan, Y. Isogai, 0. Hindsgaul, U.H. von Andrian and J.B. Lowe, The a( 1,3)fucosyltransferase fuc-TVII controls leukocyte trafficking through an essential role in L-, E-, and P-selectin ligand biosynthesis. Cell, 86 (1996) 643-653. 102. L. Biancone, M. Araki, K. Araki, P. Vassalli and I. Stamenkovic, Redirection of tumor metastasis by expression of E-selectin in vivo. J. Exp. Med., 183 (1996) 581-587. 103. Z.A. Jamjoom, A.B. Jamjoom, A.H. Sulaiman and a.R.A. Naim-Ur-Rahman, Systemic metastasis of medulloblastoma through vetriculoperitoneal shunt report of a case and critical analysis of the literature. Surg. Neurol., 40 (1993) 403-410. 104. V. Palanivel, C.H.A.M. Posey, W. Solbach, W.F. Piessens and D.A. Harn, B-cel outgrowth and ligand-specific production of IL-10 correlate with Th2 dominance in certain parasitic diseases. Exp. Parasitol, 84 (1996) 168-177. 105. Y. Guilloux, S. Lucas, V.G. Brichard, A. Van Pel, C. Viret, E. De Plaen, F. Brasseur, B. Lethe, F. Jotereau and T. Boon, A peptide recognized by human cytolytic T lymphocytes on HLA-A2 melanomas is incoded by an intron sequence of the N-acetylglucosaminyltransferase V gene. J. Exp. Med., 183 (1996) 1173-1 183. 106. S. Yagel, R. Feinmesser, C. Waghorne, P.K. Lala, M.L. Breitman and J.W. Dennis, Evidence that p 1-6 branched Asn-linked oligosaccharides on metastatic tumor cells facilitate invasion of basement membranes. Int. J. Cancer, 44 (1989) 685-690. 107. B. Korczak and J.W. Dennis, Inhibition of N-linked oligosaccharide processing in tumor cells

References

943

is associated with enhanced tissue inhibitor of metalloproteinases (TIMP) gene expression. Znt. J. Cancer, 53 (1993) 634-639. 108. R.E.B. Seftor, E.A. Seftor, W.J. Grimes, L.A. Liotta, W.G. Stetler-Stevenson, D.R. Welch and M.J.C. Hendrix, Human melanoma cell invasion is inhibited in oitro by swainsonine and deoxymannojirimycin with a concomitant decrease in collagenase IV expression. Melunomu Rex, l(1991) 43-54. 109. H. Kanayama, K. Yokota, Y. Kurokawa, Y. Murakami, M. Nishitani and S. Kagawa, Prognostic values of matrix metalloproteinase-2 and tissue inhibitor of metalloproteinase-2 expression in bladder cancer. Cuncer, 82 (1998) 1359-1366. 110. P.E. Goss, M.A. Baker, J.P. Carver and J.W. Dennis, Inhibitors of carbohydrate processing: A new class of anticancer agents. Clin. Cancer Rex, 1 (1995) 935-944. 111. P.E. Goss, J. Baptiste, B. Fernandes, M. Baker and J.W. Dennis, A study of swainsonine in patients with advanced malignancies. Cuncer Rex, 54 (1994) 1450-1457. 112. P.E. Goss, C.L. Reid, D. Bailey and J.W. Dennis, Phase IB clinical trial of the oligosaccharide processing inhibitor swainsonine in patients with advanced malignancies. Clinical Cuncer Res., 3 (1997) 1077-1086. I 13. P.A. Romero and A. Herscovics, Transfer of non-glucosylated oligosaccharides from lipid to protein in mammalian cells. J. Biol. Chem., 261 (1986) 15936-1 5940. 114. A.J. Petrescu, S.M. Petrescu, R.A. Dwek and M.R. Wormald, A statistical analysis of N- and 0-glycan linkage conformations fromcrystallographic data. Glycohiology, 9 (1999) 343-352. 1 IS. A. Kobata, Structures, function, and transformational changes of the sugar chains of glycohormones. J. Cell Biochem., 37 (1988) 79- 90. 116. J. Hodgkin, H.R. Horvitz, B.K. Jasny and J. Kimble, C. elegans: Sequence to biology. Science, 282 (1998) 201 1-2046. 1 17. S.-I. Hakomori, Aberrant glycosylation in tumors and tumor-associated carbohydrate antigens. Ado. Cuncer Rex, 52 (1989) 257-33 I . 118. K.D. Lloyd, Blood group antigens as markers for normal and abnormal differentiation and malignant change in human tissues. Am. J. Clin.Puthol., 87 (1987) 129-139. 119. G.F. Springer, Immunoreactive T and Tn epitopes in cancer diagnosis, prognosis, and immunotherapy. J. Mol. Med., 75 ( I 997) 594-602. 120. C.M. Bohm, M.C. Mulder, K. Zennadi, M. Notter, A. Schmitt-Graff, O.J. Finn, J. TaylorPapadimitriou, H. Stein, H. Clausen, E.O. Riecken and C. Hanski, Carbohydrate recognition on MUCl-expressing targets enhances cytotoxicity of a T cell subpopulation. Scand. J. Immunol., 46 (1999) 27-34. 121. I. Brockhausen, J.M. Yang, J. Burchell, C . Whitehouse and J. Taylor-Papadimitriou, Mechanisms underlying aberrant glycosylation of MUCl much in breast cancer cells. Ew. J. Biochem., 233 (1995) 607-61 7. 122. D.H. van den Eijnden, A.H.L. Koenderman and W.E.C.M. Schiphorst, Biosynthesis of blood group i-active polylactosaminoglycans. J. Bid. Chem., 263 (1988) 12461-~ 12465. 123. E.H. Holmes, S.-I. Hakomori and G.K. Ostrander, Synthesis of type 1 and 2 lacto series glycolipid antigens in human colonic adenocarcinoma and derived cell lines is due to activation of a normally unexpressed PI -3N-acetyIglucosaminyltransferase. J. Biol. Chem., 262 (1987) 15649- 15657. 124. W.C. Wang, N. Lee, D. Aoki, M.N. Fukuda and M. Fukuda, The poly-N-acetyllactosamines attached to lysosomal membrane glycoproteins are increased by the prolonged association with the Golgi complex. J. Biol. Chent, 266 (1991) 23185-23190. 125. P.A. Prieto, K.D. Larsen, M. Cho, H.N. Rivera, A. Shilatifard, J.B. Lowe, K.D. Cummings and D.F. Smith, Expression of human H-type al,2-fucosyltransferase encoding for blood group H (0) antigen in Chinese hamster ovary cells. J. Bid. Chem., 272 (1997) 2089-2097. 126. Granovsky et al., in preparation. 127. M. Granovsky, J. Fata, J. Pawling, W.J. Muller, R. Khokha, and J.W. Dennis. Suppression of tumor growth and metastasis in MagtS-deficient mice Nuture Medicine. In press, Mar (2000). 128. N.-W. Lo, J.W. Dennis, and J.T.Y. Lau. Over-expression of the a2,6-sialyltransferase, ST6Gal I, in a low metastatic variant of a murine lymphoblastoid cell line is associated with appearance ofa unique ST6Gal I mKNA. Biochem. Biophys. Rex Commun., 264 (1999)619-62 1. 129. Warren et al., in preparation.

Carbohydrates in Chemistry and Biology Beat Emst, Gerald W. Hart, Pierre Sinay copyright Q WILEY-VCH Verlag GmbH. 2000

57 Lysosomal Storage Diseases Nutlion N . Aronson, Jr.

57.1 Summary Lysosomal storage diseases are caused by pathological overloading of tissues with fragments from glycoproteins, glycolipids, glycosaminoglycans or glycogen, mostly due to mutations in genes that code the hydrolases naturally designed to degrade these polymers. In a few cases a final sugar or amino acid product is the stored material because required lysosomal transport proteins are mutated. The medical concept of lysosomal storage diseases dates to 1965 when Hers [ l ] proposed that mutation of genes for lysosomal hydrolases would result in defective catabolism that in turn would cause a progressive, metabolically detrimental compaction of the degradative organelles. Almost 40 separate lysosomal storage diseases are now known. and in the past 35 years dramatic understanding of these diseases has been gained. In a few cases medical treatments such as enzyme replacement in Gaucher disease [2] and drug therapy in the case of cystinosis [3] have been discovered and successfully applied to patients. During the past three and a half decades of scientific and clinical study of what are relatively rare genetic diseases, many important facts about general biochemistry, cell biology, molecular genetics and molecular medicine have been learned as well. Recently, a number of laboratories have developed animal models for these genetic disorders, mostly gene knock-outs in mice, and these experimentally diseased animals will undoubtedly allow further advances in explaining the pathophysiology of fragment accumulation and in development of gene-replacement or other therapies for curing storage disorders. Since there are many excellent comprehensive reviews on all aspects of the lysosomal storage diseases [4-71, including complete books on individual types [8], this Chapter will be limited to some interesting new examples, such as a protease gene-defect that causes a specific bone restorption disorder pycnodysostosis. In addition, two separate Chapters in this book give much greater detail about lysosomal degradation of glycolipids and glycoproteins. An excellent searchable electronic database is Online with Mendelian Inheritance in Man ( O M I M ) (http://www.ncbi.nlm.nih.gov/Omim)

946

57 Lysosomul Storage Diseuses

extensive, updated information about most of the lysosomal storage diseases, and when available, the OMZM entry number for a particular lysosomal storage disease is given in the Tables.

57.2 Introduction

There are approximately 40 lysosomal storage diseases, most of which involve incomplete degradation of glycoproteins, glycolipids, proteoglycans or polysaccharides. In general, these inherited metabolic disorders are heterogeneous in their pathophysiology, and often they are subclassified into infantile, juvenile and adult forms. The pathology is progressive and usually results from a gene mutation in an enzyme that normally performs hydrolysis of the storage compounds, which are fragments of various macromolecules undergoing standard intermediary catabolism [l]. Other possible causes are deficiency of various functional proteins, such as activators required for glycolipid catabolism [4], transporters necessary for lysosoma1 efflux of a final digestive product [3], enzymes involved in targeting hydrolases to the lysosomes [9], or “protective” proteins that allow survival of certain lysosoma1 hydrolases [lo]. The one example of the latter unusual case is cathepsin A, which complexes with and acts as a “protective protein” for P-galactosidase and a-neuraminidase. Both these glycosidases are rapidly degraded when certain gene defects occur in cathepsin A to cause galactosialidosis. Multiple clinical features tend to occur in patients with a particular lysosomal storage disease. There can be diversity of organs affected and variability of age of onset. Common maladies are skeletal displasia, including course faces and bone deformities, hepatosplenomegaly, telagiectasia, peripheral neuronal problems involving vision and hearing such as cherry-red spots, corneal clouding and deafness, inguinal hernias, congenital heart failure, and mental and motor retardation including progressive ataxia and seizures. There can be large differences in the severity of biochemical defects that occur. Early infantile forms of the disease tend to show the most severe pathology. Biochemically these extreme forms reflect tissues of a patient having below the critical threshold, or even complete absence, of a functional hydrolase. Many lysosomal storage diseases are autososomal recessive and heterozygous carriers with 50% enzyme activity are without symptoms. Tissue activities as low as 10% of normal may allow the individual to have healthy lysosoma1 metabolism and this behavior offers the possibility of enzyme replacement for therapy [2]. In late-onset forms of the disorders the clinical course can progress slowly with severity of an organ involvement being variable. Significant histological changes used for clinical diagnosis of lysosomal storage are vacuolated lymphocytes filled with stored material, an accumulation of lipofuscin granules, or lamellar inclusions due to glycolipid buildup. One of the most typical and useful clinical features of lysosomal storage diseases is the large urinary excretion of undigested oligosaccharide and glycopeptide fragments characteristic of each disorder.

57.4 Mucopo1ysaccharidost.s

941

57.3 Animal Models Fragment storage in most lysosomal diseases begins during development in utevo, and it is not feasible to study directly in humans many of the important biochemical and physiological problems created by these gene defects. Animal models are therefore expected to offer possibilities to develop better understanding of the causes and progression of the pathology and to create methods for successful medical treatment [Ill. Almost all of the lysosomal disorders now have an animal counterpart either produced by genetic engineering or occurring as a spontaneous disease (Table 1). One of the most extensively tested animal models for these purposes has been murine mucopolysaccharidosis type VII [ 121. Treatments of the MPS VII mice ranging from enzyme replacement to various intricate gene therapy approaches have been tried, with varied success that is still incomplete [ 13-15]. Another general difficulty often encountered with mouse models is that they fail to exhibit the severe pathophysiology seen in humans with the same genetic lysosomal storage disease. A recent example of this animal difference was observed with a knock-out mouse model for metachromatic leukodystrophy generated by homologous recombination techniques to inactivate the lysosomal arylsulphatase A gene [16]. Although the animals were deaf, the phenotype was mild and they had a normal life span. The sulfatase-deficient animals also failed to have their tissue-stored glycolipid material cleared when given bone marrow transplants transduced with a retroviral construct that expressed a replacement sulfatase.

57.4 Mucopolysaccharidoses Mucopolysaccharidoses are a set of the lysosomal storage diseases caused by deficiency of a hydrolase required for catabolism of glycosaminoglycans (Table 2). Included polymeric substrates for the disease-related hydrolases are dermatan sulfate, heparan sulfate, keratan sulfate, chondroitin sulfates and hyaluronic acid. These mucopolysaccharides in turn are components of proteoglycans, and therefore cathepsins also are necessary for the complete degradation of this class of biopolymer. As noted in earlier chapters on the lysosomal degradation of glycoproteins and glycolipids, the glycosaminoglycans also are hydrolyzed stepwise. This metabolism includes six glycosidases, five sulfatases and a single nonhydrolytic acetyl transfer reaction. The pathways for degradation of dermatan sulfate, heparan sulfate, karatan sulfate, chondroitin sulfates and hyaluronic acid are depicted in Figure 1. When a gene mutation occurs in one of the lysosomal hydrolases involved in each of these catabolisms, undigested fragments undergo large-scale accumulation in tissue lysosomes, and excess amounts are released from cells and excreted in the urine of the afflicted patient. The pathology of mucopolysaccharidoses can vary as seen in other lysosomal storage diseases, and is often multi-organ, chronic and progressive. Many naturally occurring or genetically engineered animal models

948

57 Lysosornal Storage Diseases

Table 1. Animal models for lysosomal enzyme defects. Enzyme

Animal model

Human disease

a-L-Iduronidase N- Acetylglucosamine-6-sulfatase N-Acetylgalactosamine-4-sulfatase P-D-Galactosidase

Murine Caprine Feline Murine Canine Murine Murine

Hurler MPS IIID Maroteaux-Lamy G M Gdnghosidosis ~ MPS VII Tay-Sachs

Murine

Sandhoff

P-D-Glucuronidase N-Acetyl-P-D-glucosaminidase (a-subunit) N-Acetyl-0-D-ghcosaminidase (P-subunit) N-Acetyl-B-D-glucosaminidase (a- and P-subunit) G M Activator ~ a-Neuraminidase Sphingolipid Activator Protein 0-D-Glucocerebrosidase a-D-Galactosidase Arylsulfatase A P-D-Galactocerebrosidase Sphingomyelinase Cathepsin D Cathepsin A (protective protein) Cathepsin K Glycosylasparaginase a-D-Mannosidase S-D-Mannosidase a-L-Fucosidase a-D-Glucosidase N-Acetylglucosamine-1phosphotransferase Acid phosphatase

Reference

Murine Murine Murine M urine M urine Murine Murine Murine Canine Murine Murine Murine Murine Murine Feline Bovine Caprine Bovine Canine Quail Murine Feline

-

Sialidosis ~

Gaucher Fabry Metachromatic Leukodystrophy Krabbe Niemann-Pick __

Galactosialidosis Pycnodysostosis Aspart ylglycosaminuria a-Mannosidosis P-Mannosidosis a-Fucosidosis Pompe I-Cell

Murine

exist for the mucopolysaccharidoses (Table 1) and attempts to better understand the molecular causes and progression of these disorders and to develop successful therapies for them are active areas of research. Heparan sulfates have the most complex structure of the glycosaminoglycan substrates involved in mucopolysaccharidoses. At least eight lysosomal enzymes take part in the complete digestion of this class of polysaccharide that proceeds to monosaccharides and sulfate in the sequence of enzymatic steps shown in Figure 1:

57.5 Cuthepsin K De$ciency und Pycnodysostosis

949

Table 2. Mucopolysaccharidoses. ~

~

~~~~

~

MPS#

Common name

OMIM reference

Enzyme deficiency (See Figure 1 )

MSP I MSP I1 MSP IIIA MSP IIIB MSP IIIC

Hurler Hunter Sanfilippo A Sanfilippo B Sanfilippo C

252800 309900 252900 252920 252930

MSP IIID MSP IVA

Sanfilippo D Morquio A

252940 253000

MSP IVB

Morquio B ( G MGangliosidosis) ~ Maroteaux-Lamy

253010

a-L-Iduronidase (1) Iduronate-2-Sulfatdse (2) Heparan N-Sulfatase (3) N-acetyl-a-D-Glucosaminidase (4) AcetylCoA: a-Glucosaminide Acetyltransferase (5) N-acetylglucosamine-6-Sulfatase(6) N-Acetylgalactosarnine-6-Sulfatase (7) 0-D-Galactosidase (8)

Sly Tay-Sachs/Sandhoff

253220 272800/268800

MSP VI MSP VII ~

253200

~

N-ACeylgalaCtOSdmine-4-SulfataSe (aryl sulfatase B) (9) b-D-Glucuronidase (1 0) N-Acetyl-P-D-Hexosaminidase (1 1) Glucuronic Acid-2-Sulfatase (12) Hyaluronidase (13)

iduronate 2-sulfatase, a-L-iduronidase; heparan N-sulfatase (sulfamidase); acetylCoA acetyltransferase; N-acetyl-a-D-galactosaminidase; glucuronate-2-sulfatase, pD-glucuronidase; and N-acetylglucosamine-6-sulfatase. Except for the lack of a known disease in the glucuronate-2-sulfatase, respective genetic mutations that inactivate each of the other heparan sulfate hydrolases cause MPS I1 (Hunter syndrome, a-L-iduronidase); MPS I (Hurler syndrome, iduronate-2-sulfatase); MPS IIIA (Sanfilippo A syndrome, sulfamidase); MPS IIIC (Sanfilippo C syndrome, acetyCoA: a-glucosaminide acetyltransferase); MPS IIIB (Sanfiippo B syndrome, N-acetyl-a-D-glucosaminidase); MPS VII (Sly syndrome, p-D-glucuronidase) and MPS IIID (Sanfilippo D syndrome, N-acetylglucosamine-6-sulfatase).All four forms of Sanfilippo syndrome, (MPS IIIA-IIID) cause the same basic pathology, but represent defects in four individual steps catalyzed by different kinds of enzymes in the heparan sulfate pathway.

57.5 Cathepsin K Deficiency and Pycnodysostosis There have been few discoveries of lysosomal storage diseases being related to a deficiency of a protease. Efficient lysosomal proteolysis could be such a crucial metabolic process that any defects in this system would be developmentally lethal and therefore never observed medically. For example, mice deficient in cathepsin D were generated by targeted disruption of the gene and the animals died in a state of anorexia by day 26 due to widespread intestinal necroses [42]. Since bulk lysosomal

950

57 Lysosomul Storage Diseuses

DERMATAN SULFATE Order: 2; 1; 9; 11; 10

CHONDROITIN4-SULFATE Order: 13; (lo; 6; ll),

HEPARAN SULFATE: Order: 2; 1; 3; 5; 4; 12; 10; 6

CHONDROITINCSULFATE: Order: 13; (10; 6; ll),

KERATAN SULFATE: Order: 7; 8; 6/11; 8; 6/11

HYALURONICACID Order: 13; (lo; 11)"

Figure 1. Lysosomal Digestive Pathways for Glycosaminoglycans. Each structure represents a type of repeating unit in the particular polysaccharide. Enzyme names with corresponding numbers are given in Table 2. A slash between two enzyme numbers indicates either reaction can precede the other.

proteolysis remained normal in the mice, cathepsin D must have very specific functions that are critical to overall survival. A single prominent example of a lysosomal protease defect mentioned above is galactosialidosis where the primary genetic deficiency is in cathepsin A [lo], which acts multifunctionally. Not only is cathepsin A a serine carboxypeptidase, it also becomes a protective agent by complexing with and preventing two other lysosomal hydrolases, P-D-galactosidase and a-neuraminidase, from rapid turnover in the potent proteolytic environment within lysosomes. Another example of a genetic mutation in a lysosomal cathepsin that causes a specific human disorder has recently come to light [43]. The disease is pycnodysostosis which is an autosomal recessive skeletal dysplasia. Clinical features of pycnodysostosis were first described in 1962 by Maroteaux and Lamy [60] to include bone fragility, dental abnormalities, reduced stature and skull deformities with a delay in closure of the cranial sutures. The disease experienced considerable acclaim due to the initial prediction by Maroteaux and Lamy, recently refuted [61], that this was the genetic disorder suffered by the famous French impressionist artist Toulouse-

57.6 Mouse Models,for Tuy-Such und Sundlzoff Diseases

95 1

Lautrec. Two research groups independently mapped the gene to chromosome lq21 by linkage analysis of two large consanguineous Mexican [62] and Arab [63] pedigrees. As early as 1988 Everts et al. [64] had predicted that cysteine proteases were involved in bone matrix degradation by osteoclasts, since inhibition of these activities by the well-known thiol protease inhibitors, leupeptin and E-64, yielded ultrastructural morphology similar to that found in pycnodysostosis patients. In 1996 Drake et al. [65] determined that cathepsin K uniquely among the many other lysosomal cysteine proteases (e.g. B, L or S) was abundantly expressed in human osteoclasts and rarely expressed in any other tissue. The Mexican family mutation in the cathepsin K gene causing pycnodysostosis was then identified as a C to T transition at nucleotide 862 resulting in Arg241 being changed to a STOP codon [66]. All afflicted family members were homozygous for the same cathepsin K gene mutation. Nonsense and misssense mutations in the cathepsin K gene from other pycnodysostosis patients were also discovered by Gelb et al. [67]. In addition, cathepsin K has been found uniquely capable of completely dissolving insoluble type I collagen from human cortical bone [68], an essential biochemical activity required for its role in bone resorption. Finally, a cathepsin K-deficient mouse was created recently by targeted disruption of the gene for this lysosomal thiol proteinase [43], and the animals exhibited pathophysiology similar to the human pycnodysostosis patients. All of these studies have revealed the unique role of cathepsin K in bone resorption by osteoclasts. Interestingly, this important knowledge on the role of cathepsin K in both the physiology and pathology of bone resorption has been turned to new studies on designing possible anti-cathepsin K drugs to alleviate or prevent other more prevalent bone diseases [69, 701, such as osteoporosis in which there is the opposite pathology of excess resorption. This result again emphasizes how the characterization and understanding of the relative rare lysosomal storage diseases has led biomedical scientists to learn unexpected critical facts about basic cellular biochemistry and physiology.

57.6 Mouse Models for Tay-Sachs and Sandhoff Diseases An earlier chapter details those lysosomal diseases involved in glycolipid degradation and the group is listed in Table 3 with the pathways represented schematically in Figure 2. Important information and ideas about glycolipid turnover have resulted from studying knock-out mouse models for the HexA (a-subunit) [25,27] and HexB (p-subunit) [26] genes which together encode the complete set of three N acetyl-p-D-glucosaminidase isoenzymes: p-hexosaminidase A (alp);B (p/p); and S (a/.). The HexA and HexB deficient mice respectively show pathophysiology very similar to that occurring in the human storage disorders Tay-Sachs and Sandhoff diseases. Both kinds of mice accumulate G M ganglioside ~ in their brain. The latter glycolipid is retained much more by the HexB (-/-) animals and they additionally . (-/-) animals do not suffer the fatal neurostore glycolypid G A ~HexA degenerative pathology that occurs in the HexB Sandhoff model mice. An expla-

952

57 Lysosomal Storage Diseuses

Sphlngorlne

G, GANGLIOSIDE: Order: 1; 2/3;1; 4; 5

Q GLOBOSIDE: Order: 6; 7; 1; 4; 5 0 y

3

CH- N+-CH,-CH,-O-P 5

1

II

-o - R

CH,

SULFATIDE Order: 8; 9; 5

SPHINGOMYELIN: Order: 10; 5

Figure 2. Lysosomal Digestive Pathways for Sphingolipids. Enzyme names with corresponding numbers are given in Table 3.

nation for the wide variation in the pathology seen in the two types of deficiencies is that some stored G Min~ the HexA (-/-) mice is degraded through sequential hydrolysis by neuraminidase and N-acetyl-P-D-glucosaminidase B (o/P) with G A as ~ an intermediate [71]. The Hex A (-/-) and Hex B (-/-) animals were then successfully interbred to form N-acetyl-P-D-ghcosaminidase null animals [28]. These mice interestingly began to show the disease phenotype of mucopolysaccharidoses, as large numbers of vacuolated cells that stained positively with alcian blue for glycosaminoglycans were seen in their connective tissue, corneas, heart valves and visceral organs. Since human counterparts with either Tay-Sachs or Sandhoff disease do not accumulate glycosaminoglycans in their tissue lysosomes, this new behavior in the double knock-out mouse model indicates that the redundancy of phexosaminidase forms A(a/P) [possibly S ( a / a ) ]and B(P/p) allows the normal hydrolysis of non-reducing P-D-G~cNAcand GalNAc residues encountered during the stepwise hydrolysis of glycosaminoglycans in these two human mucolipidoses.

57.7 Impuct of‘ Lysosomd Diseuses and Their Study

953

Table 3. Glycosphingolipidoses. Name

OMIM reference

Enzyme deficiency (See Figure 2)

Generalized GMIGangliosidosis Tay-Sachs

230500 272800

Sialidosis (Mucolipidosis I) Gaucher Farber Sandhoff

256550 230800 228000 268800

Fabry Metachromatic Leukodystrophy Krabbe N iemann-Pick

301500 250100 245200 257220

p-D-Galactosidase (1) N-Acetyl-P-D-Glucosaminidase A (u-subunit) (2) a-Neuraminidase (3) P-D-Glucocerebrosidase (4) Ceramidase (5) N-Acetyl-P-u-Glucosaminidase A&B (p-subunit ) (6) a-D-Galactosidase (7) Arylsufdtase A (8) 0-D-Galactocerebrosidase (9) Sphingomyelinase (1 0)

A new intriguing approach to possible therapy for the storage of glycolipids in these two lysosomal genetic disorders termed “substrate depravation” has recently been suggested independently by two research groups [72-741. Rather than improve degradation of the accumulated gangliosides, the new therapeutic concept was to decrease their concentration before storage by inhibiting G Mbiosynthesis ~ with Nbutyldeoxynojirimycin (NB-DNJ). In the Tay-Sachs mice G Maccumulation ~ was prevented in the brain by this agent and the neuropathology was greatly reduced. NB-DNJ was originally developed as an anti-HIV drug and has been through phase 11 clinical testing with relatively few side effects other than GI distress through its inhibition of disaccharidase in the gut. Although NB-DNJ failed in its HIV efficacy due to having a low concentration level in cells, the serum concentration (10-50 pM) was in the range found to be effective in treating the Tay-Sachs mice. Amazingly, a 70% depletion of peripheral glycosphingolipids caused by NB-DNJ was tolerated over a 4-month period by healthy mice.

57.7 Impact of Lysosomal Diseases and Their Study Although the occurrence of each individual lysosomal storage disease is rare, the group of approximately 40 taken together (see Tables 2, 3 and 4) can make a significant impact on the total medical commitment of an individual country. Such an analysis of patients with lysosomal storage diseases was recently reported for Australia [75],a country in which there is no genetic enrichment of any single lysosomal storage disorder as happens in Finland, where the lysosomal glycoprotein disorder aspartylglycosaminuria occurs almost exclusively [ 81. The retrospective study covered births during the 16 year period from January 1, 1980 to December 3 1, 1996 in which 27 different lysosomal diseases were found in 545 Australian individuals. Eighteen of the 27 separate disorder types had 10 or more cases. The range of oc-

954

57 Lysosomal Storage Diseases

Table 4. Lysosomal disorders involving glycoprotein catabolism and other enzymes.

Category/Enzyme

OMIM

Disease

Glycoproteins (See Chapter 56) a-Neuraminidase P-D-Galactosidase N-Acetyl-0-D-glucosaminidase a-D-Mannosidase p-D-Mannosidase Di-N-Acetylchitobiase

256550 253010 272800/268800 248800 248510 600873 230000 208400 104170

Sialidosis GMl Gangliosidosis Tay-Sachs/Sandhoff a-Mannosidosis P-Mannosidosis (evolutionary deletion from ungulates and carnivores) a-Fucosidosis Aspartylglycosaminuria Schindler

232300

Pompe

252500

I-Cell (mucolipidosis 11)

269920 21 9800

Salla Cystinosis

256540 601 105

Galactosialidosis Pycnodysostosis (mouse, knock-out [59])

a-L-Fucosidase Glycosylasparaginase N - Acetyl-a-D-Galactosaminidase Glycogen a-D-Glucosidase Glycoprotein Targeting UDP-GlcN Ac:N-acetylglucosdmine-1phosphotransferase Lysosomal Membrane Transporters Sialic Acid Transporter Cystine Transporter Proteases Cathepsin A Cathepsin K Acid Phosphatase Lipase Acid Lipase

._

278000

Wolman

currence was one per 57,000 live births for Gaucher’s disease, the most common of all lysosomal diseases world-wide, to a very rare one per 4.2 million live births for sialidosis. The prevalence of all 27 diseases in the Australian population was a surprisingly high 1 in 7,700 live births, which indicates the combined importance of lysosomal storage diseases in the Australian health care system. Such a significant medical impact for lysosomal diseases as a group is also likely to be important in other national populations. In the future molecular and clinical studies on lysosoma1 storage diseases should contribute significantly to discovery of medical treatments for many forms of human genetic affliction. References 1. Hers, G.H. Inborn Lysosomal Diseases. Gastroenterology, 1995, 48, 625-633. 2. Brady, R.O.; Pentchev, P.G.; Gal A.E.; Hibbert, S.R.; Dekaban, A S . Replacement Therapy for Inherited Enzyme Deficiency. Use of Purified Glucocerebrosidase in Gaucher’s Disease. N. Engl. J. Med., 1974, 29, 989-993. 3. Thoene, J.G.; Oshimd, R.G.; Crawhall, J.C.; Olson, D.L.; Schneider, J.A. Cystinosis. Intra-

Refkrences

955

cellular Cystine Depletion by Aminothiols in vitro and in v i m J. Clin. Inirest., 1976, 58, 180189. 4 Kolter, T.; Sandhoff, K. Recent Advances in the Biochemistry of Sphinogolipids. Brain Pathol., 1998, 8, 79-100. 5. Gieselmann, V. Lysosomal Storage Diseases. Biochim. Biophys. A d a , 1995, 1270, 103- 136. 6. Salvetti, A,; Heard, J.M.; Danos, 0. Gene Therapy of Lysosomal Storage Disorders. Br. Med. Bull., 1995, 51, 106-122. 7. Scriver, C.R.; Beaudet, A.L.; Sly, W.S.; Valle, D. (Eds) The Metabolic and Molecular Basis of Inherited Diseases, 7th edition, McGraw Hill, Inc. New York, 1995, Vol. 11, 2425-2979. 8. Mononen, I.; Aronson, N.N., Jr. (Eds) Ljjsosomal Storage Disease: Aspartyglycosanzinuria. Chapman & Hall, Austin, TX, 1997, pp. 197. 9. Hasilik, A,; Wheed, A.; von Figura, K. Enzymatic Phosphorylation of Lysosomal Enzymes in the Presence of UDP-N-Acetylglucosamine. Absence of the Activity in I-Cell fibroblasts. Biochem. Biophys. Res. Commun., 1981, 98, 761-767. 10. Galjart, N.J.; Gillemans, N.; Harris, A,; van der Horst, G.T.; Verheijen F.W.; Galjaard, H.; d’Azzo. A. Exmession of cDNA Encoding the Human “Protective Protein” Associated with Lysosomal P-Galactosidase and Neuraminidase: Homology to Yeast Proteases. Cell, 1988, 54, 755-764. 1 1 Suzuki, K., Mansson, J.E. Animal Models of Lysosomal Disease: An Overview. J. Inherit. Metab. Dis., 1998, 21, 540-586. 12 Vogler, C.; Sands, M.S.; Galvin, N.; Levy, B.; Thorpe, C.; Barker, J.; Sly, W.S. Murine Mucopolysaccharidosis Type VII: The Impact of Therapies on the Clinical Course and Pathology in a Murine Model of Lysosomal Storage Disease. J. Inherit. Metab. Dis., 1998, 21, 575-586. 13 Soper, B.W.; Pung, A.W.; Vogler, C.A.; Grubb, J.H.; Sly, S.W.; Barker, J.E. Enzyme Replacement Therapy Improves Reproductivc Performance in Mucopolysaccharidosis Type VII Mice but Does Not Prevent Postnatal Losses. Pediatr. Res., 1999, 45, 180-186. 14 O’Connor, L.H.; Erway, L.C.; Vogler, C.A.; Sly, W.S.; Nicholes, A,; Grubb, J.; Holmberg, S.W.; Levy, B.; Sands, M.S. Enzyme Replacement Therapy for Murine Mucopolysaccharidosis Type VII Leads to Improvements in Behavior and Auditory Function. J. Clin. Invest., 1998, 101, 1394-1400. 15 Taylor, R.M.; Wolfe, J.H. Decreased Lysosomal Storage in the Adult MPS VII Mouse Brain in the Vicinity of Grafts of Retroviral Vector-Corrected Fibroblasts Secreting High Levels of pGlucuronidase. Nut. Med., 1997, 3, 77 1-774. 16 Gieselmann, V.; Matzner, U.; Hess, B.; Lullmann-Rauch, R.; Coenen, R.; Hartmann, D.; D’Hooge: R.; DeDeyn, P.; Nagels; G. Metachromatic Leukodystrophy: Molecular Genctics and An Animal Model. J. Inherit. Dis., 1998, 21, 564-574. 17 Clarke, L.A.; Russell C.S.; Pownall, S.; Warrington, C.L.; Borowski, A,: Dimmick, J.E.; Toone, J.; Jirik, R.F. Murine Mucopolysaccharidosis Type I: Targeted Disruption of the Murine a-L-Iduronidase Gene. Hum. Mol. Genet. 1997, 6, 503-5 11. 18 Cavanagh, K.T.; Leipprandt, J.R.; Jones, M.Z.; Friderici, K. Molecular Defect of Caprine NAcetylglucosamine-6-Sulphatase Deficiency. A Single Base Substitution Creates a Stop Codon in the 5’-Region of the Coding Sequence. J. Inherit. Dis. 1995, 18, 96. 19 Jackson, C.E.; Yuhki, N.; Desnick, R.J.; Haskins, M.E.; O’Brien, S.J.; Schuchman, E.H. Feline Arylsulfatase B (ARSB): Isolation and Expression of the cDNA, Comparison With Human ARSB, and Gene Localization to Feline Chromosome A l . Genomics, 1992, 14, 403-41 1. 20 Matsuda, J.; Suzuki, 0.; Oshima, A,; Noguchi, Y.; Yamamoto, Y.; Asano, T.; Takimoto, K.; Sukegawa, K.; Suzuki. Y.; Naiki, M. P-Galactosidase-Deficient Mouse as an Animal Model for G M-Gangliosidosis. ~ Glycoconj. J. 1997, 14, 729-736. 21 Hahn, C.N.; del Pilar Martin, M.; Schroder, M.; Vanier, M.T.; Hara, Y.; Suzuki, K.; Suzuki, K.; d’Azzo, A. Generalized CNS Disease and Massive GM1-Ganglioside Accumulation in Mice Defective in Lysosomal Acid 8-Galactosidase. Hum. Mol. Genet. 1997. 6, 205-21 1. 22. Alroy, J.; Orgad, U.; DeGasperi, R; Richard, R.; Warren, C.D.; Knowles, K.; Thalhammer, J.G.; Raghavan, S.S. Canine Gu I -Gangliosidosis. A Clinical, Morphologic Histochemical, and Biochemical Comparison of Two Different Models. Am. J. Pathol., 1992, 140, 675-789. 23. Warner, T.G.; O’Brien, J.S. Structure Analysis of the Major Oligosaccharides Accumulating in Canine G M IGangliosidosis Liver. J. Bid. Chem. 1992, 10, 224-242.

956

57 Lysosomal Storage Diseases

24. Vogler, C.; Birkenmeier, E.H.; Sly, W.S.; Levy, B.; Pegors, C.; Kyle J.W.; Beamer, W.G. A Murine Model of Mucopolysaccharidosis VII. Gross and Microscopic Findings in P-Glucuronidase-Deficient Mice. Ant J. Pathol. 1990, 91, 9975-9979. 25. Yamanaka, S.; Johnson, M.D.; Gringerg, A.; Westphal, H; Crawley, J.N.; Taniike, M.; Suzuki, K.; Proia, R.L. Targeted Disruption of the Hexa Gene Results in Mice with Biochemical and Pathologic Features of Tay-Sachs Disease. Proc. Natl Acad. Sci. USA, 1994, 91, 99759979. 26. Sango, K.; Yamanaka, S.; Hoffmann, A.; Okuda, Y.; Grinberg, A.; Westphal, J.; McDonald, M.P.; Crawley, J.N.; Sandhoff, K.; Suzuki, K.; Proia, R.L. Mouse Models of Tay-Sachs and Sandhoff Diseases Differ in Neurologic Phenotype and Ganglioside Metabolism. Nut. Genet., 1995, 11, 170-176. 27. Cohen-Tannoudji, M.; Marchand, P.; Akil, S.; Puech, J.P.; Kress, C.; Gressens, P.; Nassogne, M.C.; Beccari, T.; Muggleton-Harris, A.L.; Evard, P.; Stirling, J.L.; Poenaru, L.; Babinet, C. Disruption of Murine Hexa Gene Leads to Enzymatic Deficiency and to Neuronal Lysosomal Storage, Similar to that Observed in Tay-Sachs Disease. Mamm. Genome, 1995, 6, 844-849. 28. Sango, K.; McDonald, M.P., Crawley, J.N.; Mack, M.L.; Tifft, C.J.; Skop, E.; Starr, C.M.; Hoffmann, A,; Sandhoff, K.; Suzuki, K.; Proia, R.L. Mice Lacking Both Subunits of Lysosoma1 P-Hexosaminiddse Display Gangliosidosis and Mucopolysaccharidosis. Nut. Genet., 1996, 14, 348-352. 29. Liu, Y.; Hoffmann, A.; Gringerg, A.; Westphal, H.; McDonald, M.P.; Miller, K.M,; Crawley, J.N.; Sandhoff, K.; Suzuki, K.; Proia, R.L. Mouse Model of G M Activator ~ Deficiency Manifests Cerebellar Pathology and Motor Impairment. Proc. Nut1 Acad. Sci. USA, 1997, 94, 81388143. 30. Rottier, R.J.; Bonten, E.; d’Azzo, A. A Point Mutation in the Neu-1 Locus Causes the Neuraminidase Defect in the SM/J Mouse. Hum. Mol. Genet. 1998, 7, 313-321. 31. Carrillo, M.B.; Milner, C.M.; Ball, S.T.; Snoek, M.; Campbell, R.D. Cloning and Characterization of a Sialidase from the Murine Histocompatibility-2 Complex: Low Levels of mRNA and a Single Amino Acid Mutation are Responsible for Reduced Sialidase Activity in Mice Carrying the Neul a Allele. Glycobiology, 1997, 7, 975-986. 32. Womack, J.E.; Yan: D.L.; Potier, M. Gene for Neuraminidase Activity on Mouse Chromosome 17 Near h-2: Pleiotropic Effects on Multiple Hydrolases. Science, 1981, 212, 63-65. 33. Oya, Y.; Nakayasu, H.; Fujita, N.; Suzuki, K.; Suzuki, K. Pathological Study of Mice with Total Deficiency of Sphingolipid Activator Proteins (SAP Knockout Mice). Actu Neuropathol., 1998, 96, 29-40. 34. Tybulewica, V.L.; Tremblay, M.L.; LaMarca, M.E.; Willemsen, R.; Stubblefield, B.K.; Winfield, s.; Zablocka, B.; Sidransky, E.; Martin, B.M.; Huang, S.P.; et al. Animal Model of Gaucher’s Disease from Targeted Disruption of the Mouse Glucocerebrosidase Gene. Nature. 1992,4, 407-410. 35. Ohshima, T.; Murray, G.J.; Swaim, W.D.; Longenecker, G.; Quirk, J.M.; Cardarelli, C.O.; Sugimoto, Y.; Pastan, I.; Gottesman, M.M.; Brady, R.O.; Kulkarni, A.B. a-Galactodsase A Deficient Mice: A Model of Fabry Disease. Proc. Natl Acad. Sci. USA, 1997, 18, 2540-2544. 36. Hess, B.; Saftig, P.; Hartmann, D.; Coenen, R.; Lullmann-Rauch, R.; Goebel, H.H.; Evers, M.; von Figura, K.; D’Hooge, R.; Nagels, G.; DeDeyn, P.; Peters, C.; Gieselmann, V. Phenotype of Arylsulfatase A-Deficient Mice: Relationship to Human Metachromatic Leukodystrophy. Proc. Nail Acad. Sci. USA, 1996, 93, 14821-13826. 37. Fujita. N.; Suzuki, K.; Vanier, M.T.; Popko, B.; Maeda, N.; Klein, A,; Hensler, M.; Sandhoff, K.; Nakayasu, H.; Suzuki, K. Targeted Disruption of the Mouse Sphingolipid Activator Protein Gene: A Complex Phenotype, Including Severe Leukodystrophy and Wide-Spread Storage of Multiple Sphingolipids. Hum. Mol. Genet., 1996, 5 , 711-725. 38. Wenger, D.A.; Victoria, T.; Rafi, M.A.; Luzi, P.; Vanier, M.T.; Vite, C.; Patterson, D.F.; Haskins, M.H. Globoid Cell Leukodystrophy in Cairn and West Highland White Terriers. J. Hered., 1999, 90, 138-142. 39. Sakai, N.; Inui, K.; Tatsumi, N.; Fukushima, H.; Nishigaki, T.; Taniike, M.; Nishimoto, J.; Tsukamoto, H.; Yanagihara, I.; Ozono, K.; Okada, S. Molecular Cloning and Expression of cDNA for Murine, Galactocerebrosidase and Mutation Analysis of the Twitcher Mouse, a Model of Krabbe’s Disease. J. Neurochem., 1996, 66, 1 1 18- 1 124.

Refevences

957

40. Suzuki, K.; Suzuki, K. The Twitcher Mouse: A Model for Krabbe Disease and for Experimental Therapies. Bruin Puthol., 1995, 5 , 249-258. 41. Otterback. B.; Stoffel, W. Acid Sphingomyelinase-Deficient Mice Mimic the Neurovisceral Form of Human Lysosomal Storage Disease (Niemann-Pick Disease). CelZ, 1995, 81, 10531061. 42. Saftig, P.; Peters, C.; von Figura, K.; Crawssaerts, K.; Van Leuven, F.; De Strooper, B. A Myloidogenic Processing of Human Amyloid Precursor Protein in Hippocampal Neurons Devoid of Cathepsin D. J. Bid. Chem., 1996. 271, 27241-27244. 43. Saftig, P.; Hunziker, E.; Wehmeyer, 0.;Jones, S.; Boyde, A,; Rommerskirch, W.; Moritz, J.D.; Schu, P.; von Figura, K. Impaired Osteoclastic Bone Resorption Leads to Osteopetrosis in Cathepsin-K-Deficient Mice. Proc. Nut/ Acud Sci. U S A , 1998, 95, 13453-~13458. 44. Kaartinen, V.; Mononen, I.: Voncken, J.W.; Noronkoski, T.: Gonzalez-Gomez, I.; Heisterkamp, N.; Groffen, J. A Mouse Model for the Human Lysosomal Disease Aspartylglycosaminuria. Nut. Med., 1996, 2, 1375-1378. 45. Jalanko, A.; Tenhunen, K.; McKinney,C.E.; LaMarca, M.E.; Rapola, J.; Autti, T.; Joensuu, R.; Manninen, T.; Sipila, 1.; Ikonen, S.; Riekkinen, P. Jr.; Ginns, E.I.; Peltonen, L. Mice with an Aspartylglucosaminuria Mutation Similar to Humans Replicate the Pathophysiology in Patients. Hum. Mol. Genet., 1998, 7, 265-272. 46. Berg, T.; Tollersrud. O.K.; Walkley, S.U.; Siegel, D.; Nilssen, 0. Purification of Feline Lysosoma1 a-Mannosidase, Determination of its cDNA Sequence and Identification of a Mutation Causing a-Mannosidosis in Persian Cats. Biochem. J., 1997, 328, 863-870. 47. Burditt, L.J.; Chotia, K.; Hirani, S.; Nugent, P.G.; Winchester, B.G.; Blakemore, W.F. Biochemical Studies on a Case of Feline Mannosidosis. Biochrm. J., 1980, 189, 467-473. 48. Tollersrud, OK.; Berg, T.; Healy, P.; Evjen, G.; Ramachandran, U.: Nilssen, 0. Purification of Bovine Lysosomal a-Mannosidase, Characterization of its Gene and Determination of Two Mutations that Cause a-Mannosidosis. Eicr. J. Biochem, 1997, 246-410-419. 49. Leipprandt, J.R.; Kraemer, S.A.; Haithcock, B.E.; Chen, H.; Dyme, J.L.; Cavanagh, K.T. Friderici, K.H.; Jones, M.Z. Caprine P-Mannosidase: Sequencing and Characterization of the cDNA and Identification of the Molecular Defect of Caprine b-Manosidosis. Gerzon7ic.r, 1996, 37, 61-56. 50. Bryan, L.; Schmutz, S.; Hodges, S.D.; Snyder, F.F. Bovine P-Mannosidase Deficiency. Biochem. Biophys. Res. Comm.. 1990, 173, 49 1-495. 51. Raben, N.; Nagaraju, K.; Lee, E.; Kessler, P.; Byrne, B.; Lee, L.; LaMarca, M.; King, C.; Ward. J.; Sauer, B.; Plotz, P. Targeted Disruption of the Acid a-Glucosidase Gene in Mice Causes an Illness with Critical Features of Both Infantile and Adult Human Glycogen Storage Disease Type 11. J. Bid. Chem., 1998, 273, 19086-19092. 52. Bijvoet, A.G.: van de Kamp, E.H.; Kroos, M.A.; Ding, J.H.; Yang, B.Z.: Visser, P.; Bakker. C.E.; Verbeet, M.P.; Oostra, B.A.; Reuser. A.J.; van der Ploeg, A.T. Generalized Glycogen Storage and Cardioniegaly in a Knockout Mouse Model of Pompe Disease. Hum. Mol. Genc,t., 1998, 7, 53-62. 53. Nunoya, T.; Tajima, M.; Mizutani, M. A New Mutant of Japanese Quail (Coturnix Coturnix Japonica) Characterized by Generalized Glycogenosis. Lab. Anim. , 1983, 17, 138-142. 54. Bosshard, N.U.; Hubler, M.; Arnold, S.; Briner, J.; Spycher, M.A.; Sommerlade, H.J.; von Figura, K.; Gitzelmann, R. Spontaneous Mucolipidosis in a Cat: An Animal Model of Human I-Cell Disease. Vet. Puthol., 1996, 33, 1-13. 55. Skelly. B.J.; Sargan, D.R.; Herrtage, M.E.; Winchester, B.G. The Molecular Defect Underlying Canine Fucosidosis. J. Med Genet., 1996, 33, 284-288. 56. Healy, P.J.; Farrow, B.R.; Nicholas, F.W.; Hedberg, K.; Ratcliffe, R. Canine Fucosidosis: A Biochemical and Genetic Investigation. Res. Vet. Sci. 1984, 36, 352-359. 57. Kelly, W.R.; Clague, A.E.: Barns, R.J.; Bate, M.J.; MacKay, B.M. Canine a-L-Fucosidosis: A Storage Disease of Springer Spaniels. Actu Neziroppathol., 1983, 60, 9-13. 58. Zhou, X.Y.; Morreau, H.; Rottier, R.; David, D.; Bonten, E.; Gillemans, N.; Wenger, D.; Grosveld, F.G.; Doherty, P.; Suzuki, K. et (11. Mouse Model for the Lysosomal Disorder Galactosialidosis and Correction of the Phenotype with Overexpressing Erythroid Precursor Cells. Genes Dev.. 1995, 9, 2623-2634. 59. Saftig, P.; Hartmann, D.; Lullmann-Rauch, R.; Evers, M.; Koster, A,; Hetman. M.; von

958

57 Lysosomal Storage Diseases

Figura, K.; Peters, C. Mice Deficient in Lysosomal Acid Phosphatase Develop Lysosomal Storage in the Kidney and Central Nervous System. J. Biol. Chem., 1997, 272, 18628-18635. 60. Maroteaux, P.; Lamy, M. La Pycnodysostose. Pres Med. 70, 999-1002. 61. Frey, J.B.; What Dwarfed Toulouse-Lautrec? Nut. Genet., 1995, 10, 128-130. 62. Polymeropoulos, M.H.; Ortiz De Luna, R.I.; Ide, S.E.; Torres, R.; Rubenstein, J.; Francomano, C.A. The Gene for Pycnodysostosis Maps to Human Chromosome lcen-q21. Nut. Genet., 1995, 10, 238-239. 63. Gelb, B.D.; Edelson, J.G.; Desnick, R.J. Linkage of Pycnodysostosis to Chromosome lq21 by Homozygosity Mapping. Nat. Genet., 1995, 20, 235-237. 64. Everts, V.; Beertsen, W.; Schroder, R. Effects of the Proteinase Inhibitors Leupeptin and E-64 on Osteoclastic Bone Resorption. CalciJ Tissue Znt., 1998, 43, 173-178. 65. Drake, F.H.; Dodds, R.A.; James, I.E.; Connor, J.R.; Debouck, C.; Richardson, S.; LeeRykaczewski, E.; Coleman, L.; Rieman, D.; Barthlow, R.; Hastings, G.; Gowen, M. Cathepsin K, but Not Cathepsins B, L and S is Abundantly Expressed in Human Osteoclasts. J. Biol. Chem., 1996,271, I251 1-12516. 66. Johnson, M.R.; Polymeropoulos, M.H.; Vos, H.L. Ortiz de Luna, R.I.; Francomano, C.A. A Nonsense Mutation in the Cathepsin K Gene Observed in a Family with Pycnodysostosis. Genome Rex, 1996, 6, 1050-1055. 67. Gelb, B.D.; Shi, G.P.; Champdn, H.A.; Desnick, R.J. Pycnodysostosis, a Lysosomal Disease Caused by Cathepsin K Deficiency. Science, 1996, 273, 1236-1238. 68. Garnero, P.; Borel, 0.; Byrjalsen, I.; Ferreras, M.; Drake, F.H.; McQueney, M.S.; Foged, N.T.; Delmas, P.D.; Delaisse, J.M. The Collagenolytic Activity of Cathepsin K is Unique Among Mammalian Proteinases. J. Bid. Chem., 1998, 273-32347-32352. 69. Inui, T.; Ishibashi, 0.;Inaoka, T.; Origane, Y.; Kumegawa, M.; Kokubo, T.; Yamamura, T. Cathepsin K Antisense Oligodeoxynucleotide Inhibits Osteoclastic Bone Resorption. J. Biol. Chem., 1997, 28, 8109-8112. 70. Votta, B.J.; Levy, M.A.; Badger, A,; Bradbeer, J.; Dodds, R.A.; James I.E.; Thompson, S.; Bossard M.J.; Carr, T.; Connor, J.R.; Tomaszek, T.A.; Szewczuk, L.; Drake, F.H.; Veber, D.F.; Gowen, M. Peptide Aldehyde Inhibitors of Cathepsin K Inhibit Bone Resorption Both in vitro and in uivo. J. Bone Miner Rex, 1997, 12, 1396-1406. 71. Yuziuk, J.A.; Bertoni, C..; Beccari, T.; Orlacchio, A,; Wu, Y.Y.; Li, S.C.; Li, Y.T. Specificity of Mouse GMZActivator Protein and P-N-Acetylhexosaminidases A and B. Similarities and Differences with Their Human Counterparts in the Catabolism of G M ~J.. Biol. Chem., 1998, 27366-72. 72. Platt, F.M.; Neises, G.R.; Karlsson, G.B.; Dwek, R.A.; Butters, T.D. N-Butyldeoxygalactonojirimycin Inhibits Glycolipid Biosynthesis but Does Not Affect N-Linked Oligosaccharide Processing. J. Bid. Chem., 1994, 269, 27108-281 14. 73. Platt, F.M.; Butters, T.D. New Therapeutic Prospects for the Glycosphingolipid Lysosmal Storage Diseases. Biochem. Pharmacol., 1998, 56, 421-430. 74. Kolter, T.; Sandhoff, K. Glycosphingolipid Degradation and Animal Models of G M ~ Gangliosidoses. J. Inhert. Metab. Dis., 1998, 21, 548-563. 75. Meikle, P.J.; Hopwood J.J.; Clague, A.E.; Carey, W.F. Prevalence of Lysosomal Storage Disorders. JAMA, 1999, 281, 249-254.

Carbohydrates in Chemistry and Biology Beat Emst, Gerald W. Hart, Pierre Sinay copyright Q WILEY-VCH Verlag GmbH. 2000

58 Genetic Diseases of Glycosylation Tomoya Akama and Michiko N. Fukuda

58.1 Introduction The studies of human genetic diseases provide insight into the function of specific molecules. This is also true in the field of glycobiology. The diseases caused by a defect in the synthesis of carbohydrates include: I-cell disease, Leukocyte adhesion deficiency type I1 (LAD 11), Galactosemia, Carbohydrate-Deficient Glycoprotein Syndrome (CDGS), and HEMPAS (Hereditary Erythroblastic Multinuclearity with Positive Acidified Serum lysis test). This chapter will describe CDGS I and 11, and HEMPAS, as these diseases have been extensively studied in recent years.

58.2 CDGS CDGSs are autosomal recessive inherited diseases due to various failures in N glycan synthesis [ 11. The phenotype of these patients are moderate to severe and symptoms appear in various tissues. So far, CDGSs are classified to four subtypes by its clinical, biological and genetic features. Recently, genes responsible for CDGS type I and I1 have been identified. These genes encode enzymes which are involved in carbohydrate metabolism and N-glycan synthesis. CDGS type 111 and IV cases are very rare and not discussed here since little is known about the biochemical defect in these patients. 58.2.1 CDGS Type I

The major population of CDGSs is identified as type I. Infant patients of CDGS type I show neurologic abnormalities including hypotonia and hyporeflexia, growth

960

58 Genetic Diseases of Glycosylution

retardation, dysmorphic features, such as high nasal bridge, prominent jaw and inverted nipple [2]. About 20% of the patients die in their first year. Ataxia, dysequilibrium and severe developmental delay are shown in childhood. Adult patients suffer from variable degrees of progressive muscular atrophy and develop severe kyphoscoliosis. Carbohydrate analysis of the serum proteins from the patients suggested that CDGS type I is basically caused by unusual N-glycan synthesis on glycoproteins [3, 41. Serum transferrin, which has two complex N-glycans on its polypeptide chain, is frequently used to evaluate the carbohydrate patterns in patients because of its well known and simple glycosylation pattern. The majority of transferrin in normal human serum has disialo-biantennary N-acetyllactosamine on both of Nglycosylation sites. However, mono-glycosylated transferrin, which has disialobianntenary type N-glycan on either of two glycosylation sites, is observed primarily in the case of CDGS type I. Non-glycosylated transferrin, which is rare in healthy serum, is also detected in the patients. Metabolic labeling analysis of fibroblast cells by radioisotopic mannose did not show any qualitative difference in carbohydrate structure on cell extract between normal and CDGS type I patients [ 5 ] . However, the incorporation of mannose into proteins and lipids in patient fibroblasts was lower than normal cells [5]. These results suggested that inefficient oligosaccharide transfer from dolichol-pyrophosphate oligosaccharide donor to acceptor protein is the major defect in CDGS type I because N-glycan synthesis on oligosaccharyl protein seems to be normal but total amount of N-glycosylated carbohydrate is decreased. Dolichol-pyrophosphate-oligosaccharidedonors are produced from several carbohydrate monomers via various intermediates converted by cytosolic enzymes (Figure 1). In this pathway, phosphomannomutase (PMM) activity almost disappeared in CDGS type I patient leukocytes [6]. This enzyme converts mannose-6phosphate to mannose-1-phosphate and vice versa. Deficiency of this enzyme causes reduction of GDP-mannose, dolichol-phosphomannose and eventually produces large amounts of immature dolichol-phospho-oligosccharides.Oligosaccharyl transferase preferably uses the mature dolichol-oligosaccharide donor, and inefficiently attaches the immature form to the acceptor protein [7]. Thus, an accumulation of lipid-donors with short oligosaccharides reduces the efficiency of glycosylation of N-linked oligosaccharides to protein. CDGS type I is mapped genetically on chromosome 16~13.3-3.12by linkage analysis [S]. In human, two PMM isozymes are known and physically mapped to different chromosomes. PMMl is located on chromosome 22q13 and PMM2 is on 1 6 ~ 1 that 3 is the same position as CDGS type I [9]. Missense mutations were found on PMM2 gene from CDGS type I patients [9]. Because the expression pattern of PMMl and PMM2 mRNA is slightly different in some tissues, decrease of PMM2 activity causes serious phenotypes only in particular tissues. Since the most frequent mutation on PMM2, which might result in a complete loss of enzymatic activity, has not been found as homozygote mutation in CDGS type I patients, the complete absence of PMM2 activity may show a lethal phenotype [ 10, 1 I]. Other enzymes involved in synthesis of oligosaccharide precursors are also found as causative genes for CDGS type 1. Mutations on phosphomannoisomerase (PMI),

58.2 CDGS

961

.Glycoprotein

Figure 1. Synthetic pathways of dolichol pyrophosphate oligosaccharide from mannose and glucose. In the cytosol, mannose and glucose are metabolized by several converting enzymes to intermediates and utilized to produce dolichol pyrophosphate oligosaccharide. Deficiencies of phosphomannomutase, phosphomannoisomerase or glucosyltransferase lead to an accumulation of immature oligosaccharide donors and eventually induce aberrant glycosylation patterns of N-linked glycoproteins, which is the major feature of the CDGS type I phenotype.

which converts mannose-6-phosphate and fructose-6-phosphate to each other, and a-1,3-glucosyltransferase (Glc-T), which adds glucose residue to dolichol-pyrophospho-oligosaccharide by using dolichol-phospho-glucose as substrate, have been found mutated in CDGS type I patients who have normal PMM2 activity [12, 131. Based on the differences in causative genes for CDGS type I, this syndrome is classified as three subtypes, type Ia for PMM mutation, type Ib for the inactivation of PMI and type Ic for Glc-T deficiency. Interestingly, phenotypes of type Ib and Ic are reported to be milder than type Ia CDGS [14]. Especially, the phenotype of PMI mutation (type Ib) does not show neurological symptoms. Because PMI mutation in CDGS type Ib does not inhibit dolichol-phospho-mannose synthesis from mannose, oral administration of mannose has been shown to be effective in type Ib [ 151. 58.2.2 CDGS Type I1

So far only two patients have been reported as CDGS type I1 [16]. Both patients showed coarse facial features, low-set ears, widely spaced nipples, ventricular septa1 defects, generalized hypotonia and limb weakness. They had a severe psychomotor retardation but no peripheral neuropathy and a normal cerebellum on nuclear magnetic resonance imaging.

962

58 Genetic Diseuses of Glycosylution

High mannose type N-glycan

,

Man’

I

GlcNAc hansferaae I Man,

+

Man

I Man/Ma~nQlcNAc-GlcNAc-Asn I GkNAC -Man’

IHEMPAS defect]

Hybrii type N-glycan

Man’a%anOlcNAc +SlaOalQkNAc -Mad

OlcNAc -A8n I

I

a-mannorldaaa II I

Ma‘ha+GkNAc-GicNAc-Am I GI~NAC-M~

+s ~ - G a l ~ ~ N-MaAr?‘c

I

n-GlcNAc-QlcNAc-Aon I

QlcNAc-Ma\ I a~GlcNAeblcNAc4an GlcNAc-MaX I

Complex type N-glycan 516-Gal-GlcNAc-Ma

1

Gal transfenme

Sllalyl t m n m I

%arrGlcNAc-GlcNAc-Aan I Sia-GaI-GlcNAc-Mad GlcNAc tranafemae

f

Galtra&nme Sllalyl transferaae

~-(GaIOlcNAc)~GaI-GIcNAc-Ma I >an-GIcNAc-GlcNAc-Am I Sla-(Ga101cNAc)~Gal-GlcNAc-Man

Poiyiactosamine

Figure 2. Biosynthesis of N-glycans and genetic diseases. In the Golgi aparatus, high mannose type oligosaccharides attached to a senescent glycoprotein are processed by several enzymes to a complex type. In CDGS type I1 and HEMPAS, a genetic defect of GlcNAc transferase I1 and a-mannosidase I1 results in the blockage of the N-glycan processing at each step, leading the accumulation of hybrid type oligosaccharides on N-linked glycoproteins.

Serum transferrin isoelectric focusing analysis showed CDGS type I1 patients have quite different spectra of carbohydrate structures from normal [16]. The major population of glycosylated transferrin in serum from normal individuals has disialobiantennary N-glycan on both of N-glycosylation sites. However, transferrin in affected jndividuals is primarily found as disialo-glycoprotein, which has monosialo-monoantennary N-glycans on both N-glycosylation sites [ 161. Structural analyses showed the N-glycan without N-acetylglucosamine, which should be added on Manal-6[GlcNAcPl-2Manal-31Manp 1-4GlcNAcPl-4GlcNAc-Asn by P-1,2-Nacetyl-glucosaminyl transferase 11 (GnT-11) (Figure 2). These results suggest that GnT-I1 may be defective in CDGS type I1 patients. Indeed, GnT-I1 activity was greatly reduced in fibroblast extracts from CDGS type I1 patients [16]. Direct sequence of GnT-I1 coding region from CDGS type I1 patients identified point mutations on the catalytic domain of GnT-I1 [17]. These mutation were inherited consistently with CDGS type I1 phenotypes. Because cell extracts expressing these mutant GnT-I1 proteins, which have the same missense

58.3 HEMPAS

963

mutations as CDGS type 11, were not only reduced GnT-I1 activity but were also decreased protein expression level compared to normal GnT-I1 protein, these mutation may induce instability of GnT-11 protein and also inactivation of the enzyme.

58.3 HEMPAS Congenital dyserythropoietic anemia (CDA) is a group of inherited disorders characterized by an ineffective erythropoiesis, bone marrow erythroid multinuclearity, and secondary tissue siderosis. In 1968, Heimpel and Wendt classified CDA into three types [18]. CDA type I1 is the most common CDA, and is also called HEMPAS, since Crookston characterized this disease as Hereditary Erythroblastic Multinuclearity with Positive Acidified Serum lysis test [ 191. HEMPAS patients suffer from a life-long anemia. Jaundice is common. In most cases the magnitude of anemia is mild; however there are severely affected patients who require constant care and frequent transfusions. The most striking diagnostic feature of HEMPAS is abnormal multinucleated erythroblasts in the bone marrow [19]. Short-lived erythrocytes in HEMPAS (7-34 days in HEMPAS versus 100 days in normal) cause splenomegaly in patients. Also common are liver hemosiderosis and cirrhosis. The incidence of diabetes is high among HEMPAS patients. Severe cases of HEMPAS show mental and sensory abnormalities [19]. Pioneering studies by Crookston [ 191 suggested that HEMPAS is caused by abnormal organization of erythrocyte membranes. In 1975, Joseph and Gockerman found an abnormality in the glycolipid profile of HEMPAS erythrocyte membranes [20]. Their analysis showed an increase in lacto-N-triaosylceramide and 1acto-Ntetraosylceramide in HEMPAS. On the other hand, in 1977 Anselstetter reported that band 3 glycoprotein from HEMPAS erythrocytes migrates slightly faster than the normal band 3 due to underglycosylation [21]. Surface labeling and endo-P-galactosidase treatment analysis demonstrated that both band 3 and band 4.5 glycoproteins, which normally are glycosylated with polylactosamines, virtually lack polylactosamines in HEMPAS [22]. Furthermore, in HEMPAS erythrocytes, polylactosamines are accumulated as polylactosaminyl ceramides. Thus it appears that a genetic factor in HEMPAS blocks the glycosylation of glycoprotein acceptors and shifts polylactosamines to lipid acceptors, resulting in an increase in the lacto-series glycolipids including polylactosaminyl ceramides. Structural analysis of HEMPAS band 3 carbohydrates showed the presence of hybrid type oligosaccharides in addition to complex type oligosaccharides. Analysis of N-glycan structures of transferrin and a1-acid glycoprotein from HEMPAS patients sera revealed that these glycoproteins have hybrid and high mannose type oligosaccharides. These results suggested disruption of N-glycan synthesis near the a-mannosidase I1 (MII) steps (Figure 2). GnT-I1 is also likely to be defective but this enzyme gene has been already identified as a causative gene for CDGS type 11. These analyses also showed that the glycosylation defect is leaky in HEMPAS and

964

58 Genetic Diseases of Glycosylution

suggested the existence of an isozyme of MII, which eventually led to identification of an MII-like enzyme gene, a-mannosidase IIx (MIIx) [23]. Clinical variations of HEMPAS suggested that this disease is a genetically heterogeneous collection, and recent analysis has indeed indicated that this is the case. A patient originating from UK was the first HEMPAS case identified as being defective in the MI1 gene. The patient showed accumulation of hybrid type oligosaccharides with the core structure GnlM5Gn2, the substrate for the Golgi N glycan processing enzyme MIL An enzyme activity assay showed a significant reduction in membrane bound MI1 activity. Northern analysis revealed that lymphoblasts from the patient express the MI1 transcript at less than 10Y0 of normal levels [24]. Recently, another HEMPAS patient originating from Ireland, having mutations in the coding region of the MI1 gene, has been identified [27]. Using gene knockout technology, a mutant mouse in which the MI1 gene is inactivated by homologous recombination has been created [25]. MI1 null mice appear normal at birth and have no obvious deformities or life threatening defects. MI1 null mice, however, show various signs of anemia and remarkable splenomegaly, which are also frequently found in human HEMPAS patients. On the other hand, Iolascon’s group employed linkage analysis using sets of microsatellite markers flanking MII, MIIx and GnT-I1 genes, which are mapped on Chromosome 5q2.1-2.2, 15q25 and 14q21, respectively. Their data excluded linkage to all three genes in HEMPAS cases from southern Italy. Furthermore, they identified the HEMPAS causative gene (CDAN2) in the long arm of chromosome 20q11.2 [26]. However, they also reported that some of HEMPAS families are not mapped on the same locus, suggesting this syndrome has genetic heterogeneity. Biochemical similarities among HEMPAS patients suggest that different gene products affect the same biochemical pathway. Further studies on HEMPAS will provide new information about factors affecting N-glycan biosynthesis.

References 1. Jaeken, J. The carbohydrate-deficient glycoprotein syndromes. Mol. Chem. Neuropathol. 1996,

27. 84-86. 2. Krasnewich, D.; Gahl, W.A. Carbohydrate-deficient glycoprotein syndrome. Adu. Pediat. 1997, 44, 109-140. 3. JaekZn, J.; van Eijk, H. G.; van der Heul, C.; Corbeel, L.; Eeckels, R.; Eggermont, E. Sialic acid-deficient serum and cerebrospinal fluid transferrin in a newly recognized genetic syndrome. Clin. Chim. 1984, 144, 245-247. 4. Yamashita, K.; Ideo, H.; Ohkura, T.; Fukushima, K.; Yuasa, I.; Ohno, K.; Takeshita, K. Sugar chains of serum transferrin from patients with carbohydrate deficient glycoprotein syndrome. Evidence of asparagine-N-linked oligosaccharide transfer deficiency. J. Biol. Chem. 1993,268, 5783-5789. 5. Krasnewich, D.M.; Holt, G. D.; Brantly, M.; Skovby, F.; Redwine, J.; Gahl, W. A. Abnormal synthesis of dolichol-linked oligosaccharides in carbohydrate-deficient glycoprotein syndrome. Glycobiology 1995, 5 , 503-510. 6. Van Schaftingen, E.; Jaeken, J. Phosphomannomutase deficiency is a cause of carbohydratedeficient glycoprotein syndrome type I. FEBS Lett. 1995, 377, 318-320.

References

965

7. Verbert, A. From Glc3Man9GlcNAc2-protein to ManSGlcNAc2-protein: transfer ‘en bloc’ and processing. In Glycoproteins; Montreuil, J.; Vliegenthart, J. F. G.; Schachter, H. Eds., Elsevier, Amsterdam, The Netherlands, 1995, Vol. 29a, 145- 152 8. Bjursell, C.; Stibler, H.; Wahlstrom, J.; Kristiansson, B.; Skovby, F.; Stromme, P.; Blennow, G.; Martinsson, T. Fine mapping of the gene for carbohydrate-deficient glycoprotein syndrome, type I (CDGI): Linkage disequilibrium and founder effect in Scandinavian families. Genomics 1997,39, 247-253. 9. Matthijs, G.; Schollen, E.; Pardon, E.; Veiga-Da-Cunha, M.; Jaeken, J.; Cassiman, J. J.; Van Schaftingen, E. Mutations in PMM2, a phosphomannomutase gene on chromosome 1 6 ~ 1 3in , carbohydrate-deficient glycoprotein type I syndrome (Jaeken syndrome). Nut1 Genet. 1997, 16, 88-92. 10. Matthijs, G.; Schollen, E.; Van Schaftingen, E.; Cassiman, J. J.; Jaeken, J. Lack of homozygotes for the most frequent disease allele in carbohydrate-deficient Glycoprotein syndrome type 1A. A m . J. Hum. Genet. 1998,62, 542-550. 1 1. Kjaergaard, S.; Skovby, F.; Schwartz, M. Absence of homozygosity for predominant mutations in PMM2 in Danish patients with carbohydrate-deficient glycoprotein syndrome type 1. Eur. J. Hum. Genet. 1998, 6, 331L336. 12. Burda, P.; Borsig, L.; deRijkvanAndel, J.; Wevers, R.; Jaeken, J.; Carchon, H.; Berger, E. G.; Aebi, M. A novel carbohydrate-deficient glycoprotein syndrome characterized by a deficiency in glucosylation of the dolichol-linked oligosaccharide. J. Clin. Invest. 1998, 102, 647 -652. 13. Jaeken, J.; Matthijs, G.; Saudubray, J. M.; DionisiVici, C.; Bertini, E.; delonlay, P.; Henri, H.; Carchon, H.; Schollen, E.; Van Schaftingen, E. Phosphomannose isomerase deficiency: A carbohydrate-deficient glycoprotein syndrome with hepatic-intestinal presentation. A m . J. Hum. Genet. 1998, 62, 1535-1539. 14. Korner, C.; Knauer, R.; Holzbach, U.; Hanefeld, F.; Lehle. L.; von Figura, K . Carbohydratedeficient glycoprotein syndrome type V: deficiency of dolichyl-P-Glc:Man9GlcNAc2-PPdolichyl glucosyltransferase. Proc. Nut1 Acad. Sci. U S A 1998, 95, 13200- 13205. 15. Niehues, R.; Hasilik, M.; Alton, G.; Korner, C.; SchiebeSukumar, M.; Koch, H. G.; Zimmer, K. P.; Wu, R. R.; Harms, E.; Reiter, K.; von Figura, K.; Freeze, H. H. Carbohydrate-deficient glycoprotein syndrome type Ib-Phosphomannose isomerase deficiency and mannose therapy. J. Clin.Invest. 1998, 101, 1414-1420. 16. Jaeken, J.; Schachter, H.; Carchon, H.; Decock, P.; Coddeville, B.; Spik, G. Carbohydrate deficient: Glycoprotein syndrome type 11: A deficiency in Golgi localised N-acetyl-glucosaminyltransferase 11. Arch. Dis.Child.1994, 71, 123-127. 17. Tan, J.; Dunn, J.; Jaeken: J.; Schachter, H. Mutations in the MGAT2 gene controlling complex N-glycan synthesis cause carbohydrate-deficient glycoprotein syndrome type 11, an autosomal recessive disease with defective brain development. Am. J. Hum. Genet. 1996, 59, 810-817. 18. Heimpel, H.; Wendt, F. Congenital Dyserythropoietic anemia with karyorrhexis and multinuclearity of erythroblasts. Helv. Med. Acta. 1968, 34, 103-1 15. 19. Crookston, J. H.; Crookston, M. C.; Burnie, K. L.; Francombe, W. H.; Dacie, J. V.; Davis, J. A,; Lewis, S. M. Hereditary erythroblastic multinuclearity associated with a positive acidified serum test; a typical congenital dyserythropoietic anemia. Brit. J. Huematol. 1969, 17, I 1 26. 20. Joseph, K . C.; Gockerman, J. P.; Alving, C . R. Abnormal lipid composition of the red cell me brane in congenital dyserythropoietic anemia type I1 (HEMPAS). J. Lab. Clin. Med. 1975, 85, 34-40. 21. Anselstetter, V. B.; Horstmann, H.-J.; Heimpel, H. Congenital dyserythropoietic anaemia, types I and 11: Aberrant pattern of erythrocyte membrane proteins in CDA IT, as revealed by two dimensional polyacrylamide gel electrophoresis. Brit. J. Huematol. 1977, 35, 209-21. 22. Fukuda, M. N.; Papayannopoulou, T.; Gordon-Smith, E. C.; Rochant. H.; Testa, H. Defect in glycosylation of erythrocyte membrane proteins in congenital dyserythropoietic anaemia type II (HEMPAS). Brit. J. Huematol. 1984, 56, 55-68. 23. Misago, M.; Liao, Y.-F.; Kudo, S.; Eto, S.; Mattei, M.-G.; Moremen, K. W.; Fukuda, M. N. Molecular cloning and expression of cDNAs encoding human a-mannosidase I1 and a novel a-mannosidase IIx isozyme. Proc. Nail Acad. Sci. U S A 1995, 92, 11766-1 1770.

966

58 Genetic Diseuses of Glycosylation

24. Fukuda, M. N.; Masri, K. A,; Dell, A,; Luzatto, L.; Moremen, K. W. Incomplete synthesis of N-glycans in congenital dyserythropoietic anemia type I1 caused by a defect in the gene encoding a-mannosidase 11. Proc. Natl Acad. Sci. USA 1990, 87, 7443-7447. 25. Chui, D.; Oh-eda, M.; Liao, Y.-F.; Penneerselvan, K.; Lal, A.; Marek, K. W.; Freeze, H.; Moremen, K. W.; Fukuda, M. N.; Marth, J. D. Alpha-mannosidase-I1 deficiency results in dyserythropoiesis and unveils an alternate pathway in oligosaccharide biosynthesis. Cell 1997, 90, 157-167. 26. Gasparini, P.; Miraglia, E.; del Giudice, M.; Delauney, J.; Totaro, A.; Granatiero, M.; Merchionda, S.; Zalante, L.; Iolascon, A. Localizaton of the congenital dyserythroblastic anemia I1 locus to chromosome 20q11.2 by genomewide search. Am. J. Hum. Genet. 1997,61, 11 12-1 116. 27. Akama et al. Unpublished data.

Carbohydrates in Chemistry and Biology Beat Emst, Gerald W. Hart, Pierre Sinay copyright Q WILEY-VCH Verlag GmbH. 2000

59 Glycobiology of Helicobacter pylori and Gastric Disease Karl-Anders Karlsson

59.1 Introduction The first isolation of Helicobacter pylori in 1982 by Marshall and Warren [ l ] marked a new era in gastroenterology and dramatically changed the diagnosis and therapy of gastritis and peptic ulcer. This newly classified bacterium is a humanspecific Gram-negative, urease-positive, curved or slightly spiral organism (see [2] for a review). In developing countries 70-90%, and in developed countries 25-50%, of the population carries H. pylori. All infected individuals develop chronic gastric inflammation, but this condition usually is asymptomatic, but in 10-20% of colonized individuals clinical disease occurs. H. pylori causes the majority of all cases of gastric and duodenal ulcers, and carriage of the bacterium also is strongly associated with the risk of development of atrophic gastritis, which is a precursor lesion to gastric cancer. This is important since gastric cancer is the second leading cause of cancer death in the world, and H. pylori was therefore defined as a class 1 carcinogen. Potential relations to other common human diseases are being investigated. Current therapy is based on two antibiotics in combination with a proton pump inhibitor, which treatment results in very low recurrence. However, extension of this treatment for a global prevention of gastric diseases is not recommendable, owing to the potential development of resistance to antibiotics. There are much current efforts on the development of vaccines, which has been successful in animal models [2]. However, the microbe has adapted for a long time in the human stomach and natural clearance of H. pylori is very rare. On the contrary, a vigorous defense response fails to resolve the infection and may contribute to the severity of the disease [ 3 ] . Therefore, much more work on the pathology behind the infection is probably required before effective vaccines may be designed . An alternative is antiadhesion therapy based on carbohydrate receptor analogs (see [4] for an overview). In the following, the present knowledge of H. pylori-related glycobiology will be briefly discussed. The surface-located lipopolysaccharides of the bacterium represent a molecular mimicry between the microbe and the host. This may be a way to

968

59 Glycobiology of Helicobacter pylori and Gastric Disease

avoid host defense mechanisms but is also a potential basis of autoimmune disease. The unique complexity of carbohydrate-binding specificities that has been demonstrated may both complicate and provide helpful alternatives for antiadhesion therapy.

59.2 The Bacterial Surface and Molecular Mimicry The classical 0 antigens of many bacteria have been successfully used for vaccine prophylaxis against infections [ 5 ] . However, in case of H. pylori the group of Aspinall reported a careful and impressive chemical study which revealed that the 0 antigen region carried an extended N-acetyllactosamine chain which was nonfucosylated or carried Lewis x, Galp4(Fuca3)GlcNAcp3, or Lewis y, Fuca2Gal~4(Fuca3)GlcNAcf33, determinants [6, 71. These are identical with determinants expressed on human cells including granulocytes and is a molecular mimicry which may be an adaptation to avoid the host immune response. A striking feature was a variation between strains in the detailed structures that are expressed. Furthermore, these differences indicate a class of lipopolysaccharides whose mechanisms of biosynthesis lead to overall architectures different from those of most lipopolysaccharides from enteric bacteria [S]. Initiated by serological screening of bacterial strains using a number of well characterized antibodies, and also by chemical analysis, a recent paper [9] documented, the presence of type 1 chains, Galp3GlcNAc, nonfucosylated (Lewis c) or substituted with Fuca2 on Gal (H-1 epitope or Lewis d) or Fuca4 on GlcNAc (Lewis a epitope). However, both these substitutions combined (Lewis b epitope) were not found chemically, contradicting the Lewis b activity detected by a monoclonal antibody. These type 1 Lewis epitopes are usually abundant on gastric mucosa cells. A separate study showed by immunoassay that the relative proportion of Lewis expression in H. pylori corresponds to the host Lewis phenotype, suggesting a selection for host-adapted organisms [ 101. A recent review discusses these data [ 113. Autoantibodies against gastric epithelial cells are detectable in up to 50% of patients with chronic, active H. pylori gastritis. However, attempts to relate these autoantibodies to Lewis antigens have failed so far [12]. Therefore, epitopes other than Lewis antigens are the autoimmune targets, and the pathogenic mechanism may be other than molecular mimicry between H. pylori and the host.

59.3 Host Surfaces and H. pylovi Recognition of Glycoconjugates: Unique Complexity Many microbes carry carbohydrate-binding specificities, but only in a few cases have these been clearly linked to pathogenicity [4]. Most notable is influenza virus,

59.3 Host Surfaces and H. pylori Recognition of Glycoconjugates

969

Table 1. Carbohydrate-binding specificities detected for Helicobacter pylori. Binding specificity

Reference

Sialic Acid Epitope 1: NeuAca3Galp4Glc or -GlcNAc Epitope 2: Unidentified on polyglycosylceramides Sulfatide Heparan sulfate Fucose (Lewis b) Gangliotetraosylceramide: Galp3GalNAc~4GalP4GlcpCer Lactosylceramide At least two additional epitopes Soluble neutrophil-activating protein (Hp-NAP): Neutrophil gangliosides and sulfatide Non-carbohydrate epitope: Phosphatidylethanolamine

for which crystal structures are known for the two carbohydrate-binding proteins, the haemagglutinin and the neuraminidase, allowing structure-based drug design of two separate sialic acid analogs, which are currently in clinical trials against influenza (see [4]). H. pylori differs from all other microbes so far characterized, concerning complexity of binding properties. About ten separate carbohydrate-binding specificities have been indicated (see Table l), the meaning of which is still largely unknown. In the following some of these specificities will be discussed before some general conclusions and proposals are provided.

59.3.1 Sialic Acid The sialic acid-dependent binding was first documented by hemagglutination studies [13], showing both abolishment of binding after treatment of red cells with neuraminidase, and inhibition of hemagglutination with sialyllactose and various sialylated glycoconjugates. The specificity was later shown restricted to 3-linked sialic acid based on inhibition studies [22] and solid-phase binding [23]. The requirement for different functional groups of sialic acid is currently under investigation [24]. The binding was expressed only after growth on agar, but not in broth, and all standard strains do not bind in a sialic acid-dependent way. Thus strain CCUG 17874 is a positive while CCUG 17875 is a negative binder. However, fresh clinical isolates were claimed to consistently express this specificity [ 251. The corresponding assumed adhesin protein on H. pylori was reported identified by genetical cloning in 1993 [26], and an antibody was raised which detected activity on bacterial cell bodies. However, a later study provided evidence for a cytoplasmic localization and a lipoprotein character of the cloned protein [27], and

970

59 Glycobiology of Helicobacter pylori and Gastric Disease

knock-out of the gene did not change adhesion properties. The lipoprotein character was confirmed by a separate study repeating the cloning and expression, but the localization as shown by a carefully defined antibody and immuno-gold labeling was neither intracellularly nor to the cell body but rather restricted to the flagellar sheath [28], and the adhesion properties were not affected by gene inactivation. Therefore, the adhesin corresponding to the sialic acid-dependent binding first reported in 1988 [ 131 still awaits identification. However, a laminin-binding protein requiring sialic acid has recently been identified in H. pylori, although not yet linked to the actual adhesin [29]. Recently, evidence was provided for a second sialic acid-dependent specificity carried by bacterial cells [14, 301. This binding is expressed also after bacterial growth in broth, when the above-mentioned binding is negative, and is apparently restricted to human polyglycosylceramides, microheterogeneous and branched polylactosamine glycolipids present in one of the target cells for H. pylori (see discussion below), the inflammatory cell [31]. The epitope being recognized (named epitope 2 to distinguish it from the first published epitope, named epitope 1, see Table 1) has not yet been identified. The binding is absolutely dependent on sialic acid and its glycerol tail, since neuraminidase hydrolysis or mild periodate oxidation completely abolish activity [ 141; a molecular dynamics simulation has been performed which indicates bisected hydrogen bonds between C9 of NeuAc and GlcNAcs of two branches of a hypothetical epitope [4].

59.3.2 Sulfatide

Sulfatide, sulfate-3-GalpCer, is abundant in gastrointestinal epithelial cells, and this glycolipid from a mucosa scraping of human stomach has been shown on a thinlayer chromatogram (tlc) to bind H. pylori strongly [15]. This was confirmed for KATO I11 cells, a gastric cancer cell line [32]. Recently, it was shown that low pH induced the binding to sulfatide, as assayed on tlc plates [33], and the stress protein hsp70 was subsequently cloned and sequenced and shown to be upregulated by low pH and expressed on the bacterial surface [34].The binding to sulfatide was blocked by anti-hsp70 antibodies. There is also a binding to sulfated mucins and sulfate-3Gal and sulfate-3-Lewis x coupled up to polyacrylamide [35]. The neutrophilactivating protein, Hp-NAP, of H. pylori was proposed to act as an adhesin which binds sulfated mucins [36], and the isolated soluble Hp-NAP was shown to bind sulfatide [20], see also Table 1.

59.3.3 Heparan Sulfate The demonstration that H. pylori binds heparan sulfate [ 161 was further studied by inhibition by various compounds, and heparin-binding proteins were detected on bacterial cells [37]. Such proteins apparently facilitate bacterial phagocytosis by polymorphonuclear cells, since bacteria with a strong heparan sulfate binding were

59.3 Host Surfaces and H. pylori Recognition of Glycoconjuyates

971

ingested in greater numbers than those with a weak binding [38]. A possible relation of heparan sulfate and sulfatide (above) binding has not been investigated. 59.3.4 Fucose-Dependent Binding (H-1 and Lewis b)

This specificity is the most extensively studied among H. pylori carbohydrate-binding properties, including cloning of the adhesin and design of a transgenic animal model. Recognition of H-1 and Lewis b glycoconjugates but not of fucosylated type 2 derivatives was documented in a careful study [ 171, and evidence that these or similar structures mediated the adherence of bacterial cells to human stomach epithelium was obtained by direct binding to biopsy samples and comparison with the distribution of labelling with various antibodies and lectins [39]. 63 out of 95 isolates tested expressed a binding to a Lewis b conjugate [40]. The Lewis b-binding adhesin, BabA, was purified by an elegant receptor-directed affinity tagging [40]. Albumin with attached Lewis b saccharide was coupled to a crosslinking reagent tagged with biotin. This probe was mixed with living bacteria for binding to the assumed adhesin, followed by UV irradiation to crosslink the biotin derivative with the adhesin, and final reduction to cleave the bond to albumin-Lewis b, which then was washed away. After SDS extraction the tagged adhesin molecules were picked up by beads coated with streptavidin. The adhesin gene was present as two slightly different alleles, babAl and babA2; inactivation of babA2, but not of babA1, resulted in loss of Lewis b-binding activity. Boren’s group is currently studying intensely the molecular genetics of this adhesin to test its role for colonization and disease. The Lewis gene, coding for Lewis a and b antigens, is restricted to primates, which excludes the use of common rodent models to test the relevance for disease of the Lewis b specificity. However, in an elegant approach the human Lewis fucosyltransferase gene was transfected to mouse and targeted to express Lewis b in mouse stomach pit cells [41]. H. pylori cells bound in vitro to these cells of the transgenic mouse but not cells of the normal littermates. Persistent infection in vivo by a clinical isolate of N.pylori occurred at comparable microbial densities in the two mice [42]. However, only in the transgenic mouse did the attachment result in the production of autoantibodies to Lewis x epitopes shared by bacteria and acid-secreting parietal cells, and in chronic gastritis, and parietal cell loss. Thus, the interaction between a bacterial adhesin and a host receptor can alter disease outcome without affecting colonization levels, thereby demonstrating that the mode of colonization is important for pathogenesis. 59.3.5 Gangliotetraosylceramide

The consistent binding of this glycolipid by H. pylori [ 181 and many other bacteria [43] is still an enigma, because this glycolipid has not been found in human stomach or other human tissues [51]. However, it exists substituted with NeuAc or Fuc and potential microbial hydrolases may expose the binding epitope. Alternatively, this represents a crossbinding with some yet unidentified epitope.

972

59 Glycobiology of Helicobacter pylori and Gastric Disease

59.3.6 Lactosylceramide Lactosylceramide is a glycolipid which also is recognized by many bacteria [43]. Therefore this binding is unlikely to mediate tropism. In case of H. pylori [ 191 the selective binding only to particular molecular species of lactosylceramide, namely ceramide composed of sphingosine/phytosphingosine and 2-~-hydroxylfatty acids (epithelial type), but not to species with sphingosine and nonhydroxyl fatty acids (non-epithelial type), is similar to many other bacteria. Binding species were detected in human stomach epithelium and shown by mass spectrometry to consist of phytosphingosine and 2-~-hydroxylfatty acids with 16-24 carbon atoms. The selective binding was illustrated by molecular modeling. Lactose, which exists in bound form only as lactosylceramide which is membrane-specific, was proposed to function as a membrane-close-anchoring site, selected after tissue-specific targeting through other specificities [ 19, 431.

59.4 The Meaning of Multiple Binding Specificities

I have speculated elsewhere on the meaning of H. pylori complexity [4]. The dependence on cultivation conditions for binding to the two sialic acid-containing epitopes [14, 301, and the stress induction of the sulfatide binding [33], may reflect a regulated expression in vivo of genes encoding bacterial adhesins based on chemical signals in the gastric microenvironments, to provide a local adaptive advantage. Bacteria are found mainly in mucus, where they are dividing, but also on the mucosal or basolateral epithelial surface, at the basement membrane or on neutrophils. Sulfatide is abundant in gastrointestinal epithelium, but is lacking in neutrophils [20].In contrast, sialic acid is very rare in human gastric epithelium but abundant in various glycoconjugates of neutrophils [52]. Therefore, sulfatide may mediate targeting to epithelium and sialic acid to neutrophils. Interaction with neutrophils is potentially suicidal for the bacterium, since contact produces a rapid inflammatory burst (within seconds) often followed by a slower phagocytosis (in minutes) [44]. However, in general the bacteria are not damaged but instead apparently use products of the inflammation for nutritional purposes [45].A potential symbiosis of bacteria and inflammatory cells is supported not only by the existence of neutrophilbinding adhesins, but also by the soluble Hp-NAP ([20], Table l), which is a neutrophil-binding protein which recruits inflammatory cells by inducing the expression of neutrophil CD11b/CD18 for an increased adhesiveness to endothelial cells [46]. Provided that all binding specificities that have been found for H. pylovi are essential for persistent colonization, a therapy based on interference with only one of them may show successful. Alternatively, a complex approach may have to be designed.

References

913

59.5 Aspects for the Future The complexity and sophistication of H. pylori glycobiology and interaction with the host will probably require large additional research efforts to result in a practical prevention of the common and important gastric diseases which are results of the presence of this microbe in a majority of the global human population. Of great help is the access to the complete genome sequence of one strain [47] and recently also a second sequence of an unrelated strain for comparison [48]. As outlined elsewhere [4], rational glycotechnology approaches will help identifying carbohydrate-binding adhesins and describing surface glycoconjugate dynamics, which may get the basis of a rational therapy. However, the diversity of various glycoforms in human and animals and the difficulties to get access to relevant human tissue material may get rate-limiting for a good progress [4]. A possible relation between susceptibility to disease and blood group phenotype of the host is still unclear. Genetic manipulation of both bacteria [40] and model animals [41] will assist in testing the biological relevance of various properties and parameters. Advanced knowledge of e.g. the regulation of adhesin expression may provide selective inhibitors of expression, which may induce an ecologically balanced clearance of the bacteria from the stomach. Of great interest are efforts within industry to produce natural oligosaccharides (25, 49, 501 and optimized synthetic analogs [4, 501 for anti-infection therapy. The outcome of current clinical tests on influenza and several gastrointestinal disorders will certainly determine the will in the near future to improve investments in basic glycobiology. References 1. B.J. Marshall, J.R. Warren, Unidentified curved bacilli in the stomach of patients with gastrititis and peptic ulceration, Lancet, 1984, i, 1311- 1315. 2. B.E. Dunn, H. Cohen, M.J. Blaser, Helicobucter pylori, Clin. Microbiol. Reo., 1997, 10, 720741. 3. J.L. Telford, A. Covacci, R. Rappuoli, P. Ghiara, Immunobiology of’ Helicobacter pylori infection, Curr. Opin. Irnrnunol., 1997, 9, 498-503. 4. K.-A. Karlsson, Meaning and therapeutic potential of microbial recognition of host glycoconjugates, Mol. Microbid., 1998,29, l l l . 5. R. Bell, G. Torrigiani (editors), Towards better carbohydrate vaccines, 1987, John Wiley & Sons; London. 6. G.O. Aspinall, M.A. Monteiro, H. Pang, E.J. Walsh, A.P. Moran, Lipopolysaccharides of the Helicobacter pylori type strain TCTC 11637 (ATCC 43504): structure of the 0 antigen chain and core oligosaccharide regions, Biochemistry, 1996, 35, 2489-2497. 7. G.O. Aspinall, M.A. Monteiro, Lipopolysaccharides of Helicobucter pylori strains P466 and M019: structures of the 0 antigen and core oligosaccharide regions, Biochemistry, 1996, 35, 2498-2504. 8. G.O. Aspinall, M.A. Monteiro, R.T. Shaver, L.A. Kurjanczyk, J.L. Penner, Lipopolysaccharides of Helicobacter pylori serogroups 0 : 3 and 0:6-structures of a class of lipopolysaccharides with reference to the location of oligomeric units of D-glycero-alpha-D-manno-heptose residues, Eur. J. Biochem., 1997,248, 592.~601. -

974

59 Glycobioloyy of Helicobacter pylori und Gastric Diseuse

9. M.A. Monteiro, K.H. Chan, D.A. Rasko, D.E. Taylor, P.Y. Zheng, B.J. Appelmelk et al., Simultaneous expression of type 1 and type 2 Lewis blood group antigens by Helicobacter pylori lipopolysaccharides. Molecular mimicry between H. pylori lipopolysaccharides and human gastric epithelial cell surface glycoforms, J. Biol. Chem., 1998,273, 11533-11543. 10. H.P. Wirth, M. Yang, R.M. Peek, K.T. Tham, M.J. Blaser, Helicobacter pylori Lewis expression is related to host Lewis phenotype, Gastroenterology, 1997, 113, 1091-1098. 11. A.P. Moran, G.O. Aspinall, Unique structural and biological features of Helicobacter pylori lipopolysaccharides, Prog. Clin. Biol. Res., 1998, 397, 37-49. 12. G. Faller, H. Steininger, B. Appelmelk, T. Kirchner, Evidence of novel pathogenic pathways for the formation of antigastric autoantibodies in Helicobacter pylori gastritis, J. Clin. Pathol., 1998, 51, 244-245. 13. D.G. Evans, D.J. Evans Jr, J.J. Molds, D.Y. Graham, N-acetylneuraminyllactose-binding fibrillar hemagglutinin of Campylobucter pylori: a putative colonization factor antigen, Infect. Immun., 1988,56, 2896-2906. 14. H. Miller-Podraza, M. Abul Milh, J. Bergstrom, K.-A. Karlsson, Recognition of glycoconjugates by Helicobacter pylori: An apparently high-affinity binding of human polyglycosylceramides, a second sialic acid-based specificity, Glycoconj. J., 1996, 13, 453- 460. 15. T. Saitoh. H. Natomi. W. Zhao. K. Okuzumi. K. Supano. M. Iwamori et al.. Identification of glycolipid receptors for Helicobacter pylori by TLC Tmmunostaining, FEBS’Lett., 1991, 282, 385-387. 16. F. Ascencio, L.-A. Fransson, T. Wadstrom, Affinity of the gastric pathogen Helicobacter pylori for the N-sulphated glycosaminoglycan heparan sulphate, J. Med. Microbiol., 1993, 38, 240244. 17. T. Bortn, P. Falk, K.A. Roth, G. Larson, S. Normark, Atachment of Helicobacter pylori to human gastric epithelium mediated by blood group antigens, Science, 1993, 262, 1892-1895. 18. B. Gold, M. Huesca, P. Sherman, C.A. Lingwood, Helicobacter mustelae and Helicobacter pylpri bind to common receptors in vitro, Infect. Immun., 1993, 61, 2632-2638. 19. J. Angstrom, S. Teneberg, M. Abul Milh, T. Larsson, I. Leonardsson, B.-M. Olsson et al., The lactosylceramide binding specificity of Helicobacter pylori, Glycobiology, 1998, 8, 297-309. 20. S. Teneberg, H. Miller-Podraza, H.C. Lampert, D.J. Evans Jr, D.G. Evans, D. Danielsson et al., Carbohydrate binding specificity of the neutrophil-activating protein of Helicobacter pylori, J. Biol. Chem., 1997,272, 19067-19071. 21. C.A. Lingwood, M. Huesca, A. Kuskis, The glycerolipid receptor for Helicobacter pylori (and exoenzyme S) is phosphatidylethanolamine, Infect. Immun., 1992, 60, 2470-2474. 22. S. Hirmo, S. Kelm, R. Schauer, B. Nilsson, T. Wadstrom, Adhesion of Helicobacter pylori strains to a-2,3-linked sialic acids, Glycoconj. J., 1996, 13, 1-7. 23. L. Johansson, K.-A. Karlsson, Selective binding by Helicobacter pylori of Ieucocyte gangliosides with 3-linked sialic acid, as identified by a new approach of linkage analysis, Glycoconj. J., 1998, 15, 713-721. 24. H. Miller-Podraza, T. Larsson, J. Nilsson, S. Teneberg, M. Matrosovich, L. Johansson, Epitope dissection of receptor-active gangliosides with affinity for Helicobacter pylori and influenza virus, Acta Biochim. Polon., 1998, 45, 439-449. 25. P.M. Simon, P.L. Goode, A. Mobasseri, D. Zopf, Inhibition of Helicobacter pylori binding to gastrointestinal epithelial cells by sialic acid-containing oligosaccharides, Infect. Immun., 1997, 65, 150-757. 26. D.G. Evans, T.K. Karjalainen, D.J. Evans Jr, D.Y. Graham, C.H. Lee, Cloning, nucleotide sequence, and expression of a gene encoding an adhesin subunit protein of Helicobacter pylori, J. Bacteriol., 1993, 175, 674-683. 27. P.W. 0 Toole, L. Janzon, P. Doig, J. Huang, M. Kostrzynska, T.J. Trust, The putative neuraminyllactose-binding hemagglutinin HpaA of Helicobacter pylori CCUG 17874 is a lipoprotein, J. Bacteriol., 1995, 177, 6049-6057. 28. A.C. Jones, R.P.H. Logan, S. Foynes, A. Cockayne, B.W. Wren, C.W. Penn, A flagellar sheath protein of Helicobacter pylori is identical to HpaA, a putative N-acetylneuraminyllactosebinding hemagglutinin, but is not an adhesin for AGS cells, J. Bacteriol., 1997, 179, 5643-5647. 29. K.H. Valkonen, T. Wadstrom, A.P. Moran, Identification of the N-acetylneuraminyllactosespecific laminin-binding protein of Helicobacter pylori, Infect. Immun., 1997, 65, 916-923.

References

975

30. H. Miller-Podraza, J. Bergstrom. M. Abul Milh, K.-A. Karlsson, Recognition of glycoconjugates by Helicohacter pylori. Comparison of two sialic acid-dependent specificities based on haemagglutination and binding to human erythrocyte glycoconjugates, Glycoconj. J., 1997, 14, 467-471. 31. H. Miller-Podraza, G. Stenhagen, T. Larsson, C. Anderson, K.-A. Karlsson, Screening for the presence of polyglycosylceramides in various tissues: Partial characterization of blood groupactive complex glycosphingolipids of rabbit and dog small intestines, Glycoconj. J., 1997, f 4 , 231-239. 32. S. Kamisago, M. Iwamori, T. Tai, K. Mitamura, Y. Yazaki, K. Sugano, Role of sulfatides in adhesion of Helicobacter pylori to gastric cancer cells, Infect. Immun., 1996, 64, 624628. 33. M. Huesca, S. Borgia, P. Hoffman, C.A. Lingwood, Acidic pH changes receptor binding specificity of Helicobacter pylori: A binary adhesion model in which surface heat shock (stress) proteins mediate sulfatide recognition in gastric colonization, Infect. Immun., 1996, 64, 26432648. 34. M. Huesca, A. Goodwin, A. Bhagwansingh, P. Hoffman, C.A. Lingwood, Characterization of an acidic-pH-inducible stress protein (hsp70), a putative sulfatide binding adhesin, from Helicobacter pylori, Inj&t. Immun., 1998, 66, 4061-4067. 35. E.C. Veerman, C.M. Bank. E. Namavar, B.J. Auuelmelk. J.G. Bolscher. A.V. NieuwAmerongen, Sulfated glycans on oral mucin as receptors for Helicohacter pylori, Glycobiology, 1997, 7. 737-743. 36. F. Namayar, M. Sparrius, E.C. Veerman, B.J. Appelmelk, C.M. Vandenbroucke-Grauls, Neutrophil-activating protein mediates adhesion of Helicobacter pylori to sulfated carbohydrates on high-molecular-weight salivary mucins, Infect. Zmmun., 1998, 66, 444-447. 37. M. Utt, T. Wadstrom, Identification of heparan sulphate binding proteins of Helicohacter pylori: Inhibition of heparan sulfate binding with sulphated carbohydrate polymers, J. Med. Microhiol., 1997, 46, 541-546. 38. M. Chmiela, A. Ljung, W. Rudnicka, T. Wadstrom, Phagocytosis of Helicohacter pylori bacteria differing in the heparan sulfate binding by human polymorphonuclear leukocytes, Zentralhl. Bakteriol., 1996, 283, 346-350. 39. P. Falk, K.A. Roth, T. Boren, U. Westblom, J.I. Gordon, S. Normark, An in vitro adherence assay reveals that Helicobucter pylori exhibits cell lineage-specific tropism in the human gastric epithelium, Proc. Nut1 Aca+ Sci. USA, 1993, YO, 2035-2039. 40. D. Ilver, A. Arnqvist, J. Ogren, I.-M. Frick, D. Kersulyte, E.T. Incecik et al., Helicobacter pylori adhesin binding fucosylated histo-blood group antigens revealed by retagging, Science, 1998,279, 373-371. 41. P.G. Falk, L. Bry, J. Holgersson, J.I. Gordon. Expression of a human u-1,3/4-fucosyltransferase in the pit cell lineage of FVB/N mouse stomach results in production of Le’-containing glycoconjugates: A potential transgenic mouse model for studying Helicobactrr pylori infection, Proc. Natl Acad. Sci. USA, 1995, 92, 1515-1519. 42. J.L. Guruge, P.G. Falk, R.G. Lorenz, M. Dans, H.-P. Wirth, M.J. Blaser et a/., Epithelial attachment alters the outcome of Helicohucter pylori infection, Proc. Natl Acad. Sci. USA, 1998, 95, 3925-3930. 43. K.-A. Karlsson, Animal glycosphingolipids as membrane attachment sites for bacteria, Annu. Rev. Biochem., 1989, 58,309-350. 44. H. Rautelin, B. Blomberg, H. Fredlund, G. Jarnerot, D. Danielsson, Incidence of Helicohucter pylori strains activating neutrophils in patients with peptic ulcer disease, Gut, 1993, 34, 599603. 45. D.E. Kirschner, M.J. Blaser, The dynamics of Helicohacter pylori infection in the human stomach, J. Theor. Biol., 1995, f76, 281-290. 46. D.J. Evans Jr, D.G. Evans, T. Takemura, H. Nakano, H.C. Lampert, D.Y. Graham et a/., Characterization of a Helicobacter pylori neutrophil-activating protein, Infect. Immun., 1995, 63, 2213-2220. 47. J.-F. Tomb, 0.White, A.R. Kerlavdge, R.A. Clayton, G.G. Sutton, R.D. Fleishmann et al., The complete genome sequence of the gastric pathogen Helicobacter pylori, Nature, 1997, 388, 539-547.

..

976

59 Glycobiology of Helicobacter pylori and Gastric Disease

48. R.A. Alm, L.-S.L. Ling, D.T. Moir, B.L. King, E.D. Brown, P.C. Doig et al., Genomicsequence comparison of two unrelated isolates of the human gastric pathogen Helicobucter pylori, Nature, 1999, 397, 176-180. 49. D. Zopf, S. Roth, Oligosaccharide anti-infective agents, Lancet, 1996, 347, 1017-1021. 50. J.C. McAuliffe, 0. Hindsgaul, Carbohydrate drugs-an ongoing challenge, Chem. and Ind., 1997, No vol., 170-174. 51. S. Teneberg et al., unpublished results. 52. H. Miller-Podraza, S. Teneberg et al. , submitted for publication. 53. K.-A. Karlsson, S. Teneberg et al. , unpublished results.

Carbohydrates in Chemistry and Biology Beat Emst, Gerald W. Hart, Pierre Sinay copyright Q WILEY-VCH Verlag GmbH. 2000

60 Immunoglobulin G Glycosylation and Galactosyltransferase Changes in Rheumatoid Arthritis John S. Axford

60.1 Introduction Oligosaccharide biochemistry has been an important field of research for many years and this area of basic science continues to flourish [1]. Glycobiology deals with the nature and role of carbohydrates in biological events [2-41. It is now clear the oligosaccharides may indeed be more than a decorative irrelevance when it comes to molecular mechanisms of disease [5-141. The last decade has witnessed an exponential interest in the associations of oligosaccharides with disease mechanisms, and significant advances have been made in the design of carbohydrate-based therapies and diagnostic techniques.

60.2 Oligosaccharide Synthesis Oligosaccharides are mediators of biospecific information by virtue of their structural complexity. As an example, the sialyl-Lewisx tetrasaccharide is one of the crucial molecules involved in cell adhesion and trafficking [4, 15, 161. The mechanisms of their synthesis are therefore crucial and the glycosyltransferases are at the centre stage of the synthesis of glycoside bonds [ 171. The most important organelle involved in glycosylation is the Golgi apparatus (GA) originally described as “appareil reticulaire interne” by Camillo Golgi in 1898 and then confirmed to be a morphological entity by electron microscopy [ 181. The GA is now clearly established as the site of glycan chain elongation and termination [19, 201. The enzyme function within the cisterne is well ordered and there is a sequential arrangement of the glycosyltransferases; which increases their biosynthetic efficiency [211.

978

60 Immunoglobulin G Glycosylution and Guluctosyltransferase Changes

Oligosaccharide structures produced by a given cell type will reflect the constitution of the multi-galactosyltransferase system [22] and hence the highly-ordered sequential action of the galactosyltransferase enzymes [23]. Synthesis of oligosaccharides can therefore be seen to depend upon a variety of factors [24]: i) the expression of functional glycosyltransferase at the correct location within the cell; ii) the presence of the sugar nucleotides within the Golgi lumen; iii) the presence of glycoprotein acceptors; iv) the absence of inhibitors of this process.

60.3 Galactosyltransferase The glycoprotein UDPp I ,4 galactosyltransferase (GTase) is an intracellular membrane-bound enzyme that can be localized to the Golgi apparatus [25],but may also be found on the cell surface [26] and in a soluble form in milk, amniotic fluid, cerebrospinal fluid, saliva, urine and serum [27]. The molecular role of GTase is to catalyze the transfer of galactose from UDPgalactose to an N-acetylglucosamine (GlcNAc) acceptor during oligosaccharide elongation, for example, in IgG glycosylation [28]. The gene-encoding human GTase is thought to be located in chromosome 9 [29], and it may specify more than one mRNA transcript [30]. GTase seems to play a multifunctional role in normal cell physiology and has been associated with sperm-egg binding [ 311, cell-cell recognition [32], embryonic maturation [33], and cell development [34].

60.4 Immunoglobulin G The glycoprotein IgG has a conserved N-linked glycosylation site in the Fc region, with variable glycosylation in the Fab depending on the presence or absence of the glycosylation motif in the variable region and whether it is conformationally possible. Investigation of the glycoforms of serum IgG has established significant differences in oligosaccharide structure between groups of patients with certain diseases compared with IgG from healthy controls. Different glycoforms arise due to the presence or absence of certain sugar residues on the oligosaccharide backbone [35]. Figure 1 shows a composite structure of the N-linked oligosaccharide chain on human IgG and indicates how some structural variations can arise.

60.6 Quant@cation qf IgG sugars in RAY

Figure 1. A composite structure of the N-linked oligosaccharide chain on human IgG. Common structural variations include change in content of sialic acid, galactose and bisecting N-acetylglucosamine.

979

I -FUCOSE I ASN ( 2 9 ) GIcNAc

60.5 Rheumatoid Arthritis Rheumatoid arthritis (RA) is a multisystem disorder in which immunological abnormalities characteristically result in symmetrical joint inflammation, articular erosions and extra-articular complications. It is the most common and disabling autoimmune arthritis, and genetic susceptibility is well defined. The production of increased amounts of IgG and the presence of rheumatoid factor (immunoglobulins directed against the Fc portion of IgG) are two characteristic features associated with RA [36].

60.6 Quantification of IgG sugars in RA Several different chromatographic methods and a lectin-based assay have been compared for the quantitation of oligosaccharides released from IgG (Figure 2) [ 3 7 ] .The analysis of a series of IgG samples purified from the serum of RA patients was carried out by these methods to evaluate the percentage of the glycoforms having 0, 1 or 2 galactose residues (GO, G1 and G2) in order to:

i) identify the method that can be most widely used for quantitation; ii) accurately define the range of GO values found in patients with RA; iii) make available a series of characterized standards for distribution to clinical chemistry laboratories. The chromatographic methods involve: i) release of oligosaccharides by glycosamidase A after protease digestion followed by HPLC analysis of aminopyridine derivatives on reverse phase and normal phase columns;

980

-

60 Immunoglobulin G Glycosylation and Galactos~ltvansferaseChanges Purified lgG

I

Lectin binding assay

Chemical or enzymatic release of glycans

2-AP labelling

Sialidase

HPLC

2-AB labelling

Anion exchange

Normal phase HPLC

\

!

Sialidase

Biogel P4 HPAEC chromatography (PA-1 column)

(Gl ycosep-N)

Figure 2. Summary of the different methods employed to study a panel of IgG molecules of rheumatoid sera.

ii) hydrazinolysis treatment with exoglycosidases (GO mix) and Biogel P4 chromatography of 2-aminobenzamide (2-AB) derivatives; iii) hydrazinolysis and weak anion exchange or normal phase HPLC of 2-AB derivatives; iv) release of oligosaccharides by PNGase F and either Biogel P4 chromatography of 2-AB derivatives or HPAEC-PAD analysis of native oligosaccharides. The GO values given by these methods compared favorably with each other and a dot blot assay of denatured IgG interaction with Ricinus communis agglutinin and Bundeiruea simplicifoliu lectin 11. The HPLC and HPAEC methods give additional information that may be important in less routine assays.

60.7 RA and Pregnancy 60.7.1 Galactosylation of IgG To examine the relevance of glycosylation changes to the pathogenesis of RA, we have examined the oligosaccharide composition of IgG in relation to the amelioration of RA during pregnancy, which occurs in 75% of RA patients [38]. Serum IgG levels of terminal GlcNAc, galactose and a2,6 sialic acid were compared, in pregnant RA patients (n = 23) with active disease. Patients were divided into three separate groups depending on whether their disease activity remained unchanged, relapsed or went into remission. Longitudinal lectin analysis revealed unique patterns of IgG oligosaccharide variation in the three groups examined. Although the observed microheterogeneity

60.7 RAr and Pvegnuncy

98 1

Table 1. Analysis of the inter-relationship between Gal and GlcNAc.

Remission Unchanged Re1apse

Correlation

p value

-0.80 -0.43 +0.87

<0.05 NS <0.05

involved both the terminal galactose as well as the levels of exposed GlcNAc, the remission group had a distinct pattern and was specifically associated with a significant decrease ( p < 0.05) in mean IgG GlcNAc levels. Overall, the temporal changes in galactose and GlcNAc and the time frame at which they occurred were found to be critical discerning factors associated with the clinical outcome of the disease during pregnancy. In addition, analysis of the inter-relationship between Gal and GlcNAc in individual IgG samples, during the initial 13 weeks of gestation, demonstrated further important differences between the three groups. There was a significant negative correlation in the remission group, which was in direct contrast to the positive correlation found for the relapse IgG samples (Table 1). In conclusion, our results indicate that IgG glycosylation changes, or the factors that control them, may be an integral component of the mechanisms determining the clinical course, in particular remission, during pregnancy [ 381.

60.7.2 a3-Fucosylation of al-Acid Glycoprotein

In all three categories changes in the plasma level and glycosylation of al-acid glycoprotein (AGP) were determined longitudinally in comparison to those occurring in pregnancy of healthy women (Figure 3 ) [39]. In healthy pregnancy, we observed: i) a peak in the plasma concentration at week 18 and a minimum at week 30; ii) a continuous increase in the degree of branching of the glycans during the entire pregnancy period; iii) a decrease in the degree of a3-fucosylation of AGP-glycans with a minimum occurring at week 25. Comparable pregnancy-induced changes in glycosylation were found for two other acute-phase proteins al-protease inhibitor (Pl) and al-antichymotrypsin (ACT). Increased oestrogen levels, known to occur during pregnancy, may be one of the factors that induce these changes, because the increased branching and decreased a3-fucosylation is in agreement with our earlier findings regarding an involvement of this hormone in the regulation of acute phase glycosylation in estrogen-treated males as well as females. In all three clinical categories in RA, pregnancy also induced a continuous increase in the degree of branching of the glycans of AGP. However, similar changes in concentration and fucosylation were only found during

982

60 Immunoglobulin G Glycosylation and Galactosyltransferase Changes

co

'

cw

'

''

cs

r

AO

'

Aw

' As I

C

Figure 3. Reactivity of AGP (a, d ) , ACT (b, e ) and PI (c, f ) from healthy individuals (HSPC) with ConA (left panel) and AAL (right panel). Sera were subjected to CAIE. Only the second dimension gels are shown. The lower right corner of each pattern corresponds to the site of application in the first dimension gel. Electrophoresis was performed from right to left in the first dimensions, and from bottom to top for the second dimension, CO, AO: AGP-glycoforms non-reactive with ConA or AAL respectively; Cw, Aw: AGP-glycoforms weakly reactive with ConA and AAL, respectively; Cs, As: AGP-glycoforms strongly reactive with ConA and AAL, respectively. No fractionation was obtained when samples were analyzed in the absence of lectin, resulting in recovery of all APPprotein at the position of CO or AO.

remission of the disease symptoms. In the relapse and unchanged categories in RA, the degree of fucosylation and the plasma concentration of AGP remained constant throughout pregnancy. This indicates a relationship between changes in a3fucosylation of AGP and RA disease activity.

60.8 Agalactosyl-IgG and Rheumatoid Factor Binding We have used a series of IgG preparations differing in their content of oligosaccharide chains lacking galactose by 18-86% to determine whether changes in sugar content affect the binding of rheumatoid factor [40]. Five of 16 monoclonal

60.9 Tissue-specijic Guluctosyltrunsf~ruseAbnormalities

983

- 0 0

20

4060 8 0 1 0 0 %C(O)

/-:,

. . ,

w 80

2.0

:::p* OO

20

40

100

0

20 4 0 6 0 8 0 1 0 0

%(O)

%GO)

0.5

Figure 4. Binding of five synovial tissuederived IgM monoclonal RFs to IgG preparations varying in GO. The horizontal axis represents GO and the vertical axis represents optical density at 450 nm wavelength. A line of best fit is shown for each graph.

0.2

0.1

OO

20 40 60 80 100 % G(O)

rheumatoid factors prepared from synovial tissue, from patients with juvenile or adult rheumatoid arthritis, bound better to IgG, which was deficient in galactose (Figure 4). Six of the 16 rheumatoid factors from the same patients bound independently on the galactose content. Four of the 16 rheumatoid factors could not be absolutely grouped in this manner but seemed to demonstrate a preference for agalactosyl-IgG. One rheumatoid factor bound better to fully galactosylated IgG. There was an association between enhanced binding to galactose-deficient IgG and mono-reactivity and a very strong association between the functional affinity of the rheumatoid factors and the dependent binding.

60.9 Tissue-specific Galactosyltransferase Abnormalities in an Experimental Model of Rheumatoid Arthritis We have investigated tissue-specific galactosyltransferase abnormalities in an experimental model of rheumatoid arthritis [41]. The study focused on whether the observed pathophysiological similarities that develop in both the collagen induced experimental model of arthritis (CIA) and RA are associated with similar glyco-

984

60 Immunoglobulin G Glycosylution and Guluctosyltvansfevasr Chunges

sylation changes, and evaluated possible differences in the relative activity of GTase within various tissues. This gave us a new insight into the potential pathogenic mechanisms controlling glycosylation changes. Lymphocytic membrane-bound GTase activity was examined in 10 mice with CIA, 10 age-matched controls and ten adjuvant treated non-arthritic DBA/l mice. Tissue-specific changes were assessed by comparison of GTase activitiy in peripheral (P.GTase) and paired splenic lymphocytes. In addition, the study investigated the effect that these changes may exert on the overall extracellular level of this enzyme, by assaying serum GTase (SGTase) activity in these and a further group of 27 arthritic and 20 control mice. To analyze this synthetic abnormality in greater depth and to investigate the relevance of these glycosylation changes to the pathogenesis of arthritis, we also examined the humoral regulatory component associated with this system by assaying for both anti-collagen as well as anti-GTase antibodies. The induction of arthritis in DBA/l mice results in a marked reduction in P.GTase activity, compared with age-matched unimmunized mice and the adjuvant controls (Figure 5). In contrast to the P.GTase, splenic GTase activity was found to be similar in all the groups examined (Figure 6). Correspondingly, serum GTase activity was also found to be significantly lower in the collagen-induced arthritic mice (Figures 7 and 8). This overall reduction in 1-4 GTase activity reflects the clinical severity of arthritis and is associated with increased levels of naturally occurring anti-GTase antibodies. This demonstrates that the GTase defect seen in the peripheral B and T cells in rheumatoid arthritis is also evident in the arthritic DBA/ 1 mouse model of RA. This may indicate a common pathological process in both

0

:I

I

0 0 0 0

I

uninuaunised n = 10

I.."

0 I I

Cdlimmunised n =10

Figure 5. Effect of collagen immunization on peripheral lymphocytic GTase activity. Results from individual mice are shown, with standard errors of mean, for each group. The two Coll immunized mice with the highest GTase activity had the lowest clinical score (CS = 4). * p = 0.05.

60.10 Glycosylation Homeostasis within RAY Lymphocytes is Abnormal

985

h

81

T

0

A 0

Unimmunised CFA

mi m:&

Unimmunised

CFA

immunised Colt

Figure 6. Comparison of GTase activities of peripheral and paired splenic lymphocytes and the effect of adjuvant treatment on enzyme activity. a) Peripheral lymphocytes: The marked reduction in pooled P.GTase activity is only evident in the Coll induced arthritic and not the age-matched CFA treated group ( p < 0.03). b) Splenic lymphocytes: GTase activity is markedly lower in the splenic compared with the paired peripheral lymphocytes, and is unaffected by CFA treatment or Coll immunization.

rheumatoid arthritis and CIA, in which changes in glycosylation are dependent on the aberrant modulation of GTase in circulating, but not splenic lymphocytes. The relative expression and activity of glycosyltransferases within various tissues may not only contribute to immunoglobulin G (IgG) glycosylation changes, but perhaps also the aberrant expression of cell surface carbohydrates and thus cell trafficking.

60.10 Glycosylation Homeostasis within RA Lymphocytes is Abnormal To investigate potential mechanisms controlling protein glycosylation we have studied the inter-relationship between lymphocytic GTase activity and serum agalactosylated immunoglobulin G levels, GO in healthy individuals and patients with RA and non-autoimmune arthritis [42]. In RA there was reduced GTase activity (Figure 9) and increased GO. A positive linear correlation between B and T cell

986

60 Immunoglobulin G Glycosylution and Guluctosyltransferuse Changes

. .

w

.

1.

unimmunised

coi

immunised

Figure 7. Comparison of IgM and IgG serum anti-GTase antibodies in Coll immunized and unimmunized mice. a) IgM anti-GTase antibody levels are significantly higher in the serum of the Coll induced arthritic mice, and appear to be associated with arthritis (the solid bar represents the mean antibody level + 2 standard deviations of the unimmunized control mice). b) IgG anti-GTase antibodies levels are raised in only a few of the Coll immunized mice and do not appear to be associated with arthritis. * p < 0.05.

GTase was found in all individuals (Figure lo). The relationship between GTase and GO was found to be positive and linear in the control population (Figure 11) and negative and linear in the RA population (Figure 12). These data describe a defect in RA lymphocytic GTase, with associated abnormal GO changes. A possible regulatory mechanism controling galactosylation in normal cells is suggested. This is disrupted in RA, where the positive feedback between GTase and GO is lost. We suggest that these mechanisms are of relevance to the pathogenesis of RA, and that their manipulation may form part of a novel therapeutic approach (Figures 13 and 14).

60.11 Are the Rheumatoid Arthritis Associated Glycosylation Abnormalities Unique? We have investigated the relationship between exposed galactose and GlcNAc on IgG in RA, juvenile chronic arthritis (JCA) and Sjogren’s syndrome (SS) [43]. This

987

60.11 Are the Rheumatoid Arthritis Associated Glvcosylution Abnormalities Unique?

I."

, 0

i0

i0

0

i0

Serum GTase Activity 6 -1 cpmx10 ml (MkSEM)

Figure 8. Scatter plot demonstrating the inverse correlation (dashed line of best fit; r value = 0.77; p < 0.05) between IgG anti-Collagen antibodies and GTase activity in the serum of collagen immunized arthritic DBA/l mice.

NS

a*

a**

80

.-

h

m'

Figure 9. B lymphocytic galactosyltransferase activity (mean cpm/mg) protein lo5 +/- SEM) in the total RA population and divided according to drug therapy. Control: 35 +/- 4. All RA 21 +/- 3. Gpl (NSAID or no drugs): 14 +/- 5. Gp2 (SASP):28 +/- 8. Gp.3 (other second/third line therapy): 21 +/- 4. Bars define the mean #- SEM. (Significance of difference from controls: *p < 0.01; **p < 0.002; ***p < 0.05).

0

0

0

0

o

8 -?-

00 00

0 Control

AURA

Cpl

CpZ

Cp3

1

988

60 Immunoglobulin G Glycosylution and Galuctosyltransferase Changes

20

0

40

80

60

100

T cell plactosylhmsferrse activity cprn/mg protein x 1V

Figure 10. Regression analysis of control (closed diamonds and broken line, Y = 0.770, p < 0.001 and RA (open diamonds and unbroken line, Y = 0.691, p < 0.001) paired B and T lymphocytic galactosyltransferase activities.

."" I

A

A

l

A

?

A

-

40-

A

10

20

M

40

50

I

%SerumIgC G(0)

Figure 11. Regression analysis of paired B (open triangles and unbroken line) and T (closed triangles and dotted line) lymphocytic galactosyltransferase activities and serum agalactoimmunoglobulin G (percent serum IgG GO) in the control population (Y = 0.263 and 0.21 1 for B and T cells, respectively, p < 0.05).

60.I 1 Are the Rheumatoid Arthritis Associuted Glycosylution Abnormulities Unique?

989

80 A

I

A

40

20

1 I

.\

-

-

' A

A

w

AA

,

0

0

%Serum IgG G( 0) Figure 12. Regression analysis of paired B lymphocytic galactosyltransferase activity and serum agalacto-immunoglobulin G values (percent serum IgG G(0)) in the rheumatoid arthritis population (r = -0.338, p < 0.004). There is a significant difference ( p < 0.01) when this value is compared with the control regression value ( r = 0.263)

60000

50000 A

I

n

40000

.-* c

>

I

30000

0

a 0)

cn

c" 0

20000

10000

0 4

5

PH Figure 13. A profile of GTase activity plotted against pH following solution phase isoelectric focusing of serum from a healthy individual. There is a large peak forming at pH 4.55 and much smaller peak at pH 5.20.

990

60 Immunoglobulin G Glycosylution and Guluctosyltransferase Changes

4

5

PH Figure 14. This is a profile of GTase activity plotted against pH following solution phase isoelectric focusing of serum from a patient with RA. A large peak is seen at pH 4.65 and a smaller peak forms at pH 5.05. There is a left shift of the second peak, and it constitutes a significantly greater proportion of the total activity when compared to the healthy population.

was achieved using IgG isolated from serum where the levels of galactose and GlcNAc were detected using biotinylated lectins. Galactose and GlcNAc on IgG from patients with RA and JCA are inversely related, but in contrast, in SS, galactose expression on IgG decreased while GlcNAc expression remained similar to normal levels (Figure 15). Alterations of IgG glycosylation are closely associated with the development of RA, JCA and SS, but the changes involved are different in RA compared with SS, suggesting that the precise pattern of exposed sugars is associated with different rheumatological changes.

60.12 Sugar Printing Rheumatic Disease is Possible To look for oligosaccharide structural variants of IgG that may be unique to specific rheumatic diseases, normal phase HPLC was used (Figure 16) and a comparison made of the oligosaccharide pools released from serum IgG from patients with systemic lupus erythematosus (SLE) ( n = lo), ankylosing spondylitis (AS) ( n = lo), primary Sjogren’s syndrome (PSS) ( n = 6), JCA ( n = 13), PsA ( n = 9), RA (a = 5) and healthy individuals ( n = 19) [44].

60.12 Suyur Printing Rheumatic Diseuse is Possible

-1

a

991

.

IC'

Figure 15. (a) Correlation between galactose and GlcNAc levels on RA IgG ( n = 29) (b) Correlation between galactose and GlcNAc levels on JCA IgG ( n = 11) (c) Correlation between galactose and GlcNAc levels on SS IgG ( n = 37) (d) Correlation between galactose and GlcNAc levels on normal IgG (n = 100).

mV

75

80

85

90

95

100

105

110

ii5

iio

Minutes

Figure 16. An example of a normal phase HPLC profile from a healthy control, annotated to show the oligosaccharide structures represented by each peak. Normal phase resolves sixteen oligosaccharides into thirteen peaks. There are four agalactosyl and four digalactosl oligosaccharide peaks and monogalactosyl oligosaccharides form five peaks with co-elution of different oligosaccharides according to their arm specificity.

992

60 Immunoglobulin G Glycosylution and Guluctosyltrunsferuse Changes

The oligosaccharide pools were resolved into 13 peaks and the relative proportions of the peaks in each disease group was significantly different from those in healthy controls ( p < 0.001-0.05). A characteristic serum IgG oligosaccharide profile or sugar print for each of the rheumatic diseases was found. The sugar prints exhibited a range of glycosylation patterns whereby all RA (p < 0.001) and JCA (p < 0.006) patients had predominantly agalactosylated structures, while SLE (p < 0.03-0.0001) and AS (p < 0.025-0.0001) patients had predominantly digalactosyl structures. The data suggest that each disease is associated with a specific mechanism giving rise to alterations in the normal glycosylation pattern of IgG. Sugar printing of IgG is therefore a potential means for the differentiation of rheumatic diseases and may provide insight into disease pathogenesis.

60.13 Rapid Profiling of IgG N-Glycans by Fluorophore-coupled Oligosaccharide Electrophoresis has the Potential of Differentiating Rheumatic Diseases A rapid cost-effective test which discriminates between rheumatic diseases in their early stages is still not available to the Rheumatologist. One clinical feature of the rheumatic diseases, which may be exploited in the development of a diagnostic test, is the change in the glycosylation profile of IgG and IgG sugar printing may be potentially useful in this respect. However, a more rapid investigation than HPLC analysis is required. To enable rapid profiling of IgG sugars, a fluorophore-coupled oligosaccharide electrophoresis (FCOE) system was developed to analyze IgG oligosaccharides derived from patients with rheumatic disease [45]. The glycans from patients with RA ( n = 20), PsA ( n = 5), AS (n = 14), JCA ( n = 9), SLE ( n = 19) and healthy controls ( n = 10) were released by PNGaseF and labeled with 2-amino benzoic acid. They were resolved on 30"A polyacrylamide gels using tris-glycine in nine bands. Oligosaccharide profiles were obtained for each group and comparison made between digalactosyl-fucosylated (g2f) structures (Band 1) and agalctosyl-fucosylated (gOf) structures (Band 4). Significant reduction (g2f) and elevation (gOf) in band intensity was noted in RA (g2f and gOf p < O.OOl), PSA (gOf, p < O.OOl), JCA (g2f p < 0.002) compared to healthy controls. RA was significantly different from AS (gof, p < 0.001), JCA (gOf, p < 0.001) and SLE (gOf and g2f, p < 0.001). These data would indicate that FCOE is a rapid test that appears to have the potential of differentiating rheumatic diseases.

60.14 In What Way could GTase Enzymatic Control be Abnormal? We have investigated possible specific p1,4 GTase isoenzyme changes in serum of patients with RA [46]. Using solution phase isoelectric focusing, we have deter-

60.15 Conclusion

993

mined the isoenzyme profiles in GTase in healthy individuals (n = 9) (Figure 13) and patients with psoriatic arthritis (PsA) ( n = 9) and RA ( n = 8) (Figure 14). GTase activity was determined using a previously reported assay, in which 3H galactose is transferred to ovalbumin. Comparison of GTase isoenzyme activity profiles demonstrated that the RA group (Figure 13) was significantly different from both the PsA patients and healthy individuals (Figure 14) ( p < 0.01). In seven patients with RA, eight healthy individuals and six with psoriatic arthritis, two fractions showed distinct peaks of activity. The first peak of activity formed at pH 4.5 (range 4.3-4.7), pH 4.6 (range 4.4-5.0) and 4.65 (range 4.35-5.0), whilst the second peak of activity focused at pH 5.0 (range 5.0-5.1), pH 5.0-5.4) and 5.20 (range 5.0-5.50) in RA, healthy individuals and PsA populations respectively. There was a significant shift in the PI position of the second peak when comparing the RA group to the healthy individuals ( p = 0.006). A significant increase in GTase activity was noted in the RA second peak ( p = 0.008) when compared to healthy individuals and PsA. Two main peaks of serum GTase activity have therefore been identified and there is thus evidence that RA associated serum GTase isoenzymes occur. The RA GTase isoenzyme profile is significantly different from the healthy individuals and PsA as there is acidic skewing of the first peak, a prominent second peak concentrated over a narrow pH band (pH 4.97-5.11) and a loss of activity in the pH range 5.20-5.60. The significance of these findings now needs to be determined with reference to RA pathogenesis and the enzyme contained within these peaks fully characterized.

60.15 Conclusion

To date our work concerning the glycoimmunology of RAr and other rheumatic diseases has demonstrated a significant defect in the GTase enzyme that results in a profound change in the galactosylation of IgG. This change has been demonstrated to be integrally associated with pathogenic mechanisms associated with inflammation in RAr, as well as in the arthritic DBA/l mouse model of RAr. Observing pregnancy-associated IgG glycosylation changes has enabled us to determine that the temporal changes in galactose and GlcNAc are critical discerning factors associated with clinical activity. It is hypothesized that there are major disruptions in glycosylation homeostasis associated with the pathogenesis of RAr and that perhaps in part these involve the generation of unique GTase isoenzymes that differ functionally from those present in healthy individuals. It is not thought that these changes are unique to RAr but it is thought that there may be subtle changes in the disruption of glycosylation homeostasis causing a unique sugar change to be associated with a number of other rheumatic diseases. This is referred to as “sugar printing the rheumatic diseases” and may be a concept both useful diagnostically and therapeutically.

994

60 Immunoglobulin G Glycosylation and GaluctosyltrunsJerase Changes

Acknowledgements

I should like to thank Azita Alavi for editorial assistance. References 1. Glycoimmunology I1 Advunces in experimental medicine und biology. Edited by J.S. Axford, Published by Plenum Press Vol. 435 (1996). 2. Dwek RA. Glycobiology: Towards understanding the function of sugars (Wellcome Trust Award Lecture 1994) Biochem Soc Trans (1995) 23(1) 1-25. 3. Feizi T. Demonstration by monoclonal antibodies that carbohydrate structures of glycoproteins and glycolipids are onco-developmental antigens. Nature (1985) 314: 53-57. 4. Crocker PR & Feizi T. Carbohydrate recognition systems: functional triads in cell-cell interactions. Curr Ouin Struct Biol(1996) 6 (5): 679-91. 5. Axford JS et ai. 2nd Jenner Internakonai Glycoimmunology meeting. Ann Rheum Dis (1992) 1269- 1985. 6. Axford JS et al. 3rd Jenner International Glycoimmunology meeting. Glycosylut Dis (1994) 3: 197-227. 7. Axford JS. 31d Jenner International Glycoimmunology meeting. Immunol Toduy (1993) 14(3): 104- 106. 8. Axford JS. Trends. 2"d Jenner International Glycoimmunology meeting. Zmmunol Today (1995) 16(5): 213-215. 9. Axford JS. Oligosaccharides: An optional extra or of relevance to disease mechanisms in rheumatology? J. Rheumatol(1991) 18(8): 1124-1 127. 10. Axford JS. Glycosylation and rheumatic disease: more than icing on the cake. J. Rheumatol (1994) 21(10): 1791-1795. 1I . Alavi A & Axford JS. Chapter: The glycosyltransferases from abnormalities of IgG glycosylation and immunological disorders. Edited by Isenberg DA & Rademacher TW. Published by John Wiley & Sons (1996). 12. Alavi A & Axford JS. Glycoimmunology I. Published by Plenum Press (1995). 13. Axford, JS. Glycosylation and Autoimmune Disease. Glycoconj J. (1996) 13393. 14. Axford JS. 4th Jenner International Glycoimmunology Meeting: A Review. Immunol Today 1997; 18(11): 511-513. 15. Varki A. Biological roles of oligosaccharides: all of the theories are correct. Glycobiology (1993) 3(2): 97-130. 16. Feizi T. Oligosaccharides that mediate mammalian cell-cell adhesion. Curr Opin Struct Biol (1993) 3: 701-710. 17. Hagopian A & Eylar EH. Glycoprotein biosynthesis: studies on the receptor specificity of the polypeptide: N-acetylglucosaminyl-transfcrase from bovine submaxillary glands. Arch Biochem Biophys (1968) 128(2): 422-433. 18. Farquhar M G & Palade G. The golgi apparatus (complex)-(1954-81) from artifact to center stage. J Cell Biol (1981) 91(3); 77-103. 19. Roth J & Berger EG. Immunocytochemical localisation of galactosyltransferase in HeLa cells: co-distribution with thiamine pyrophosphatase in trans-Golgi cisternae. J Cell Biol (1982) 93(1): 223-229. 20. Roth J, Taatjes DJ, Lucocq JM, Weinstein J, Paulson JC. Demonstration of an extensive transtubular network continuous with the Golgi apparatus stack that may function in glycosylation. Cell (1985) 43( l): 287-295. 21. Rabouille C, Hui N, Hunte F, Kieckbusch R, Berger EG, Warren G, Nilsson T. Mapping the distribution of golgi enzymes involved in the construction of complex oligosaccharides. J Cell Sci (1995) 108(4) 1617-1627. 22. Roseman S. The synthesis of complex carbohydrates by multi-glycosyltransferase systems and their potential function in intercellular adhesion. Chem Phys Lipids (1970) 5( I): 270-297.

References

995

23. Nilsson T & Warren G. Retention and retrieval in the endoplasmic reticulum and the golgi apparatus. Curr Opin Cell Biol(1994) 6(4): 517-521. 24. Berger EG & Thurnher M. Clues to the cell-specific synthesis of complex carbohydrates. News Physiol Sci (1993) 8: 57-60. 25. Berger EG, Thurnher M & Muller U. Galactosyltransferase and sialyltransferase are located in different sub-cellular compartments in HeLa cells. Exp Cell Res (EPB) (1987) 173: 267-273. 26. Shur BD & Roth S. Cell surface glycosyltransferases. Biochem Biophys Actu (1975)415: 473-512. 27. Schachter H & Roden L. The biosynthesis of animal glycoproteins. In: Metabolic Conjugation und Metabolic Hydrolysis. WH Fishman, Editor: Academic Press New York pp. 1-149. 28. McGuire EJ, Kerlin R, Cebra JJ & Roth S. A human milk galactosyltransferase is specific for secreted, but not plasma IgA. J Inimunol(1989) 143(9): 2933-2938. 29. Shaper NL, Shaper JH, Bertness V, Chang Hg, Kirsch IR and Hollis GF. The human galactosyltransferase gene is on chromosome 9 at band p13. Somatic Cell Mol Genet (1986) 12(6) 633-6. 30. Russo RN, Shaper NL and Shaper JHI. Bovine 114 galactosyltransferase: two sets of mRNA transcripts encode two forms of the protein with different amino-terminal domains. In vitro translation experiments demonstrate that both the short and the long forms of the enzyme are type I1 membrane-bound glycoproteins. J Biol Chem (1990) 265(6): 3324-3331. 31. Shur BD & Hall NG. Sperm surface galactosyltransferase activities during in vitro compaction. J Cell Biol(1982) 95: 567-573. 32. Bayna EM, Runyan RB, Scully NF, Reichner J, Lopez LC and Shur BD. Cell surface galactosyltransfer as a recognition molecule during development. Mol Cell Biochem (1986) 72: 141- 15 I , 33. Bayna ME, Shaper JH & Shur BD. Temporally specific involvement of cell surface b1,4 galactosyltransferase during mouse embryo morula compaction. Cell (1988) 53: 145-1 57. 34. Begovac PC and Shur BD. Cell surface galactosyltransferase mediates the initiation of neutite outgrowth from PC12 cells on laminin. J. B i d Chem (1990) llO(2): 461-470. 35. Parekh RB, Dwek RA, Sutton BJ et al. Association of rheumatoid arthritis and Drimary osteoarthritis with changes in the glycosylation pattern of total serum IgG. Nature (1985) 316: 452-457. 36. Axford JS. Medicine Published by Blackwell Science 1996. 37. Routier FH, Hounsell EF, Rudd PM, Takahashi N, Bond A, Hay FC, Alavi A, Axford JS and Jefferis R. Quantitation of the different oligosaccharides of human serum IgG from patients with rheumatoid arthritis: A critical evaluation of different methods. J Immunol Meths 1998; 213: 113-130. 38. Alavi A, Arden N, Spector T D & Axford JS. IgG glycosylation changes are associated with the clinical outcome of rheumatoid arthritis in pregnancy. Arth Rheum 1997; 40(9) Suppl. S149 No. 771. 39. Havenaar EC, Axford JS, Brinkman-van der Linden ECM, Alavi A, Van Ommen ECR, van het Hof B, Spector T, Mackiewicz A & Dijk WV. Severe rheumatoid arthritis prohibits the pregnancy-induced decrease in a3-fucosylation of CLI -acid glycoprotein. Glycoconj J (1998) 15: 723-729. 40. Soltys AJ, Hay FC, Bond A, Axford JS, Jones MG, Randen I, Thompson K & Natvig J. The binding of synovial tissue-derived human monoclonal immunoglobulin M rheumatoid factor to immunoglobulin G preparations of differing galactose content. Scan J Zmmunol (1994) 40(2): 135-143. 41. Alavi A, Axford JS, Hay FC, Jones MG. Tissue-specific galactosyltransferase abnormalities in an experimental model of rheumatoid arthritis. Ann Med Intern (1998) 149(5): 251-260. 42. Axford JS, Sumar N, Alavi A, Isenberg DA, Young A, Bodman KB, Roitt IM. Changes in normal glycosylation mechanisms in autoimmune rheumatic disease. J Clin Inoest (1992) 89(3): 1021-1031. 43. Bond A, Alavi A, Axford JS, Youinou P & Hay FC. The relationship between exposed galactose and N-acetylglucosamine residues on IgG in RA, JCA and Sjogren’s syndrome. Clin Exp Zmmunol(1996) 105: 99-103. 44. Watson M, Rudd P, Dwek RA & Axford JS. Sugar printing rheumatic diseases: A potential method for diagnosis and differentiation using immunoglobulin G oligosaccharides. Arth Rheum (1996) 39(9) Suppl. 216 No. 1133.

996

60 Immunoglobulin G Glycosylution and Guluctosyltrunsjeruse Changes

45. Frears ER, Merry AH and Axford JS. Rapid profiling of IgG N-glycans by fluorophorecoupled oligosaccharide electrophoresis has the potential of differentiating rheumatic diseases. Arth Rheum 1998; 41(9) Suppl. S85 No. 317. 46. Pool AJ, Alavi A & Axford JS. 01,4 galactosyltransferase isoenzyme changes in serum of patients with rheumatoid arthritis. Br J Rheuinutol(l996) 35(1): Suppl. 174 No. 335.

Carbohydrates in Chemistry and Biology Beat Emst, Gerald W. Hart, Pierre Sinay copyright Q WILEY-VCH Verlag GmbH. 2000

61 Calnexin, Calreticulin and Glycoprotein Folding Within the Endoplasmic Reticulum Michael R. Leach and David B. Williams

Proteins destined for secretion or for various locations along the secretory pathway are initially translocated wholly or in part into the endoplasmic reticulum (ER). Within this compartment the vast majority of proteins are glycosylated at Asn residues with a preassembled oligosaccharide of structure Glc3MangGlcNAc2. These Asn-linked glycoproteins subsequently fold and assemble with the aid of a diverse array of protein folding catalysts and molecular chaperones. The protein folding catalysts include the thiol oxidoreductases protein disulfide isomerase (PDI), ERp57, and ERp72, that catalyze the formation and reshuffling of disulfide bonds, as well as peptidyl prolyl isomerases that catalyze the interconversion between cis and trans forms of peptide bonds preceding proline residues. By contrast, molecular chaperones enhance the efficiency of protein folding by binding transiently to hydrophobic segments of folding intermediates thereby preventing the formation of irreversible aggregates. Molecular chaperones of the ER also participate in a process termed quality control in which their interactions cause non-native proteins to be retained in this organelle either until they are degraded or until a native conformation is attained. Multiple molecular chaperones reside within the ER, including Grp94, a member of the HSP90 family of heat shock proteins, and BiP, an HSP70 family member. In addition, two other molecular chaperones, calnexin and calreticulin, appear to have evolved specifically to participate in the biogenesis of Asnlinked glycoproteins. In this review, we discuss the molecular basis for the glycoprotein specificity of these unique chaperones, their structural and functional relationship, current models for their mechanism of action, and their interplay with other ER chaperones.

61.1 Structure and Properties of Calnexin and Calreticulin Calnexin, calreticulin, and the testis-specific homolog calmegin constitute a family of abundant ER proteins. Calnexin is a type I transmembrane protein that is pres-

998

calnexin

61 Calnexin, Culreticulin and Glycopvotein Folding

w+ 592 K4

calmegin H ~ N +

oltgosaccharide binding

4

C--- N domain -----I- P domain 4- C domain -4

/I

Figure 1. Structural features of dog calnexin, mouse calmegin, and rabbit calreticulin. Segments of sequence similarity are represented by the white boxes and the two tandemly repeated sequence motifs are indicated by the numbers 1 and 2. Sites of high affinity calcium binding (Ca2+),oligosaccharide binding, and phosphorylation by casein kinase I1 (C.K. 11) are shown. ER localization sequences are indicated at the C-terminus of each protein.

ent in all eukaryotes so far examined, including the yeasts Succhuromyces cerevisiue and Schizosucchuromyces pombe, plants, insects, worms, and mammals (Figure 1; reviewed in [1, 21). No prokaryotic homologs of calnexin have been described. Canine calnexin consists of 574 amino acids with a predicted molecular weight of 65.4 kDa, but runs as a 90 kDa band on SDS-PAGE, presumably due to its acidic character (PI 4.5 [3]). The cytoplasmic tail of canine calnexin contains two potential phosphorylation sites for casein kinase 11, has been shown to be phosphorylated in vivo [4], and possesses a C-terminal ER localization signal (-RKPRRE). Calnexin also contains a Ca2+ binding site, unrelated to either the calmodulin EF-hand or the annexin Ca2+ binding sequences [5] located in its luminal segment and centered on a set of unusual tandemly repeated sequence motifs [6]. The tandem repeats consist of four copies of motif 1 [I-DP(D/E)A-KPEDWD(D/E)]followed by four copies of motif 2 [G-W-P-IN-P-Y] (Figure 1). Using cross-linking to [a-32P]ATP, direct binding of calnexin to ATP has been shown, but no detectable ATPase activity has so far been demonstrated. Both the removal of Ca2+ and the addition of ATP have been shown to promote the oligomerization of the lumenal domain of calnexin in vitro and to increase its sensitivity to proteolysis [7]. Calnexin has also been shown to bind in vitro to a thiol oxidoreductase of the ER termed ERp57 [8]. The ER luminal portion of calnexin has been crystalized and heavy metal derivatives have been prepared [9], but to date the crystal structure has not been solved due to poor resolution data. By contrast with calnexin, calreticulin is a soluble 46 kDa resident of the ER lumen that consists of 401 amino acids. Due to its acidic character (PI 4.6) calreticulin runs aberrantly on SDS-PAGE with an apparent size of 60 kDa. Calreticulin is not

61.1 Structure and Properties of Culnexin und Culreticulin

999

found in yeasts, but is present in all other eukaryotes examined. Mammalian calreticulin shares -39% overall sequence identity with calnexin and this rises to 66% in the region of the tandem repeat motifs that it has in common with calnexin. However, only three copies of each motif are present compared to the four of calnexin (Figure I ) . Calreticulin has been divided into three putative structural domains based on secondary structure prediction [lo, I l l . The N-terminal N-domain (amino acids 1-186) is essentially neutral with respect to charge and hydropathy and is predicted to be a globular domain composed of approximately eight antiparallel pstrands. The central P-domain contains the acidic, proline rich repeat motifs predicted to be mostly random coils and p-turns followed by a short hydrophobic stretch. A highly acidic C-terminal C-domain is the site of high capacity, low affinity Ca2+ binding ending in a -KDEL ER localization sequence. Calreticulin possesses a nuclear localization signal and there is ongoing controversy regarding possible extra-ER localizations. As with calnexin, the repeat motifs of calreticulin are the site of high affinity Ca2+ binding (Kd = 10 pM [ 5 ] ) .Calreticulin has also been shown to have two Zn2+ binding sites in the N-domain, which has no known Zn2+binding motifs [ 121. Zn2+-binding results in a conformational change that exposes hydrophobic segments to the solvent. Like calnexin, calreticulin binds to the thiol oxidoreductase ERp57 [ 131 and also to protein disulfide isomerase (PDI) in vitro under low Ca2' conditions. Both binding sites have been mapped to the N-domain [13, 141. Calmegin is a 592 testis-specific calnexin homolog (56% sequence identity) with a predicted molecular weight of 69.5 kDa. Like calnexin, it is a type I transmembrane protein with four copies each of repeat motifs 1 and 2 and potential casein kinase [I phosphorylation sites in its cytoplasmic tail [15, 161. Calmegin is also acidic with a PI of 5.2 and contains a C-terminal ER localization signal. The central repeat motifs of calmegin have been shown to be the site of high affinity Ca2+ binding [ 171. Perhaps the most characteristic feature of calnexin and calreticulin is that they are lectins which bind to monoglucosylated oligosaccharides of composition Glcl ManS-9 GlcNAc2 (18-20). This oligosaccharide is a glucose-trimming intermediate of the N-linked oligosaccharide Glc3MansGlcNAc2, which is added en bloc to nascent polypeptide chains in the ER. Detailed in vitro binding studies have revealed that calnexin and calreticulin possess essentially identical oligosaccharide binding specificities and affinities, with the single terminal glucose residue being a crucial determinant, i.e. little binding was observed with oligosaccharides possessing 0, 2, or 3 glucose residues. However, binding competition experiments indicated that additional mannose residues on the glucosylated branch of the oligosaccharide do participate in the binding interaction since GlcuOCH3 was not a competitor whereas Glcu3Man. Glcu3Manu2Man, and Glcu3Manu2Manu2Man exhibited progressively increasing potency as competitors. By contrast, mannose residues on non-glucosylated branches did not participate substantially since binding was observed with monoglucosylated oligosaccharides possessing 5-9 mannoses (GlclMans-9GlcNAcz [18, 191). As with the site of Ca2+ binding, the site of oligosaccharide binding in calnexin and calreticulin is centered on the repeat motifs and includes residues flanking both sides (see Figure 1 [20, 861). The lectin binding function of calnexin and calreticulin

1000

61 Calnexin, Calreticulin and Glycoprotein Folding

is Ca2+ dependent and the proteins themselves adopt a less compact conformation in the absence of Ca2+ as assessed by protease sensitivity 17, 201. The lectin function of calmegin has not been tested in vitro. However, it is also likely to bind oligosaccharide since treatment of cells with processing inhibitors that prevent the formation of the monoglucosylated oligosaccharide inhibits the binding of calmegin to various glycoproteins [21]. Although many acidic residues are present as potential Ca2+ and oligosaccharide ligands, there is no strong sequence similarity between the repeat motifs of calnexin, calreticulin, and calmegin and the consensus carbohydrate recognition domain of the C-type lectins. Thus calnexin, calreticulin and calmegin are members of a novel class of Ca*+-dependent lectins.

61.2 Biological Functions Initial evidence that calnexin and calreticulin might be molecular chaperones came from the observation that they interact transiently with a wide array of newly synthesized glycoproteins and that their release frequently correlates with completion of a folding or oligomerization event (reviewed in [ l , 221). For example, calnexin and calreticulin bind to incompletely oxidized transferrin [23, 241 and influenza hemagglutinin (HAg 1251) but dissociate at about the time disulfide bond formation is completed. In the case of class I histocompatibility molecules, calnexin or calreticulin associate with partially assembled heavy chain-0,-microglobulin heterodimers but dissociate following acquisition of the final peptide subunit 126, 271. Calnexin and calreticulin also exhibit prolonged interaction with misfolded or incompletely assembled proteins [ 1, 221. That the binding of calnexin is largely selective for Asn-linked glycoproteins was first shown by treating human hepatoma cells with tunicamycin which resulted in a profound reduction in proteins that could be co-immunoprecipitated as complexes with calnexin 1231. Subsequent studies made use of castanospermine and 1-deoxynojirimycin that inhibit the action of glucosidase I, which removes the first glucose from Glc3MangGlcNAcz oligosaccharides, and glucosidase 11, which removes the next two glucoses. Both inhibitors prevented in vivo associations between either calnexin or calreticulin and the bulk of glycoproteins with which they normally associate [28, 291. Mutant cells that are deficient in either glucosidase I or I1 exhibited a similar phenotype 130, 311. These observations, coupled with the demonstration of monoglucosylated oligosaccharides on glycoproteins that associate with calnexin [32] or calreticulin [29] led to proposals that calnexin and calreticulin are ER-localized lectins with specificity for monoglucosylated (Glcl MangGlcNAc2) oligosaccharides [28, 291. As discussed in the previous section, this prediction has been substantiated for both calnexin and calreticulin by direct in vitro binding experiments with purified oligosaccharides [ 18-20]. Glucosidase inhibitors, particularly castanospermine (CST), have proven to be useful tools to study the in vivo effects of blocking the interactions of calnexin and calreticulin with glycoproteins. Most studies support the view that these molecules function to promote the correct folding of glycoproteins. For example, treatment

61.2 Biological Functions

1001

with CST doubled the rate of disulfide oxidation and oligomerization of influenza HAg, but decreased folding efficiency by increasing aggregate formation [33]. For class I histocompatibility molecules, CST resulted in increased aggregate formation and reduced assembly efficiency in murine systems [34] and slowed disulfide formation [35] and prevented interactions with the accesssory protein TAP in human systems [27]. CST treatment has also been reported to abolish expression of tyrosinase activity in Cos 7 cells [36], to cause premature dimerization and misfolding of the insulin receptor [37], to inhibit folding of the VSV G glycoprotein [38], and to decrease folding, assembly and surface expression of the nicotinic acetylcholine receptor [ 391. Furthermore, in a heterologous Drosophilu cell expression system, canine calnexin and rabbit calreticulin have been shown to enhance the assembly of murine class I histocompatibility molecules. This enhancement was due to a reduction in aggregation and an increase in proper folding of the class I heavy chain 134, 401. The advantage of this system is that calnexin- and calreticulin-specific effects can be assessed in a drug-free assay that does not involve altering the oligosaccharide strucure of all N-linked glycoproteins. Overall, the data are consistent with a molecular chaperone function for calnexin and calreticulin although the mechanism of how this is accomplished is the subject of considerable debate (see next section). It is important to note that some glycoproteins can fold properly in the presence of CST, exhibiting only minor changes in folding kinetics. For example, calnexin association is required for surface expression of membrane immunoglobulins if they are expressed without Ig-as dimers with which they normally associate, but is not required if all components of the complex are co-expressed [41]. Likewise CST treatment merely slows folding of the HN glycoprotein of Newcastle disease virus to half the normal rate, but the folding pathway is unchanged [42] and tissue-type plasminogen activator folds to an active species in the presence of CST [43l. Recently, a new aspect of calnexin and calreticulin function has arisen from chemical cross-linking experiments in canine pancreatic microsomes. Newly synthesized soluble and membrane-bound glycoproteins were found to interact in a carbohydrate-dependent manner not only with calnexin and calreticulin but also with the thiol oxidoreductase, ERp57 [44, 451. As with calnexin and calreticulin, the association of ERp57 with newly synthesized glycoproteins is transient and dependent on glucose trimming [46-491. Since ERp57 has no lectin properties of its own, the carbohydrate specificity of its interactions appears to be the result of an association with calnexin or calreticulin. Indeed, in vitro studies have demonstrated that ERp57 binds to calnexin [ S] and calreticulin [ 131 and furthermore, that calnexin and calreticulin enhance the rate of ERp57-catalyzed oxidative refolding of reduced, nionoglucosylated RNase B [8]. Enhanced folding was not observed for non-glucosylated RNase B showing that the effect was due to lectin engagement. Presumably this effect is the result of a ternary complex formed when calnexin or calreticulin bring ERp57 into proximity with a glucosylated substrate. Thus the question arises as to what extent the in vivo efTects of calnexin and calreticulin are due to intrinsic chaperone functions or to their abilities to recruit foldases such as ERp57 into complexes with folding glycoproteins. In addition to their involvement in the folding of glycoproteins, calnexin and calreticulin participate in the retention of non-native glycoproteins within the ER

1002

61 Calnexin, Culreticulin and Glycoprotein Folding

(quality control). When expressed in Drosophila cells in the absence of their partner subunits, free class I heavy chains or peptide-deficient heavy chain-P,-microglobulin heterodimers are aberrantly transported to the cell surface. Co-expression of calnexin or calreticulin results in retention of these incompletely assembled molecules within the ER [34, 40, 501. Conversely, if normally retained subunits of the class I histocompatibility molecule or T cell receptor are co-expressed in mammalian cells with a calnexin mutant lacking its ER localization sequence, the subunits undergo a relocalization with calnexin to the Golgi and cell surface [51, 521. Furthermore, disruption of the calnexin gene in Succhuromyces cerevisiae results in enhanced secretion of a fusion construct consisting of lysozyme and a hydrophobic pentapeptide [ 531. This retention of non-native glycoproteins by calnexin or calreticulin may be relevant to the ER retention phenotype of mutant proteins associated with human diseases such as AF508 CFTR in cystic fibrosis and al-antitrypsin Z in adult-onset emphysema. Calnexin and calreticulin also influence the degradation of non-native glycoproteins that they retain in the ER, although conflicting effects have been reported. When interactions with calnexin and calreticulin are reduced by expression of glycoproteins in insect cells or in glucosidase deficient cells or by treatment of cells with glucosidase inhibitors, the degradation of most non-native glycoproteins is accelerated [30, 33, 39, 54, 551. By contrast, studies in S. cerevisiue have shown that disruption of the gene encoding the calnexin homologue, Cnelp, results in impaired degradation of a model misfolded glycoprotein suggesting some role for calnexin in the degradation process [56]. Consistent with this view are studies documenting degradation of the T cell receptor a subunit in immature thymocytes despite association with calnexin [ 571 and also that calnexin becomes polyubiquitinated following its association with mutant a,-antitrypsin [58]. A model to explain these disparate findings has been offered by Sifers and co-workers who among other groups have noted a correlation between the action of ER a-mannosidase on a misfolded glycoprotein and its rapid degradation [59, 601. They provide evidence suggesting that the formation of Glcl MansGlcNAc2 oligosaccharides on misfolded mutant al-antitrypsin by the action of ER a-mannosidase I is required for degradation and also results in a prolonged association with calnexin, since deglucosylation of this specific oligosaccharide by glucosidase I1 is inefficient. Furthermore, they demonstrate that the degradation of mutant a1 -antitrypsin occurring in the in the presence of bound calnexin proceeds via the proteasome whereas in the absence of calnexin (CST treatment) degradation occurs via a non-proteasomal pathway [60]. Therefore, calnexin may play a role in targeting proteins to the proteasomal pathway and the disparate effects observed upon inhibiting calnexin interactions in various systems may reflect inherent differences in the abilities of individual glycoproteins to be degraded by proteasomal and non-proteasomal systems.

61.3 Mechanism of Action Two models have been proposed to describe the interactions of calnexin and calreticulin with newly synthesized glycoproteins (Figure 2). In the first model [28, 321,

61.3 Mechanism ojAction

1003

Lcctin-only model

Dual-bind ing model

h Figure 2. Current models for the mechanism of action of calnexin and calreticulin.

calnexin and calreticulin function solely as lectins. They bind to Glci MangGlcNAc2 oligosaccharides that are formed as an intermediate in the trimming of glucose residues from the precursor Glc3MangGlcNAc2 oligosaccharide. Following the initial association of calnexin and calreticulin with the monoglucosylated oligosaccharide on glycoproteins, it is proposed that cycles of release and re-binding occur that are regulated by the removal and re-addition of the single glucose residue. Glucose removal is catalyzed by glucosidase I1 and re-addition is catalyzed by UDP-glucose: glycoprotein glucosyltransferase, a resident ER enzyme that has the interesting property of reglucosylating only those glycoproteins that possess a non-native conformation [61].Thus the glucosyltransferase is thought to function as a folding sensor that terminates the cycle only when the glycoprotein is fully folded/assembled. Many features of this model have received experimental support. Treatment of cells with the glucosidase inhibitor CST, which blocks the formation of monoglucosylated oligosaccharides, prevents the association of calnexin and calreticulin with most glycoproteins. Furthermore, the addition of CST after complexes of glycoproteins and calnexin/calreticulin have formed prolongs the lifetime of the complexes, consistent with the view that glucosidase I1 action is involved in complex dissociation [32]. The latter treatment also interferes with the productive folding of influenza hemagglutinin [33] and transferrin [24] suggesting that cycles of glycoprotein binding and release with these chaperones are necessary for proper folding

1004

61 Calnexin, Calreticulin and Glycoprotein Folding

to occur. Several studies have also demonstrated that deglucosylation and reglucosylation of glycoproteins do indeed occur both in living cells and in intact microsomal systems and that the capacity to reglucosylate is important for interaction with calnexin and calreticulin [24, 32, 621 and for the efficient folding of transferrin [24]. Finally, studies on the binding of calnexin with monoglucosylated ribonuclease B have shown that the association occurs only via lectin-oligosaccharide interactions and, unlike other molecular chaperones, calnexin bound equally well to unfolded versus native conformational states [63, 641. Thus, in the “lectin only” model, calnexin and calreticulin do not function like classical molecular chaperones. Rather, their observed abilities to prevent aggregation and promote folding may arise from their recruitment of other molecular chaperones and protein folding catalysts such as ERp57 [8, 13, 44-49] and PDI [14] to the vicinity of the unfolded glycoprotein substrate. In the second “dual-binding” model, calnexin and calreticulin function both as lectins and as classical molecular chaperones [ 181. This model incorporates all of the features of the “lectin-only” model but, in addition, both calnexin and calreticulin are proposed to possess a second site that is capable of recognizing non-native polypeptide segments. In order to accomplish cycles of regulated binding and release that are characteristic of molecular chaperones, this model requires a change in the conformation of calnexin/calreticulin’s polypeptide binding site that, in addition to the action of glucosidase 11, permits dissociation of the non-native glycoprotein from the chaperone. Folding of the glycoprotein occurs upon dissociation and cycles of rebinding and release occur until a native conformation is achieved, i.e. when neither the chaperone’s polypeptide site nor the glucosyltransferase recognizes the glycoprotein. Thus, both calnexin/calreticulin and the glucosyltransferase act as folding sensors in this model. There is abundant evidence to support the basic tenet of the “dual-binding model”, i.e. that calnexin and calreticulin are capable of binding to glycoproteins via protein-protein interactions. For example, endoglycosidase H digestion of complexes between calnexin and class I histocompatibility molecules [ 18, 651, class I1 histocompatibility molecules [66], class I1 invariant chain [66], T cell receptor a and 6 chains [67], and a1-antitrypsin [IS] results in complete deglycosylation of each glycoprotein, yet it remains associated with calnexin. It has been suggested that the glycoproteins in these experiments have been isolated as aggregates and hence are binding non-specifically to calnexin [68], however in the case of class I molecules no such aggregates could be detected [ 341. Furthermore, the class I1 invariant chain [35] and acid phosphatase from Schizosaccharomycespombe [69] are able to bind to calnexin following castanospermine treatment. In addition, point mutations in mouse [70] and human [71] class I histocompatibility molecules have been described that are accompanied by a failure to bind to calreticulin. This appears to be due to a loss of protein-protein contacts since no change in glycosylation state nor substantial misfolding could be detected in the mutants. In vitro binding studies have also shown that calreticulin is capable of binding to denatured but not to native states of various proteins and that this binding is not dependent on the presence of N-linked glycans. Optimum binding of denatured proteins by calreticulin was observed near its isoelectric point and in the presence of either Ca2+ or Mg2+, suggesting that

61.4 Functional Relationship Between Culnexin und Calrrticulin

1005

Ca2+ binding to its acidic C-domain may regulate the polypeptide binding function [72]. Finally, calreticulin has been shown to bind to unglycosylated peptides both in vitro and in vivo [ 7 3 ] . The most compelling support for a polypeptide binding component in calnexin and calreticulin’s interactions with glycoproteins comes from in vitro experiments designed to test directly their molecular chaperone functions [74, 871. Using inhibition of aggregation as a test of chaperone function, calnexin and calreticulin were shown to be potent inhibitors of the thermal aggregation not only of unfolded soybean agglutinin (SBA) that possesses Glcl Mans GlcNAc;! oligosaccharides but also SBA that possesses Man9GlcNAcz oligosaccharides and even deglycosylated SBA. Similarly, calnexin and calreticulin were capable of inhibiting, in stoichiometric amounts, the thermally induced aggregation of citrate synthase (CS) and malate dehydrogenase (MDH) which do not possess N-linked glycans. Both chaperones formed stable complexes with unfolded CS and MDH but did not interact with the native enzymes, thereby demonstrating their capacity to discriminate between unfolded and folded conformations. Furthermore, calnexin and calreticulin were shown to enhance the refolding of thermally inactivated CS to an enzymatically active form. Overall, these findings indicate that calnexin and calreticulin do function as bona fide molecular chaperones and that their ability to bind to polypeptide segments of proteins is important for this function. The current challenge is to understand how their polypeptide binding capacity interrelates with their lectin function which is clearly crucial for their association with most glycoproteins in vivo. It is also unclear why previous in vitro studies with unfolded RNAse B failed to detect lectin-independent interactions with calnexin [64]. It is possible that during the folding of some glycoproteins there is minimal exposure of hydrophobic segments and in these cases interactions with calnexin or calreticulin’s lectin site (and consequent recruitment of foldases such as ERp57) may be sufficient to reach the native state.

61.4 Functional Relationship Between Calnexin and Calreticulin Based on their similar mechanisms of action, the question arises as to why the ER contains two related molecular chaperones, one membrane bound and the other soluble. Perhaps calreticulin and calnexin possess distinctly different functions in vivo or, alternatively, their functions may be similar but the labour of chaperoning the biogenesis of nascent glycoproteins is divided between them. One approach to address this issue has been to determine if calnexin and calreticulin are functionally interchangeable in vivo. Using the mouse class I histocompatibility molecule as a model glycoprotein, the individual effects of calnexin and calreticulin on the biogenesis of the class I molecule in a heterologous Drosophila expression system were compared. It was found that calnexin and calreticulin are largely interchangeable in their abilities to enhance the folding of class I heavy chains, to promote the assembly of heavy chains with the j3,-microglobulin subunit, to prevent rapid degradation

1006

61 Culnexin, Culreticulin and Glycoprotein Folding

of unassembled heavy chains, and to retain incompletely assembled class I molecules within the ER [MI. Given their apparently interchangeable functions, is there evidence that they divide the labour of chaperoning glycoprotein biogenesis? An examination of their glycoprotein binding specificities suggests that this is the case. Although there are several examples of glycoproteins that bind to both calnexin and calreticulin (in some instances simultaneously; reviewed in [22]), inspection of the overall spectrum of glycoproteins that co-isolate with calnexin and calreticulin reveals overlapping but distinctly different patterns [29, 751. Furthermore, the vesicular stomatitis G glycoprotein has been shown to bind to calnexin but not to calreticulin [29], calreticulin dissociates more rapidly than calnexin during the biogenesis of both influenza hemagglutinin (HA) and the T cell receptor [75, 761, and calnexin and calreticulin bind at distinctly different stages during the assembly of class I histocompatibility molecules [27]. Most attempts to understand the basis for this difference in binding specificity have focused on the number and location of Asnlinked oligosaccharide chains that are present on a glycoprotein substrate. For example, Helenius and co-workers studied the effects of removing different combination of the seven oligosaccharides of influenza HA and determined that calreticulin binds preferentially to oligosaccharides in the membrane distal portion of the molecule whereas calnexin binds equally well to oligosaccharides in the membrane distal and proximal portions of HA [76]. These plus additional studies [70, 771 indicate that the number and/or location of N-linked oligosaccharides can influence the choice of which chaperone is used and to some degree this may correlate with the luminal versus membrane locations of calreticulin and calnexin, i.e. calreticulin may bind preferentially to more exposed oligosaccharides whereas calnexin may bind to oligosaccharides located closer to the membrane. Additional experiments with point mutants of class I molecules suggest that polypeptide determinants may also influence interactions with calnexin and calreticulin. Specifically, substitution of Thr 134 by Lys in the human HLA-A2 molecule and substitution of Asp 227 by Lys in the mouse Ld molecule resulted in loss of calreticulin association but calnexin binding was unaffected. The mutations did not affect the glycosylation state of the molecules nor did they induce substantial conformational changes [70, 711. The simplest explanation is that the mutations result in the loss of a site of interaction with calreticulin’s polypeptide binding site. Further insight into the basis for the differential glycoprotein binding specificity of calnexin and calreticulin comes from mutational analysis of the chaperones themselves. When calreticulin was anchored to the ER membrane either via calnexin’s transmembrane and cytoplasmic segments or by the transmembrane and cytoplasmic segments of the adenovirus E3/19K glycoprotein, the pattern of glycoproteins that the chimera bound resembled that of calnexin rather than calreticulin [78, 881. Conversely, conversion of calnexin to a soluble molecule by deletion of its cytoplasmic tail and transmembrane segments resulted in a shift in its pattern of associated glycoproteins to that of calreticulin [40]. Combined with the results of selectively mutating membrane distal versus proximal glycosylation sites, the data indicate that the different topological orientations of calnexin and calreticulin play an important role in determining which glycoproteins each binds.

61.5 Relationship with other E R Chaperones arzd Folding Catalysts

1007

61.5 Relationship with other ER Chaperones and Folding Catalysts Although calnexin and calreticulin bind to many different glycoproteins it is apparent that a cell can survive in the absence of these interactions. For example, glucosidase I1 deficient cells are viable and cells can be cultured for extended periods in the presence of glucosidase inhibitors. This may be due in part to the fact that the folding of some glycoproteins does not depend on calnexin and calreticulin [41431. However, it is also clear that there is redundancy between the functions of calnexin, calreticulin, and other chaperones of the ER. When interactions with calnexin and calreticulin are prevented in glucosidase I1 deficient cells, the expression of the hsp70 chaperone BiP increases markedly [79]. Furthermore, in castanospermine treated cells, glycoproteins such as influenza HA and the insulin receptor that normally interact with calnexin and calreticulin now associate with BiP [37, 801. This redundancy also appears to extend to protein folding catalysts since studies in a cell free microsomal system have shown that when the interaction of ERp57 with a model glycoprotein is blocked by castanospermine treatment it is replaced by enhanced association with protein disulfide isomerase [44]. Although cultured cells can survive in the absence oi' chaperones that bind monoglucosylated glycoproteins, recent work has demonstrated their importance in whole organisms. Disruption of the calreticulin gene in mice results in an embryonic lethal phenotype and the presence of abnormalities in the embryonic heart suggests that calreticulin plays an important role in cardiac development [Sl, 821. Similar gene disruption studies performed on calmegin, the testis-specific homolog of calnexin, revealed that calmegin is required for sperm fertility and that calmegin-deficient sperm do not adhere to the egg zona pellucida. It was speculated that calmegin might be a chaperone for sperm surface glycoproteins [21]. It is becoming apparent that that there is some degree of organization and cooperation between ER folding factors. Studies in which influenza infected cells were chemically cross-linked in situ revealed that newly synthesized HA is present transiently in very large complexes with sedimentation coefficients ranging from 8s to more than 40s. Multiple ER chaperones including calnexin, calreticulin and BiP could be detected in these complexes [83]. Similarly, during heat stress, calreticulin has been detected in large 400-600 kDa complexes that also contain GRP94 and BiP [84]. The recruitment of ERp57 by calnexin and calreticulin to enhance the oxidative folding of monoglucosylated glycoproteins provides another example of the organization of folding factors [8]. There also appears to be temporal as well as spatial cooperation between ER chaperones, such as the initial interaction of VSVG glycoprotein with BiP and then subsequent association with calnexin [ 381, initial binding of reduced thyroglobulin to calnexin followed by association with BiP [ 851: and the transfer of human class I histocompatibility molecules from calnexin to calreticulin at the point of heavy chain assembly with p2-microglobulin [27]. One of the major challenges for the future is to understand how and to what extent folding factors within the ER are organized. Given the ability of calnexin and calreticulin to recruit ERp57 and of calreticulin to bind PDI, it is possible that these chaperones represent the core of a multimeric folding machine that binds to unfolded glyco-

1008

61 Calnexin, Culreticulin and Glycoprotein Folding

proteins, prevents their aggregation, and also presents them to other ER chaperones and protein-folding catalysts. In addition, our understanding of how calnexin and calreticulin interact with unfolded glycoproteins is far from complete. In vitro studies have revealed the presence of both oligosaccharide and polypeptide binding sites in these chaperones but how these two sites are utilized during folding cycles is unknown. Can both sites be occupied simultaneously or does occupancy of one site control access to the other? Preliminary evidence also suggests that different glycoproteins may have different requirements for binding to the polypeptide site. Clearly, detailed in vitro analysis of the interactions of a variety of glycoproteins with calnexin and calreticulin will be required to address these issues. References 1. Williams, D.B. Calnexin: A Molecular Chaperone with a Taste for Carbohydrate Biochem. Cell Biol. 1995 73, 123-132. 2. Bergeron, J.J.M., Brenner, M.B., Thomas, D.Y., and Williams, D.B. Calnexin: A Membranebound Chaperone of the Endoplasmic Reticulum Trends Biol. Sci. 1994 19, 123-128. 3. Wada, I., Rindress, D., Cameron, P.H., Ou, W.-J., Doherty 11, J.-J., Louvard, D., Bell, A.W., Dignard, D., Thomas, D.Y., and Bergeron, J.J.M. SSRa and Associated Calnexin are Major Calcium Binding Proteins of the Endoplasmic Reticulum Membrane J. Biol. Chem. 1991 266, 19599-19610. 4. Cala , S.E., Ulbright, C., Kelley, J.S. and Jones, L.R. Purification of a 90-kDa Protein (Band VII) from Cardiac Sarcoplasmic Reticulum J. Biol. Chem. 1993 268, 2969- 2975. 5. Baksh, S. and Michalak, M. Expression of Calreticulin in Escherichiu coli and Identification of Its Ca2' Binding Domains J. Biol. Chem. 1991 266, 21458-21465. 6. Tjoekler, L.W., Seyfried, C.E., Eddy, R.L. Jr., Byers, M.G., Shows, T.B., Calderon, J., Schreiber, R.B., and Gray, P.W. Human, Mouse, and Rat Calnexin cDNA Cloning: ldentification of Potential Calcium Binding Motifs and Gene Localization to Human Chromosome 5 Biochemistry 1994 33, 3229-3236. 7. Ou, W-J., Bergeron, J.J.M., Li, Y . , Kang, C.Y., and Thomas, D.Y. Conformational Changes Induced in the Endoplasmic Reticulum Luminal Domain of Calnexin by Mg-ATP and Ca2+ J. Biol. Chem. 1995 270, 18051-18059. 8. Zapun, A,, Darby, N.J., Tessier, D.C., Michalak, M., Bergeron, J.J.M., and Thomas, D.Y. Enhanced Catalysis of Ribonuclease B Folding by the Interaction of Calnexin or Calreticulin with ERp57 J. Biol. Chem. 1998 273, 6009-6012. 9. Hahn, M., Borisova, S., Schrag, J.D., Tessier, D.C., Zapun, A,, Tom, R., Kamen, A.A., Bergeron, J.J.M., Thomas, D.Y., and Cygler, M. Identification and Crystallization of a ProteaseResistant Core of Calnexin That Retains Biological Activity J. Struct. Biol. 1998 123, 260-264. 10. Smith, M.J., and Koch, G.L.E. Multiple Zones in the Sequence of Calreticulin (CRP55, Calregulin, HACBP), a Major Calcium Binding ERjSR Protein EMBO J . 1989 8, 3581-3586. 11. Fliegel, L., Burns, K., MacLennan, D.H., Reithmeier, R.A.F., and Michalak, M. Molecular Cloning of the High Affinity Calcium-binding Protein (Calreticulin) of Skeletal Muscle Sarcoplasmic Reticulum J. Biol. Chem. 1989 264, 21522-21 528. 12. Baksh, S., Spamer, C., Heilmann, C., and Michalak, M. Identification of the Zn2+ Binding Region in Calreticulin FEBS Lett. 1995 376, 53-57. 13. Corbett, E.F., Oikawa, K., Francois, P., Tessier, D.C., Kay, C., Bergeron, J.J.M., Thomas, D.Y., Krause, K.H., Michalak, M. Ca2+ Regulation of Interactions between Endoplasmic Reticulum Chaperones J. Biol. Chem. 1999 264, 6203-621 1. 14. Baksh, S., Bums, K., Andrin, C., and Michalak, M. Interaction of Calreticulin with Protein Disulfide Isomerase J. Biol. Chem. 1995 270, 31338-31344. 15. Ohsako, S., Hayashi, Y . , and Bunick, D. Molecular Cloning and Sequencing of Calnexin-t J. Biol. Chem. 1994 269. 14140-14148.

References

1009

16. Watanabe, D., Yamada, K., Nishina, Y., Tajima, Y., Koshimizu, U., Nagata. A., and Nishimune, Y. Molecular Cloning of a Novel Ca'+-binding Protein (Calmegin) Specifically Expressed during Male Meiotic Germ Cell Development J. Bid. Chem. 1994 269, 7744-7749. 17. Ohsako, S., Janulis, L., Hayashi, Y., and Bunick, D. Characterization of Domains in Mice of Calnexin-t, a Putative Molecular Chaperone Required in Sperm Fertility, with Use of Glutathione S-transferase-fusion Proteins Biol. Reprod. 1998 5Y, 1214-23. 18. Ware, F.E., Vassilakos, A,, Peterson, P.A., Jackson, M.R., Lehrman, M.A., and Williams, D.B. The Molecular Chaperone Calnexin Binds Glcl MangGlcNAc2 Oligosaccharide as an Initial Step in Recognizing Unfolded Glycoproteins J. Biol. Chem. 1995 270, 4697-4704. 19. Spiro, R.G., Zhu, Q., Bhoyroo, V., and Soling, H.-D. Definition of the Lectin-like Properties of the Molecular Chaperone Calreticulin, and Demonstration of its Copurification with Endomannosidase from Rat Liver Golgi J. Bid. Chern. 1996 271, 11588-1 1594. 20. Vassilakos, A,, Michalak, M., Lehrman, M.A., and Williams, D.B. Oligosaccharide Binding Characteristics of the Molecular Chaperones Calnexin and Calreticulin Biochemistry 1998 37, 3480-3490. 21. Ikawa, M., Wada, I., Kominami, K., Watanabe, D., Toshimori, K., Nishimune, Y., and Okabe, M. The Putative Chaperone Calmegin is Required for Sperm Fertility Nature 1997 387, 607-61 1. 22. Helenius, A,, Trombetta, E.S., Hebert, D.N., and Simons, J.F. Calnexin, Calreticulin and the Folding of Glycoproteins Trends Cell Biol. 1997 7, 193-200. 23. Ou, W.-J., Cameron, P.H., Thomas, D.Y., and Bergeron, J.J.M. Association of Folding Intermediates of Glycoproteins with Calnexin During Protein Maturation Nature 1993 364, 771 776. 24. Wada, I.; Kai, M., Imai, S., Sakane, F.; and Kanoh, H. Promotion of Transferrin Folding by Cyclic Interactions with Calnexin and Calreticulin EMBO J . 1997 16, 5420-5432. 25. Tatu, U.? Hammond, C., and Helenius, A. Folding and Oligomerization of Influenza Hemagglutinin in the ER and the Intermediate Compartment EMBO J . 1995 14, 1340-1348. 26. Degen, E., Cohen-Doyle, M.F., and Williams, D.B. Efficient Dissociation of the p88 Chaperone from Major Histocompatibility Complex Class I Molecules Requires Both P2Microglobulin and Peptide J. Exp. Med. 1992 175, 1653 -1661. 27. Sadasivan, B., Lehner, P.J., Ortmann, B., Spies, T., and Cresswell, P. Roles for Calreticulin and a Novel Glycoprotein, Tapasin, in the Interaction of MHC Class I Molecules with TAP Immunity 1996 5, 103-1 14. 28. Hammond, C., Braakman, I., and Helenius, A. Role of N-linked Oligosaccharide Recognition, Glucose Trimming, and Calnexin in Glycoprotein Folding and Quality Control Proc. Nut1 Acad. Sci. USA 1994 91, 913-917. 29. Peterson, J.R., Ora, A,, Van, P.N., and Helenius, A. Transient, Lectin-like Association of Calreticulin with Folding Intermediates of Cellular and Viral Glycoproteins Mu/. Biol. Cell 1995 6, 1173-1 184. 30. Kearse, K.P., Williams, D.B., and Singer, A. Persistence of Glucose Residues on Core Oligosaccharides Prevents Association of TCRa and TCRP Proteins with Calnexin and Results Specifically in Accelerated Degradation of Nascent TCRa Proteins within the Endoplasmic Reticulum EMBO J . 1994 13, 3678-3686. 31. Ora, A., and Helenius, A. Calnexin Fails to Associate with Substrate Proteins in Glucosidasedeficient Cell Lines J. Biol. Chem. 1995 270, 26060-26062. 32. Hebert, D.N., Foellmer, B., Helenius, A . Glucose Trimming and Reglucosylation Determine Glycoprotein Association with Calnexin in the Endoplasmic Reticulum Cell 1995 81, 425433. 33. Hebert, D.N., Foellmer. B., and Helenius, A. Calnexin and Calreticulin Promote Folding, Delay Oligomerization and Suppress Degradation of Influenza Hemagglutinin in Microsomes EMBO J . 1996 15, 2961L2968. 34. Vassilakos, A,, Cohen-Doyle, M.F., Peterson, P.A., Jackson, M.R., and Williams, D.B. The Molecular Chaperone Calnexin Facilitates Folding and Assembly of Class I Histocompatibility molecules EMBO J . 1996 15, 1495-1506. 35. Tector, M., and Salter, R.D. Calnexin Influences Folding of Human Class I Histocompatibility Proteins but not Their Assembly with P2-Microglobulin J. Bid. Chem. 1995 270, 19638-19642. -

1010

61 Calnexin, Calreticulin and Glycoprotein Folding

36. Toyofuku, K., Wada, I., Hirosaki, K., Park, J.S., Hori, Y., and Jimbow, K. Promotion of Tyrosinase Folding in Cos 7 Cells by Calnexin J. Biuchem. 1999 125, 82-89. 37. Bass, J., Chiu, G., Argon, Y., and Steiner, D.F. Folding of Insulin Receptor Monomers Is Facilitated by the Molecular Chaperones Calnexin and Calreticulin and Impaired by Rapid Dimerization J. Cell Biol. 1998 141, 636-646. 38. Hammond, C., and Helenius, A. Folding of VSV-G Protein: Sequential Interaction with BiP and Calnexin Science 1994 266, 456-458. 39. Chang, W., Gelman, M.S., and Prives, J.M. Calnexin-dependent Enhancement of Nicotinic Acetylcholine Receptor Assembly and Surface Expression J. Biol. Chem. 1997 272, 2892528932. 40. Danilczyk, U., Cohen-Doyle, M.F., and Williams, D.B. Functional Relationship between Calnexin, Calreticulin and Calnexin Domains ,4401. Bid. Cell 1998 9, 345a. 41. Wu, Y., Pun, C., and Hozumi, N. J. Roles of Calnexin and Ig-ap Interactions with Membrane Igs in the Surface Expression of the B Cell Antigen Receptor of the IgM and IgD Classes Immunity 1997 158, 2762-2770. 42. McGinnes, L.W., and Morrison, T.G. Role of Carbohydrate Processing and Calnexin Binding in the Folding and Activity of the HN Protein of Newcastle Disease Virus Virus Res. 1998 53, 175-185. 43. Allen, S., and Belleid, N.J. Calnexin and Calreticulin Bind to Enzymically Active Tissue-type Plasminogen Activator During Biosynthesis and are not Required for Folding to the Native Conformation Biochem. J . 1997 328, 113-1 19. 44. Elliott, J.G., Oliver, J.D., and High, S. The Thiol-dependent Reductase ERp57 Interacts Specifically with N-glycosylated Integral Membrane Proteins J. Bid. Chem. 1997 272, 21, 1384913855. 45. Oliver, J.D., van der Wal, F.J., Bulleid, N.J., and High, S. Interaction of the Thiol-Dependent Reductase ERp57 with Nascent Glycoproteins Science 1997 275, 86-88. 46. van der Wal, F.J., Oliver, J.D., High, S. The Transient Association of ERp57 with Nglycosylated Proteins is Regulated by Glucose Trimming Eur. J. Biochem. 1998 256, 51-59. 47. Morrice, N.A., and Powis, S.J. A Role for the Thiol-dependent Reductase ERp57 in the Assembly of MHC Class I Molecules Curr. Biol. 1998 8, 713-716. 48. Lindquist, J.A., Jensen, O.N., Mann, M., and Hammerling, G.J. ER-60, a Chaperone with Thiol-dependent Reductase Activity Involved in MHC Class I Assembly E M B O J . 1998 17, 2 186-2 195. 49 Hughes, E.A., and Cresswell, P. The Thiol Oxidoreductase ERp57 is a Component of the MHC Class I Peptide-loading Complex Curr. Biol. 1998 8, 709-712. 50 Jackson, M.R., Cohen-Doyle, M.F., Peterson, P.A., Williams, D.B. Regulation of MHC Class 1 Transport by the Molecular Chaperone, Calnexin (p88, IPYO) Science 1994 263, 384-387. 51 Rajagopalan, S., Xu, Y., and Brenner, M.B. Retention of Unassembled Components of Integral Membrane Proteins by Calnexin Science 1994 263, 387-390. C? J L . Rajagopalan, S., and Brenner, M.B. Calnexin Retains Unassembled Major Histocompatibility Complex Class I Free Heavy Chains in the Endoplasniic Reticulum J. Exp. Med. 1994 180, 407-41 2. 53. Arima, H., Kinoshita, T., Ibrahim, H.R., Azakami, H., Kato, A. Enhanced Secretion of Hydrophobic Peptide Fused Lysozyme by the Introduction of N-glycosylation Signal and the Disruption of Calnexin Gene in Sacrharomyces cerevisior FEBS Lett. 1998 440, 89-92. 54. Moore, S.E.H. and Spiro, R.H. Inhibition of Glucose Trimming by Castanospermine Results in Rapid Degradation of Unassembled Major Histocompatibility Complex Class I Molecules. J. Biol. Clzem. 1993 268, 3809-3812. 55. Chen, Y., Le Caherec. F., Chuck, S.L. Calnexin and Other Factors that Alter Translocation Affect the Rapid Binding of Ubiquitin to apoB in the Sec61 Complex J. Biol. Chem. 1998 273, 11887-1 1894. 56. McCracken, A.A., Brodsky, J.L. Assembly of ER-Associated Protein Degradation In Vitro: Dependence on Cytosol, Calnexin, and ATP J. Cell Biol. 1996 132, 291 -298. 57. Bennett M.J., van Leeuwen, J.E., and Kearse, K.P. Calnexin Association is not Sufficient to Protect T Cell Receptor a Proteins from Rapid Degradation in CD4 + CD8+ Thymocytes J. Biol. Chem. 1998 273, 23674-23680.

References

101 1

58. Qu, D., Teckman, J.H., Omura, S., Perlmutter, D.H. Degradation of a Mutant Secretory Protein, ul -antitrypsin Z, in the Endoplasmic Reticulum Requires Proteasome Activity J. Bid. Cliem. 1996 271, 22791-22795. 59. Jakob, C.A., Burda, P., Roth, J., and Aebi, M. Degradation of Misfolded Endoplasmic Reticulum Glycoproteins in Succharomyces ceveuisiae Is Determined by a Specific Oligosaccharide Structure J. Cell Biol. 1998 142, 1223-~1233. 60. Liu, T., Choudhury, P., Cabra, C.M., Sifers, R.N. Oligosaccharide Modification in the Early Secretory Pathway Directs the Selection of a Misfolded Glycoprotein for Degradation by the Proteasome J. Biol. Chem. 1999 274, 5861-5867. 61. Sousa, M. and Parodi, A.J. The Molecular Basis for the Recognition of Misfolded Glycoproteins by the UDP-G1c:glycoprotein Clucosyltransferase EMBO J . 1995 14, 41 96-4203. 62. van Leeuwen, J.E.M., and Kearse, K.P. Reglucosylation of N-linked Glycans is Critical for Calnexin Assembly with T Cell Receptor (TCR) u Proteins but not TCRP Proteins J. Bid. Chenz. 1997 272, 4179-4186. 63. Rodan, A.R., Simons, J.F., Trombetta, E.S., and Helenius, A. N-linked Oligosaccharides are Necessary and Sufficient for Association of Glycosylated Forms of Bovine RNase with Calnexin and Calreticulin EMBO J . 1996 15, 24, 6921-6930. 64. Zapun, A,, Petrescu, S.M., Rudd, P.M., Dwek, R.A., Thomas, D.Y., and Bergeron, J.J.M. Conformation-Independent Binding of Monoglucosylated Ribonuclease B to Calnexin Cell 1997 88, 29-38. 65. Zhang, Q., Tector, M., and Salter, R.D. Calnexin Recognizes Carbohydrate and Protein Determinants of Class I Major Histocompatibility Complex Molecules J. Biol. Chem. 1995 270> 3944-3 948. 66. Arunachalam, B., and Cresswell, P. Molecular Requirements for the Interaction of Class I1 Major Histocompatibility Complex Molecules and Invariant Chain with Calnexin J. Biol. C h e w 1995 270,2784-2790. 67. van Leeuwen, J.E.M., and Kearse, K.P. Calnexin Associates Exclusively with Individual CD3y and T Cell Antigen Receptor (TCR) a Proteins Containing Incompletely Trimmed Glycans that are not Assembled into Multisubunit TCR Complexes J. Bid. Chem. 1996 271, 96609665. 68. Cannon, K.S., Hebert, D.N., and Helenius, A. Glycan-dependent and -independent Association of Vesicular Stomatitis Virus G Protein with Calnexin J. Biol. Chem. 1996 271, 1428014284. 69. Jannitipour, M., Callejo, M., Parodi, A.J., Armstrong, J., and Rokeach, L.A. Calnexin and BiP Interact with Acid Phosphatase Independently of Glucose Trimming and Reglucosylation in Schizosuccharomyccs powbe Biochemistry 1998 37, 17253-~ 17261. 70. Harris, M.R., Yu, Y.Y.L., Kindle, C.S., Hansen, T.H., and Solheim, J.C. Calreticulin and Calnexin Interact with Different Protein and Glycan Determinants During the Assembly of MHC Class I J. Immun. 1998 160, 5404-5409. 71. Lewis, J.W. and Elliott, T. Evidence for Successive Peptide Binding and Quality Control Stages During MHC Class I Assembly Curr. Biol. 1998 8, 71 7-720. 72. Svzrke, C. and Houen, G. Chaperone Properties of Calreticulin Actu Chenz. Scund. 1998 52, 942-949. 73. Basu, S., and Srivastava, P.K. Calreticulin, a Peptide-binding Chaperone of the Endoplasmic Reticulum, Elicits Tumor- and Peptide-specific Immunity J. Exp. Med. 1999 189, 797-802. 74. Ihara. Y., Cohen-Doyle, M.F., and Williams, D.B. Calnexin Functions in uitro as a Chaperone for Non-glycosylated Proteins Mol. B i d . Cell 1998 9, 345a. 75. van Leeuwen, J.E.M., and Kearse, K.P. The Related Molecular Chaperones Calncxin and Calreticulin Differentially Associate with Nascent T Cell Antigen Receptor Proteins with the Endoplasmic Reticulum J. Biol. Chem. 1996 271, 25345-25349. 76. Hebert, D.N., Zhang, J-X., Chen, W., Foellmer, B., Helenius, A. The Number and Location of Glycans on Influenza Hemagglutinin Determine Folding and Association with Calnexin and Calreticulin J. Cell Biol. 1997 139, 613-623. 77. Zhang, A,, and Salter, R.D. Distinct Patterns of Folding and Interactions with Calnexin and Calreticulin in Human Class I MHC Proteins with Altered N-Glycosylation J. Immun. 1998 160, 831-837.

1012

61 Calnexin, Culseticulin and Glycopsotein Folding

78. Wada, I., Imai, S.-I., Kai, M., Sakane, F., and Kanoh, H. Chaperone Function of Calreticulin when Expressed in the Endoplasmic Reticulum as the Membrane-anchored and Soluble Forms J. Bid. Chem. 1995 270, 20298-20304. 79. Balow, J.P., Weissman, J.D., and Kearse, K.P. Unique Expression of Major Histocompatibility Complex Class I Proteins in the Absence of Glucose Trimming and Calnexin Association J. Biol. Cliem. 1995 270, 29025-19029. 80. Hammond, C., and Helenius, A. Quality Control in the Secretory Pathway: Retention of a Misfolded Viral Membrane Glycoprotein Involves Cycling bewteen the ER, Intermediate Compartment. and Golgi Apparatus J. Cell Biol. 1994 126, 41-52. 81. Coppolino, M.G., Woodside, M.J., Demaurex, N., Grinstein, S. St.-Arnaud, R., and Dedhar, S. Calreticulin is Essential for Integrin-mediated Calcium Signalling and Cell Adhesion Nature 1997 386, 843-847. 82. Mesaelin, N., Nakamura, K., Zvaritch, E., Dickie, P, Opas, M., MacLennan, D.H., and Michalak, M. Impaired Cardiac Development in Calreticulin (CRT) Knockout Mouse Mol. Bid. Cell 1998 9, 496a. 83. Tatu, U., and Helenius, A. Interactions between Newly Synthesized Glycoproteins, Calnexin and a Network of Resident Chaperones in the Endoplasmic Reticulum J. Cell Biol. 1997 136, 555-565. 84. Jethmalani, S.M., and Henle, K.J. Calreticulin Associates with Stress Proteins: Implications for Chaperone Function During Heat Stress J. Cell Biochem. 1998 69, 30-43. 85. Kim, P.S., and Arvan, P. Calnexin and BiP Act as Sequential Molecular Chaperones during Thyroglobulin Folding in the Endoplasmic Reticulum J. Cell Bid. 1995 128, 29-38. 86. Leach, M., and Williams, D. Unpublished data. 87. Ihara, Y., Saito, Y., and Williams, D. Manuscript in preparation. 88. Danilczyk, U., and Williams, D. Manuscript submitted.

Carbohydrates in Chemistry and Biology Beat Emst, Gerald W. Hart, Pierre Sinay copyright Q WILEY-VCH Verlag GmbH. 2000

62 Glycobiology of The Nervous System Ronald L. Schnaar

62.1 Introduction Nerve tissue is unique in its cellular complexity. A multitude of neurons characterized by different morphologies and neurotransmitters interact with one another, and with a variety of glial support cells, in a complex yet highly specific cellular tapestry. The precise spatial orientation of neurons and glia is key to proper nervous system function. Through mechanisms which have yet to be fully elucidated, neural cell surface molecules interact to ensure the establishment and maintenance of appropriate cell-cell interactions. Because glycoconjugates are highly enriched at the cell surface, where they are predominant determinants on all neural cells, it is inviting to speculate that they function in cell-cell recognition, among other roles. Although nervous system glycoconjugates share many properties with glycoconjugates of other tissues, there are qualitative and quantitative differences worth noting when considering their functions. This chapter presents a brief overview of nervous system glycoconjugates, touching on just a few of their unique features and potential functions. A more comprehensive overview of nervous system glycobiology can be found elsewhere [ 11.

62.2 Nervous System Glycoconjugates-Overview Nervous system glycoconjugates are notable for the preponderance of glycolipids (Table 1). Over half of the conjugated saccharide in the brain, representing a remarkable 1.9% of the brain’s total fresh weight, is accounted for by two closely related glycolipids, galactosylceramide (GalCer) and its 3-0-sulfated form, sulfatide [2]. Galactolipids are enriched in myelin, the multilayered membrane structure which ensheathes and insulates nerve cell axons [3].The second most abundant class

1014

62 Glycobioloyy of The Nervous System

Table 1 . Major glycoconjugates of adult rat brain Glycoconjugate Class

Concentration"

Reference

Glycolipids Galactolipids Gangliosides Glycoproteins N-linked 0-linked Glycosaminoglycans

36.8 (81%) 25.4h (57U/ir) 10.9' (24%) 7.gd (17%) 6.9 (15%)) 0.88' (1.9%) 0.89' (2%)

~

Total

45.5 (lOOO/;l)

5

1

PI PI [71 1761

PI

pmol of constituent monosaccharides per g fresh weight of brain: calculated from data in the indicated primary references as detailed in footnotes b-f. 1.94% of brain fresh weight (144-day-old rat); weight averaged molecular weight (76% GalCer/ 24% sulfatide) = 747.6 daltons 1.15 mg ganglioside sialic acid per g fresh weight, average of 2.94 ganglioside saccharide units per ganglioside sialic acid. 71 pmol total monosaccharide per g lipid-free dry weight; 9.09 g fresh weight per g lipid-free dry weight '2.5 pmol alkali-labile N-acetylgalactosamine per g lipid-free dry weight; 9.09 g fresh weight per g lipid-free dry weight; average of 3.2 0-linked saccharide units per 0-linked GalNAc 4.05 pmol hexosamine per g lipid-free dry weight; 2 glycosaminoglycan saccharides per hexosamine; 9.09 g fresh weight per g lipid-free dry weight a

of glycoconjugates in the brain, and the most abundant brain sialoglycoconjugates, are gangliosides, a large and varied family of sialylated glycosphingolipids [4-61. Glycoproteins carry about one sixth of the brain's total conjugated saccharides, mostly as N-linked oligosaccharides [7]. Finally, 2% of the brain's conjugated saccharides are found as proteoglycans [8]. In addition to the distinctive distribution of saccharides among these three glycoconjugate classes, there are individual oligosaccharide structures which are unique to or highly enriched in the nervous system. Specific neural lectins have also been identified. Glycoconjugates are likely to serve multiple biochemical and cell biological functions in the complex cellular environment of the nervous system, including functioning in neural cell-cell recognition.

62.3 Nervous System Glycolipids 62.3.1 Galactosylceramides

Galactolipids are the most abundant conjugated saccharides in the brain (Table l), and are particularly enriched in myelin [2]. Galactosylceramide (GalCer), the major galactolipid in the brain, is the simplest of glycoconjugates, comprised of a single

OH

Sphingorine .............................................................

i . .

,

HO \

GalCer, R=H Sufatide, R=SO,

Figure 1. Structures of galactosylceramides, the major glycoconjugates of brain. The fatty acid amide is shown with variable length (dotted line) and optional a-hydroxylation (parentheses).

galactose linked in P-linkage to ceramide (Figure 1). Structural variability in brain galactolipids derives from two sources. A portion ( ~ 2 5 %in adult rodent brain) [9, 101 exist as sulfatide, the 3-0-sulfated form of GalCer. Additional structural variation is found in the ceramide moieties of GalCer and sulfatide, which are characterized by long chain fatty acid amides, over half of which may be a-hydroxylated [9]. Fatty acid hydroxylation alters the presentation and thereby the recognition properties of GalCer, as evidenced by an anti-GalCer antibody which requires the ceramide fatty acid hydroxylation for binding [ l 11. Through sulfation and ceramide structural variations, even this simplest of glycoconjugates, GalCer, presents various structural determinants. What role do galactolipids play in the nervous system? Based on their abundance in myelin, one might propose that GalCer plays a role in myelination and/or myelin function. This is the case, as demonstrated by genetic disruption of the gene responsible for synthesis of brain galactolipids, UDPga1actose:ceramide galactosyltransferase [ 12, 131. Myelin, the multilamellar axon sheath formed by the plasma membrane of oligodendrocytes (in the central nervous system) or Schwann cells (in the peripheral nervous system), facilitates rapid saltatory nerve impulse conduction, in which the action potential jumps along the axon due to clustering of sodium channels at spaced, unmyelinated nodes of Ranvier [14]. Loss of myelin results in a sharp decrease in nerve conduction velocity, with severe loss of nervous system function, such as that found in the major demyelinating neuropathies multiple sclerosis and Guillain-Barre syndrome. Mice engineered to lack galactosylceramide demonstrate the phenotypic hallmarks of dysmyelination, including a reduction in conduction velocity (to unmyelinated levels), tremor, ataxia, and early death [ 12, 131. Surprisingly, Even though myelin function is greatly diminished, myelin is still formed throughout the nervous system and has near-normal ultrastructure [ 121. However, myelin formed in the absence of galactolipids is both diminished in function and unstable; extensive demyelination is apparent as the animals age. Based on these and subsequent studies, it appears that nervous system galactolipids are required for:

1016 i) ii) iii) iv)

62 Glycobioloyy of The Nevuous System

proper lipid packing in the myelin membrane [ 101; proper structure of the nodes of Ranvier [3, 151; long-term myelin maintenance [ 12, 131; and appropriate myelin gene expression [ 161.

The mechanism(s) by which galactolipids ensure proper myelin function and stability have yet to be determined. An interesting biochemical observation in galactolipid knock-out mice is the novel appearance of glucosylceramide with a-hydroxyl fatty acid chains [ 121. This lipid, which appears only in mice lacking GalCer expression, may provide some structural (although apparently little functional) compensation for the loss of galactolipids.

62.3.2 Gangliosides and Related Anionic Glycosphingolipids Gangliosides are defined as glycosphingolipids bearing at least one sialic acid residue. In the brain, gangliosides are expressed most frequently as components of the outer leaflet of the plasma membranes of neural and glial cells, with their ceramide portion embedded in the plasma membrane and their oligosaccharides extending into the extracellular space. The ganglioside family is comprised of about 100 known molecular species based on their varied carbohydrate structures alone, and many more when variations in their ceramide structures (which modify the carbohydrate presentation at the cell surface) are considered [4]. The simplest of these is carries of just two saccharides (NeuAc a2 Gal p1 Cer, GM4, a human myelin component) whereas the most complex may have 15 or more saccharides in branched arrays (see Table 2 for glycosphingolipid nomenclature). Gangliosides are expressed in anatomic, cell type, and developmentally specific patterns [ 17-19]. Although gangliosides are found in all vertebrate tissues, they are particularly abundant in the brain, where they represent a quarter of the total conjugated saccharides and 75-80% of the conjugated sialic acid [5]. This is in contrast to other tissues, which express gangliosides at O.2-8% the level found in the brain, depending on the tissue [20]. The unique abundance of gangliosides in the brain has led to considerable interest in their structure, metabolism, and function. The major brain gangliosides and their biosyntheses are shown in Figure 2 [21]. Gangliosides are synthesized step-wise, starting with the transfer of glucose from UDP-Glc to ceramide. Biosyntheses of GM 1, G D 1a, GD 1b, and GTl b, which together represent >90% of brain gangliosides, require the action of seven glycosyltransferases, all of which have been cloned (numbered reactions in Figure 2). Genetic and biochemical studies are beginning to clarify the roles of gangliosides in the nervous system. Disruption of GM2/GD2 synthase (UDP-GalNAc:GM3/GD3 pN-actetylgalactosaminyltransferase, enzyme “5” in Figure 2) resulted in mice with similar total brain ganglioside levels, but lacking all of the major brain gangliosides, and instead expressing only the simpler gangliosides GM3 and GD3 [22, 231. Despite this striking change in brain ganglioside structures, the mice appeared to develop normally and showed only modest initial nervous system deficits. A thorough study of these mice, however, demonstrated increased axon degeneration in the

62.3 Nervous System Glycolipids

1017

Table 2. Glycosphingolipid nomenclature. Designation”

Schematic structure

GalCer Sulfatide LacCer GM4 GM3 GD3 GM2 GD2 GM 1 3’-LM1 LK1 (SGGL) GDla GDlb GTlb GTlaa GQlba

Gal pl Cer 3-SOj-Gal B1 Cer Gal p4 Glc pl Cer NeuAc a3 Gal pl Cer NeuAc a3 Gal 84 Glc pl Cer NeuAc a8 NeuAc a3 Gal 84 Glc 81 Cer GalNAc p4 (NeuAc 1x3) Gal p4 Glc 81 Cer GalNAc p4 (NeuAc a8 NeuAc a3) Gal p4 Glc pl Cer Gal p3 GalNAc p4 (NeuAc 1x3)Gal 84 Glc 01 Cer NeuAc a3 Gal p4 GlcNAc p3 Gal p4 Glc 81 Cer 3-SO3-GlcU p3 Gal 84 GlcNAc p3 Gal 04 Glc p l Cer NeuAc a3 Gal p3 GalNAc p4 (NeuAc a3) Gal p4 Glc pl Cer Gal p3 GalNAc p4 (NeuAc a8 NeuAc a3) Gal 84 Glc 01 Cer NeuAc a3 Gal 03 GalNAc p4 (NeuAc a8 NeuAc a3) Gal p4 Glc p l Cer NeuAc a3 Gal 83 (NeuAc a6) GalNAc p4 (NeuAc a3) Gal b4 Glc p l Cer NeuAc a3 Gal p3 (NeuAc a6) GalNAc p4 (NeuAc a8 NeuAc a3) Gal 84 Glc 01 Cer

Ganglioside nomenclature according to [77]. Saccharides in parentheses indicate branches.

central (CNS) and peripheral (PNS) nervous systems, along with reduced myelination in the CNS and increased demyelination in the PNS [24]. One can conclude that the expression of the major brain gangliosides (GM1, GDla, GDlb, GTlb) is not required for gross nervous system development, but is required for nervous system maintenance. A more complete study of this model and other ganglioside glycosyltransferase gene disruption models is required to provide insights into the physiological functions of these major neural cell surface determinants. There are two general mechanisms by which gangliosides modulate neural cell biochemistry and physiology: cis regulation of proteins in the same membrane and trans recognition by complementary ganglioside binding proteins (lectins) on apposing cell surfaces [25, 261. Gangliosides and other sphingolipids, along with GPIanchored glycoproteins and certain intracellular signaling molecules (e.g. Src-family members), spontaneously cluster in the plane of the membrane [27]. An association between brain gangliosides and signaling molecules in the same membrane was evidenced by selective co-immunoprecipitation of the Src-family non-receptor tyrosine kinase Lyn using antibodies against the ganglioside GD3 [28]. A potential signaling cascade was evidenced by rapid activation of intracellular Lyn upon antibody-crosslinking of nerve cell surface GD3 via a mechanism that has yet to be defined. Lateral association of gangliosides with receptor tyrosine kinases has also been shown to modulate tyrosine kinase activity via a cis-association [29]. The potential for gangliosides to serve as ligands in cell-cell recognition is evidenced by studies of myelin-associated glycoprotein (MAG). MAG is expressed on the inner-

1018

62 Glycobiology of The Nervous System

most wrap of the myelin membrane, directly apposed to the axon surface [30]. MAG is a lectin (see below) which binds to gangliosides bearing a terminal a2,3linked sialic acid on a gangliotetraose core, such as the major brain gangliosides G D l a and GTlb [31, 321. Evidence that gangliosides are physiological ligands for MAG comes from studies on GM2/GD2 synthase knock-out mice (see above), which are lacking G D l a and G T l b (among other gangliosides) and display similar neuropathological changes as MAG-deficient mice [24]. It is possible that MAG on periaxolemmal myelin binds to ganglioside in the axolemma to maintain appropriate myelin-axon interactions. Although recent studies indicate potential roles for gangliosides in the nervous system, and mechanisms by which they may exert influence, the range of functions and physiological importance of gangliosides in the nervous system remain important matters for future investigation. In addition to their physiological functions, gangliosides are neuropathological targets. The neurotropism of tetanus toxin is due, at least in part, to specific binding of a region in its C-terminus to “lb” gangliosides (e.g. GDlb, GTlb) [33, 341. Gangliosides are also appear to be the targets of autoimmune attack, the basis for a Guillain-Barre-like motor neuropathy which appears in seasonal epidemics among children in certain areas of China [35, 361. Epidemiological and biochemical evidence indicate that particular strains of Campylobacterjejuni carry ganglioside-like structures on their lipopolysaccharide coats [ 371. Campylobacter infection leads to generation of anti-ganglioside antibodies (e.g. anti-GM 1 and anti-GD la), which initiate pathogenesis leading to motor axonal neuropathy. Anti-ganglioside antibodies are also found in more common forms of Guillain-Barre syndrome, which are a major cause of flaccid paralysis [36]. A unique class of anionic glycosphingolipids, sulfoglucuronyl glycolipids (SGGLs [38, 391, see Table 2) were discovered as immune targets in a related demyelinating peripheral neuropathy. Some patients with high levels of circulating monoclonal IgM (macroglobulinemia) develop an associated demyelinating peripheral neuropathy. The target for the monoclonal IgM in most such patients is a carbohydrate epitope terminated in a 3-sulfated glucuronic acid residue (3-SO3GlcU p3 Gal p4 GlcNAc) [40]. This epitope, which is also recognized by antihuman natural killer cell antibody (HNK-I), is found both on peripheral nerve glycolipids and as a glycoform on some cell recognition glycoproteins (see next section). Complementary binding proteins which recognize the SGGL determinant have been reported both in the CNS and PNS, raising speculation that it may be involved in physiological cell-cell recognition [41, 421. SGGL is preferentially expressed in the PNS, a finding which may relate to the core oligosaccharide on which it is carried (neolactotetraosylceramide, Gal p4 GlcNAc 83 Gal p4 Glc pl Cer). This core saccharide is preferentially expressed outside of the CNS, and is the core for another common PNS ganglioside, 3’-LM1 (see Table 2) [43]. In addition to the major nervous system glycosphingolipids, there are less abundant molecular species, some with intriguing and highly specific cellular distributions. One class of such structures are the “Chol-1” gangliosides, identified by an antibody raised against the purely cholinergic electric organ of the electric fish, T murmoruta. The Chol-l antibody is so named since it binds specifically to the minority of neurons in mammalian brain which produce acetylcholine as their neuro-

62.4 Nervous System Glycoproteins

1019

Cer

ECer

Key: Glucose 0 GalNAc 0Galactose V Sialic Acid

Figure 2. Biosynthetic pathways to the major brain gangliosides. Enzymes responsible for biosynthesis of the major brain gangliosides (and their immediate precursors) are numbered as follows: (1) UDP-G1c:ceramide P-glucosyltransferase [78]; (2) UDP-Ga1:glucosylceramide p I ,4galactosyltransferase [79]; (3) CMP-NeuAc:lactosylceramide a2,3-sialyltransferase [80]; (4) CMPNeuAc:GM3 a2,8-sialyltransferase [81]; ( 5 ) GM2/GD2 synthase [82]; (UDP-GalNAc:GM3/GD3 ~1,4-N-acetylgalactosaminyltransferase);(6) GM l/GD1 b synthase [83]; (UDP-Gal:GM2/GD2 ~1,3-galactosyltransferase); and (7) CMP-NeuAc:(Gal P3 GalNAc) a2,3-sialyltransferase (multiple enzymes possible) [84].

transmitter, but not to neurons producing other neurotransmitters [44]. The targets of Chol-I antibodies are a class of gangliosides bearing an “extra” sialic acid branch, linked a2,6 to the GalNAc residue in the gangliotetraose core (see Table 2) [45, 46). Thus, the Chol-l ganglioside “GQlba” is akin to GTlb (Figure 2), but with one additional sialic acid a2,6-linked to the GalNAc residue (also compare GTlacl to GDla). The functions of these and other gangliosides which are highly restricted to particular neuronal sub-classes are, as yet, unknown.

62.4 Nervous System Glycoproteins Glycoproteins carry a minority of conjugated saccharides in the brain (Table 1), most as N-linked glycans. As in other tissues, glycoproteins in the nervous system are particularly abundant on cell surfaces and exocytotic (synaptic) vesicles [I]. Among functionally important glycoproteins in the nervous system are cell-cell recognition glycoproteins, as well as glycoproteins involved in neuronal pathfinding, synaptic transmission, myelination, and other nervous system functions. Glycoproteins in the nervous system share many structural properties with glycoproteins in other tissues. This section will briefly review two glycoprotein

1020

62 Glycobioiogy of The Nervous System

modifications which are preferentially expressed in the nervous system. The reader is referred to more comprehensive compendia for general and thorough reviews of nervous system glycoproteins [ 11. 62.4.1 Polysialic Acid The neural cell adhesion molecule (NCAM) is a glycoprotein which is broadly distributed in the nervous system and in some non-neural tissues 1471. It exists in both membrane-spanning and glycophosphatidylinositol anchored forms. During embryonic development, NCAM is carries a large amount of polysialic acid, linear chains of a2,8-linked sialic acid numbering 50-200 residues, linked to the termini of N-linked oligosaccharides [48]. Whereas polysialic acid is abundant on NCAM during embryonic development, it is sharply down-regulated postnatally, becoming restricted to regions capable of neuronal plasticity (remodeling). Evidence has accumulated that the bulky, anionic polysialic acid chains function to reduce cell-cell contact, and thereby enhance migration and axon pathfinding 1491. Loss of polysialic acid, either experimentally or late in development, results in close apposition of cell membranes, enhanced cell-cell adhesion, and reduced cell or axon migration. These data implicate polysialic acid on NCAM, the predominant polysialic acid carrier in the nervous system, as a key regulator of cell-cell interactions during nervous system development. 62.4.2 The HNK-1 Determinant Several glycoproteins in the nervous system may display a determinant recognized by the antibody HNK-1, which also binds the SGGL glycosphingolipids described in the previous section. Many HNK- 1 reactive glycoproteins are cell-cell recognition proteins, including MAG, PO, NCAM, L1, tenascins and some proteoglycans [50].Because of they react with the HNK-1 antibody, they are presumed to carry an epitope akin to the SGGL glycosphingolipid. Oligosaccharides released from bovine PO, a PNS myelin protein required for stable myelination [51],revealed the expected saccharide (3-SO3-GlcU 133 Gal 84 GlcNAc) as the terminal sequence on N-linked oligosaccharide chains [52]. Whereas the HNK-1 determinant was reported to be expressed prominently on PO from bovine PNS, it is a minor glycoform on PO from other species and on other adhesion molecules. The functional significance of the HNK-1 determinant on adhesion glycoproteins has yet to be resolved.

62.5 Nervous System Glycosaminoglycans As in other tissues, brain glycosaminoglycans are found both as the non-sulfated unconjugated hyaluronic acid ([GlcU 133 GlcNAc B4In), and as the sulfated and

62.6 Lectins in the Bruin

1021

protein-conjugated proteoglycans, chondroitin sulfate and heparan sulfate. In the adult rat brain, glycosaminoglycans are comprised of 63Y0 chondroitin sulfates, 25% hyaluronic acid, and 12% heparan sulfates [53]. The major brain chondroitin sulfate proteoglycans are part of the family of proteoglycans akin to the major cartilage proteoglycan, aggrecan. Two family members, neurocan and brevican, appear to be specifically expressed in brain, whereas versican is abundant in brain and other tissues [54]. The name “lectican” has been proposed for this group of chondroitin sulfate proteoglycans, based on the presence of a lectin-like domain in the primary sequence of their protein core. Several lecticans have been cloned and sequenced. They are large polypeptides (brevican, ~ 1 0 kDa; 0 neurocan, ~ 1 4 kDa; 0 and versican, ~ 3 7 kDa) 0 containing an N-terminal hyaluronic acid-binding domain and a C-terminal C-type lectin motif. In addition they contain C-terminal EGF-like and complement regulatory protein repeats, similar to the selectin sub-family of Ctype lectins. The central portion of lecticans contains serine/proline rich glycosaminoglycan attachment sites. Lecticans and hyaluronic acid may be responsible for structuring the brain extracellular matrix, which (in the adult) is relatively devoid of other classes of extracellular matrix components [ 541. As multifunctional binding proteins, lecticans may cross-link cellular and extracellular components. The C-terminal lectin-like domain may bind both to cell surface glycoconjugates (see next section) and/or to the multimeric extracellular matrix glycoprotein, tenascin-R, whereas the N-terminal domain binds to hyaluronic acid. One can envision these components cooperating to generate a stable, three dimensional hydrated matrix. Proteoglycans in the brain are likely to share other functions with their counterparts elsewhere, including modulating cell migration (and thereby axon outgrowth), cell-cell adhesion (either directly or by modifying cell adhesion proteins), and polypeptide growth factor binding.

62.6 Lectins in the Brain As in other tissues, the functions of carbohydrates in cell recognition are likely to be mediated by complementary carbohydrate binding proteins [55, 561, although data is also emerging implicating carbohydrate-carbohydrate recognition [ 571. Evidence for specific nervous system lectins is briefly reviewed in this section. 62.6.1 Myelin-Associated Glycoprotein Oligodendrocytes in the CNS and Schwann cells in the PNS ensheathe neuronal axons with myelin, an insulating structure of multi-layered membranes [ 141. Myelin’s insulation is required for efficient nerve impulse conduction, but myelin has other profound biological effects. The inability of nerves to regenerate after CNS injury is largely due to the nerve’s inability to extend axons when in contact with

1022

62 Glycobiology of The Nervous System

myelin [58]. Myelin’s structural integrity and neuromodulatory effects may be due, in part, to myelin-associated glycoprotein (MAG). MAG, a minor constituent of CNS (1%) and PNS (0.1%) myelin, is implicated in myelin-axon interactions based on its in vivo location, in vitro functions, and deficits in MAG-deficient animals. In the CNS, MAG is found exclusively on the periaxonal membrane, directly across from the axon surface [59], and MAG incorporated into liposomes specifically binds to neuronal processes in cell culture [60, 611. Studies of genetically engineered MAG-deficient mice indicate that MAG plays a role in myelination, axon-myelin integrity, axon cytoarchitecture, and axon outgrowth [62-661. Presumably, these effects are dependent on MAG binding to specific target molecules on the axon surface. The nature of those targets emerged from seemingly unrelated studies on sialic acid binding lectins. MAG, a glycoprotein belonging to the immunoglobulin (Ig-like) superfamily of molecular recognition proteins, contains five Ig-like domains, a single transmembrane domain, and a short cytoplasmic tail [30]. Based on sequence similarity to other sialic acid-dependent Ig-like family member lectins, sialoadhesin and CD22, it was tested for lectin function and found to bind to saccharides in a sialic acid-dependent manner [67]. These lectins are now recognized as a new lectin family, the “siglecs” [68]. MAG was found to bind to a2,3-linked but not a2,6-linked sialic acids, and to preferentially bind to [NeuAc a3 Gal p3 GalNAc] termini [67]. Since these termini are most abundantly expressed on gangliosides in the brain, MAG was tested and found to bind avidly to the major brain gangliosides G D l a and GTlb [31]. This binding was highly specific, in that MAG did not bind to closely related gangliosides (e.g. GMl or GDlb, see Figure 2 for structures). Modification of the carboxylate, N-acetyl group, or any of the hydroxyl groups of ganglioside sialic acids blocked MAG binding, indicating a remarkably high level of ligand specificity and structural recognition [32, 691. The functional significance of MAG-sialylglycoconjugate binding is emerging from studies in vitro and in vivo. Mice engineered to lack complex gangliosides such as G D l a and G T l b (GM2/GD2 synthase knockout mice, see above) demonstrate neuropathology similar to that found in MAG knockout mice [24]. In vitro studies indicate that the ability of MAG to inhibit nerve regeneration is a sialic acid dependent event [70]. These data support the hypothesis that MAG, an important mediator of glial-neuronal interactions, initiates intercellular recognition via its sialic acid-dependent lectin activity. 62.6.2 Other Nervous System Lectins Lecticans, the major chondroitin sulfate proteoglycans in the brain, are characterized by a C-type (calcium dependent) lectin motif in their primary structure [54]. The additional presence of EGF-like and complement regulatory protein repeats is shared by the selectin sub-family of C-type lectins, leukocyte and endothelial cell lectins which recognize anionic (sialylated or sulfated) glycoconjugates and mediate leukocyte cell-cell recognition [ 7 11. A recent study demonstrated that brevican binds to sulfatides and HNK- 1 reactive glycolipids (SGGLs) in a calcium-specific

Rejeuences

1023

manner, and that this interaction could mediate nerve cell adhesion and neurite outgrowth [721. Other proteins which specifically bind to the HNK-1/SGGL structure have been found in both the CNS and the PNS. A soluble 30 kDa SGGLbinding lectin akin to the neurite-promoting protein amphoterin was purified from neonatal rat brain [42]. Its complementary localization to cellular structures apposing SGGL indicate possible interactions between the two [73]. A menibranebound, calcium-dependent specific binding activity for SGGL was also discovered on PNS myelin membranes [41]. Galectins are soluble, non-calcium requiring lectins which have a broad tissue distribution and share affinity for terminal galactose residues on a variety of glycoconjugates [74]. They are synthesized as cytoplasmic proteins, but are secreted by non-classical mechanisms into the extracellular space where they may bind to target glycoconjugates and, through multivalent interactions, mediate molecular recognition events. Selective expression of galectin-1 in a subset of sensory and motoneurons, together with selective expression of potential galectin-1 targets on the same and associated neurons, suggests that this lectin may function in nerve cell development or cell-cell interactions [75].

62.7 Concluding Remarks Glycoconjugates, predominant cell surface determinants on all neural cells, are likely to serve multiple functions in neurobiology, including roles in regulating the remarkably complex intercellular interactions required for appropriate neural development and function. The brain is unusual in the remarkable relative abundance of its glycolipids compared to its glycoproteins and proteoglycans. The roles of the major brain glycoconjugates, galactolipids and gangliosides, in the function and structural maintenance of myelin is clear from gene disruption models in mice. Full appreciation of the functions of these other glycoconjugates selectively expressed in the brain will benefit from the expanding knowledge of their metabolism and the genetic, biochemical, and cell biological tools available for their manipulation. References 1. R. U. Margolis, R. K. Margolis, Eds. Neurobioloyy of Glycoconjuyutes. Plenum Press, New York, 1989; 453 pp. 2. Norton, W. T.; Poduslo, S. E. Myelination in rat brain: Changes in myelin composition during brain maturation. J. Neurochem. 1973, 21 759-773. 3. Coetzee, T.; Suzuki, K.; Popko, B. New perspectives on the function of myelin galactolipids. Trends Neurosci. 1998, 21 126-130. 4. Stults, C. L. M.; Sweeley, C. C.; Macher, B. A. Glycosphingolipids: Structure, biological source, and properties. Methods Enzymol. 1989, 179 167-214. 5. Tettamanti, G.; Bonali, F.; Marchesini, S.; Zambotti, V. A new procedure for the extraction, purification and fractionation of brain gangliosides. Biochim Biophys. Acta 1973,296 160-170.

1024

62 Glycobiology of The Ncvvous System

6. Svennerholm, L.; Fredman, P. A procedure for the quantitative isolation of brain gangliosides. Biochim. Biophys. Acta 1980, 617 97-109. 7. Margolis, R. K.; Preti, C.; Lai, D.; Margolis, R. U. Developmental changes in brain glycoproteins. Brain Res. 1976, 112 363-369. 8. Margolis, R. U.; Margolis, R. K.; Chang, L. B.; Preti, C. Glycosaminoglycans of brain during development. Biochemistry 1975, 14 85-88. 9. De Haas, C. G.; Lopes-Cardozo, M. Hydroxy- and non-hydroxy-galactolipids in developing rat CNS. Intl J. Dev. Neurosci. 1995, 13 447-454. 10. Bosio, A,; Binczek, E.; Haupt, W. F.; Stoffel, W. Composition and biophysical properties of myelin lipid define the neurological defects in galactocerebroside- and sulfatide-deficient mice. J. Neurochem. 1998, 70 308-315. 11. Nakakuma, H.; Arai, M.; Kawaguchi, T.; Horikawa, K.; Hidaka, M.; Sakamoto, K.; Iwamori, M.; Nagai, Y.; Takatsuki, K. Monoclonal antibody to galactosylceramide: discrimination of structural difference in the ceramide moiety. FEBS Lett. 1989, 258 230-232. 12. Coetzee, T.; Fujita, N.; Dupree, J.; Shi, R.; Blight, A,; Suzuki, K.; Popko, B. Myelination in the absence of galactocerebroside and sulfatide: normal structure with abnormal function and regional instability. Cell 1996, 86 209-219. 13. Bosio, A,; Binczek, E.; Stoffel, W. Functional breakdown of the lipid bilayer of the myelin membrane in central and peripheral nervous system by disrupted galactocerebroside synthesis. Proc. Nut1 Acad. Sci. U.S.A. 1996, 93 13280-13285. 14. Lemke, G . Myelin and myelination. In An Introduction to Molecular Neurobiology; Hall, Z., Ed.; Sinauer: Sunderland, MA, 1992; pp. 281-309. 15. Dupree, J. L.; Coetzee, T.; Blight, A.; Suzuki, K.; Popko, B. Myelin galactolipids are essential for proper node of Ranvier formation in the CNS. J. Neurosci. 1998, 18 1642-1649. 16. Coetzee, T.; Dupree, J. L.; Popko, B. Demyelination and altered expression of myelinassociated glycoprotein isofoms in the central nervous system of galactolipid-deficient mice. J. Neurosci. Res. 1998, 54 613-622. 17. Kotani, M.; Tai, T. An immunohistochemical technique with a series of monoclonal antibodies to gangliosides: their differential distribution in the rat cerebellum. Brain Res. Brain Res. Protoc. 1997, 1 152-156. 18. Yu, R. K.; Macala, L. J.; Taki, T.; Weinfeld, H. M.; Yu, F. S. Developmental changes in ganglioside composition and synthesis in embryonic rat brain. J. Neurochem. 1988,50 1825-1829. 19. Schwarz, A,; Futerman, A. H. The localization of gangliosides in neurons of the central nervous system: the use of anti-ganglioside antibodies. Biochim. Biophys. Acta 1996, 1286 247-267. 20. Svennerholm, L. Gangliosides and synaptic transmission. Adu. Exp. Med. Bid. 1980, 125 533544. 21. van Echten, G.; Sandhoff, K. Ganglioside metabolism. Enzymology, topology, and regulation. J. Biol. Chem. 1993,268 5341-5344. 22. Takamiya, K.; Yamamoto, A,; Furukawa, K.; Yamashiro, S.; Shin, M.; Okada, M.; Fukumoto, S.; Haraguchi, M.; Takeda, N.; Fujimura, K.; Sakae, M.; Kishikawa, M.; Shiku, H.; Aizawa, S. Mice with disrupted GM2/GD2 synthase gene lack complex gangliosides but exhibit only subtle defects in their nervous system. Proc. Nut1 Acad. Sci. U.S.A. 1996, 93 10662-10667. 23. Liu, Y.; Wada, R.; Kawai, H.; Sango, K.; Deng, C.; Tai, T.; McDonald, M. P.; Araujo, K.; Crawley, J. N.; Bierfreund, U.; Sandhoff, K.; Suzuki, K.; Proia, R. L. A genetic model of substrate deprivation therapy for a glycosphingolipid storage disorder. J. Clin. Invest. 1999, 103 497-505. 24. Sheikh, K. A.; Sun, J.; Liu, Y.; Kawai, H.; Crawford, T. 0.; Proia, R. L.; Griffin, J. W.; Schnaar, R. L. Mice lacking complex gangliosides develop Wallerian degeneration and myelination defects. Proc. Natl. Acad. Sci. U.S.A. 1999, 96 7532-7537. 25. Hakomori, S. Bifunctional role of glycosphingolipids: Modulators for transmembrane signaling and mediators for cellular interactions. J. Biol. Chem. 1990, 265 18713- 18716. 26. Schnaar, R. L. Glycosphingolipids in cell surface recognition. Glycobioloyy 1991,l 477-485. 27. Simons, K.; Ikonen, E. Functional rafts in cell membranes. Nature 1997, 387 569-572. 28. Kasahara, K.; Watanabe, Y.; Yamamoto, T.; Sanai, Y. Association of Src family tyrosine kinase Lyn with ganglioside GD3 in rat brain. Possible regulation of Lyn by gl ycosphingolipid in caveolae- like domains. J. Biol. Chem. 1997,272 29947-29953.

Rejeuences

1025

29. Igarashi, Y.; Nojiri, H.; Hanai, N.; Hakomori, S. Gangliosides that modulate membrane protein function. Methods Enzymol. 1989, 179 521-541. 30. Trapp, B. D. Myelin-associated glycoprotein. Location and potential functions. Ann. N. Y Acad. Sci. 1990, 605 29-43. 31. Yang, L. J. S.; Zeller, C. B.; Shaper, N. L.; Kiso, M.; Hasegawa, A.; Shapiro, R. E.; Schnaar, R. L. Gangliosides are neuronal ligands for myelin-associated glycoprotein. Proc. Nut1 Acad. Sci. U.S.A. 1996, 93 814--818. 32. Collins, B. E.; Kiso, M.; Hasegawa, A.; Tropak, M. B.; Roder, J. C.; Crocker, P. R.; Schnaar, R. L. Binding specificities of the sialoadhesin family of I-type lectins. Sialic acid linkage and substructure requirements for binding of myelin-associated glycoprotein, Schwann cell myelin protein, and sialoadhesin. J. Biol. Chem. 1997, 272 16889-16895. 33. Rogers, T. B.; Snyder, S. H. High affinity binding of tetanus toxin to mammalian brain membranes. J. Biol. Chem. 1981,256 2402-2407. 34. Shapiro, R. E.; Specht, C. D.; Collins, B. E.; Woods, A. S.; Cotter, R. J.; Schnaar, R. L. Identification of a ganglioside recognition domain of tetanus toxin using a novel ganglioside photoaffinity ligand. J. Biol. Chem. 1997,272 30380-30386. 35. Griffin, J. W.; Li, C. Y.; Ho, T. W.; Tian, M.; Gao, C. Y.; Xue, P.; Mishu, B.; Cornblath, D. R.; Macko, C.; McKhann, G. M.; Asbury, A. K. Pathology of the motor-sensory axonal Guillain-Barre syndrome. Ann. Neurol. 1996, 39 17-28. 36. Sheikh, K. A.; Nachamkin, I.; Ho, T. W.; Willison, H. J.; Veitch, J.; Ung, H.; Nicholson, M.; Li, C. Y.; Wu, H. S.; Shen, B. Q.; Cornblath, D. R.; Asbury, A. K.; McKhann, G. M.; Griffin, J. W. Campylobacter jejuni lipopolysaccharides in Guillain-Barre syndrome: molecular mimicry and host susceptibility. Neurology 1998, 51 371-378. 37. Yuki, N.; Taki, T.; Inagaki, F.; Kasama. T.; Takahashi, M.; Saito, K.; Handa, S.; Miyatake, T. A bacterium lipopolysaccharide that elicits Guillain-Barre syndrome has a GM 1 ganglioside-like structure. J. Exp. Med. 1993, 178 1771-1775. 38. Quarles, R. H.; Ilyas, A. A,; Willison, H. J. Antibodies to glycolipids in demyelinating diseases of the human peripheral nervous system. Chem. Phys. Lipids 1986, 42 235-248. 39. Quarles, R. H. Myelin-associated glycoprotein in demyelinating disorders. Crit. Rev. Neurobiol. 1989, 5 1-28. 40. Ariga, T.; Kohriyama, T.; Freddo, L.; Latov, N.; Saito, M.; Kon, K.; Ando, S.; Suzuki, M.; Hemling, M. E.; Rinehart, K. L.; Kusunoki, S.; Yu, R. K. Characterization of sulfated glucuronic acid containing glycolipids reacting with IgM M-proteins in patients with neuropathy. J. Biol. Chem. 1987,262 848-853. 41. Needham, L. K.; Schnaar, R. L. Carbohydrate recognition in the peripheral nervous system: A calcium-dependent membrane binding site for HNK- 1 reactive glycolipids potentially involved in Schwann cell adhesion. J. Cell Biol. 1993, 121 397-408. 42. Nair, S. M.; Jungalwala, F. B. Characterization of a sulfoglucuronyl carbohydrate binding protein in the developing nervous system. J. Neurochem. 1997, 68 1286-1297. 43. Svennerholm, L.; Bostrom, K.; Fredman, P.; Jungbjer, B.; Lekman, A,; Minsson, J.-E.; Rynmark, B.-M. Gangliosides and allied glycosphingolipids in human peripheral nerve and spinal cord. Biochim. Biophys. Acta 1994, 1214 115-123. 44. Jones, R. T.; Walker, J. H.; Richardson, P. J.; Fox, G. Q.; Whittaker, V. P. Immunohistochemical localization of cholinergic nerve terminals. Cell Tissue Res. 1981, 218 355373. 45. Hirabayashi, Y.; Nakao, T.; Irie, F.; Whittaker, V. P.; Kon, K.: Ando, S. Structural characterization of a novel cholinergic neuron-specific ganglioside in bovine brain. J. Bid. Chem. 1992,267 12973-12978. 46. Ando, S.; Hirabayashi, Y.; Kon, K.; Inagaki, F.; Tate, S.; Whittaker, V. P. A trisialoganglioside containing a sialyl a2-6 N-acetylgdlactosamine residue is a cholinergic-specific antigen, Chol-la-a. J. Biochem. 1992, I l l 287-290. 47. Rutishauser, U.; Acheson, A.; Hall, A. K.; Mann, D. M.; Sunshine, J. The neural cell adhesion molecule(NCAM) as a regulator of cell-cell interactions. Science 1988, 240 53-57. 48. Troy, F. A. Polysialylation: from bacteria to brains. Glycobiology. 1992, 2 5-23. 49. Rutishauser, U.; Landmesser, L. Polysialic acid in the vertebrate nervous system: a promoter of plasticity in cell-cell interactions. Trends Neurosci. 1996, 19 422-427.

1026

62 Glycobiology of The Nervous System

50. Schachner, M.; Martini, R. Glycans and the modulation of neural-recognition molecule function. Trends Neurosci. 1995, 18 183-191. 51. Martini, R.; Zielasek, J.; Toyka, K. V.; Giese, K. P.; Schachner, M. Protein zero(P0)-deficient mice show myelin degeneration in peripheral nerves characteristic of inherited human neuropathies. Nut. Genet. 1995, 11 281-286. 52. Voshol, H.; van Zuylen, C. W.; Orberger, G.; Vliegenthart, J. F.; Schachner, M. Structure of the HNK-1 carbohydrate epitope on bovine peripheral myelin glycoprotein PO. J. Biol. Chem. 1996,271 22957-22960. 53. Margolis, R. K.; Margolis, R. U. Structure and localization of glycoproteins and proteoglycans. In Neurobiology of Glycoconjugutes; Margolis, R. U., Margolis, R. K., Eds.; Plenum Press: New York, 1989; pp. 85-126. 54. Ruoslahti, E. Brain extracellular matrix. Glycobiology. 1996, 6 489-492. 55. Sharon, N.; Lis, H. Lectins-proteins with a sweet tooth: functions in cell recognition. Essays Biochem. 1995,30:59-75 59-75. 56. Weis, W. I.; Drickamer, K. Structural basis of lectin-carbohydrate recognition. Annu. Rev. Biochem. 1996,65 441-473. 57. Iwabuchi, K.; Yamamura, S.; Prinetti, A,; Handa, K.; Hakomori, S. GM3-enriched microdomain involved in cell adhesion and signal transduction through carbohydrate-carbohydrate interaction in mouse melanoma B16 cells. J. Biol. Chem. 1998,273 9130-9138. 58. Schwab, M. E.; Kapfhammer, J. P.; Bandtlow, C. E. Inhibitors of neurite growth. Annu. Rev. Neurosci. 1993, 16 565-595. 59. Trapp, B. D.; Andrews, S. B.; Cootauco, C.; Quarles, R. The myelin-associated glycoprotein is enriched in multivesicular bodies and periaxonal membranes of actively myelinating oligodendrocytes. J. Cell Biol. 1989, 109 2417-2426. 60. Sadoul, R.; Fahrig, T.; Bartsch, U.; Schachner, M. Binding properties of liposomes containing the myelin-associated glycoprotein MAG to neural cell cultures. J. Neurosci. Res. 1990,25 1-13. 61. Johnson, P. W.; Abramow-Newerly, W.; Seilheimer, B.; Sadoul, R.; Tropak, M. B.; Arquint, M.; Dunn, R. J.; Schachner, M.; Roder, J. C. Recombinant myelin-associated glycoprotein confers neural adhesion and neurite outgrowth function. Neuron 1989, 3 377-385. 62. Yin, X.; Crawford, T. 0.;Griffin, J. W.; Tu, P.; Lee, V. Y.; Li, C.; Roder, J.; Trapp, B. D. Myelin-associated glycoprotein is a myelin signal that modulates the caliber of myelinated axons. J. Neurosci. 1998, 18 1953-1962. 63. Fruttiger, M.; Montag, D.; Schachner, M.; Martini, R. Crucial role for the myelin-associated glycoprotein in the maintenance of axon-myelin integrity. Eur. J. Neurosci. 1995, 7 51 1-515. 64. Lassmann, H.; Bartsch, U.; Montag, D.; Schachner, M. Dying-back oligodendrogliopathy: a late sequel of myelin-associated glycoprotein deficiency. GIiu 1997, 19 104-1 10. 65. Bartsch, S.; Montag, D.; Schachner, M.; Bartsch, U. Increased number of unmyelinated axons in optic nerves of adult mice deficient in the myelin-associated glycoprotein(MAG). Bruin Res. 1997, 762 231-234. 66. Schafer, M.; Fruttiger, M.; Montag, D.; Schachner, M.; Martini, R. Disruption of the gene for the myelin-associated glycoprotein improves axonal regrowth along myelin in C57BL/WldS mice. Neuron 1996, 16 1107-1 113. 67. Kelm, S.; Pelz, A,; Schauer, R.; Filbin, M. T.; Song, T.; de Bellard, M. E.; Schnaar, R. L.; Mahoney, J. A.; Hartnell, A.; Bradfield, P.; Crocker, P. R. Sialoadhesin, myelin-associated glycoprotein and CD22 define a new family of sialic acid-dependent adhesion molecules of the immunoglobulin superfamily. Curr. Biol. 1994, 4 965-972. 68. Crocker, P. R.; Clark, E. A,; Filbin, M.; Gordon, S.; Jones, Y.; Kehrl, J. H.; Kelm, S.; Le Douarin, N.; Powell, L.; Roder, J.; Schnaar, R. L.; Sgroi, D. C.; Stamenkovic, K.; Schauer, R.; Schachner, M.; van den Berg, T. K.; van der Merwe, P. A,; Watt, S. M.; Varki, A. Siglecs: a family of sialic-acid binding lectins. Glycobiology 1998, 8 v. 69. Collins, B. E.; Yang, L. J. S.; Mukhopadhyay, G.; Filbin, M. T.; Kiso, M.; Hasegawa, A,; Schnaar, R. L. Sialic acid specificity of myelin-associated glycoprotein binding. J. Biol. Chem. 1997,272 1248-1255. 70. DeBellard, M.-E.; Tang, S.; Mukhopadhyay, G.; Shen, Y.-J.; Filbin, M. T. Myelin-associated glycoprotein inhibits axonal regeneration from a variety of neurons via interaction with a sialoglycoprotein. Mol. Cell. Neurosci. 1996, 7 89-101.

References

1021

7 1. Drickamer, K. Evolution of Ca2+-dependent animal lectins. Proy. Nucleic Acid Res. 1993, 45 207-232. 72. Miura, R.; Aspberg, A,; Ethell, 1. M.; Hagihara, K.; Schnaar, R. L.; Ruoslahti, E.; Yamaguchi, Y . The proteoglycan lectin domain binds sulfated cell surface glycolipids and promotes cell adhesion. J. Biol. Chem. 1999,274 11431-11438. 73. Nair, S. M.; Zhao, Z.; Chou, D. K.; Tobet, S. A,; Jungalwala, F. B. Expression of HNK-1 carbohydrate and its binding protein, SBP-I, in apposing cell surfaces in cerebral cortex and cerebellum. Neuroscience 1998, 85 759-77 1. 74. Barondes, S. H.; Cooper, D. N.; Gitt, M. A,; Leffler, H. Galectins. Structure and function of a large family of animal lectins. J. Biol. C%em.1994, 269 20807-20810. 75. Hynes, M. A.; Gitt, M.; Barondes, S. H.; Jessell, T. M.; Buck, L. B. Selective expression of an endogenous lactose-binding lectin gene in subsets of central and peripheral neurons. J. Neurosci. 1990,lO 1004-1013. 76. Finne, J. Structure of the 0-glycosidically linked carbohydrate units of rat brain glycoproteins. Biochim. Biophys. Acta 1975, 412 317-325. 77. Svennerholm, L. Designation and schematic structure of gangliosides and allied glycosphingolipids. Prog. Bruin Res. 1994, 101 xi-xiv. 78. Ichikawa, S.; Sakiyama, H.; Suzuki, G.; Hidari, K. I.; Hirabayashi, Y. Expression cloning of a cDNA for human ceramide glucosyltransferase that catalyzes the first glycosylation step of glycosphingolipid synthesis. Proc. Nutl Acad. Sci. U.S.A . 1996, 93 4638-4643. 79. Nomura, T.; Takizawa, M.; Aoki, J.; Arai, H.; Inoue, K.; Wakisaka, E.; Yoshizuka, N.; Imokawa, G.; Dohmae, N.; Takio, K.; Hattori, M.; Matsuo, N. Purification, cDNA cloning, and expression of UDP-Gal: glucosylceramide beta-I ,4-galactosyltransferase from rat brain. J. Biol. Chem. 1998,273 13570-13577. 80. Ishii, A.; Ohta, M.; Watanabe, Y.; Matsuda, K.; Ishiyama, K.; Sakoe, K.; Nakamura, M.; Inokuchi, J.; Sanai, Y.; Saito, M. Expression cloning and functional characterization of human cDNA for ganglioside GM3 synthase. J. Biol. Chem. 1998,273 31652-31655. 81. Nara, K.; Watanabe, Y.; Maruyama, K.; Kasahara, K.; Nagai, Y.; Sanai, Y . Expression cloning of a CMP-NeuAc:NeuAca 2- 3GalpI -4Glcpl-1’Cer a2,8-sialyltransferase (GD3 synthase) from human melanoma cells. Proc. Nutl Acad. Sci. U.S.A. 1994, 91 7952-7956. 82. Nagata, Y.; Yamashiro, S.; Yodoi, J.; Lloyd, K. 0.; Shiku, H.; Furukawa, K. Expression cloning of pl,4 N-acetylgalactosaminyltransferase cDNAs that determine the expression of GM2 and GD2 gangliosides. J. Biol. Chem. 1992, 267 12082-12089. 83. Miyazaki, H.; Fukumoto, S.; Okada, M.; Hasegawa, T.; Furukawa, K. Expression cloning of rat cDNA encoding UDP-galactose:GD2 betal,3-galactosyltransferasethat determines the expression of GDlb/GMl/GAI. J. Biol. Chem. 1997, 272 24794-24799. 84. Lee, Y. C.; Kojima, N.; Wada, E.; Kurosawa, N.; Nakaoka, T.; Hamamoto, T.; Tsuji, S. Cloning and expression of cDNA for a new type of Gal beta 1,3GalNAc alpha 2,3sialyltransferase. J. Bid. Chem. 1994, 269 10028-10033.

Carbohydrates in Chemistry and Biology Beat Emst, Gerald W. Hart, Pierre Sinay copyright Q WILEY-VCH Verlag GmbH. 2000

63 Glycobiology of the Immune System Elizabeth F. Hounsell

Protein and lipid glycosylation can be said to be at the very heart of immunology. The first line defense of the innate immune system is based on recognition of carbohydrates. The acute phase response and classical, immunoglobulin regulated, immunity are mediated by glycoproteins and some of the early antigens to be characterized were the carbohydrate blood groups. Blood group related structures are the first signals in attracting leucocytes to the sites of infection (via selectins) and the infectious microorganisms themselves are highly glycosylated. Recently glycopeptides and GPI membrane anchors of glycoproteins have been implicated in the T cell arm in so-called, non-classical MHC immune reactions. Some of the best candidates for cancer immunotherapy are sugar-based, but on the other hand antibodies to an oligosaccharide antigen are a major obstacle in xenotransplantation. This Chapter tracks the fate of the invading organism, concentrating on bacteria and the immune reaction against them. It reviews new information on the resulting pathology (microbial pathogenesis, inflammation and autoimmunity) and addresses how the immune system is controled to lead to opsonization and apoptosis rather than toxic shock. It raises the question of how we can pervert this control to increase immune reactions against cancer cells, but reduce them in xenotransplantation. For both B and T cell responses there is a distinct role of interacting molecules which are highly glycosylated (mucin and mucin-type molecules). Immunoglobulins are glycosylated too and for the reasons to be discussed one needs to be aware of this in their use in immunotherapy.

63.1 Infection and Pathogenesis Although there are many infectious organisms: viruses, mycobacteria, trypanosomes and other parasites, yeasts, prions etc, bacteria hold a major historical centerstage position. Pathogenic bacteria are categorized into two groups, Gram-negative

1030

63 Glycobiology of the Immune System

and Gram-positive, by the eponymous staining technique. These have distinct antigenic coats containing the glycoconjugate lipopolysaccharide (LPS) and S-layer polysaccharides, respectively. The oligosaccharide sequences are strong candidates for the cross-reaction with mammalian tissues which gives rise to the blood group antibodies from recognition of the gastrointestinal (GI tract) microflora, eg. blood group A persons have the blood group A antigen (Table 1) on their red blood cells and they raise antibodies to the blood group B antigen of GI tract microflora so that they can not accept blood group B red blood cells. More recently new pathologies based on cross-reactivity have been implicated for example against the bacterium Helicobacter pylori, infection with which in the stomach leads to gastritis and gastric cancer 111. One of the components of the LPS of this organism has the Le" structure (Table 1) which is well represented in the gastric mucosa including the gastric proton pump and may raise cross-reacting auto-antibodies [l]. Much is now also known about the colonization of bacteria based on binding to carbohydrate sequences of glycoproteins, eg. the mucins which line the GI tract and are also present on cell membrane glycoproteins and glycolipids. Once the bacteria have adhered to the epithelial cells by their carbohydrate binding-adhesins, they secrete invasins [2] and proteases which help establish infection in the tissues. A host inflammatory response is then mobilized to fight infection which involves a set of reactions aimed at preventing ongoing tissue damage, isolating and destroying the infectious agents as well as organizing the repair of damaged tissue (Figure 1). As shown in Figure 1, initially, an invading pathogen must be held in check by the innate immune system until a specific immune response can be mounted. Bacterial components discussed in the next section stimulate local tissue macrophages to secrete cytokines which in excess can cause toxic shock 121. Under control they alert the liver to the presence of infection which then secretes acute phase proteins (APPs). One of these, mannose-binding lectin (MBL), a serum C-type lectin can activate complement for direct bacterial opsonization. Deficiencyin the lectin results in recurrent bacterial infections [3]. Another, LPS-binding protein (LBP), regulates LPS induced cytokine release [4] via the GPI-anchored glycoprotein CD14 (discussed below). Local reaction caused by cytokines (or bacterial chemokines) induce the expression of endogenous mammalian adhesion molecules (discussed in detail in this book) on the endothelium among which are the E- and P-selectins which bind to sialyl Le" (Table 1) on circulating lymphocytes. Thus begins the specific, adaptive immune response mobilizing B and T cells. This is furthered by L-selectin on lymphocytes which binds to mucin-like molecules (see below) on endothelium. Leukocyte adhesion deficiency I1 (LAD 11) results in a low level of fucosylation of the ligands for E and P-selectins. The disease is characterized by severe vulnerability to infections [ 5 ] . On the other hand for the APP alpha-1 acid glycoprotein (AGP), an increase in fucosylation and the sialyl LeXdeterminant occurs during acute inflammation [6] which is suggested to increase its binding to selectins thus dampening down the immune response. Thus, AGP can protect mice from toxic shock induced by TNF or LPS [7]. Except for MBL and LBP, the generally assumed function of APPs is to limit the damaging affects of tissue injury by inhibition of released proteinases and by transport of proteins with anti-oxidant activity. With the exception of C-reactive protein (CRP), all APPs are glycosylated and this glycosylation is

63.1 Infection and Pathogenesis

7(

1031

(7)

Toxins and structural components (eg. LPS, LAM,UAsetc)

I

I

Neutrophil extravasecation and attack

I

e.g. MBL e.g. LBP

i8)

Dampen down acute response eg AGP

APPS

CD14 positive tissue macrophages

(2)

Liver

E and P selectins on endothelial cells )-

Selectin inhibition

binds, GlyCAh4 1 CD34, MadCAh4 1

\ / L

sgp200

kl

Inappropriatechronic response

Figure 1 . The glycobiology of infection.

subject to change during the inflammatory response. Because of their high serum concentrations these were amongst the first glycoproteins to be characterized, e.g. Setuin (from the cow, the human homolog of which is cystatin, an anti-trypsin) and AGP (previously known as orosomucoid because of its extent of glycosylation) which have long been used as standard N - , and in the case of Setuin 0-glycosylated mammalian proteins. Although glycosylation of eukaryotic proteins and their interactions have been the subject of three decades of extensive research, the presence of glycoproteins in prokaryotes has only just becoming appreciated. The oligosaccharides can be attached through either Asn, Ser/Thr or Tyr amino acids [S]. Although the first two

1032

63 Glycobiology of the Immune System

Table 1. Blood group and related antigens. Horizontal sequences act as precursors for vertical additions i.e. H for A/B, Sialyl Type I/Type I1 for fucosyltransferase. Structure on red blood cells/other tissues, lymphocytes etc Precursor Blood group 0 (H-gene encodes a fucosyltrans ferase) Blood group A (A-gene encodes a GaiNActransferase) Blood group B (B-gene encodes a Galtransferase) Lewis (Le) Leb/LeY for Type I/Type I1 Lea/Lex for Type I/Type I1

Sialyl Leb/LeX

Type I Gal~1-3GlcNAcj31-

Type I1 Galpl-4GlcNAcpl-

Fucal

+ 2Galpl-3GlcNAcpl-

Fucal

+

2Galpl-4GlcNAcpl-

Fucal

+ 2Galpl-3GlcNAcpl-

Fucal

+

2Galpl-4GlcNAc~1-

tl,3 GalNAca Fucal + 2Galpl-3GlcNAcpl11,3 Gala Fucal + 2Galfil-3GlcNAcpltl,4 Fuca Galpl-3GlcNAcplr1,4 Fuca

Tl>3

GalNAca Fucal + 2Galpl-4GlcNAcpl11,3 Gaia Fucal + 2Galpl-4GlcNAcpl1L3 Fuca Gal Pl-4GlcNAcplT1,3 Fuca NeuAca2-3Galpl-3GlcNAc~l- NeuAca2-3Galpl-4GlcNAc~I TL4 T1,3 Fuca Fuca

linkages approximate to those found in mammals, there are distinct differences making characterization difficult because it is not possible to rely on any profiling methods now available commercially. Similarly, the oligosaccharide sequences of LPS have many unique monosaccharides and non-sugar substituents. In Porphyromonas gingivalis, a bacterium associated with chronic inflammatory disease, we have detected glycosylation of bacterial proteases which we have shown to carry LPS-type monosaccharides [9]. Where these proteases have been shown to be virulence factors in microbial pathogenesis, our new data show that the glycosylation appears to be important in “cloaking” the protease activity from host immune recognition, i.e. in evading the host immune response. Another type of glycosylation of microorganisms which may mimic the host are of high mannose type. Although N-linked oligosaccharide chains of eukaryote glycoproteins exist, these have only 5-9 mannose residues and none terminating in the sequence Manal-3Manal-2Man. The latter sequence has been implicated in the allergic reaction of bakers and brewers to the yeast to which they are exposed in their work and also as a possible cross-reactive epitope with mycobacteria in Crohn’s disease [ 101. We have yet to find out whether protein glycosylation in general is important in allergenicity (see the last section herein). However, in general, highly polyvalent mannose structures of bacteria (and other microorganisms) are important in fighting infection because they are the first line of surveillance in the

63.2 Control of the Irnrnune Rtcsponse

1033

innate immune system resulting in macrophage opsonization via C l q or the collectin receptor found in phagocytes which opsonize the MBL-pathogen complex.

63.2 Control of the Immune Response In the case of Gram-negative bacteria, a major stimulator of the innate immune system is LPS which, for a rapid induction of an inflammatory response, needs to be bound by LBP [ l l ] . LBP has two distinct domains, the amino terminal 200 amino acids which bind LPS has no consensus glycosylation sites, whereas the carboxyl terminus has multiple N-glycosylation. It is the latter part of the LPS/LBP complex that then binds to a high affinity cell surface glycoprotein receptor, CD14, which is anchored in the mammalian cell membrane via a GPI moiety [ 121. The next stage of cell signal transduction [ 13, 141 involves the interaction of GPI-anchored glycoproteins with transmembrane glycoproteins (those having a hydrophobic amino acid sequence spanning the lipid bilayer). The resulting inflammatory response results in secretion of cytokines, which in excess cause endotoxic shock [2] and where CD14 binding has a key role. However at least two, and probably more, independent pathways exist for triggering lethal shock as this occurs with both Gram-negative and Gram-positive (LPS lacking) bacteria. For the latter, exotoxins (superantigens acting directly at the TCR), cell wall peptidoglycan (PTG) and lipoteichoic acids can also be the mediators [15]. CD14 will also bind to lipoarabinomannan (LAM) of Mycobacterium tuberculosis and polymannuronic acid [16, 171 for example. It must also be said that membrane GPI-CD14-independent pathways must function in patients with paroxysmal nocturnal hemoglobinuria (PNH) where GPI-anchored glycoproteins are absent [18]. PNH is an acquired abnormality of hemopoietic cells affecting GPI-tail biosynthesis (or attachment) which leads to the circulating immune complexes and susceptibility to bacterial infections associated with this disease [18]. In these patients soluble, sCD14 may have a role [ 191 as LBP catalyzes the binding of endotoxin to CD14 both in serum (sCD141 and at the monocyte cell surface (GPI-CD14 on macrophages or to a lesser extent on neutrophils and granulocytes). The GPI-anchored ADP-ribosyltransferase has also been implicated in correct mobilization of neutrophils in infection [20]. After binding of bacterial components, the GPI-CD 14-LBP complex signals across the cell membrane. As has been highlighted recently [ 141, GPI-anchored glycoproteins most probably associate with transmembrane glycoproteins in order to cell signal. N-glycosylation has been implicated in this interaction which may occur through C-type carbohydrate binding (lectin) domains of the transmembrane glycoproteins, in particular of integrins [ 13, 211, but the details remain controversial. The N-glycans and the glycan domain of the GPI, are each the same size as protein domains of 50-100 amino acids and, as we have shown for the GPI-anchored glycoprotein Thy-1, can have several influences on conformation and recognition [22]. In addition the leucine rich repeat (LRR) regions such as those in CD14 are known to have high affinity interactions which may be influenced by glycosylation (e.g the

1034

63 Glycobiology o j the Immune System

tumor associated antigen 5T4 [23])and the LRR-type transmembrane glycoproteins called Toll-like receptors (TLRs) have also recently been implicated in LPS-CD14 signaling [ 141. Two protein tyrosine kinases of the Src kinase family, Lck and Fyn, which are anchored on the cytoplasmic side of the membrane by lipid tails have also been implicated in this process 1241. The signaling is thus likely to involve both membrane lipid rafts 1251 and transmembrane glycoproteins. The cytokines released by the above mechanisms also signal to the liver to release the APPs which probably act to modulate the immune response as described above. Once the immune cells have been mobilized another significant part of the control is in ordered cell death (apoptosis). Distinct parallels can be drawn in apoptosis with the network of molecules implicated in the CD14 mechanism of initial response to bacterial macromolecules. Indeed, CD14 has now been shown to mediate recognition and phagocytosis of apoptotic cells 1261. Also CD16, a GPI-anchored glycoprotein with related interactions to CD 14, has been implicated in neutrophil apoptosis [27]. Thy-1 is involved in the membrane lipid channel-forming induced by the toxin aerolysin 1281 and, together with another GPI-anchored protein decay accelerating factor (DAF), in triggering thymocyte apoptosis 1291. These also use the Lck and Fyn Src tyrosine kinases in signalling in thymocytes [24]. Apoptosis can be seen as primarily a phenomenon of lipid rearrangement which exposes, on the outside of apoptotic vesicles, the normally internal phosphatidyl serine (PS) to which antibodies are directed as in phospholipid antibody disease [30]. In addition to exposure of PS on apoptotic cells, distinct differences in the surface carbohydrate have been found in particular loss of terminal sialic acid exposing a ligand for a Man/Fuc specific lectin [31]. This lectin interaction and integrin interactions with their glycoprotein receptors such as vitronectin and thrombospondin [30, 311 form a highly regulated network to ensure apoptosis rather than necrosis. It is important to understand the complex regulation of the networks described here not only in mounting an appropriate immune response against infection, but also in cancer. The new consensus is that oncogenesis is mediated, not necessarily by cell proliferation, but by suppression of naturally occurring apoptosis [32]. In cancer, as in infection, when the ordered regulation breaks down, there results a fatal cytokine imbalance. In both cancer and infection we need to understand these factors to manipulate the responses to antigens discussed next. Not to be forgotten here are also the role of the anionic, proteoglycans in orchestrating cell interactions and controling hemostasis 133, 341.

63.3 Bacterial and Tumor Antigens, Mucins and Mucin-like Molecules The carbohydrate blood group antigens were the first to be characterized at the structural level (reviewed in [35]).The antibodies (sera) and plant lectins which were shown to recognize them were than used as reagents to look at the distribution of blood group and related oligosaccharides in normal and tumor tissue [35, 361. These and anti-bacterial antibodies have now been characterized by X-ray crystallography

63.3 Bacterial and Tuntor Antigens, Muc'ins and Mucin-like Molecules

1035

initially within the complex with the carbohydrate binding protein, but more recently of the oligosaccharide alone [ 3 7 ] .Even a superficial discussion of key papers would be beyond the scope of the present Chapter. However, a few references are mentioned here. In a classic series of papers, Lemieux and his associates described their studies of the interactions of human blood group oligosaccharides with a variety of lectins and antibodies. These investigations are highlighted in an X-ray crystal structure analysis of the lectin IV of Griffonia simplicifolia which binds Le blood group oligosaccharides or their modified derivatives [38]. Bundle and his group reported on an X-ray crystal structure analysis of an antibody Fab fragment complexed with an oligosaccharide determinant from a Salmonella LPS [39] which is the paradigm for studying antibody responses to bacterial carbohydrate antigens in the design of conjugate vaccines. In another example, Sharon and his associates determined the crystal structure of the complex between lactose and the lectin from Erythrina corallodendron [40]. Plant lectins have homologous carbohydrate binding domains to those in mammals and may function in the plant as an early immune system. Their conformational analysis therefore is useful for studies of oligosaccharide-protein interactions in man. Mammalian proteins are also now being studied, such as the X-ray crystal structure analysis performed of a selectinlike, mutant mannose-binding protein with the bound sialyl-LeXtetrasaccharide or its modified derivatives [41]. The recent X-ray structures of the T-antigen disaccharide binding to Amuranthus caudaatus agglutinin [42] and of glycophorin-binding wheat germ agglutinin [43] marks a turning point in understanding the recognition of glycopeptide antigens at the core of glycoproteins, in particular of 0-linked chains. Compared to N-linked chains (see the next section), there is restricted movement around the GalNAcwlSer/Thr linkage resulting in distinct glycopeptide motifs rather than more independently mobile N-linked glycosylation. Another major factor in biological recognition of glycoproteins is the presentation of oligosaccharides on 0-linked chains in a multivalent fashion. Thus the large molecular weight highly 0-glycosylated mucins which line the GI mucosa are the major source of antigenic activity in tumour diagnosis [36, 441. Antibodies to the oligosaccharide sequences have been explored in cancer immunotherapy. However it may be that at the core, both the oligosaccharide structure and its distribution on several near-neighbor amino acids is involved in recognition [45]. Equally important are the recent hopes of raising cytotoxic T-cell responses to the glycopeptides of mucins as it has been shown that these can be presented via MHC molecules to T cells [46].In infection specific T cell response have been shown against non-peptide epitopes and GPI-anchors [47]. Several lymphocyte surfdce molecules also have multiple 0-linked domains which are the major forms implicated in binding to cell surface glycoproteins via selectins. This is important in inflammatory responses and has been associated with LPSinduced activation of neutrophils via binding of lymphocyte selectin [48]. The molecules quoted in Fig 1 as binding to L-selectin are all highly 0-glycosylated, mucin-like molecules. In addition to presenting multivalent oligosaccharide recognition motifs at the end of the chains, the interactions at the glycopeptide core provide an extended structure to the molecules which are thence projected out into the intercellular space to promote cell interaction [49]. Interestingly mucins theni-

1036

63 Glycobiology of the Immune System

selves are now shown to have transmembrane domains which opens up studies of diverse recognition functions of these multi-domain macromolecules and exploitation of the tumor-associated changes that occur.

63.4 Immunoglobulins and Pathology Immunoglobulins (Ig), MHC molecules, and the receptors with which they interact are all glycoproteins (an important IgG receptor, FcyR3 or CD16, is also GPIanchored). Their N-linked oligosaccharides are primarily important in biosynthesis and folding [50] rather than specific recognition (IgA and IgM also have 0-linked glycosylation of unknown function). However, there is evidence that the glycosylation of the most studied of these, IgG, may be involved in pathology via recognition of exposed GlcNAc residues (Figure 2) in both the Fc [51] and Fab [52] regions. Fc glycosylation is one of the few N-linked glycosylations which can be seen as electron density in X-ray crystallography [53] and thus is expected to be more intimately involved in Fc folding and function. In rheumatoid arthritis this glycosylation is skewed towards glycoforms which have less terminal Gal (Figure 2) thus exposing penultimate GlcNAc residues [51, 541. In Fab regions the variability and hypervariability of amino acids leads to the introduction of consensus glycosylation sites which in those cases which have been studied are occupied. In three out of three which we have characterized these were bisected bi-antennary chains and molecular modeling showed that the bisecting GlcNAc was accessible away from protein [52] for possible interactions as has been proposed in metastatic cell [55]. The importance of this finding is that the Fab characterized were found in plaques of deposited Ig light chains in myeloma patients. This syndrome is called primary systemic amyloidogenesis, a fatal complication of myeloma which is also one of the family of amyloid diseases which includes Alzheimer’s disease, chronic dialysis syndrome and Creutzfeldt-Jakob disease via deposition of different (g1yco)peptides/proteins. Additionally, the possibility of associated pathology with both Fc and Fab glycosylation suggests strongly that glycosylation should be controlled in Ig preparations being used in immunotherapy [ 561. A second complication of modern therapy based in glycosylation is in xenotransplantation. Pigs offer the best hope of providing organs for transplantation to humans, but their tissues bear an oligosaccharide antigen to which humans have large amounts of circulating antibodies. This together with the presence of pig complement receptors leads to hyperacute rejection of xenografts. The antigen is structurally related to the blood group B antigen (Table l), but unlike in humans, the glycosyltransferase involved in its synthesis can add Galal-3 to GalP in the absence of the blood group H, Fucal-2. As with blood group A people with antibodies to blood group B erythrocytes, due we presume to bacterial cross reactivity (discussed above), so all humans have antibodies to the pig blood group B-like antigen (Gala1-3GalP). Several possible avenues are available to circumvent this problem. First, the glycosyltransferase gene in the pig can be knocked out. Second,

63.4 Immunoglobulins and Pathology

1037

Majw Fc glycoqlatioa *G.If31-4GlcNAcf3l-2M~a.l\ *Fucal\ 6 6 iGlcNAcf31-4 Manf3l-4GlcNAcf3l-4GlcNAc~l-Asn 3 +G.I~1-4GlcNAcf3l-2Manal/

Majw Fab glycosylation

NeuSAccrZ~~~l-4GkNAc~l-ZM~al\ F’ucal\ 6 6 tGIcNAcf31-4M.n~l-4ClcNAef31-4GlcNAcf3l-Asn 3 NeuSArcrZdCdf3l~kNAcf31-2M.aal/ GlcNAcf31-4is the bisecting GlcNAc

Fab

Fc

Figure 2. Immunoglobulin G glycosylation N-linked to the nitrogen of Asn in the consensus sequence AsnXaaSer/Thr (adapted from an original by F. Routier with thanks).

the blood group H fucosyltransferase could be over-expressed in the pig which would lead to a product which could not be a substrate for the pig enzyme (and in humans antibodies do not exist against blood group H, called 0).Third, the offending antibodies can be removed from the recipient (but they come back very quickly). Gene transfer of the human complement receptor to the pig may help, but in overcoming the problems of tissue rejection, we may be increasing our risk of infection from pig viruses. Thus one mechanism of restriction of viruses between species is that they carry the glycoprotein and oligosaccharide antigens of the host and are thus rejected. “In other words, virus inactivation occurs by precisely the same mechanism as hyperacute rejection of xenografts” [57].

1038

63 Glycobioloyy of the Immune System

The last area to be discussed here where glycoproteins are implicated in immunopathology is as allergens. Already mentioned above are the adverse reactions to yeast mannans. These are primarily IgG or IgM responses rather than classical IgE mediated hypersensitivity. Similarly, we have characterized the IgG responses to both 0- and N-linked avian and mammalian glycoprotein allergens. The first are the mucins of pigeon GI tract [58]which in pigeon fanciers cause an alveolitis, an acute inflammatory reaction of the lung. In the second, the glycosylation of animal proteins such as those in horse dander [59] may be important in their allergenicity. Thus oligosaccharides need to be investigated more thoroughly in many areas of inflammation: that mediated by our response to infection (Figure 1); periodontitis caused by organisms such as Porphyromonas gingivalis; and, in the lung, GI tract (see section 63.1) and joints. References 1. Moran, A.P., Appelmelk, B.J. and Aspinall, G.O. (1996) Molecular mimicry of host structures by lipopolysaccharides of campylobacter and Helicobacter spp: implications in pathogenesis J. Endotoxin Res. 3, 521-531. 2. Henderson, B., Poole, S. and Wilson, M. (1996) Bacterial modulins: a novel class of virulence factors which cause host tissue pathology by inducing cytokine synthesis. Microbiol. Rev. 60, 316-341, 3. Summerfield, J.A., Ryder, S., Sumiya, M., Thursz, M., Gorchein, A., Monteil, M.A., Turner, M.W. (1995) Mannose binding protein gene mutations associated with unusual and severe infections in adults. Lancet 345, 886-9. 4. Jack, R.S., Fan, X., Bernheiden, M., Rune, G., Ehlers, M., Weber, A., Kirsch, G., Mentel, R., Fiirll, B., Freudenberg, M., Schmitz, G., Stelter and Schiitt, C. (1997) Lipopolysaccharidebinding protein is required to combat a murine Gram-negative bacterial infection. Nature 389, 742-745. 5. Etzioni, A,, Frydman, M., Pollack, S., Avidor, I., Phillips, M.L., Paulson, J.C., GershoniBaruch, R. (1992) Brief report: recurrent severe infections caused by a novel leukocyte adhesion deficiency. New Eng. J. Med. 32, 1789-1792. 6. Van Dijk, W., Havenaar, E.C. and Brinkman-Van der Linden, E.C.M. (1995) Alpha 1-acid glycoprotein (orosomucoid): pathophysiological changes in glycosylation in relation to its function. Glycoconj. J. 12, 227-233. 7. Libert, C., Brouckaert, P. and Fiers, W. (1994) Protection by al-acid glycoprotein against tumor necrosis factor-induced lethality. J. Exp. Med. 180, 1571-1575. 8. Messner, P. (1997) Bacterial Glycoproteins. Glycoconj. J., 14, 3-1 1. 9. Curtis, M.A., Thickett, A,, Slaney. J. M., Rangdrajan, M., Aduse-Opoku, J., Shepherd, P., Paramonov, N. and Hounsell, E.F. (1999) Variable carbohydrate modifications to the catalytic chain of the RgpA and RgpB proteases of Porphyromonas gingivalis W50. Infect. Immun. 67, 3816-3823. 10. Young, M., Davies, M. J., Bailey, D Smestad-Paulsen, B., Wold, J., Barnes, R.M.R. and Hounsell, E. F. (1998) Characterization of oligosaccharides from an antigenic mannan of Saccharomyces cerevisiae. Glycoconj. J., 15, 815-822. 11. Wright, S.D., Ramos, R.A., Tobias, P.S., Ulevitch, R.J. and Mathison, J.C. (1990) CD14, a receptor for complexes of lipopolysaccharide (LPS) and LPS binding protein. Science 249 1431 1433. 12. Cauwels, A,, Wan, E., Leismann, M. and Tuomanen, E. (1997) Coexistence of CD14Dependent and Independent pathways for stimulation of human monocytes by gram-positive bacteria. Infect. and Immun. 3255-3260. 13. Petty, H. R . and Todd, R.F. 111. (1996) Integrins as promiscuous signal transduction devices. Immunol. Today 17, 209-2 1 1. -

References

1039

14. Gerard, C, (1998) For whom the bell tolls. Nature 395, 217-219. 15. Blank, C., Luz, A., Bendigs, S., Erdmann, A,, Wagner, H. and Heeg, K. (1997) Superantigen and endotoxin synergize in the induction of lethal shock. Eur. J. Immunol. 27, 825-833. 16. Espevik, T., Otterlei, M., Skjak-Braek, G., Ryan, L., Wright, SD., Sundan, A. (1993) The involvement of CD14 in stimulation of cytokine production by uronic acid polymers. Eur. J. Immunol. 23,255-61. 17. Yu, W., Soprano, E., Cosentino, G., Volta, M., Lichcnsteiu, H.S., Viale, G. and Vercelli, D. (1998) Soluble CD141-, 52 confers responsiveness to both lipoarabinomannan (LAM) and LPS in a novel HL-60 cell bioassay. J. fmmunol. In Press. 18. Selvaraj, P., Rosse, W.F., Silber, R. and Springer, T.A. (1988) The major Fc receptor in blood has a phosphatidylinositol anchor and is deficient in paroxysmal nocturnal haemoglobinuria. Nature 333, 565-567. 19. Vita, N., Lefort, S., Sozzani, P., Reeb, R., Richards, S., Borysiewicz, L.K., Ferrara, P. and Labeta, M.O. (1997) Detection and biochemical characteristics of the receptor for complexes of soluble CD14 and bacterial LPS. J. Immunol. 158, 3457-3462. 20. Kefalas. P., Allport, J.R., Donnelly, L.E., Rendell, N.B., Murray, S.: Taylor, G.W., Lo, G., Yadollah-Farsani, M. and MacDermot, J. (1997) Arginine specific mono(ADP-ribosy1)transferase activity in human neutrophil polymorphs. A possible link with the assembly of filamentous actin and chemotaxis. Adv. Exp. Med. B i d 419, 241-244. 21. Stockinger, H. (1997) Interaction of GPI-anchored cell surface proteins and complement receptor type 3. Exp. Clin. Imm. 14, 5-10, 22. Barboni, E., Rivero, B.P., George, A.J.T., Martin, S.R., Renouf, D.V., Hounsell, E.F., Barber, P.C. and Morris, R.J. (1995) The glycophosphatidylinositol anchor affects the conformation of the Thy-I protein. J. Cell Science. 108, 487-497. 23. Starzynska, T., Marsh, P.J., Schofield, P.F., Roberts, S.A., Myers, K.A. and Stem, P.L. (1994) Prognostic significance of 5T4 oncofetal antigen expression in colorectal carcinoma. Brit. J. Cancer 69, 899-902. 24. Brown, D.A. (1992) Interactions between GPI-anchored proteins and membrane. Trends Cell Biol. 2, 338-343. 25. Schnitzer, J.E., McIntosh, D.P., Dvorak, A.M., Liu, J. and Oh, P. (1995) Separation of caveolae from associated microdomains of GPI-anchored proteins, Science, 269, 14351439. 26. Devitt A., Moffatt, O.D., Raykundalia, C., Capra, D.J.: Simmons, D.L. and Gregory, C.D. (1998) Human CD14 mediates recognition and phagocytosis of apoptotic cells. Nature, 392, 505 -509. 27. Dransfield, I., Buckle, A.M., Savill, J.S., McDowall. A., Haslett, C. and Hogg, N. (1994) Neutrophil apoptosis is associated with a reduction in CD16 (Fc gamma RIII) expression. J. Immunol. 153, 1254-63. 28. Nelson, K.L., Raja, S.M. and Buckley, J.T. (1997) The glycosylphosphatidylinositol-anchored surface glycoprotein Thy-I is a receptor for the channel-forming toxin aerolysin. J. Bid. Chem. 272, 12170&12174. 29. Hueber, A-O., Raposo, G., Pierres. M. and He, H-T. (1994) Thy-I triggers mouse thymocyte apoptosis through a bcl-2-resistant mechanism. J. Exp. Med. 179, 784-796 30. Fadok, V.A., Savill, J.S., Haslett, C., Bratton, D.L., Doherty, D.E., Campbell, P.A. and Henson, P.M. (1992) Different populations of macrophages use either the vitronectin receptor or the phosphatidylserine receptor to recognize and remove apoptotic cells. J. Immunol. 149, 4029-4035. 31. Hall, S.E., Savill, J.S., Henson, P.M. and Haslett, C. (1994) Apoptotic neutrophils are phagocytosed by fibroblasts with participation of the fibroblast vitronectin receptor and involvement of a mannose/fucose-specific lectin. J. Immunol. 153, 3218-27. 32. Evan, G. and Littlewood, T (1998) A matter of life and death. Science 281, 1317-1322. 33. Jackson, D.G. (1997) Human leucocyte heparan sulphate proteoglycans and their roles in inflammation. Biochem. SOC.Trans. 25, 220-224. 34. Heyderman, R.S., Klein, N.J., Shennan, G.I., Levin, M. (1992) Reduction of the anticoagulant activity of glycosaminoglycans on the surfxe of the vascular endothelium by endotoxin and neutrophils: Evaluation by an amidolytic assay. Thrombosis Rex 67, 677-685.

1040

63 Glycobiology of the Immune System

35. Hounsell, E.F. (1994) Physicochemical analysis of oligosaccharide determinants of glycoproteins. Ado. Curbohydr. Chem. Biochem., 50, 31 1-350. 36. Hounsell, E.F. Young, M. and Davies, M.J. (1997) Glycoprotein changes in tumours: a renaissance in clinical applications. Clin. Sci. 93, 287-293. 37. Perez, S., Mouhous-Riou, N., Nifant’ve, N.E. and Imberty, A. (1996) Crystal and molecular structure of a histo-blood group antigens involved in cell adhesion: the Lewis x trisaccharide. Glycobiology 6, 537-542. 38. Vandonselaar, M., Delbaere, L.T.J., Spohr, U. and Lemieu, R.U. (1987) Crystallization of the lectin IV of Crifloniu simplicijolia and its complexes with the Lewisb and Y human blood group determinants. J. Biol. Chem., 262, 10848-10849. 39. Cygler, M., Rose: D.R. and Bundle, D.R. (1991) Recognition of a cell-surface oligosaccharide of pathogenic Salmonella by an antibody Fab fragment. Science, 253, 442-445. 40. Shaanan, B. Lis, H. and Sharon, N. (1991) Structure of a legume lectin with an ordered Nlinked carbohydrate in complex with lactose. Science, 254, 862-866. 41. Ng, K . K . 3 . and Weis, W.I. (1997) Structure of a selectin-like mutant of mannose-binding protein complexed with sialylated and sulfated Lewis’ oligosaccharides. Biochem. 35, 979988. 42. Transue, T.R., Smith, A.K., Mo, H., Goldstein, I.J. and Saper, M.A. (1997) Structure of benzyl T antigen disaccharide bound to Amurunthus caudutus agglutinin. Nature Struct. Biol. 4., 779783. 43. C.S Wright and Jaaeger (1993) J. Mol. Biol 232, 620-638. 44. Hounsell, E.F., Davies, M.J. and Renouf, D.V. (1996) 0-linked protein glycosylation structure and function. Glycoconj. J. 13, 19-26. 45. Haavik, S., Nilsen, M., Thingstad, T., Barsett, H., Renouf, D.V. Hounsell, E.F. and Codington. J.F. (1999) Specificity studies of an antibody developed against a mucin type glycoprotein. Glycoconj. J. 16, 229-236. 46. Meijohanns, E., Meldal, M., Jensen, T., Werdelin, 0, Galli-Stampino, L., Mouritsen, S. and Bock, K. (1997) Versatile soli-phase thiolytic reduction of azido and N-Dts groups in the synthesis of haemoglobin (67-76) 0-glycopeptides and photoaffinity labelled analogues to study glycan T-cell specificity. J. Chem. Soc., Perkin Trans. 1, 871-884. 47, Schofield, L., McConville, M.J., Hansen, D., Campbell, A S , Fraser-Reid, B., Grusby, M.J. and Tdchado, S.D. (1999) CDld-Restricted immunoglobulin G formation to GPI-anchored antigen mediated by NKT cells. Science, 283, 255-229. 48. Malhotra, R., Priest, R. and Bird, M.I. (1996) Role of L-selectin in lipopolysaccharide-induced activation of neutrophils. Biochem. J. 320, 589-593. 49. Barclay, N.A., Birkeland, M. L., Brown, M.H., Beyers, A,, Davis, S. J., Somoza, C., Williams, A.F. (1993) The lymphocyte antigen facts book, Acudemic Press Ltd. 50. Helenius, A. (1994) How N-linked oligosacchardies affect glycoprotein folding in the endoplasmic reticulum. Molec. Cell Biol. 5 , 253-265. 5 1 . Malhotra, R, Wormald, M.R., Rudd, P.M., Fischer, P.B., Dwek, R.A. (1995) Glycosylation changes of IgG associated with rheumatoid arthritis can activate complement via the mannosebinding protein. Nuture Medicine, I , 237-241. 52. Omtvedt, L.A., Bailey, D., Renouf, D.V., Hounsell, E.F., Davies, M.J., Paramonor, N.A., Haavik, S., Husby, G. and Sletten K. Glycosylation of immunoglobulin light chains in primary systemic amyloidogenesis: implications for plaque deposition. Amyloid. Int. J. Exp. Clin. Invest. In mess. 53. Dekenhofer, J. (1981) Crystal refinement and atomic models of a human Fc fragment and its complex with fragment B of protein A from Staphylococcus uureus at 2.9 and 2.8A resolution. Biochemistry, 20, 2361-2370. 54. Routier, F.H., Hounsell, E.F., Rudd, P.M., Takahashi, N., Bond, A,, Axford, J.S., and Jefferis, R. (1998) Quantitation of human IgG glycoforms isolated from rheumatoid sera: A critical evaluation of chromatographic methods. J. Immunol. Methods. 213, 113-130. 55. Yosimura, M., Ihara, Y. Ohnishi, A,, Ijuhin, N., Nishiura, T., Kanakura, Y., Matsuzawa, Y. and Taniguchi, N. (1996) Bisecting N-acetylglucosamine on K562 cells suppresses natural killer cytotoxicity and promotes spleen colonisation. Cancer Res., 56, 412-418.

References

1041

56. Routier, F.H., Davies, M.J., Bergemann K. and Hounsell, E.F. (1997) The glycosylation pattern of a humanised IgGI antibody (D1.3) expressed in CHO cells. Glycoconj. J. 14, 201207. 57. Weiss, R.A. (1998) Transgenic pigs and virus adaptation. Nutztvr, 391, 327-328. 58. Baldwin, C.I., Calvert, J.E., Renouf, D.V. Kwok, C. and Hounsell, E.F. Analysis of pigeon intestinal much allergens using a novel dot blot assay. C’ovhohydv. Rrs. In press. 59. Johnsen, J.K., Rademaker, G.J., Thomas-Oates, J., Hounsell, E. F., Bailey, D., Smestad Paulsen, J and Wold, J.K. Charactization of the N-glycan of Equ c 1, the major glycoprotein allergen in horse dander. In preparation.

Carbohydrates in Chemistry and Biology Beat Emst, Gerald W. Hart, Pierre Sinay copyright Q WILEY-VCH Verlag GmbH. 2000

64 Metabolic Engineering Glycosylation: Biotechnology's Challenge to the Glycobiologist in the Next Millennium Thomas G. Warner"

64.1 Introduction Over the course of the past two decades remarkable advances in recombinant DNA technology have made possible the commercial development of highly efficacious glycoprotein therapeutics for treatment of many debilitating and life threatening diseases. As biotechnology has matured, so has the realization that the proteinattached glycans are important and sometimes functionally critical components of these therapeutic molecules. Several excellent reviews have focused on the unique challenges that the complex, multistep process of protein glycosylation presents in the design, manufacture and development of glycoprotein pharmaceuticals [ 1-41. Some of the major issues that have been identified in these earlier summaries are still of concern to the biotechnology industry today. These include:

1) an awareness of the need to maintain consistency of carbohydrate content over the course of development of high-yield production; 2) the importance of insuring the complete extension of oligosaccharide chains during production and product isolation; 3 ) the need to eliminate potentially immunogenic carbohydrate residues; and 4) the need to more closely match the carbohydrates of human glycoproteins. Clearly, there has been a great deal of progress defining the difficulties involved in the production of recombinant glycoprotein therapeutics; however, implementing practical strategies for their solution has only recently begun. Metabolic engineering of the cell expression systems provides a means to address many of these problem areas. The definitions of metabolic engineering vary considerably [ 5-81 but the general concept is to genetically modify the metabolic pathways of an organism in order to increase production of a metabolite of interest. In this case metabolic engineering provides a mechanism for improving glycosylation, e.g. productivity, quality or consistency of recombinant proteins.

* email: [email protected]

1044

64 Metabolic Engineering Glycosylation

Many additional challenges have also arisen in the past few years. Market demands for more cost effective and more efficiently developed processes have dictated the need for alternate or improved expression hosts which are more easily manipulated and give significantly higher productivity than is possible with present technology. Newly discovered aspects of metabolism of several carbohydrate components of mammalian glycoproteins present novel avenues for metabolic engineering in animal cell expression hosts to address these challenges. In addition, insect cells and plants are two expression systems that have been recognized as offering great potential for facile, high productivity recombinant glycoprotein production. However, in their present form, they are not ideal for generation of protein pharmaceuticals because they produce molecules with glycans that may not be compatible with therapeutic applications in humans. It seems timely, therefore, to focus this review on these emerging research areas which represent biotechnology’s current and future challenges for glycobiology. First is a summary of new developments in carbohydrate biosynthetic pathways of animal cell expression hosts and avenues for enhancing glycosylation by intervention at specific sites within these pathways. Second is a summary of efforts currently underway to metabolically engineer glycosylation in insect cells and plants to develop these expression hosts so that they will be suitable for recombinant therapeutic glycoprotein production.

64.2 Recent Developments in Carbohydrate Biosynthesis in Animal Cells 64.2.1 Optimizing Sialylation of Recombinant Proteins by Metabolic Engineering Sialic Acid Biosynthesis The biosynthetic pathway for sialic acid formation in liver was well established nearly thirty years ago, by the elegant work from the groups of Roseman [9-111 and Warren [ 121 (Figure 1). The enzymatic conversion of UDP-N-acetyl glucosamine to N-acetylmannosamine is the critical step in the overall synthesis of sialic acid because the activity of this epimerase is controlled downstream in the pathway by feedback inhibition by CMP-N-acetylneuraminic acid [ 131. Thus, production of ManNAc as a precursor for sialic acid biosynthesis is a closely regulated cellular process. Recent studies have demonstrated that cellular glycoproteins and possibly glycolipids are ‘undersialylated’ in some cultured cell systems. Levels of sialic acid conjugates of cultured aortic cells [14] and human fibroblasts [ 151, increased dramatically in cultures that contained high concentrations of ManNAc. The interpretation of these results is that, down regulation of ManNAc synthesis by the UDP-GlcNAc2-epimerase is circumvented when exogenously added aminosugar is available for conversion into sialic acid. When the rate limiting step in sialic acid biosynthesis is by-passed, sialylation of proteins and gangliosides increases. Along similar lines, Gu and Wang [16] have recently shown significant increases in sialic acid content

64.2 Recent Developments in Curhohydrute Biosynthesis in Animal Cells

1045

CMP-aidlc acid transporter

CMP-NeuSAc

Neu5Ac

Feed beck lnhlbnron \

\

UDP-GlcNAc

Figure 1. Metabolic pathways for sialylation of recombinant proteins expressed in animal cells. Although the precise pathway in CHO or BHK cells has not yet been clearly defined, shown are potential routes for sialic acid biosynthesis based on metabolic studies of other tissues and somatic cell types. Enzymes shown in bold type may be gene targets for overexpression in animal cells to improve sialylation under high productivity conditions.

(- 15% increase) in the glycans on human recombinant interferon produced in CHO cells when an external source of ManNAc is available. These results suggest that, under normal conditions, the UDP-GlcNAc-2-epimerase maintains an intracellular pool size of CMP-sialic acid that is not sufficient to give complete coverage of the oligosaccharide chains of newly synthesized recombinant protein with sialic acid. Even though sialylation of the recombinant interferon increased in the presence of added ManNAc, it was still not complete. Partially capped structures with exposed galactose residues remained on the recombinant protein. Although it could be argued that steric factors account for the inability of the sialyltransferase to act on the remaining uncapped glycans, it is also possible that, in addition to the UDPGlcNAc-2-epimerase, other steps in the sialylation pathway, such as the transport of CMP-sialic acid into the Golgi compartment or the sialyltransferase, may also be partially rate limiting under some conditions. The antiporter protein has been cloned and the effect of overexpression of the nucleotide carrier on recombinant protein sialylation should be explored [ 17-19]. Overexpression of the sialyltransferase has been carried out in some systems, but thus far, significant enhancements in sialic acid levels on recombinant protein have not been observed [20-23). This may not be unexpected, if the basal sugar nucleotide level in the cells lines used in these experiments was not adequate to give complete sialylation prior to over-

1046

64 Metabolic Engineering Glycosylution

expression of the siayltransferase, then enhancing the transferase levels may not have a significant impact on the overall sialic acid content of the protein. Overexpression of several genes involved in sialic acid biosynthesis in combination with the a2,3 or 2,6 sialyltransferase may be most effective at achieving more complete termination of oligosaccharide bi0synthesis.t UDP-GlcNAc-2-epimerase activity [24], has not been detected in many tissues other than liver suggesting that there may be alternate routes for sialic acid formation. Recently the gene for the epimerase has been used as a probe for Northern blot analysis of mRNA from various tissues. The lack of an epimerase mRNA signal from many of the tissues confirms the observations of earlier enzymatic studies [25, 261. There are at least two possible alternate pathways for sialic acid biosynthesis in mammals. The first is the direct conversion of free GlcNAc to ManNAc by a cytosolic GlcNAc-2-epimerase which has been found in nearly all tissues including liver [24, 271 (Figure 1). Phosphorylation of ManNAc at C-6 (ManNAc kinase), and subsequent condensation with phosphoenol pyruvate will give 9-phosphoneuraminic acid. A second alternate route for sialic acid formation is by way of the acetylneuraminate pyruvate-lyase, which has usually been assumed to be a catabolic enzyme breaking down sialic acid into ManNAc and pyruvate [28]. However, the lyase, under appropriate conditions, can act reversibly and the equilibrium shifted toward the formation of sialic acid from these two substrates. ManNAc produced by the GlcNAc-2-epimerase can condense with pyruvate via the lyase catalyzed reaction by-passing the CMP-sialic acid regulated pathway entirely. It is important to note that the mechanism of sialic acid biosynthesis in CHO cells is not known at present. All of these enzymes may contribute to varying degrees in the formation of cellular sialic acid. In addition, their individual levels of expression in cultured cells may be cell line dependent or culture condition dependent. Compounding these ambiguities are recent observations we have made monitoring sialylation, by two different methods, of a recombinant immunoadhesion molecule, GP1-IgG [29], over the course of a CHO cell culture under high productivity conditions in which sodium butyrate is added as an inducer of protein synthesis (Figure 2). A 10 1 bioreactor culture of CHO cells secreting GP1-IgG was established and samples of the culture were withdrawn every 24 hr. The recombinant protein was isolated, the titers of the protein quantified, and the sialic acid content determined by chemical analysis. With these data the sialic acid levels of the newly synthesized protein could also be calculated for each 24 hr interval. In parallel experiments, samples of the culture were withdrawn and the sialic acid content of the protein was monitored by metabolically labeling sialic acid on the protein by including 'H-ManNAc in the culture media [30]. Remarkably, the two methods for monitoring sialylation gave dramatically differing results. Direct measurement of sialic acid by chemical analysis indicates that early in the culture, the sialylation of the recombinant protein is at near maximal levels -7 mol sialic acid/mol protein. However, after the butyrate addition, the recombinant protein synthesized had progressively lower levels of sialic acid. The actions of the CHO cell sialidase cannot account for these results because the viability of the cells was maintained at high levels over the course of the experiment

64.2 Recent Developments in Carhohydrute Biosynthesis

2 3

4

5 6 7

1047

8 9 10

24 hr periods Figure 2. Metabolic sialylation profile of an immunoadhesion molecule, GPI -1gG under high productivity conditions. A 10 I bioreactor culture of CHO cells expressing GPI-IgG was sampled every 24hr. The recombinant protein was isolated and sialic acid content determined by chemical analysis (+). Concomitantly, samples of the culture were incubated with 3H-ManNAc. The amount of metabolically incorporated sialic acid was determined on the recombinant protein as radioactive On day 3 of the culture sodium butyrate was added to be 12 mM to stimulate protein counts (0). production. Up to day 8, the viability of the culture was 80% or greater. Modified from [29].

[31]. Metabolically incorporated sialic acid showed an even more dramatic drop to near undetectable levels as the time in culture progressed. It is apparent from these results that the actions of butyrate on sialylation of recombinant proteins are complex and quite profound. Although a precise understanding of the butyrate effects is not as yet available, these experiments suggest that the overall pathway of sialylation is significantly altered or, that a different pathway is employed after induction. Sialyltransferase activity has been shown to be stimulated in CHO cells in the presence of butyrate [32, 331. Thus, some step(s) other than sialyltransferase, must be rate limiting which gives rise to the progressive decrease in sialic acid levels on the protein. Induction of an alternate biosynthetic pathway which does not utilize ManNAc as a sialic acid precursor and does not have the capacity to give complete sialylation under high productivity conditions may account for both the dramatic decline in metabolically incorporated sialic acid as well as the observed decrease in sialic acid content of the protein. Clearly, testing of this hypothesis will require that the mechanism for biosynthesis of sialic acid in CHO cells be carefully investigated. Only in this way can rational approaches for augmenting sialylation of recombinant proteins be designed. In summary, feeding studies with ManNAc in CHO cells and other cell types provide strong circumstantial data that it will be possible to improve sialylation of recombinant proteins by the metabolic engineering of one or more steps in the sialylation pathway (see also experiment described below, and Figure 4). Several key enzymes that may enhance sialylation or improve consistency of glycosylation when overexpressed in animals cells are listed in Table 1. Many of the genes for these

phosphoglucomutase glucose-6-phosphatase P-galactosyltransferase CMP-sialic acid hydroxylase sialidase

GlcNAc-2-epimerase or UDPGlcNAc-2-epimerase CMP-sialic acid transporter a-2,3 or 2,6 sialyltransferase UDP-glucose-4-epimerase UDP-glucose pyrophosphorylase

N-acetylgalactosaminyltransferase a-I ,3 fucosyltransferase P-N-acetylglucosaminidase

N-acetylglucosminyltransferaseI1 P-galactosyltransferase GlcNAc-2-epimerase N-acetylneuraminyl lyase CMP-sialic acid synthetase a-2,3-sialyltransferase

Insect Cells

a-1,3 fucosyltransferase p-1,2 xylosyltransferase

a-2,3-sialyltransferase

a-mannosidase I and I1 and/or N-acetylglucosaminidase I1 P-galactosyltransferase GlcNAc-2-epimerase N-acetylneuraminyl lyase CMP-sialic acid synthetase

Plants

a

Listed are all potential gene candidates for overexpression. However, it may only be necessary to overexpress individual genes or a selection of several genes, in combination, to enhance glycosylation or provide consistency of glycosylation under a variety of culture conditions. Many microbial genes for these reactions are available and they may be employed in either of these hosts.

Genes that need to be inhibited or deleted

Genes needed to supplement or enhance exisiting glycosylation machinery."

CHO cells

Table 1. Genetic Alterations Needed To Improve N-Glycosylation Of Recombinant Proteins Produced In CHO cells, Insect Cells And Plants.

g. 5

ff

G

$

z'i

Q

5'

3

n

9. s

P

0

s

s0-

2 $

%

0

-

64.2 Recent Deue1opment.r in Curhohydrate Biosynthesis

1049

UDP-Glucose pyrophosphorylase

UDP-Glc f t Glc-I-P 7-

Phosphoglucomutase

UTP

3

Glc-6-P

@

Glucose-6-phosphalase

UDP-Glucose-4-epimerase

UDP-Gal

/

Glc

Figure 3. Metabolic pathway for galactosylation of recombinant proteins in CHO cells. Overexpression of the glucose-4-epimerase, alone or in combination with several other enzymes along this pathway, may be helpful in enhancing galactosylation (and potentially sialylation) of recombinant proteins in this host.

proteins have been cloned and are readily available from microbial sources. Adding ManNAc as a medium component to enhance sialylation is not practical on a production scale because of cost concerns.

64.2.2 Optimizing Galactosylation of Recombinant Proteins by Metabolically Engineering Galactose Biosynthesis As with sialic acid levels, galactose content in many animal tissues is carefully regulated. Three critical enzymes, galactokinase [ 341, galactose- 1-phosphate uridyltransferase [ 351, and UDP-galactose-4-epimerase [36] catalyze reactions for conversion of galactose to glucose, thereby, reducing tissue or cellular levels of galactose. Deficiencies of either of the enzymes results in the accumulation of toxic levels of galactose and galactitol. In humans a number of inherited disorders (e.g. galactosemia) have been described that result from enzyme defects in the galactose pathway [37, 381. It is not surprising therefore that CHO cells carry with them, as genetic baggage from the animal from which they were derived, this series of detoxifying enzymes. In cultured cells the UDP galactose-4-epimerase (see Figure 3) serves as the major metabolic source of galactose using UDP-glucose as substrate. However, the reaction is reversible, the equilibrium favors the formation of glucose (Keq= 3.5), which is in keeping with the enzyme’s intended functional role of maintaining galactose at physiologically safe levels. For this reason, galactose available for recombinant glycoprotein synthesis is dependent on the conversion rate of UDP glucose. ‘Undergalactosylation’ of a recombinant, humanized anti-CD 20-IgG, has been observed in high productivity cultures when exogenous galactose is not added to the medium. This resulted in an enrichment of glycans on the protein that terminate in GlcNAc residues. Unexpectedly, the bioactivity of the mole-

1050

64 Metabolic Engineering Glycosylution

cule in complement dependent killing assays was reduced proportionately to the number of exposed GlcNAc residues [39]. These results are significant not only because they demonstrate that galactose available for protein biosynthesis is limiting in CHO cell cultures, but also because they illustrate the important functional role of galactose on bioactivity of the antibody molecule. In addition, correlations between biological activity and exposed GlcNAc residues on the heavy chain glycans of humanized IgGl antibody directed against CD52 have also been noted [40]. In related studies, we examined the effect of including galactose in the culture medium of CHO cells producing the immunoadhesion molecule GPl-IgG (Figure 4) [41]. The negatively charged glycans in the variable region of the immunoglobin were released with N-glycanase treatment and their masses determined by matrixassisted-laser desorption ionization-time of flight mass spectrometry, MALDITOF. The oligosaccharide structures on the control protein were extraordinarily heterogeneous. However, when an external source of galactose was available in the culture, the heterogeneity decreased substantially (Figure 4, second panel). In some cases, complete elimination of truncated structures was observed. For example, the biantennary oligosaccharide in the control protein with one antenna terminating in sialic acid and the other antenna terminating in GlcNAc ( m / z = 1,914) was almost completely eliminated in the presence of added galactose. When ManNAc along with galactose was included in the culture medium, further reduction in heterogeneity was observed with an additional decrease in incompletely capped structures ( m / z = 2,076 and 2,732; Figure 4 bottom panel). Overall, the number of antennae terminating in sialic acid increased from about 75”/0 in the control to about 84% in the culture containing both ManNAc and galactose. The results of this experiment strongly indicate the value of increasing the intracellular levels of these two carbohydrates, either by providing them as an external source or by overexpressing the biosynthetic machinery for their production. Unlike ManNAc, galactose is a relatively inexpensive carbohydrate, and its use as a media component at production scale is feasible, although significant additional indirect costs are incurred because each additional media component complicates the overall production process, generating an extensive ‘paper trail’ of documentation of purity and production history. In addition, there is the additional risk that any added reagent may harbor an adventitious virus which could con-

r

Figure 4. Etkct of adding galactose and ManNAc in CHO cell cultures on the glycans of recombinant immunoglobin, GP1-IgG. Shown are the masses of N-linked, negatively charged glycans released from recombinant GPI -1gG. Two liter bioreactor cultures of CHO cells expressing the protein were grown under different conditions. Upper Panel: CHO cell cultures were grown in serum free medium in presence of glucose (2 g/l). Second panel: Cultures were grown in the presence of glucose and added galactose (2 g/l). Bottom Panel: Cultures were grown in the presence of glucose, galactose and ManNAc (2 g/l). Recombinant protein was isolated after 8 days of culture, purified and treated with N-glycanase. The released structures were examined in the negative mode with MALDI-TOF mass spectrometry. Vertical arrows identify structures that decreased in the presence of added sugar. (+ = sialic acid, A = galactose, 0 = GlcNAc, 0 = Man, A = Fucose) [411.

64.2 Recent Drveloprnents in Carbohydrate Biosynthesis

I05 I

Control

h

n^h

+ Galactose al

U J K 0 Q UJ

a,

[r

5

c

0

a

c

a,

n

+ Galactose & + ManNAc

2000

2500

3000

Mass (mk)

3500

4000

1052

64 Metabolic Engineering Glycosylation

taminate cultures [42]. Given these considerations, it seems cogent to construct a CHO cell line with enhanced metabolic capabilities to produce galactose from glucose efficiently and eliminate the need for external galactose. Overexpression of selected enzymes involved in galactose biosynthesis, either independently or collectively as an entire metabolic pathway, may be a route to increased levels of galactosylation in CHO cells. These include the UDP-glucose-4epimerase [ 361, the phosphoglucomutase [43], and the UDP-glucose pyrophosphorylase [44] and the UDP-galactose transferase [45]. All of the genes for these enzymes have been isolated from either mammalian or microbial sources. In summary, incomplete sialylation of proteins may be the result of inadequate galactosylation and not just from deficiencies of sialic acid biosynthesis. Ultimately, engineering a ‘metabolic net’ [7] of several enzymes to elevate both sialic acid and galactose levels may be necessary to prepare a universal production host cell capable of giving complete extension of the oligosaccharide side chains for a variety of recombinant glycoproteins under high productivity conditions with robust protein synthesis.

64.2.3 Mannose Biosynthesis and Mannosylation of Recombinant Proteins Until recently, it has generally been assumed that mannose for glycoprotein synthesis was indirectly derived metabolically from glucose. Surprisingly, a mannose specific uptake mechanism has recently been identified in many mammalian cells [47]. The mannose transporter has a sufficiently low Km (-30-70 pM) to permit internalization of mannose even in the presence of high concentrations of glucose. And, externally derived mannose serves as the major source of the carbohydrate for glycoprotein synthesis (471. In humans, impaired mannose uptake and biosynthesis leads to the lethal disease, carbohydrate-deficient glycoprotein syndrome type I [48]. One of the major biochemical features of the disease is reduced pool sizes of dolichol-oligomannosyl conjugates and, more importantly, aberrant glycosylation of many serum and membrane proteins. In some cases, entire N-linked chains are not added to the molecule. The glycoproteins in this disease serve as examples of the types of glycosylation structures that result when the amount of mannose is limiting. These observations could have significant implications for the production of recombinant proteins in cultured cells, since mannose is not normally added as a medium component. It would be very interesting to determine the effect of added mannose in animal cell cultures on the glycans of recombinant proteins. The resulting structures may be more like those of the native protein which utilizes mannose at physiological concentrations taken up from the blood, rather than relying on the limited supply of mannose derived metabolically from glucose. Such studies may identify additional avenues for metabolic engineering to insure sufficient levels of mannose for protein biosynthesis and further enhance glycoprotein quality and consistency of production.

64.3 Glycosylutioiz Engineering Alternute Expression Hosts

1053

64.3 Glycosylation Engineering Alternate Expression Hosts For Recombinant Protein Therapeutic Production Many of the medical conditions that were initially targeted for treatment with recombinant protein therapeutics required low dosages of the pharmaceutical. In contrast, current immunotherapies that utilize recombinant chimeric anitbodies for treatment of life-threatening diseases such as breast cancer or non-Hodgkins lymphoma require several hundred milligrams of protein for each dose, with several administrations to achieve maximal efficacy. At the present, recombinant pharmaceuticals of this type far outnumber all other recombinant therapeutics being developed or in clinical trials. High productivity processes are needed to produce, in a cost effective manner, the large amounts of recombinant material that will be necessary to meet market demands for these and future molecules. Although animal cell expression hosts continue to be improved upon, other expression systems which have the potential for even greater productivity would be invaluable. Plants and insect cells are alternate hosts that would be especially attractive if the glycosylation machinery were altered to give glycans that would be suitable for application as human therapeutics. The following section is a summary of the current efforts underway to modify glycosylation processes in these hosts as well as identification of specific targets for intervention that will be required in order to give proteins with glycans compatible with these applications.

64.3.1 Engineering Glycosylation of Recombinant Proteins Expressed in Baculovirus-Insect Cells Recombinant baculoviruses and their lepidopteran insect cell hosts have become one of the most widely used expression systems for the production of recombinant proteins [49, 501. Manipulation of the system is straightforward, and high yields of recombinant material can be readily obtained in a matter of a few weeks compared with the several months that are required with stable animal cell expression systems. Generally, the insect-produced proteins are intended for research applications rather than as pharmaceuticals. The glycans on insect proteins differ significantly from those found on animal cell proteins, and this has limited the development of an attractive expression system for production of human therapeutics. Efforts to metabolically engineer glycosylation pathways in the baculovirusinsect cell system have been recently summarized [51-53]. Insect cells lack many of the enzymes required to give complete extension of N-linked oligosaccharides. In addition, insect proteins contain carbohydrate residues not found on mammalian glycoproteins and these could be potentially immunogenic in humans. Many of the genes needed to supplement insect cell glycosylation along with deleterious genes that may need to be deleted or inhibited are summarized in Table 1.

1054

64 Metabolic Engineering Glycosylation

The initial steps in the N-glycosylation pathway of proteins in insect cells are similar to those of mammalian cells [ 54, 551. The dolichol-linked -Glc3-Man9GlcNAc2 glycan moiety is transferred, en bloc, to the nascent polypeptide. Following ‘trimming’ reactions by the a-glucosidase I and I1 and a-mannosidase I, further extension of the resulting oligosaccharide product takes place by the activities of N-acetylglucosaminyltransferase I and N-acetyglucosaminyltransferase 11. The presence of these enzymes has been inferred from structural analysis of the glycans on insect proteins [55]. Biochemical assays for these enzymes have confirmed their presence in insects; however, the activities are extremely low compared with those found in animal cells [56, 571. Thus, extension of the N-linked oligosaccharides is not an efficient process with this level of enzyme expression. In order to supplement the existing Golgi transferase, Wagner and co-workers coexpressed the human N-acetylglucosaminlytransferase 1 along with the fowl plague virus hemagglutinin in Spodoptera frugiperda (Sf9) cells, which gave a recombinant product with much greater levels of the extended glycans containing terminal GlcNAc residues [58]. Similar to the animal cell pathway, fucose addition takes place after the action of N-acetylglucosaminyltransferase I, along with the continued processing of the mannose residues by a-mannosidase I1 (Figure 5). Both a-1,6 and a-1,3 linked fucose residues have been identified on insect proteins [54, 551. Genes needed to supplement glycosylation of recombinant proteins in insect cells

Glycosylation in insect cells deviates substantially from the animal cell pathway with the absence or extremely low levels of galactosyltransferase and sialyltransferase. Addition of these genes would be required to complete oligosaccharide biosynthesis. Stable insect cells expressing the galactosyltransferase have recently been made [59]. Although galactose has been identified on the resulting insect proteins in these cells, detailed structural characterization of the glycan structures remains to be carried out to determine the efficiency of transfer of galactose onto recombinant proteins. Insect cells lack the ability to synthesize sialic acid. At least three additional genes would be needed to provide CMP-sialic acid substrate for the sialyltransferase. Overexpression of the GlcNAc-2 epimerase is an attractive enzyme to provide a source of ManNAc from GlcNAc [28]. The resulting ManNAc can be condensed with phosphenolpyruvate with the sialic acid synthetase [26, 271, or with pyruvate using the N-acetylneuraminyl lyase from animal [28] or microbial sources [60]. The latter combination has recently been successfully employed to produce large quantities of sialic acid in Escherichia coli grown in the presence of pyruvate [61]. Finally, the CMP-sialic acid synthetase would be needed to provide the activated sugar nucleotide [62]. Deleterious genes may need to be deleted or inhibited to enhance recombinant glycoprotein biosynthesis in insect cells

Deletion or inhibition of several insect genes may be necessary to insure that potentially immunogenic glycans are not incorporated into the recombinant thera-

64.3 Glycosylution Engineering Alternate Expression Hosts

I

I

1055

al,Z - gluwsidase I a1,Z- glucosidase / I

MangGl~NAcfiSN a-mannosidase I a-mannosidase /

N-Glycosylation pathway similar with animal systems

I

Man5GlcNAc2ASN

I

N-acefylgluwsaminyitransferase I

'

M-M-M\ M-GIcNAc~ASN GlcNAc -M a-mannosidase I / 1,6or 1,3*fumsy/transferase

M\ N-acetylglucosaminidase'

/

GlcNAc - M

I Fuc 1,6or 1,3'

/

M\ ,WGICNAC~FUC-ASN

1

N-a~lgl~samlnyltransteraae /I

M Metabolicpathway needed to complete N-glycan synthesis

1

M-GICNACGICNACASN

M-GlCNACpFucASN GlcNAc - M'

GlCNAc M,

GlcNAc 2-11neraae

ManNAc

+

pyruvate neuram/ny//yaae

4

Neu5Ac

+

CTP

-

NeuBAc-Gal-GlcNAc M, M-GIcNAc~FucASN

NeuBAc-Gal-GlcNAc- M'

aiallc acld aynthe tam

CMP-Neu5Ac galactesyttransfuase

2,3 or 2,6 slalylfranafuase

Figure 5. Glycosylation pathway in insect cells and potential sites for metabolic engineering. Enzymes shown in bold type are those genes needed to be overexpressed to give recombinant proteins with more human-like glycans. The sialylation pathway is needed to supply galactose and sialic acid. These genes may be of microbial origin. Enzymes with an asterisk are those that may need to be inhibited or deleted.

peutic or to maximize complete oligosaccharide extension, Table 1. The insect a1-3 fucosyltransferase is a likely candidate for elimination or inhibition since the resulting fucosylated structures on insect glycoproteins have been found to be allergenic [63]. Similarly, unlike animal cells, insect cells contain N-acetylgalactosyl transferase which adds GalNAc residues onto GlcNAc residues of N-linked glycans

1056

64 Metabolic Engineering Glycosylation Man.

GaINAcp 1,&GlcNAc-Man I

a1.3

/

MwI-GIcNAc-GIcNAc 1 a1.3

Fuc

Fuc

Figure 6. An N-linked glycan structure found on honeybee venom protein [64]. The GalNAc residue attached on N-linked oligosaccharides is not found in animal systems. The corresponding transferase gene may need to be deleted to avoid this potentially immunogenic residue on recombinant therapeutics expressed in insect cell hosts.

(Figure 6) [64]. These structures may also be recognized as foreign glycans in animals. Also unique to insect glycoprotein biosynthesis is the presence of a membrane bound N-acetylglucosamindase [65]. This enzyme has been suspected to cleave the GlcNAc residues from the 1-3 branch of nascent oligosaccharide chains, giving rise to truncated hybrid structures and prohibiting further glycan extension (see Figure 5). Elimination of this enzyme may be needed to allow for more complete and efficient oligosaccharide biosynthesis.

64.3.2 Engineering Glycosylation of Recombinant Proteins Expressed in Plants Transgenic plants capable of producing recombinant proteins were reported as early as 1983 [66]. It has become increasingly apparent that plants offer significant advantages for recombinant protein production compared with animal or human cell lines. The single most important advantage is the cost of plant production. Extraordinary large-scale recombinant protein production (e.g. metric tons) is possible at minimal cost and with minimal investment in manufacturing facilities. For example, the cost for producing recombinant protein in soybeans is estimated to be about $4.95 per kg of crude recombinant material [67]. This is at least 10-50-fold less than protein produced in bacterial expression systems and several hundred-fold less than protein derived from animal cell culture. Thus, plants are ideal production sources for recombinant therapeutic proteins such as antibodies or other proteins that have high dosage requirements [68-701. However, one of the major concerns with plant-derived recombinant glycoproteins are the glycan structures on the therapeutic molecule which are significantly different than those on animal glycoproteins. The biosynthetic steps of N-glycosylation of plant proteins has recently been reviewed [71]. Similar to what has been observed with insects, the initial events along the glycosylation pathway in plants are conserved with those in animals, as shown in Figure 7. Unlike insect glycosylation, plant glycan processing is more complete and some plant proteins have extended oligosaccharide chains terminating with GlcNAc residues, and in some cases (although infrequently) with galactose. Glycosylation of plant proteins diverges from the animal pathway after the action of N-acetylglucosaminyltransferase 11, by the addition of xylose and fucose residues (Figure 6). The xylose unit is attached in a 8-1,2 linkage to the interior mannose residue of biantennary oligosaccharides to give bisected structures. Fucose is found

64.3 Glycosylation Engineering Alternate Expression Hosts

1057

Endoplasmic reticulum and early Golgi processing of N-glycans is similar to animals and insects

N-acetylgluwsaminy/trmsferase I

'

M-M-M\

Hybrid-typeN-glycans

M-GIcNAc~SN /

a-mnnddase I/ M\

N-aceiylglmsaminylfransferase /I

M-GIcNAcGIcNAcASN

GlCNAc - M /I

XYl 8-1.2'

I

GIcNAc - M\

FUCa-1,3'

M-GIcNAc~ASN GlcNAc - M' p-xylosyItmnsferase* a-fucosyltransferase' GlcNAc - M, M-GIcNAc~ASN I GlcNAc-M') Xylp1.2' FUCa-1.3'

Genes to synthesize and transfer sialic acid must be added to give sialylated glycans.

\

\ -

Gal- GlCNAC M\ Complex-typeN-glycans

M-GlCNAcfiSN Gal- GlcNAc - M' I I xyl p-1,2' FUCa-1.3"

Figure 7. Glycosylation pathway in plants and potential sites for metabolic engineering. Enzymes in bold type are those that may need to be overexpressed. Enzymes with an asterisk are those that may give immungenic structures in animals and may need to be inhibited or deleted. The sialic acid pathway may also be needed, see Figure 5.

in w-1,3 linkages at the initial core GlcNAc and in a-1,4 linkages in the outer chain GlcNAc residues. Sialic acid is not present in plant proteins, since the biochemical machinery to synthesize this critical constituent of mammalian glycoproteins is absent. The glycans on a recombinant IgG antibody, eg. 'plantibody', produced in transgenic tobacco plants have recently been compared with the carbohydrates on the complementary IgGl molecule, (Guy's13 antibody) produced in a murine monoclonal cell line (Figure 8) [72] Four major biantennary glycan structures were

1058

64 Metabolic Engineering Glycosylation Man-GlcNAc-GlcNAc

GlcNAcp 1,2

XY I

Fuc

Man\ Man’

Manal,6 \

Man, Man-GlcNAc-GlcNAc /

Man

Figure 8. The major N-linked glycans found in recombinant antibody, IgG, expressed in tobacco plants. Eight different glycan structures were identified on the molecule [72]. The more highly processed structures contained xylose and fucose, which are known to be immunogenic. Over 40% of the glycans were high mannose-type.

identified on the Fc portion of the mouse Mab which were typical of the IgG structures from many animal species [73, 741. Nearly 60% of the oligosaccharide chains terminated with either one or two galactose units. All of the structures contained fucose in 1,6 linkages to the internal core GlcNAc. As with many other animal IgG molecules, oligosaccharides with sialic acid were a minor component of the Fc glycans. In contrast, the recombinant plant antibody contained a heterogeneous array of eight distinct oligosaccharides, none containing sialic acid. About 40% of the structures were incompletely processed, ‘high mannose’-type molecules, containing 5-8 mannose residues. The remaining molecules terminated in either one or two GlcNAc residues, but no structures terminating with galactose were detected. These more highly-processed sidechains also contained xylose and a- 1,3 fucose residues (see Figure 8). Genetic addition and supplementation needed to improve plant recombinant protein glycosylation

Analysis of the carbohydrate components of the plantibody immunoglobin molecule has provided invaluable insight into the types of genetic modifications of plant glycosylation machinery that may be needed to give more desirable or ‘human like’ glycans on other recombinant therapeutic proteins or antibodies. The presence of the high mannose type structures had no detectable adverse effect on epitope recognition or binding by the recombinant form of the Guy’s 13 antibody. However, these types of structures could have significant impact on the efficacy of other recombinant therapeutic molecules, since they would accelerate clearance from the blood stream by receptor mediated processes [75].The Guy’s 13 antibody is intended for oral administration for the treatment of Streptococcus mutans promoted dental caries so that plasma clearance issues are not of concern with this molecule [72]. However, with other proteins the preponderance of mannan structures may be disadvantageous, limiting bioavailabilty and reducing serum residence time of the therapeutic. More extensive oligosaccharide processing of recombinant proteins may be possible by devising a transgenic plant to overexpress N-acetylglucosaminidase I

64.4 Summary

1059

in combination with a-mannoside 11. Similarly, overexpression of P-galactosyl transferase, at least in the tobacco plant system, may insure more efficient extension of the oligosaccharide chain with terminal galactose. In specific cases where sialylation of the recombinant protein is critical, the construction of an entire metabolic pathway for sialic acid biosynthesis, similar to that we have suggested for the insect cell system, may be necessary (Figure 5). As an alternate to the transgenic approach for modifying glycosylation, success at identifying glycosylation mutants of Arabidopsis thaliana has been achieved by mutagenesis followed by screening for the appropriate glyco-phenotype [76]. Although this strategy has proven feasible, it is limited because it requires the availability of the appropriate screening probes in order to effectively identify the desired genetic mutation, which can be expected to be a low frequency event. The transgenic approach of overexpressing key biosynthetic enzymes is more direct and may have greater flexibility, especially if enhancements of multiple genetic loci are necessary. Inhibition or deletion of plant glycosylation genes

Endogenous plant glycoproteins as well as recombinant proteins produced in this host contain xylose and u1,3 fucose residues. Xylose is foreign to mammalian glycoproteins and fucose, although a normal constituent of animal glycoproteins, is not normally found with this linkage. The presence of these sugars on recombinant therapeutics could be of concern since they may elicit an immunogenic reaction with parenteral administration. Antibodies specifically directed against these sugar residues on plant glycoproteins have been raised in rabbits when compounded with a stimulating adjuvant [77]. Moreover, evidence that protein-linked xylose and fucose are IgE epitopes in humans and that they are primarily responsible for the allergenicity of plant proteins has been presented [78]. Given the anticipated risks of therapeutics containing these carbohydrates, it is prudent to investigate the potential for inhibition or deletion of the glycosyl transferase genes responsible for the addition of these sugars to glycoproteins. Gene targeting and disruption by homologous recombination has been used extensively with embryonic stem cells in animals and with some success in somatic cells [79-811. At present, gene knock-outs in plants is a field in its infancy, although some attempts have been made [82, 831. However, inhibition of gene expression with constitutively expressed antisense has been applied successfully in plants, controlling expression of a variety of enzymes [84, 851. Application of either of these techniques for controlling the levels of the xylosyl transferase or the fucosyl transferase will require some knowledge of the DNA sequences of the respective genes. As yet, neither of these proteins has been purified from plants.

64.4 Summary The mandate of biotechnology to develop cost effective, robust recombinant glycoprotein production processes with consistent, complete, and physiologically tol-

1060

64 Metabolic Engineering Glycosylution

erated glycosylation on therapeutic molecules presents the glycobiologist with nearHerculean challenges for the future. This article summarizes some of the recent developments in glycoprotein carbohydrate metabolism and identifies specific genes in these glycosylation pathways that are potential targets for augmentation in order to enhance recombinant protein productivity in currently employed animal cell expression hosts. Similarly, with more prolific production hosts such as insect cells or plants, we have detailed the extensive metabolic engineering of the glycosylation pathways that may be necessary to provide the desired glycan structures on the therapeutic. The potential cost benefits offered by these new expression systems is clear justification for pursuing this challenging effort. In this review we have focused on glycosylation in these three expression systems because they are the least complex and most likely to be readily manipulated. Similar metabolic engineering can be considered for yeast and transgenic animals which also have potential as high level expression hosts. In addition, we have defined only those enzyme systems that relate to biosynthesis of simple, biantennary N-linked glycans. Similar approaches for metabolic engineering 0-linked glycans as well as more highly branched Nlinked structures are also important modifications that should be considered. Although achieving these goals will not be trivial, the resulting effort will undoubtedly be of great benefit, not only for biotechnology, but also for furthering our understanding of fundamental issues in glycobiology. Acknowledgments

I acknowledge Lydia Santell, Jeff Ferrari, Martin Gawlitzek, and Thomas Ryll for their invaluable contributions to this work. Also, I thank Mike Mulkerrin, Shantha Raju, Don Jarvis, and Loic Faye for making available preprints of their work. I acknowledge the generous support of Genentech, Inc. for a portion of this research. Finally, I thank Dr. Willie Vann and D. Amanda Warner for critically reading this manuscript. Note Added in Proof:

Recently, over expression of both the sialyltransferase and galactosyl transferase in CHO cells resulted in a near-complete extension of the oligosaccharides on recombinant proteins expressed in this host. This supports the notion that the glycosylation processes are rate limiting in this expression system under high productivity conditions. (Weikert, S. et al., Engineering Chinese hamster overy cells to maximize sialic acid content of recombinant glycoproteins. Nat. Biotech. 1999, 17, 11161121). References 1. Cumming, D. A. Clycosylation of recombinant protein therapeutics: control and functional implications. Glycobiology, 1991, 1 , 115-1 30.

References

1061

2. Jenkins, N. and Curling, E.M.A. Glycosylation of recombinant proteins: problems and prospects. Enzyme Microh. Technol., 1994, 16, 354-364. 3. Jenkins, N., Parekh, R. B., James, D. C. Getting the glycosylation right: Implications for the biotechnology industry. Nature Biotech., 1996, 14, 975-961. 4. Goochee, C. F. and Monica, T. Environmental effects on protein glycosylation. Biotechnology, 1990, 8, 421-427. 5. Bailey, J. E. Toward a science of metabolic engineering. Science, 1991, 252, 1668-1675. 6. Jacobsen, J. R. and Khosla, C. New directions in metabolic engineering. Curr. @in. Chem. Bid., 1998, 2, 133-137. 7. Stephanopoulos, G. Metabolic engineering. Curr. Biol., 1994, 5, 196-200. 8. Cameron, D. C. and Tong, U. Cellular and metabolic engineering. Appl. Biochem. Biotechnol., 1993,38, 105-140. 9. Kundig, W., Ghosh, S., Roseman, S. The sialic acids: N-acyl-D-mannosamine kinase from rat liver. J. Biol. Chem., 1966, 241, 5619-5626. 10. Watson, D. R., Jordian, G. W., and Roseman, S. The sialic acids: Sialic acid 9-phosphate synthetase. J. Biol. Chem., 1966, 241, 5627--5636. 11. Kean, E. L. and Roseman, S. The sialic acids: Purification and properties of cytidine 5’monophosphosialic acid synthetase. J. Bid. Chem., 1966, 241, 5643-5650. 12. Warren, L. and Felsenfeld, H. The biosynthesis of sialic acids. J. Biol. Chem., 1962, 237, 14211431. 13. Kornfeld, S., Kornfeld, R., Neufeld, E. F., O’Brien, P. J. The feedback control of sugar nucleotide biosynthesis in liver. Proc. Natl Acad. Sci. USA, 1964, 52, 371-379. 14. Hagemeier, C., Vischer, P., and Buddecke, E. Studies on the biosynthesis of N-acetylneuraminic acid in cultured arterial wall cells. Biochem. Soc. Trans., 1986, 14, 1173-1 174. 15. Thomas, G. H., Reynolds, L. W., Miller, C. S., Overproduction of N-acetylneuraminic acid by sialuria fibroblasts. Pediatr. Res, 1985, 19, 451-455. 16. Gu, X., Wang, D. I. C. Improvement of interferon-y sialylation in Chinese hamster ovary cell culture by feeding of N-acetylmannosamine. Biotech. and Bioeng., 1998, 58, 642-647. 17. Eckhardt, Muhlenhoff, M., Bethe, A,, Gerardy-Schahn, R. Expression cloning of the Golgi CMP-sialic acid transporter. Proc. Nut1 Acad. Sci, USA, 1996, 93, 7572-7576. 18. Ishida, N., Ito, M., Yoshioki, S., Sun-Wada, G. H., Kawakita, M. Functional expression of human Golgi CMP-sialic acid transporter in the Golgi complex of a transporter-deficient Chinese hamster ovary cell mutant. J. Biochem. (Tokyo), 1998, 124, 171-178. 19. Eckhardt, M., Gerardy-Schahn, R. Molecular cloning of the hamster CMP-sialic acid transporter. Eur. J. Biochem., 1997, 248, 1878-192. 20. Minch, S. L., Kallio, P. T., Bailey, J. E., Tissue plasminogen activator coexpressed in Chinese hamster ovary cells with alpha (2,6)-sialyltransferase contains NeuAc alpha (2,6)Gal beta (1,4) Glc-N-AcR linkages. Biotechnol. Prog., 1995, 11, 348-35 1. 21. Grabenhorst, E., Hoffman, A,, Nimtz, A. M., Zettlmeissl, G., Conradt H. S. Construction of stable BHK-2 1 cell co-expressing human secretory glycoprotein and human Gal(P 1-4)GlcNAca2,6 sialyltransferase.u2,6 NeuAc is preferentially attached to the Gal(P-l-4)GlcNAc(P-l,2)Man(a-l,3)-branch of diantennary oligosdcchdrides from secreted recombinant p-trace. Eur. J. Biochem., 1995,232, 718-125. 22. Jenkins, N., Buckberry L., Marc, A,, Monaco, L. Genetic engineering of alpha 2,6sialyltransferase in recombinant CHO cells. Biochem. Soc. Trans., 1998, 26, S115. 23. Zhang, X., Lok, S. H. and Kon, 0. L. Stable expression of human alpha-2,6-sialytransferase in Chinese hamster ovary cells; funtional consequences for human erythropoietin expression and bioactivity. Biochiem. Biophys. Acta, 1998, 1425, 441-452. 24. Van Rinsum, J., Van Dijk, W., Hooghwinkel, G. J. M., Ferwerda, W. Subcellular localization and tissue distribution of sialic acid precursor forming enzymes. Biochem. J. 1983,210,21-28. 25. Stasche, R., Hinderlich, S., Weise, C., Effertz, K., Lucka, L., Poormann, P., Reutter, W. A bifunctional enzyme catalyzes the first two steps in N-acetylneuraminic acid biosynthesis of rat liver. J. Biol. Chem., 1997, 272, 24319-24324. 26. Hinderlich, S., Stasche, R., Zeitler, R., Reutter, W. A bifunctional enzyme catalyzes the first two steps in N-acetyneuraminic acid biosynthesis of rat liver. J. Biol.Chem., 1997, 272, 2431324318.

1062

64 Metabolic Engineering Glycosylation

27. Maru, I., Ohta, Y., Murata, K., Tsukada, Y. Molecular cloning and identification of N-acyl-Dglucosamine-2-epimerase from porcine kidney as a renin binding protein. J. Biol. Chem. 1996, 271, 16294-16299. 28. Schauer, R. and Wember, M. Isolation and characterization of sialate lyase from pig kidney. Biol. Chem. Hoppe Seyler, 1996, 377, 293-239. 29. Santell, L., Ryll, T., Etcheverry, T., Santoris, M., Dutina, G., Wang, A,, Gunson, J., Warner, T. G. Aberrant metabolic sialylation of recombinant proteins expressed in Chinese hamster ovary cells in large-scale batch cultures, Biochem. Biophys. Rex Comm. 1999, 258, 132-137. 30. Diaz, S. and Varki, A. Metabolic labeling of sialic acids in tissue culture cell lines: Methods identify substituted and modified radioactive neuraminic acids. Anal. Biochem., 1985, 150,32-46. 31. Warner, T. G., Chang, J., Ferrari, J., Harris, R., McNerney, T., Bennett, G., Burnier, J., Sliwkowski, M.B. Isolation and properties of a soluble sialidase from the culture fluid of Chinese hamster ovary cells. Glycobiology, 1993, 3, 455-463. 32. Chotigeat, W., Watananpokasin, Y., Mahler, S., Gray, P.P. Role of environmental-conditions on the expression levels, glycoform pattern and levels of sialyltransferase for hFSH expression produced by recombinant CHO cells. Cytotechnology, 1994, 15, 217-221. 33. Gebert, C. A. and Gray, P.P. Expression of FSH in ChO cells.2. stimulation of hFSH expression levels by defined medium supplements. Cytotechnology, 1994, 14, 39-45. 34. Stambolina, D., Ai, Y., Sidjanin, D., Nesburn, K., Stahe, G., Rosenberg, M., Bergsma, D.J. Cloning of the galactokinase cDNA and identification of mutations in two families with cataracts. Nat. Genet., 1995, 10, 307-12. 35. Flach, J. E., Reichardt, J. K., Elsas, E. J. Sequence of a cDNA encoding human galactose-lphosphate uridyl transferase. Mol. Biol. Med., 1990, 7 , 365-369. 36. Daude, N., Gallaher, T. K., Zeschnigk, M., Starzinski-Powitz, A,, Petry, K. G., Haworth, I. S., Reichardt, J. K. Molecular cloning, characterization, and mapping of a full-length cDNA encoding human UDP-galactose 4’epimerase. Biochem. Mol. Med., 1995, 56, 1-7. 37. Segal, S. Galactosaemia today: the enigma and the challenge. J. Inherit. Metah. Dis., 1998, 21, 455- 471. 38. Wang, B. B., Xu, Y. K., Ng, W. G., Wong, L. J. Molecular and biochemical basis of galactosemia. Mol. Genet. Metab., 1998, 63, 263-269. 39. Mulkerrin, M.G., Raju, S., Gazzano-santoro, H., Gray, R., Parker, M., Bjork, N., Raffo, W., Borge, S., Chen, A. B., Ultsch, M., DeVos, A. M., Antibody mediated cell killing depends on the structure of the Fc oligosaccharide. Second symposium on analysis of well characterized biotechnology pharmaceuticals, Sun Francisco, 1998. 40. Boyd, P.N., Line, A.C., Patel, A.K. The effect of the removal of sialic acid, galactose and totae carbohydrate on the functional activity of campdth. In. Mol. Zmmunol. 1995, 32, 1311-1318. 41. Gawlitzek, M., Santell, L., Papac, D., Warner, T. G., unpublished obsevoations. 42. Garnick, R. L. Raw materials as a source of contamination in large-scale cell culture. Deu. Biol. Stand., 1998, 93, 21-29. 43. Lu, M. and Kleckner, M. Molecular cloning and characterization of the pgm gene encoding phosphoglucomutase of Escherichia coli. J. Bacteriol., 1994, 176, 5847-5851. 44. Hossain, S. A,, Tanizawa, K, Kazuta, Y., Fukui., T. Overproduction and characterization of recombinant UDP-glucose pyrophosphorylase from Escherichia coli K-12. J. Biochem. (Tokyo), 1994, 115, 965-912. 45. Narimatsu, H., Sinha, S., Brew, K., Okayama, H., Qasba, P.K. Cloning and sequencing of cDNA of bovine N-acetylglucosamine (beta 1-4) galactosyltransferase. Proc. Natl Acad. Sci. USA, 1986, 83, 4720-4724. 46. Panneerselvam, K and Freeze, H.H. Mannose enters mammalian cells using a specific transporter that is insensitive to glucose. J. Bid. Chem., 1996, 271, 9417-9421. 47. Alton, G., Hasilik, M., Niehues, R., Panneerselvam, K., Etchison, J. R., Fana, F., Freeze, H. H. Direct utilization of mannose for mammalian glycoprotein biosynthesis. Glycobiology, 1998, 8, 285-295. 48. Stibler, H., Holzbach, U., Kristiansson, B. Isoforms and levels of transferrin, antithrombin, alpha (1)-antitrypsin and thyroxine-binding. Scand. J. Clin. Lah. Invest., 1998, 58, 55-61. 49, O’Reilly, D. R., Miller, L. K., Luckow, V. A. Baculovirus Expression Vectors, Freeman Press, New York 1992.

References

1063

50. Miller, L. K. Baculoviruses as gene expression vectors. Annu. Reu. Microbiol., 1988, 42, 177199. 51. Jarvis, D. L., Kawar, Z . S., Hollister, J. R. Engineering N-glycosylation pathways in the baculovirus-insect cell system. Curr. Opin. Biotech., 1998, 9, 528-533. 52. Altmann, F. More than silk and honey-or, can insect cells serve in the production of therapeutic glycoproteins? Glycoconj. J., 1997, 14, 643-646. 53. Jarvis, D. L, and Finn, E. E. Modifying the insect cell N-glycosylation pathway with immediate early baculovirus expression vectors. Nut. Biotech., 1996, 14, 1288-1292. 54. Jarvis, D. L. and Finn, E. E. Biochemical analysis of the N-glycosylation pathway in baculovirus-infected lepidopteran insect cells. Virology, 1995,212, 500-5 1 1. 55. Altmann, F. N-glycosylation in insects revisited. Trends Glycosci. Glycotech., 1996, 40, 101114. 56. Altmann, G . Kornfeld, Dalik, T., Staudacher, E., Glossl, J. Processing of asparagine-linked oligosaccharides in insect cells. N-acetylglucosaminyltransferase I and I1 activities in cultured lepidopteran cells. Glycobiology, 1993, 3, 619-625. 57. Velardo, M.A., Bretthauer, R. K., Boutaud, A,, Reinhold, B., Rienhold, V.N., Castellino, F. J. The presence of UDP-N-acetylglucosaminyltransferase I activity in Spodoptera frugiperda cells (IPLB-SF-21AE) and its enhancement as a result of bculovirus infection. J. Biol. Chem., 1993, 268, 17902-17907. 58. Wagner, R., Liedtke, S., Kretzschmar, E., Geyer, H., Geyer, R., Klenk, H. D. Elongation of the N-glycans of fowl plauge virus hemagglutinin expressed in Spodoptera frugiperda (Sf9) cells by coexpression of human beta 1,2-N-acetylglucosaminyltransferaseI. Glycobiology, 1996, 6, 165- 175. 59. Hollister, J. R., Shaper, H., Jar& D. L., Stable expression of mammalian beta 1,4galactosyltransferase extends the N-glycosylation pathway in insect cells. Glycobiology, 1998, 8, 472-480. 60. Traving, C., Roggentin, P., Schauer, R., Cloning, sequencing and expression of the acylneuraminate lyase gene from Clostridium perfringens A99. Glycoconj. J., 1997, 14, 821-830. 61. Maru, I., Ohnishi, J., Ohta, Y., Tsukada, Y. Simple and large-scale production of Nacetylneuraminic acid from N-acetyl-D-glucsamine and pyruvate using N-aceyl-D-glucosamine 2-epimerase and N-acetylneuraminate lyase. Curbohydr. Res., 1998, 306, 575-578. 62. Munster, A. K., Echkhardt, M., Potvin, B., Muhlenhoff, M., Stanley, P., Gerardy-Schahn, R.. Mammalian cytidine 5’-monophosphate N-acetylenuraminic acid synthetase: a nuclear protein with evolutionarily conserved structural motifs. Proc. Nutl Acud. Sci. USA, 1998, 95, 91409145. 63. Tretter, V., Altmann, F., Kubelka, V., Marz, L., Becker, W.M. Fucose alpha 1,3-linked to the core region of glycoprotein N-glycans creates an important epitope for IgE from honeybee venom allergic individuals. Int. Arch. Allergy bnmunol., 1993, 102, 259-266. 64. Van Die, I., Van Tetering, A., Bakker. H., Van den Eijnden, D. H., Joziasse, D. H. Glycosylation in lepidopteran insect cells: identification of a beta l -4-acetylgalactosaminyltransferase involved in the synthesis of complex-type oligosaccharide chains. Glycobiology, 1996,6,157-164. 65. Altmann, F., Schwihla, H., Staudaucher, E., Glossl, J., Marz, L. Insect cells contain an unusual, membrane-bound beta-N-acetyglucosaminidase probably involved in the processing of protein N-glycans. J. Biol. Chern, 1995, 270, 17344-17349. 66. Fraley, R. T., Rogers, S. G., Horsch, R . B., Sanders, P. B., Flick, J. S., Adams, S. P., Brand, M.L., Fink, C. L., Fry, J. S., Galluppi, G. R., Goldberg, S. P., Hoffmann, N. L., Woo, S. C. Expression of bacterial genes in plant cells. Proc. Nutl Acud. Sci. USA, 1983, 80, 1090-1093. 67. Larrick, J. W., Yu, L., Chen, J., Jaiswal, S. Production of antibodies in transgenic plants. Res. Imrnunol., 1998,149, 603-608. 68. Conrad, U. and Fiedler, U. Expression of engineered antibodies in plant cells. Plunt Mol. Bid., 1994,26, 1023-1030. 69. Ma, J. K. and Hein, M. B. Plant antibodies for immunotherapy. Plunt Physiol. 1995, 109, 341346. 70. Evangelista, R. L., Kusnadi, A. R., Howard, J. A,, Nikolov, Z. L. Process and economic evaluation of the extraction and purification of recombinant beta-glucoronidase from transgenic corn. Biotechnol. Prog. 1998, 14, 607-614.

1064

64 Metabolic Engineering Glycosylution

71. Lerouge, R., Cabanes-Machetearu, M., Rayon, C., Fischetter-Laine, A., Gomord, V., Faye, L. N-glycoprotein biosynthesis in plants: recent developments and future trends. Plant Mol. Biol., 1998,38, 31-48. 72. Cabanes-Macheteau, M., Fitchette-Laine, A,, Loutelier-Bourhis, C., Lange, C., Vine, N. D., Ma, J. K. C., Lerouge, P., and Faye, L. N-glycosylation of a mouse IgG expressed in transgenic tobacco plants. Glycobioloyy, 1999, 9, 365-372. 73. Hamako, J., Matsui, T., Ozeki, Y., Mizuochi, T., Titani, K. Comparative studies of asparagine-linked sugar chains of immunoglobulin G from eleven mammalian species. Comp. Biochem. Physiol., 1993, 106B, 949-954. 74. Raju, T. S., Briggs, J., Borge, S., Wilks, M., Lerner, L. Species specific glycosylation of IgG. Glycobioloyy, 1998, 8, A104. 75. Ashwell, G . and Morell, A.G. The role of surface carbohydrates in the hepatic recognition and transport of circulating glycoproteins. Adu. Enzymology, 1974; 41, 99- 128. 76. Von Schaewen, A,, Sturm, A,, O'Neill, J., Chrispeels, M. J. Isolation of a mutant Arabidopsis plant that lacks N-acetyl glucosaminyl transferase I and is unable to synthesize Golgi-modified complex N-linked glycans. Plant Physiol., 1993, 102, 1109 -1 118. 77. Faye, L., Gomord, V., Fitchetter-Laine, A. C., Chrispeels, M. J. Affinity purification of antibodies specific for Asn-linked glycans containing alpha 1,3 fucose or beta 1,2 xylose. Anal. Biochem., 1993,209, 104-108. 78. Garcia-Casado, G., Sanchez-Monge, R., Chrispeels, M. J., Armentia, A,, Salcedo, G., Gomez, L. Role of complex asapargine-linked glycans in the allergenicity of plant glycoproteins. Glycobiology, 1996, 6 , 471-477. 79. Accili, D. and Taylor, S. I. Targeted inactivation of the insulin receptor gene in mouse 3T3-Ll fibroblasts via homologous recombination. Proc. Nut1 Acud. Sci USA, 1991, 88, 4708-4712. 80. Brown, J. P., Wei, W., Sedivy, J. M. Bypass of senescence after disruption of p21C'P'/WAF'gene in normal diploid human fibroblasts. Science. 1997, 277, 831-834. 81. Zheng, H. and Wilson, J. H. Gene targeting in normal and amplified cell lines. Nature, 1990, 344, 170-173. 82. Halfter, U., Morris, P. C.,Willmitzer, L. Gene targeting in Arabidopsis thaliana. Mol. Gen. Genet., 1992,23, 186-193. 83. Thykjaer, T., Finnemann, J., Schauser, L., Christensen, L., Poulsen, C., Stougaard, J. Gene targeting approaches using positive-negative selection and large flanking regions. Plant Mol. Biol., 1997, 35, 523-530. 84. Schuch, W. Using antisense RNA to study gene function. Symp. Soc. Exp. Biol., 1991,45, 117127. 85. Hamilton, A. J., Fray, R. G., Grierson, D. Sense and antisense inactivation of fruit ripening genes in tomato. Curr. Top. Microbial Immunol., 1995, 197, 77-89.

Carbohydrates in Chemistry and Biology Beat Ernst, Gerald W. Hart, Pierre Sinay copyrightQ WILEY-VCH Verlag GmbH, 2000

Index Roman figures attached to the page numbers refer to Part I (vols. 1 and 2) and Part I1 (vols. 3 and 4) respectively: eg. 111478 refers to page 478 in Part 11: Biology of Saccharides (volume 3 or 4).

ABAKAN IU412 ABO blood groups 111314 Acanthamoeba IU878 acceptor analog IU300, IV304 f acceptor mapping IV300, I11308 accumulation in Tay-Sachs and Sandhoff disease IU951

2-acetamido-2-deoxy - fi-D-mannopyranosides U33.5 acetimidyl-2-deoxy-hexopyranoses U394 acetobacter IV794 - xylinum IU793 - - cellulose synthase genes I11793 N-acetyl-4-0-acetylneuraminicacid I11235 N-acetyl-9-0-acetylneurarninicacid I11228 0-acetylation IU323 Di-N-acetylchitobiase 111475, IU479 ff - evolutionary deficiency IU954 acety1CoA:a-glucosaminidase acetyltransferase - in mucopolysaccharidoses I11949 acetyl-coenzyme A:sialate-4-0-acetyltransferase I11234 acetyl-coenzyme A:sialate-7(9)-0acetyltransferase IU234 3,4-di-O-acetyl-L-fucal I1374 - NIS-promoted glycosylation I1374 - ciclamicin 0 I1376 - synthesis V376 N-acetylgalactosamine (Tn-antigen) U274, I/287 N-acetylgalactosamine-6-sulfatase - in mucopolysaccharidoses IVY49 N-acetyl-a-D-galactosaminidase - in lysosomal disorder 1119.54 a-N-acetylgalactosaminidase IU.501 P4-acetylgalactosaminyltransferase I11265 P4-N-acetylgalactosaminyltransferase II/263, IU266 ff - family V61.5

N-acetylgalactosaminyltransferase 111334, IU85.5 9-0-acetyl GD3 I11238 3,4,6-tri-O-acetyl-D-glucal I1369 f - addition of alcohol U368 - chlorine addition I1370 N-acetylglucosamine 111717 -bisecting IV853 - 4-sulfatase - - in mucopolysaccharidoses IVY49 - 6-sulfatase - - in rnucopolysaccharides I11949 N-acetylglucosamine, 0-linked 11279, I11652 ff, IV661 - dynamic nature 111653 f - galactosyltransferase IU652 - hexosarnine biosynthetic pathway I11657 - insulin resistance I11657 - 0-GlcNAcase IU6.53 - 0-GlcNAc transferase IU6.53 - protein stability IU655 f - protein translation W6.56 - RNA polymerase I1 11165.5, I11657 - transcriptional regulation 111657 - transcription factors 111652, IU6.55 f - type I1 diabetes II/657 N-acetylglucosaminidase 111904 N-acetyl-P-D-glucosaminidase - lysosomal 11147.5 f - HexA and HexB deficient mice 11/948,11/9.51 ff - in lysosomal disorder 111954 - mucopolysaccharidoses in null mice IV952 N-acetylglucosaminyltransferase 111146, IU262, 111336, 111412 N-acetylneurarninic acid-Yphosphate I11228 N-acetyl-P-D-glucosaminidaseB - degradation of GMZ by IU952 1,2-N-acetylglucosaminyltransferaseI IU148 fi- I ,2-N-acetylglucosaminyltransferase I1 (GnT-IT)

P-

I2

Index

111152, IU962 ~-1,4-N-acetylglucosaminyltransferase 111 IU155 p-~,4-N-acetylglucosaminyltransferase IV IU157 p-1,4-N-acetylglucosaminyltransferaseVI 11/161 p- 1,6-N-acetylglucosaminyltransferaseIU337 P-I ,6-N-acetylglucosaminyltransferaseV IIll.58 ~-3-acetylgh1cosaminyltransferase(i-enzyme) IV262 P-3-N-acetylglucosaminyltransferase(i-enzyme) I11268 P-4-N-acetylylglucosaminyltransferase 11126.5 ff P-6-N-acetylglucosaminyltransferasefamily 11613 N-acetyI-a-D-hexosaminidase - in mucopolysaccharidoses IU949 N-acetyl- or N-glycolyl-8-0-sulfoneuraminic acid 111239 1-0-acyl-2-deoxy-hexopyranosesI1394 P-N-acetylhexosaminidase W501 N-acetyllactosamine IU265, I117 17 - galectins I11639 - IacNAc IU261 N-acetyllactosaminoglycans U6.56 - a3-sialylation I1656 N-acetylmannosamine IV229 - formation of IV1044 f - metabolic labeling sialic acid I111046 f N-acetylneuraminic acid 111228, I11485 ai-acid glycoprotein - lysosomal degradation IU476,11/479 acid phosphatase - lysosomal disorder IU948, IU954 acremonium a-N-acetylgalactosaminidase II/502 acrosome reaction IU896 - gamete interaction IU896 acrosome reaction W904 acrosome reaction IU906 f activation IU318 active specific immunotherapy I11679 f - animal models IU679 - carbohydrate antigens II/680 - clinical studies IU680 - MUC 1 tandem repeat II/680 - vacinnia virus IU680 acute phase proteins IU1030 acyl camer protein (ACP) I11439 acyloxy II/441 ff acyloxyacyl units I11437 acyl-protected nucleophiles U18 1 - anomeric 0-alkylation with primary triflates V181 lipoteichoic acid fragment Ul82 - synthesis via anomeric 0-alkylation U182 acyltransferases IU441 addition of carbohydrate side chains IV424 -Gal IV424 - GalNAcPl-4 IU424 - mammalian cells IU424

- Man I11424 - Taxoplasma gondii 111424 - T. brucei I11424 - yeast W424 S-adenosyl-L-methionine:sialate-9-0-methyltransferase IU238 Ad~2-mechanism U369 adhesin W969 f, IU973, IU1030 adhesion molecule IU590 f - acessory molecule I11591 - COS cells IU590 - length of Sn extracellular region IU590 - macrophages IV590 adhesion, cell II/721 affinity IV553 - chromatography V1082 - enhancement 1V.554 - labeling 111457 - tagging I11971 agalactosyl-IgG and rheumatoid factor binding IU982 aggrecan IV375, IV377 ff, IU382, IU384, IU386. IV390, 111409, IU411, I11722 aggregation of de novo complexes U1081 - integration of glycans into liposomers U1081 - modification of inert beads I/108 1 aglycon IU455 AGP IU1030 agrin 111712 -a-dystroglycan IU7 12 - neuromuscular junction IU712 - renal tubular Bms 111712 - synaptic differentation IU7 12 AIDS, acquired immuno deficiency syndrome 11/851, 1V862 alanes U171 - formation using diazirines U171 albumen gland IU266 aldolase V638 alg (asparagine-linked glycosylation 1u47 ALG genes IU134 ff alginate network I11065 alkaloids, naturally occuring IV.514 alkyl glycosidases V811 ff - enzymatic synthesis (table) 1/811 alkylations of reducing sugars U177 - anomeric 0-alkylation V177 allantoic I11722 allergenicity 1111032 allergens IV1036 allosamidin U66 - synthesis via iterative assembly U66 alternative splicing IU155 Alzheimer’s disease IU721 amide-linked carbohydrate oligomers U574 - solution synthesis US74 amino acids U40

Index

- glycosylation 1140 aminoglycosides V1 I 13 aminosugar trichloroacetimidates U S ff amipurimycin Y567 amylases IV498 a-amylases 111498 P-amylases I1/498 anarchimeric stabilization V436 anemia IU963 2,6-anhydro-2-thio sugars V389 - glycosyl donors U389 2,7-anhydro-N-acelylneuraminicacid IY228 animal models I11467 ankylosing spondylitis IU990 anomeric a$-S-phosphorodithioates I1396 - from glycalepoxide V396 anomeric ally1 carbamate U223 anomeric bromides V484 - preparation U484 anomeric radicals 11402 - glycosylation reactions U402 anomeric sulfates U232 antagonists I11446 anthracyclins I/1106 antiadhesion therapy I11967 antibacterial agents IV430 antibiotics I11967 antibodies 11/560, 111737 ff - anti-T I11858 - anti-Tn IV858, 111860 - blood groups 111858,111859 - cross-reactions IU854 - 0-glycan recognizing 11/857,111858 - hybridoma II/854, W858 - mannose-related IU857 - natural 111853, IU858,111859 - neutralizing 111855 - serum IU85 1, W859 anti-cancer agents I11935 - carbohydrate processing inhibitors IV935 antigen - masking by carbohydrate II/856 - recognition by T lymphocytes I11856 - carbohydrate 11/85 1 0-antigen IU968 - repeat II/435 ff 02-antigen I1226 antigenicity and glycosylation of gp 120 I11856 antigens - blood group 1V858, IV859, IV860 - I i IV856 - T and Tn IU860 anti-infection therapy IV973 anti-influenza drugs I11235 anti tumor - agents Y1096 - antibiotics I11097

I3

apical surface IU776 apoprotein B IU746 apoptosis IU40,1111034 - galectins II/638, II/640, I11641 Matrix assisted laser desorption ionization-time of flight-mass spectrometry (MALDI-TOF-MS) I/917, V926 - application in glycobiology U926 APS kinase IU245 f AR see acrosome reaction IU896 ff arabidopsis 111786, I11793 - cellulose synthesis IV793 arabinans II/790 - RG-I W790 arabinogalactans IU790 -RG-I IU790 arabinoxylans W787 - cross-linking I11787 - ferulic acid W787 - glucuronarabinoxylans IU787 arachidonic acid I11446 armed-disarmed - concept I1230 - principle 11203, I1230 arterial smooth muscle 111746 Arthrobacter sialidase I11504 arylp-D-mannopyranosides I1336 arylsulfatase A IU460, I11465 - in metachromatic leukodystrophy 111947, II/953 ascidian IIB96, IU903, IU899, IY902 - sperm-egg coat binding IV899 ascidran IV896 asialofetuin - lysosomal digestion 111479 asialoglycoprotein receptor II/476, I1/549 ff asialoglycoproteins W262 - catabolism I11475 f, I11479 asialoorosomucoid - lysosomal digestion IU475 f, I11479 ASN-GlcNAc - hydrolysis by glycosylasparaginase I11477 ASN-linked glycoproteins - lysosomal degradation 111473, W475 ff Asn-linked oligosaccharides - biological roles 11/82 - processing 11/82 - quality IU85 - control IU85 - mannose trimming 11/85 - ER mannosidase I 11/85 Asn-X-Thr1Ser IU48 asparagine N-glycosides I1271 - formation I1271 asparagine-linked glycoproteins IV37 asparagine-linked glycosylation see N-glycosylation aspartylglycosaminuria IU477

I4

Index

- animal model IV948 - in Finland II/953 Asp-box motif I11490 assignment of mass values V92 1 Asx-turn IV49 atherosclerosis IV743 - extracellular matrix I11743 ATP sulfurylase 111245 f attachment of the GPI precursor to a protein - o site 111425 - transamidase IU425 aureolic acids U1111 Austin disease IV465 autoantibodies IV968 autoantigen IV856 autoimmune arthritis W979 avermectin a disaccharide I/203 - synthesis with exo-enol ethers 1/203 avidity IV861 azatrisaccharide IV203 2-azido---imidates V475 - reactivity I1475 1-azi-tetra-0-benzylglucopyranose7 1/162 - glycosidation of phenols I1162 Bacillus circulans P-galactosidase W499 bacteria IU1029 bacterial - lipopolysaccharide fragments 111016 - interaction with monoclonal antibodies Ul016 - sialidases I11490 baculovirus IV1053 - enzymes needed to improve recombinant protein production W1048, IVI 054 - deleterious carbohydrate structures 1111056 baculoviruslSf9 insect cell system IV15 1 bamacan IU377, II/381 band3 IV963 Bandeiraea simplicifolia lectin I1 II/980 Barton deoxygenation V73 basalioma II/238 basolateral surface IV776 batroboxin I11267 betaglycan 111377,111381 f IV707 - TGFP type I11 receptor I11707 biacore 111046 ff - affinity determination I11053 - application areas 1/1054 ff - determination of kinetic rate constants I/1052 - experimental procedures I/1048 - immobilization of biomolecules at the sensor surface I/1048 - in situ modification of immobilized carbohydrates V1056 - interaction analysis and controls 111051 - oligosaccharide characterization V1055

- real-time analysis by surface plasmon resonance U1045 - selectin binding to a glycoprotein ligand U1054 - sensor chip V1046 - sensorgram U1047 - steps for applying the biosensors U1046 - surface regeneration I/1050 - system I11045 bidentate ligand V223 - as a leaving group V223 biglycan IU377,111380, IV384, IU748 BI-GP IV554 ff binding specificity IU553 biological recognition IV230 biosynthesis IU791 ff, IV968 - cellulose IV79 1, IV793 - cellulose synthase genes IV793 - glucuronoarabinoxylan IU791 - glycosyl transferase IV792 - Golgi apparatus IV791 f - methyl esterified homogalacturonan 1u791 - nucleotide sugars IU792 - of hyaluronan IV363 ff - of nod factors U847 - pectic polysaccharides IV792 - plasma membrane IU791 - RG-I I11791 - RG-I1 IU791 - wall glycoproteins IV791 - xyloglucans 11/79I f biosynthetic - mechanism of HA W367 - precursors IV364 - - nucleotide sugars IU364 bis(monoacylg1ycero)phosphate IV461, Ill464 bisected oligosaccharides IV148 bisecting GlcNAc 1V148 - residue IU155 bisubstrate analog IV300 blastocyst 11/723 B-lipoproteins II/745 blood-group IV1029 - antigens IV853, IV856, IU857, I1/858 - i-type V652 - - enzymatic synthesis U652 -A - - active glycoproteins Ill502 - - galectins IV639 - A and B I11321 blood group Lewis: FucT 111, V and VI W204 - cancers I11205 - chronic liver diseases IV205 - digestive tract I11205 - Lewis enzyme W204 -lung IV205 - plasma-type enzymes I11204 - leukocyte enzyme: FucT VII IV206

Index

- controll by cytokines IV206 trafficking of leukocytes IU206 blood-type-related oligosaccharides V419 - synthesis by the ortogonal strategy I1419 boranes I1171 - formation using diazirines V17 I boron I11795 - cross-link W795 - deficiency IU795 - wall pectin IV795 - wall pore size W795 bovine mammary gland 11/264, IV266f, I11269 bovine milk glycoprotein 111269 brain IV412 - KS proteoglycans I11410 - tumor I11238 branch specificity IV157 branches 111317 breast cancer I11675 ff - cells IV318,I11323 - cell lines IV676 breast carcinomas W238 brefeldin A IU386 f brevican 111377,I11380 bromine addition to D-glucal V37 1 - influence of protecting groups I1371 bromoalkoxylation V372 N Butyldeoxynojirimycin - inhibition of G Mbiosynthesis ~ IV953 -

C. elegans IV154 cadherin see E-cadherin caenorhabditis elegans I11276 - galectins IV625,11/627, 111632,IV63.5 calicheamicins I11100 calmegin IU997 calnexin 111835,IV997 - role in glycoprotein folding IU123 ff calorimetric data I1882ff - interpretation U882 - solvation/desolvation I/882 - solvation entropy U883 - translational/rotational entropy V884 calreticulin Ill997 - role in glycoprotein folding 111123 ff calustrin I11412 calystegines IV514 - A3 I11514

B I II/514 IV514 - C I 111514 -

- B2

Camillo Golgi I11977 cancer IV1.57,IV696 f - associated antigens I11315 -cells IV318 - galectins 111637,IU641 , IV642

I5

- immunotherapy 1111029 - initiation and progression I11926 - metastasis IV637,I1/641, IU642 carbamate family of participatinggroups I1475 carbohydrate I1269 ff carbohydrate amino acids U566 ff, U572 - as important construction elements I1566 - conensation I1572 - natural I1566 - synthetic I/567 carbohydrate-antigen IV851 - xeno-antigen 111853 -binding proteins 111853,IV857,IV861,IU862 - in parasitic protozoans IV869 carbohydrate-binding specificities IV861 f, 111968 f, carbohydrate-carbohydrate interactions I/] 061 ff - affinity interactions V1079 ff - agarose I11063 - alginate I11063 - arrangement of motifs I11075 - atomic force microscopy I/ 1083 - carbohydrate motifs V1073 - carrageenan I11063 -cells V1088 - cell walls Ill064 - crystallography V1084 - electron microscopy U1083 - experimental approaches I11078 - extracellular matrix of seaweeds I11063 - mammalian extracellular matrix components

V1066 - mass spectrometry I11085 - measurements of interactions I/1078 - microscopy I11083 - molecular aspects I/1074 - molecular basis V1076 - molecular modelling I11086 - nuclear magnetic resonance V108.5 - part of recognition keys U1068 - polyvalence I/1074 - recognition V1063 - repulsion I11072 - schematic overview M062 - structural components to cell recognition

I11063 -tools I11086 - weak interactions I11074 carbohydrate-deficiency gl ycoprotein syndrome

11120,11129 carbohydrate-deficient glycoprotein syndrome (CDGS) 111154,111959 carbohydrate-interactions 111069,I11071 - embryonal cells I11071 - invertebrates I11069 - sponges U1069 -tumor cells I11071

I6

Index

- vertebrates I11071 carbohydrate-nucleic acid interactions 111096 ff - carbohydrates binding to DNA I/1096 - carbohydrates binding to RNA I11112 carbohydrate-polymers US72 carbohydrate-protecting groups I1269 carbohydrate-receptor analogs I1/967 carbohydrate-recognition IV584 ff - A crystal structure 111584 - binding site IU584 -NMR 1U586 - 3’-sialyllactose IV584 - V-set domain IU584 - X-ray crystallography IV584 -recognition domain (CRD) 11/55I , 111599ff - galectins IV626, IV627, IU629, I1/630, IU635 - ligand binding IU600 carbohydrate-specificity IV.537 f - concanavalin A IU537 - D. biflorus lectin IV537 - EcorL lectin IV.537 - G. simplicifolia lectin IU537 - L. tetragonolobus lectin 111538 - pea lectin IU537 - peanut lectin I11537 - phaseolus vulgaris lectin 111538 -potato lectin IV538 - ricinus communis agglutinin I11537 - soybean agglutinin IU537 - S. nigra lectin 111538 - snow drop lectin IV537 - U. europaeus lectin I11538 - wheat germ agglutinin I11537 carbohydrate-sulfotransferases II/249,11/252 ff - cDNAs I11250 - - 3-OSTs I11250 - - chondroitin 6-sulfotransferase I11250 - - GlcA-3-0- sulfotransferase I11250 - - IduA-2-0- sulfotransferase IV250 - - keratan sulfate sulfotransferase 111250 - galactosyl ceramide 3-0- sulfotransferase I11252 - GSTS 1V252 ff - HEC-GlcNAc6ST 1112.55 - HNK-1 sulfotransferase IV252 - NSTS I11252 - 3-OSTs IV2.52 - 3-OST- 1 111255 - 3-OST-2 II/255 - 3-OST-3A W255 - purification 111249 - SPLAG domain 111252 - stem domain IV252 - 2-0-sulfotransferases IU252 - 2-0-sulfotransferase domain I11252 carboxylic acids V44 - glycosylation 1144 carcinomas 111285,I11560

carnitine - formation by proteolytic release of trimethyllysines IU474 cartilage 111409, IU722, IV731, IV733,111735 f castanospermine IV480, IU93.5, II/IOOO cathepsins -A - - in galactosialidosis II/957 - - protective protein, for P-galactosidase and a-neuraminidase II/946 -D - - deficiency in mice II/948, IU950 -K - - in pycnodysostosis IU945, I11948 ff - role in degradation of Asn-linked glycoproteins IU475 ff - role in degradation of thyroglobulin 1u478 f caveolae IU764 - GPI IU764, 111766 caveolae II/772 ff CCAAT IV151 CD4 II/857,11/860, 111862 CD14 IV446, I111030 CD22 W233 CD33 111231 CD43 - galectins IU639 CD44 1111.56, IV377,11/381, IV411, I11687 f, 111694, 111696,II/708,11/721, I11747 - CD44E I11708 - hematopoiesis I11708 - hyaluronan receptor IV708 - lymphocyte homing I11708 - inflammation I11708 - metastasis IU708 - RANTES I11708 - tissue inhibitor of metalloproteases-l (TIMP-I) Ill708 - tumor progression I11708 CD44H 11/72] CD4.5 - galectins I11639 CDA 111963 - type I1 IV963 CDGS IU88 - type I I11959 ff - type I1 II/959, IV961 ff CD-MPR IV565, 111567, IV572 - crystal structure II/572 - - disulfide bond pairing II/572 - - interaction with Man-6-P IU572 - - oligomeric state IU572 - - polypeptide fold IV572 - intracellular trafficking pathways 1v565 - - cellular machinery IU565 - -targeting sequences IU56.5 - co and post-translational modifications IU568

Index

- mutant forms I11571 oligomeric state IU568 - structure IV567 CDw60 I11237 cell - adhesion - - molecules, see also selectins 111862, IV912 ff - - galectins IU638, IV640, I11641 - - integrin ligand (aspi) I11916 - - L I W914 - - laminin IV916 - - NCAM II/9 1 2 ff - - PSA-NCAM 111913 -binding studies I11097 - - cell adhesion to substrate V1097 - - cell aggregation V1097 - cycle IVl39 - fusion activity of HIV-1, see also syncytium IV855, I11857 - migration W930 - - galectins I11640 - proliferation - - galectins IU638, IU640 - recognition and adhesion U1061, I111013 - surface heparan sulfate proteoglycans IU705 - -cell recognition 111l014, IUl017, IV1019 cellobioside 11734 - glucosynthase-catalyzed synthesis V734 a-cellobiosyl phosphate U629 - synthesis V629 cellodextrins I1991 - conformations I1991 P-D-cellotetraose V993 - molecular structure V993 cellular - adhesion IY237 ff - aspects I11552 - communication 111229 cellulose IV787 - xyloglucan interactions IV793 f - - arabinosylated xyloglucans 111794 - - computer modeling IV793 - - cross-linkages II/794 - - fucosylated xyloglucans IU794 - - hydrogen-bonding I11793 - - in muro I11794 - - in vitro I11794 - - microfibrils IV793 - - modification - xyloglucan interactions IV801 - - cell elongation IU801 - xyloglucan network I11798 ceramidase IU465 ceramide IV455 ff, IV465, W972 - p-galactosyltransferase IYI 87 cerebrosides IV340 cervix I11722 -

I7

chain elongation I11368 - non-reducing end IV368 - reducing end IV368 chaperone proteins, calnexin and calreticulin 111852 chemotaxis - galectins IV640 Chinese hamster ovary (CHO) cells 111853, IU854 chitin V731 - enzymatic synthesis U73 1 - oligosaccharides US52 - - production in E. coli I1852 chitobiase see-Di-N-acetylchitobiase chlaydia trachomatis I11441 chlorate I11245 chol-l II/1018 cholesterol IV761 - detergent insolubility I1/761 f, I1/764 -GPI IV761 - membrane domains I11761 ff chondroitin sulfate IV1021 chondroitin 4-sulfate - digestive pathway in lysosomes I11947 f, I11950 chondroitin 6-sulfate - digestive pathway in lysosomes I11947 f, IU950 - proteogl ycan-LDL complexes I11746 chromomycin Ul 1 1 1 ciclamycin 0 1/64 - synthesis via iterative assembly 1/64 CIDNP - analysis of glycoproteins Yl040 - method VI 025 ff - - applications V1027 - - detection of surface exposed Tyr-, His- and Trp-residues of a protein I11025 - related molecular modelling I11027 CI-MPR I11565 ff - co- and post-translational modifications IU.567 - intracellular trafficking pathways IV565 - - cellular machinery 1V565 - - targeting sequences IV565 - mutant forms I11571 - oligomeric state IU567 - structure I11566 circulating anodic antigen 111881 - Schistosoma mansoni 111881 class I histocompatibility IV1000 class I mannosidases - ER mannosidase I 11193 - golgi mannosidase I 11195 - fungal secreted mannosidases IV97 - new genes with unknown functions IU98 class 2 mannosidases IV98 ff - golgi mannosidase I1 11/99 - lysosomal mannosidase 111101 - epididymal1sperm mannosidase IV103 - heterogeneous cluster of mannosidase homologs

I8

Index

I11104 classification of mannosidases IU89 ff - class 1 mannosidases (family 47 glycosylhydrolases) IU92 - class 2 mannosidases (family 38 glycosylhydrolases) IU92 claustrin IU720 cloning strategy U849 cluster glycoside effect V910 - molecular basis U910 CMP-N-acetylneuraminic acid W1044 f - Golgi transporter IU1045 CMP-Neu5,9Ac2 I11235 CMP-Neu5Ac hydroxylase I11230 f CMP-NeuSAc monooxygenase I11232 - gene cloning 111232 CMP-NeuSGe IU23 1 CMP-sialic acid 1/631 - synthesis V63 1 - derivatives U63 1 - synthesis V63 1 CoA transporter IV235 collagen I11983 - fibrillogenesis IV748 - type I, digestion by cathepsin K 111951 collectin 111601 ff - domain organisation IU60 1 collectin-43 I11608 colon cancer tissue 1113 19 colon tumor W238 competition IU267 ff competitive inhibitors IU15 1 , 111305 f complement, activity, pathway 111859, 111862. complement, glycoproteins C3 and iC3b IU862 complementation group 11123 -Lec2 IU23 - Lec8 11/23 - Had-1 11/23 complex N-glycans IVI 45 complex-type N-glycans IV884 - Acanthocheilonema vitae I11884 - Trichinella spiralis IU884 complex-type oligosaccharide I/607 - IacdiNAc pathway I/607 complex-type oligosaccharide IU261 - synthesis V602, IU261, I11264 f, W267 - - chitobio pathway 111265, IU267 - - lacdiNAc pathway I11264 f - - lacNAc pathway IU261, IV265 - - terminal reactions I1602 concanvalin A IV536, IU542 -Con A I11857 congenital dyserythropoietic anemia (CDA) IU963 conglutinin IU608, IU857, IU861 consensus sequon 11/46 ff control of sugar nucleotide levels IU13

conversion of 6-OH into 6-NH2 111558 co-operative effects of GalNAc-transferase isoforms IU281 corbohydrate-protein interactions U1025 - laser photo chemically induced dynamic nuclear polarization (C1DNP)-NMR technique U1025 core a-1,6-fucosyltransferase IU148 core 1 W323 f - P3-Gal-transferase IU3 16 core 2 P6-GlcNAc-transferase 11/3 17 f core3 IV319 - P3-GlcNAc-transferase I113 19 core4 IU319 - P6-GlcNAc-transferase II/3 19 core5 IY320 - 1x3-GlcNAc-transferaseactivity IU320 core6 IU320 core 7...IU320 core 8 111320 core glycosylation 11145 core protein IU73 1, IU734 f - hyalectans 111731 - proteoglycans, small leucine-rich 11/731, IU733 core region of 0-glycans Y648 - in vitro synthesis U648 core sugars I11435 ff cornea IU407, IU7 18 coronaviruses IU238 coulombic stabilization U870 coupling of sugar phosphates I1630 Creutzfeldt-Jakob I111036 cryotosporidium parvum IU878 crystalline oligosaccharides V987 - hydrogen bonding I/987 Csk IU772 CSPGs IU720 cyclic sulfites U233 cyclization U401 - of acyclic sugars V401 - Wittig reactions U401 cystinosis 11/474, IU945 cytokeratins - galectins IU639 cytokines IV370, 111446, II/8S2 -1FN-)I IU370 - IL-1 I11370 - TNF-a W370 cytopathogenic effects, HIV-1 111852 cytoplasmic 111658, W661 - glycosylation 1116.58 cytosolic sulfotransferase IU245 - neurotransmitters IV24.5 - steroids IU245 5D4 IU722 DAD1 (defender against apoptotic death) IV57

Index

daunorubicin U1107 deacetylase I1/439 deacylase 11638 decarboxylative glycosylation U22 1 - method I/221 decorin IU375, IU377 ff, W390 decorinhiglycan (DSPGsj IV748 delivery 111559 dendritic cells IU856 3-deoxy-C-cellobiose U507 - preparation I/507 2’-deoxy-~-disaccharide V378 - synthesis with a glycal epoxide as glycosyl donor I/378 2’-deoxy-~-glycosides U385 - synthesis with cycloadduct glycosyl donors 11385 2-deoxygl ycopyranosides I/198 - enol ether synthesis U198 2-deoxyglycosides V13, V202, V367 - synthesis U13, U367 - - with endo-glycals U202 2-deox yglycosyl bromides I/395 - glycosylation reactions U395 2-deoxyglycosyl donors U393 2-deoxygl ycosyl fluorides V395 - glycosylation reactions I/395 2-deoxyglycosyl phosphate U397 - for glycosylations V397 2-deoxyglycosyl phosphites UI 27, V397 - for glycosylations U397 - glycosylation U127 2-deoxyglycosyl phosphoramidites V397 - for glycosylations 1/397 S-(2-deoxyglycosyljphosphorodithioates I/396 - glycosylation reactions V395 2-deoxyglycosyl sulfoxides U399 - armeddisarmed concept 11399 - for glycosylations U399 2-deoxyhexopranoses U394 2-deoxyhexopyranoside 11369 - synthesis U369 3-deoxy-D-manno-octul IU44 1 ff 3-deoxy-D-munno-octulosonicacid (Kdoj W437 deoxynojirimycin II/408 I -deoxynojirimycin IVlOOO 2’-deoxy 2’-phenylseleno-a- and P-disaccharides U383 - stereoselective syntheses U383 2-deoxy-2-phtalimido glycosides U472 2-deoxy sugars U367 ff 2-deoxy thioglycosides U398 - glycosylation reactions V398 dermatan sulfate - digestive pathway in lysosomes II/947 f, II/95O development IU729, II/735 ff, IV911 ff - auditory organ IV9 13

I9

- axonal growth and synaptogenesis II/916 - blastocyst, trophnectodenn and inner cell mass IU9 17 - brain microvasculature IV912 - capillary morphogenesis IU917 - cartilage and mesenchyme IU916 - chick embryo IU915 - ciliary ganglion motoneurons in chicken W915 - cultured rat calvarial osteoblasts IV916 - early-postnatal mouse cerebellum II/9 14 - embryo development II/912 - fertilization II/9 15 - fetoembryonic defense mechanisms IV914 - formation of myotubes 111917 - hamster embryos 1V916 - heart, kidney and hair-follicule II/914 - human neuroblastoma cell line SK-N-MC IU917 - human prosencephalon IV911 - human sublingual salivary glands IU911 - immune systems W914 - implantation IV917 -kidney IV916 - malformed tissues IU912 - motoneuron and myotube IU9 13 - mouse brain IU915 - mouse embryo IU915 - mouse Leydig cell IV9 12 - mouse olfactory system IU916 - mouse neuroblastoma cells IU9 13 -nervous systems IU914, IU916 - neurogenesis 1V913 - of multi-cellular animals 111154 - preimplantation embryos IV917 - rabbit fallopian tubes, uterus and blastocysts II/9 12 - rat brain IU913 - rat cerebellum II/917 - sensory systems and neuronal plasticity IV9l2 - thymus of fetal and newborn animals IU917 N,N’-diacetylchitobios 111265 N,N’-diacetyllactosediamine IV265 N,N’-diacetyllactosediamine,IacdiNAc 11/26 1, II/263 diaziridine V157 - formation V157 - oxidation U157 - oxidation with 12 diazirines 1/171 - exploratory use U17 1 Dictyostelium discoideum II/419 2,2’-dideoxy-P-disaccharide B-C U387 - synthesis I/387 differentation IV3 18 diffraction methods I/969 1,l-difluorides 11171 - formation using diazirines Y171 dimethyldioxirane epoxidation U20.5

I 10

Index

2,2-dimethyldioxirane (DMDO) I1377 - for epoxidation of glycals I1377 disaccharide synthase IV440 disaccharides 1/973 - crystalline conformations I1973 dispersive interactions 11872 distribution between compartments V1082 boyden chamber V1082 dog pancreas I115 1 dolichol 11/47. IV960 f - chain length II/132 - phosphate glucose IV39 -phosphate mannose IV39, IV13If, IU135f - - synthase IV132, IV135 - pyrophosphate IV131, 111134 - - oligosaccharides - - - degradation IV480 Dol-P-Man see dolichol phosphate mannose Dol-PP see dolichol pyrophosphate domain structure II/148 domains, N-linked I111035 DPM synthase 11141 dual glycon specificity IV501 dylinositol (GP1)-anchored variable surface glycoprotein (VSG) IV873 E-64 IV951 Ebola virus II/832 E-cadherin IU156 echinoderms IV238 ff EGCases IV507 egg cells IV239 electrospray-mass spectrometry (ES-MS) I/Y 18, V926 - application in glycobiology I1926 embryonic fibroblasts 111932 - focal adhesions IU932 embryonic lethality IV154 endoD 11/505f endo H I11505 f endo-P-galactosidase IV507 endo-P-galactosidase IV963 endo-a-N-acetylgalactosaminidase 111506 endo-P-N-acetylglucosaminidase IV505 endocytosis IV456 ff, IIh52, IV761 endoglucosarninidase - degradaton of polymannose oligosaccharides in ER IU480f endoglycoceramidase I11507 endo-glycosidase IV505 endomannosidase - absence in Chinese hamster ovary cells I117 1 - alternate deglucosylation pathway 11/73 f - assay 11/72 - cloning 11171 f - concerted action with other deglucosylating

enzymes IV72 ff inhibitors 11/67,11/72 - molecular characteristics 11/67, IU71 - phylogenetic distribution IU67, IV71 - purification IV67, II/7 1 - processing of N-linked oligosaccharides IU66 - properties IV67,11170 ff - specificity IV66f, 11/69 ff - subcellular location 11167, IV7 1 endomembrane assembly line 111146 endometrial eoithelial W723 endometrium I11722 endoplasmic reticulum (ER) IV134, IU376, IV997 - role in degradation of polymannose oligosaccharides IV479 ff endothelial cells 111862 endotoxin IV435 enediyne antibiotics I11096 Entamoeba histolytica I11878 Enterobacter W439 envelope glycoprotein, see also gp41, gp120, gp 160 11/851, IV852,11/861 - viral II/832 - - intracellular transport IU832 - - pathogenicity 111832 - retroviral I11830 - - carbohydrate structure I11830 enzymatic glycosylations V647 ff, 11663, Y685 ff - N-acetylglucosaminylation I/696 - N-acetylglucosaminyltransferase I, I1 and 111 U696 - N-acetylglucosaminyltransferase V V698 - fucosylations 11690 -FucTV U692 - FucT 111 and IV V692 - galactosylations Y686 - al,3-galactosyltransferase V688 - ~1,4-galactosyltransferase U686 - human milk -1,3/4-fucosyltransferase I/690 - non-natural acceptors I1658 - non-natural donors I1658 - recycling of sugar necleotides I1663 - sialylations 1/692 - a2,3-sialyltransferase I/692 - P2,6-sialyltransferase Y692 enzymes - catalyzed additions I/370 - glycals V370 - linked lectin assay (ELLA) V878 - modifying NeuSAc I11230 - replacement W469 E-PHA 1111.56 epidermal growth factor (EGF) IV156 epiphycan IU377, IU380 epithelia IV860 epithelial membrane mucins IV674 ff - MUCl I11675 -

epithelial mucins IV670, I11672 - expression of epithelial mucins IV672 epithelium II/972 epitope IU968, I11970 f epoxidation U200 Epstein Barr virus 111856 ER see also endoplasmic reticulum IV378 f, I11383 ff - mannosidase I IV85 - retrieval - - sequence 11128 ERGIC-53 IU86, IU.545 Erp57 IIllOOl erythroleukemia IV830 erythromycin A U391 - synthesis I1391 erythropoietin receptor TV831 Escherichia coli I11437 - K26 antigen U226 E-selectins II/322,II/613 f, 111619, 1111030 ES-MS see electrospray-mass spectrometry (ES-MS) esperamicins V1096 I7-P-estradiol I11722 Ets transcription factors I11153 eubacteria - lipopolysaccharide IU130, W139 - mannosyltransferase IU139 - mycobacterium IV130, IV135, IY139 everninomicin V388 - synthesis V388 evolution 111154, IV239 ff, 111266, IV276 - of GPI biosynthesis W426 - - archaea I11426 - - GlcNccl-6myo-inositol- 1-phosphodialkylglycerol 111426 - - Methanosarcina barkeri IU426 evolutionary considerations I11588 f - disulphide bond 111589 - hypervariable interstrand loop regions IV589 - immunoglobulins I11589 - primitive multicellular organisms 111589 - primordial Ig-like domain II/589 - prokaryotes IV589 - protein-protein interactions Ill589 exo-glycosidases IV497 exon I11266 f expansins I11799 - cell expansion IU799 - wall stress relaxation I11799 expression patterns 111286 expression systems I1849 extracellular HSF'Gs, other. IV713 - collagen XVIII 111713 - endostatin IU713 - heparin IV713 extracellular matrix ILnOl

extracellular matrix heparan sulfate proteoglycans II/7 10 - aggrecan family IU7 10 - agrin II/7lO - basement membrane IV7 10 - decorin IV7 10 - leucine-rich repeats 111710 - perlecan W710 extrahepatic tissues IV.550 ezomycin Ai I1567 FAB-MS see fast atom bombardment-mass spectrometry fabry disease - animal model IU953 FAK 111772, I11777 Farber disease IU465 Fasciola hepatica IV868 fast atom bombardment-mass spectrometry (FAB-MS) .I/915, V926 - application in glycobiology U926 fertilization IU895, I11901 ff - sperm-egg binding IU901 f fibril, collagen IU719 fibromodulin II/409, II/7 17 fibronectin - galectins IV638, IV640 filarial parasites I11267 fluorophore-coupled IV992 F-MCFV see Friend mink cell focus-inducing virus F-MuLV see Friend murine leukemia virus formation of complex-type I1594 - branching US94 - committed steps I1594 forskolin IV156 Forsmann - galectins IU639 Forssman glycolipid synthetase I11191 free GPIs 111417,111758, 111760 - trafficking I11760 Friend mink cell focus-inducing virus I11830 Friend murine leucemia virus IU82.5, IV83 1 - glycosylation mutants 11/831 Friend spleen focus-forming virus I11830 frog II/899f - integumentary mucin I11673 fructosides I1201 - enol ether synthesis I1201 fruit ripning IV794, IU801 - enzymic activities IV801 - gene expression I11801 - pectins IV801 - wall loosening I11801 - xyloglucan I11801 F-SFFV see Friend spleen focus-forming virus

I 12

Index

fucose IU410 - -dependent binding 111971 - inhibition of glycosylasparaginase IU476 f - N-linked U295 a-L-fucosidase 111502, IU902, IU903 - lysosomal II/475, IU477 - lysosomal disorder IU948, 111954 P-D-fucosidase IU503 fucosidosis IU477 a-fucosidosis - animal model IU948 a-fucosylation U27 1 a3-fucosylation of a1 -acid glycoprotein 11/981 fucosyl phosphites U125 - glycosylation U125 fucosyltransferases IUI 97 ff, II/621 - enzymatic reaction mechanism IU201 f - - deoxyfuconojirimiycin IU20l - - ethylmaleimide (NEM) IU20 1 - GDP-Fucose: Fuca 1(Fuca 1,2Fuc)a2-fucosylransferase.. .W203 - - Schistosoma mansoni IU203 - - Tricobilharzia occelata IU203 - - general characteristics II/198 fucosy1transferase - nomenclature IV198 - protein structure and topology IU200 - - structural organization IV200 - sequence peptide motifs IU199 - - catalyc site IV199 - - GDP binding domain IU199 - specificity IU199 - - hypervariable region IUI 99 - unconventional types of fucosylation: FucPI-P-Ser and cytoplasmic Fucal,2-GalP1,3-GlcNAc-Pro (Dictyostelium discoideum) IU207 - - Dictyosteliumdiscoideum IU207 - - FucP 1-P-Ser IU207 - - Fucal,2-Gal~l,3-GlcNAc-Pro I11208 - - - cytosolic IU208 al,3-fucosyltransferase IU148, IU331 a2-fucosyltransferase 111264, IV268 - family U611 a3-fucosyltransferase U602, IU264, IU268 a3/4-fucosyltransferase family U6 12 P4-fucosyltransferase IU266 GAZ - accumulation in Tay-Sachs and Sandhoff disease IU95 1 - intermediate in GMZ degradation IU952 GAG IU717 - chains IU376 ff, IU381 ff, IU388 f - - biosynthesis II/379, IU388 ff - - disaccharides II/388

- - epimerization 111376,111383, II/389 ff - - formation, repeating W388 - - formation, repeating disaccharides II/376, IU378 - - functions IV382 - -heterogeneity IU382 f - - initation IU379, IU381 - - lyases W379 - - microstructures IU379, W383 - - precursors 111376, 111384 - - - nucleotide sugars IU384 - - regulation IU390 - - specification IU382,11/385, IU388 - - structure IV379, IV382 - - sulfation IU376, IU378, IU382, I11390 f - - types IV376 ff, IU385, IU388 f - - - CS IU376 ff, IU381 ff, IU389 ff - - - DS IV376 ff, IU381 ff, IU389 - - - heparin I1/381 ff - - - HS IU377 ff, II/385, IU388 ff - - - HSiheparin IU389 ff GaINAc - galectins W639 galactans IU790 -RG-I IU790 galactocerebrosidase IU464, I11466 galactopyranoses V180 - anomeric 0-alkylation with primary triflates II180 - result of the reactive anomeric p-anion U180 galactose - addition to cultured cells IU1050 - biosynthesis in animal cells IU1049 - effect on complement fixation IU1049 - levels on recombinant proteins IU1050 f galactose-mediated uptake of antigens I1/857 galactose-recognizing receptors IU230 galactosialidosis IV467 - animal model IU946, IU948 a-galactosidase IU498 P-galactosidase IU464, 111499, U727 P-D-galactosidase - in galactosialidosis IU946 - in G Mgangliosidosis ~ IU948, IU953 - in mucopolysaccharidoses I11949 - in degradation of globoside W952 - in degradation of GMI IU952 - lysosomal II/475 f P-galactosides - galectins IU625 a3-galactosyl/N-acetylgalactosaminyl transferase family I1613 galactosyl phosphites U125 - glycosylation V125 galactosylation of IgG IU980 galactosylceramide IV466, IU1013 ff galactosylsphingosine IU466

Index

galactosyltransferases 111175, IV977 f - database inquiries IV177 - nomenclature I11177 - reaction catalyzed IV175 a( 1-3)galactosyltransferase I1642 a3-galactosyltransferase V602, IV175 - gene family IV188 - - evolution of 111191 - a-gal epitope IV190 - - xenotransplantation IV190 - ABO blood group IV190 a4-galactosyltransferase IVI 92 P-galactosyltransferase 11/41 2 p( 1-4)galactosyltransferase V642 P3-galactosyltransferase I11264 - family 116I7 P4-galactosyltransferase 11/I 75, 11/26 I , IV263, 111265, II/267, IV269 - family 11615, II/267 - P4-galactosyltransferase-I I11178 - - biosynthetic reactions IV178 - - lactose biosynthesis IV178,11/182 - - protein domain structure 111181 - - gene organization I11181 - - gene regulation 11/181 - gene family IV182 - - evolution of IVI 84 -gene family W185 - protein domain structure II/I 86 - acceptor sugar substrates IV186 galectins I111023 - candidates 1V627, IV629, II/630ff, 1V635 - expression 111637, I11638 - functions IV638, W640, IV641 - gene knockout IV638, II/641 - genes 111629, II/630,111632, 111633, IV635, 111637 - ligands W639 ff - regulation IV637 - specificity I11639 - structure - - acetylation 111628, II/635 - - carbohydrate recognition domain IV626, I11627 - - crystallography I11626 - - linker I11626 - - phosphorylation IV628 - - repetitive domain 111628, 111634 - subcellular targeting - - nucleus 111628,111636,IV642 - - secretion 111635, IV636 GalNAc - benzyl 111316 - glycosylation IV276 - peptide I11314 - transferases IV274, I113 15 - - catalytic domains IV274 - - - DxD-like IV274

I 13

- - - sequon I11274 GalNAca-SerlTnr-linked oligosaccharides (0-glycans) IV313 a3-Gal-transferase I11321 P3-Gal-transferase W321 gamete interaction see spern-egg interaction ganglio-series gangliosides V305 ff - retrosynthetic analysis U305 - synthesis V305 gangliosides 11305, V345, 111456 ff, 11110l4, I111016 f - classification V305 - comparison with GDla 1/311 - GD3 111235, 111237 ff - GMI 111458, - - galectins II/639 - GM2 IL'458, IU462 - G Q l b V311 - synthesis I/305 ff gangliotetraosylceramide 111969, IV97 1 gastric disease W967 ff - atrophic gastritis I11967 - duodenal ulcer W967 - gastric cancer 111967 - gastritis 111967 - Helicobacter pylori I11967 - iniflammation W967 - peptic ulcer IV967 - therapy W967 gastric mucosa 1111030 gastritis 111967 Gaucher disease IV464, II/460 - animal model IV948, IV953 - enzyme replacement IV946 - in Australian population I11954 GDla 1111016ff, 1111022 GDlb II11016 f GDP-deoxyfucoses V634 - preparations V634 GDP-Fucose: GlcNAc-N(Fucal,6GlcNAc) a6-fucosylransferasesIV207 - hepatocellular carcinoma 11/207 GDP-Fucose: 0-Ser(Fuca1-0-Ser)GlcNAc polypeptide fucosylransferases. ..11/207 - EGF domains I11207 - 0-glycosidic bond IU207 GDP-Fucose: Gal0 1 (Fucal,2Gal)a2-fucosyItransferase.. ,111204 - xenograft rejection W204 GDP-Fucose: Gal~l,3GlcNAc(Fucal,3GlcNAc) bacterial (Helicobacter pylon) a3-fucosyltransferase ...111207 - Helicobacter pylori IV207 - leucine zipper 11/207 GDP-Fucose: GalPl,4/3GlcNAc (Fucal,3/4GlcNAc)a3/4-fucosylransferases I11204

I 14

Zndex

gel-forming mucins W672 ff - processing and function W672 ff gene - disease IV959 - knockout IV964 - therapy II/469 - transfer IV229 gene structure - evolution IV199 - Caenorhabditis elegans II/199 genetic manipulation 11/973 genome sequence II/973 genomic organization W266 Giardia lamblia 111878 GlcNAc U183 - anomeric 0-alkylation I/183 - -benzyl IV322 - -PI deacetylation 11/42] - - GIcNAc-PI 11/421 - - GlcN-PIS IV42 1 - - GTP IU421 - - PIG-L 11/42 1 - - sub-compartment of the ER IV421 - -PI synthesis W420 - - GlcNAc transferase W420 f - -GPII IU420 - - P I IV420 - - PIG-A II/420 - - PIG-C W420 - - PIG-H IV420 - - PIG-a 111420 - - UDP-GIcNAc W420 - transferase I1 1U962 - -2 epimerase - - in animal cells IV1045 - - in insect cells IV1054 - - in plants IV1057 - -TV gene expression IV929 - - up-regulation Ill929 0-GlcNAc I/279 - synthesis V279 - side chains V294 Pl,3-GlcNActransferase 1V337 P6-GlcNAc-transferase IU320 f -family IV319 GlcT I1/961 GlcT-1 IU342 globoH V82 - synthesis via iterative assembly 1/82 globoside - lysosomal digestive pathway W952 D-glucal V368 - addition of alcohol 11368 a-glucan-active enzymes V560 glucoamylases IU498 glucocerebrosidase IV457, IV464 glucopyranoses U180 f

anomeric 0-alkylation 1/18] anomeric 0-alkylation with primary triflates U180 - result of the reactive anomeric p-anion 11180 P-glucopyranosides Y329 - epimerization at C-2 V329 a-D-glucopyranosyl 1-thio-a-D-mannopyranose I/557 glucosamine IU437 glucose I/ 188 - anomeric 0-alkylation with cyclic sulfates V188 glucosidase I IVlOOO - assay 11/68 - cloning 11/67 f - concerted action with other deglucosylating eniymes 11/72 ff - inhibitors IV67 f - molecular characteristics IV67 f - mutants 11/67, 11/74 f - phylogenetic distribution 11/67 f - processing of N-linked oligosaccharides 1v66 - properties 11/66 ff - purification 11/67 f - specificity 11/66 ff - subcellular location 11/67 f glucosidase I1 IV1000 -assay 11/70 - cloning II/67,1V70 - concerted action with other deglucosylating enzymes 11/72 ff - inhibitors IV67, IV69 f - molecular characteristics IV67; IV69 ff - mutants IV67, IY74 f - phylogenetic distribution IV67, IV69 - processing of N-linked oligosaccharides 1v66 - properties IU67 ff - purification 11/67, IU69 f - role in glycoprotein processing 1V119 - role in glycoprotein folding II/123 ff - sequence homology IV67,11/70 - specificity IV66 ff - subcellular location 11/67, IU69 glucosidase inhibitors IV5 14, IM22 ff - australine IV5 14, 111524 - castanospermine IV522 - deoxynojirirnycin IV524 -

2,6-diamino-2,6-imino-7-O-(~-D-gluco-

pyranosy I)-D-glycero-guloheptit01 (MDL 25,637) W524 - 2,5-dihidroxymethyl-3,4-dihydroxypyrrolidine IV514 - DMDP (2,5dihydroxymethyl-3,4-dihydroxypyrrolidine) IV524 - glucosidase I IU522 - glucosidase I1 IV522 - lentiginosine 1V514

MDL 25,637 II/5 14 processing glucosidases W522 - trehazolin 111524 a-glucosidase IU497 a-D-glucosidase - in biosynthetic quality control I1/480 f - in pompe diasease IU954 P-glucosidase W498 glucosylation IV47 glucosylceramide IU464 glucosyltransferase (Glc-T) 11/41, II/341, 11/961 glucuronic acid-2-sulfatase - in mucopolysaccharidoses IV949 P-D-glucuronidase - in mucopolysaccharidoses IU949 glucuronyltransferases II/340 P3-glucuronyltransferase family 1/615 glycals I/6 I, U369 f, U376 ff - addition of a phenylsulfenate ester 11382 - addition of PhSCl to D-glucal derivatives U380 - addition of selenium based electrophiles U382 - addition of sulfonium salts and sulfenates I/381 - addition of sulfur based electrophiles 11379 - assembly V62 ff - conversion into spiro-ortholactones U383 - derivatives 1/61 - - iterative assembly U6 1 - epoxidation I/377 - epoxides V377 - - nucleophilic attack I/377 - fluoroglycosylation U38.5 - for the assembly of ohgosaccharides and glycoconjugates I/376 - halogenation I/370 - iterative assembly 1/61 - phenylsulfenyl chloride addition 11379 - protonation U369 - stereochemistry 11380 GlyCAM- 1 IV6 16 glycans 1/1068,11/881 - biosynthesis II/238 - - N-linked 11192.5 - carbohydrate-carbohydrate interactions I/1 068 - processing I1/924 - Schistosoma SD. 111881 N-glycans 11/26,11/314, II/856, IU859, IU861, 111880, lI/1033 - Schistosoma sp. 111880 N-glycans, di-, tri-, tetra-antennary IV853 N- and 0-glycan changes in cancer cells II/924 - summary 0-glycan IV20, IU273,11/314, IV322, II/3 15, IU3 19 f, II/323, II/676, II/854, II/856 - biosynthesis I/648, IU323 - - extension of core 3 and core 4 glycans 1/651 - - initialization U648 - - in vitro extension of core 1 glycans V650 -

in vitro extension of core 2 glycans V65 1 - - synthesis of core 1 1/648 - - synthesis of core 2 U649 - - synthesis of core 3 and core 4 V650 - core 1 II/3 I5 - core 2 II/3 16 ff - c o r e 3 IV316 - core 4 II/3 16 - core 5 II/3 16 - core 6 II/3 16 - c o r e7 IV316 - core 8 IV316 - in cercerial glycocalyx II/881 - - Schistosoma mansoni 111881 - Schistosoma sp. II/880 glycocalix IU4.55 glycoconjugates 1/715, II/239, II/9 1 1 ff - N-acetylgalactosamine II/9 12 - N-acetylglucosamine IU912 - acetylglucosaminoside IU912 -P-N-acetylglucosaminoside II/9 I2 - amine 111912 - diversification 111239 - enzymatic synthesis on soluble supports U715 ff - galactose 11/912 - P-D-galactose-BSA - 0-galactoside IV9 1 I f - --galactoside-_2,6-~iaIyltransferaseI1/9 13 - ganglioside sialidase mediated GM(I) 111916 - lactose-BSA IU912 - N-linked oligosaccharides 111914 - 2,8-linked sialic acid units W913 - long-chain polysialic acid (PSA) 111913 - mannosides II/9 12 -NOC-3 IU913 - NOC-4 11/913 - sialyl Lewis-A 11/917 - sialyl Lewis-X IV917 glycodelin A IV262 ff glycoforms I/591, II/979 glycogen IV349 - biosynthesis IU349 - branching enzyme 111349 f, IV356 - bulk synthesis II/354 - glycogenin IV349 ff - initiation IU349 - intermediates II/357 f - proglycogen IV349,11/357f - synthase IU349 f, II/354 f - synthesis W350 f glycoglycerolipids V305 - synthesis V305 glycolipids 1/305,1V229, 111455 ff, IU809, II/963, IV970 ff, II/1013 - bacterial pathogenesis II/809 - fucosyltransferases 11/330 -GA2 IU462 --

I 16

Index

- galactosyltransferases I11333 receptors 111812 - - stress response I11812 - receptor function 111810 - - modulation I118 10 - recognition 111810 - - via the stress response 1118I2 - sialyltransferases I11338 - synthesis 11305 N-glycolylmannosamine I11233 N-glycolylneuraminic I11228 N-glycolylneuraminic acid 1V231 ff, I11323 - biological roles 111231 - great apes I11232 - human tumours I11233 glycomimetics I1495 ff - definition V495 glycopeptides V267 ff, V7 12 - biological role I1267 ff - definition V267 ff - enzymatic solid-phase synthesis 1/712 - formation 11271 - 0-glycopeptides U274 - N-glycopeptides I1280 - glycosylation methods I1271 - structural diversity I/267 ff - synthesis V274, I1280 - - in solution 11274 N-glycopeptide libraries V298 N-glycopeptides 11297,11280, V285 - Lewis-type saccharide side-chains I1285 - natural saccharide side-chains I1280 - synthesis on the solid phase I1297 0-glycopeptides 11274, I1287 - synthesis 11274 - synthesis on the solid phase I1287 glycopolymer mass spectrometry V923 - fragmentation pathways U923 glycoproteins U268, IV229,11/669,11/791 - arabinose IV79 1 - catabolism - - oligosaccharide release from the ER 11187 - - failure of quality control 11187 - - cytosalic oligosaccharide degradation IV87 - cytoplasmic 111658, I11661 - - glycogenin IU659 - - parafusin IV659 - - phosphoglucomutase 111659 - - SKPl 111658- galactose I11791 - hydroxypyroline IV79 1 - hydroxypyroline-rich glycoproteins 11/791 - linkage regions I1268 - N-linked IU37 - lysosomal degradation 111473 ff - mitochondria1 IV661 - serine IV791 - viral IV821 ff -

- - biosynthesis I11826 - - carbohydrate substituents 111826, IV828 - - functions IU825 f - - - enzymatic activities IV826 - - - immune response I11826 - - - membrane fusion IV825 - - - receptor binding I11825 - - - virion assembly IV825 - - glycosylation 111822, IU826 f - - - epitopes, antigenic I11827 - - - folding of proteins I11826 - - - oligosaccharide diversity IV827 - - - oligosaccharides, N-linked 111827 - - - oligosaccharides, 0-linked IV827 - - - structure analysis 11/828 N-glycoproteins - endoplasrnic reticulum processing IV119 - quality control of folding IV123 ff - synthesis IV119 glycosaminoglycans (GAG) IV375, IV701, IV717 ff, 1V729 ff, IV744, IV1020 - catabolism in lysosomes I11947 ff - chains IV385 ff - - precursors I11385 - - - nucleotide sugars IV385 ff - - - phosphoadenosine 5’-phosphosulfate (PAPS) 111385 - - specification IV387 - - types IV386 f - - - CS I11387 - - - D S IU386 - - - HS 111387 - chondroitin 4 IV744 - chondroitin sulfate IV729 - dermatan sulfate IU729 ff, IU734, IV737 ff, IV744 - disaccharides 111730 f - - chondroitin sulfate D IV730, W737 - - chondroitin sulfate E IV730 - - chondroitin 4-sulfate IV736 - - chondroitin 6-sulfate 111730, IV736 f - - 2-sulfated iduronic acid IV739 - - unsulfated chondroitin IU730 - heparin 111729, I1/739, IY745 - heparan sulfate 111729, IV739, IV744 - heterogeneity 111375 - hyaluronan 111729, IU73 I , 111733, IU736 - hyaluronic acid I11744 - keratan sulfate 111729,111732,111734, 111736, I11745 - nuckar IV660 - 6 sulfate (C-43, C-6-S) I11744 -types IV375 - - chondroitin sulfate (CS) IV375 - - dermatan sulfate (DS) IV375 - - heparan sulfate (HS)/heparin 111375 glycosidases V723, IV456

Index

- amyloglucosidase IU517 - aryl-glycosidases IU517 - cytosolic IU480f - ER I1/480f - ER mannosidase 111517 - families I/723 - P-glucosidase 111517 - glucosidase I I11517 - glucosidase I1 IU517 - glycoprotein processing enzymes 1115 17 - indolizidines IU515 - inhibitors IU515 - inhibitory activity II/517 - intestinal maltase IU517 - known number I1723 - mannosidase I IU517 - nortropanes IU5 I5 - piperidines I1/51S - pyrrolidines IV515 - pyrrolizidines IU515 - sucrase IU517 - trehalase IU517 glycoside - bond formation U5 - - through anomeric oxygen-exchange reactions

1/5 - - through retention of the anomeric oxygen I15 - cluster W549 - preparation I/197 - - via enol ethers W197 C-glycosides U5 1 N-glycosides 1/51 glycosidic bond formation W408 - control of stereochemistry U408 glycosidic mechanism U428 ff - Lemieux’s glycosidic mechanism I/428 glycosignaling domain (GSD) IU772ff glycosolic stereocontrol V408 - axial glycoside formation U408 - axial glycoside formationby SN’ displacement U408 - equatorial glycoside formation U408 - P-mannosylation by intramolecular aglycone delivery U408 - P-mannosylation by means of insoluble silver salts I/408 - stereochemistry I/408 - 1,2-trans glycoside formationby means of a C-2 participating group U408 glycosphingolipids (GSLs) U305,11/329, IU455 ff, IU761,IU810,IU1014,IV1016 f - functional domains I1/8 10 - synthesis U305 - detergent insolubility IU761 - GPI IW761 - membrane domains IW761 glycosphingolipidoses

I 17

- substrate depravation therapy II/953 a-glycosylation W355 gl ycosyl - donors with a C-2heteroatom U386 ff - - 2,6-anhydro-2-thio-glycosyl donors 11388 - - 2-bromo-2-deoxyglycosyl bromides V386 - - 2-deoxy-2-(thiophenyl)-glycosylfluorides

11387 esters I/216 - - precursors for other glycosyl donors Y216 - phosphatidylinositols (GPIs) IU417,IY757 - phosphines V171 - - formation using diazirines U17 1 - phosphorodithioates U233 - - preparation V233 - phosphoroselenoates V233 - - preparation V233 - phosphorothioates U233 - - preparation U233 - transfer Y384 - - via cycloaddition I/384 - transferases IU382,IU385 ff - - GacA transferases IU387 - - GalNAc transferases IU388 - - Gal transferases IU386 - - GlcA transferases IU389 - - GlcAT-P IU387 glycosylation If724 - basic mechanisms If724 - N-linked I11840f, IU844 - synthesis vs. hydrolysis U724 N-glycosylation IV45,IW134ff, 111315,US93 - gp 120,HIV-I IW851 - HIV-1 IU851, IU854 - host-cell dependent IU853,IU854 - inhibitors, antiviral effects IV852 - reactions in the rough endoplasmic reticulum -

V593 site II/851,11/855 0-glycosylation 111273,IW315,IU671,I11676 f - chain extension IU671 - chain termination IU671 - changes in cancer IU677 - core 1 IU671 - core 2 IU671, IW676 - terminal sugars IU671 - inhibitors IU316 1,2 cis glycosyl donors I1108 a-glycosyltransferase IUl33 P-glycosyltransferase IV133 glycosylasparaginase IU475 ff, IU479 - knock-out mouse IW949 - lysosomal disorder IU954 glycosylated natural products 1/61 - synthesis U61 glycosylation U443,I11912ff, IU1029 - N-acetylglucosamine IU912 -

- acetylglucosaminoside IV9 12 - P-N-acetylglucosaminoside IV9 12 - amine IV912 - galactose IV9 12 - P-D-galactose-BSA - P-galactoside 111911 f - P-galactoside-cl2,6-sialyltransferase 11/913 - ganglioside sialidase mediated GM(1) IU916 - heterogenity IV859 - homeostasis IU985 - lactose-BSA I11912 - 2,8-linked sialic acid units IV913 - Winked oligosaccharides IV9 14 - long-chain polysialic acid (PSA) W913 - mannosides IV912 - methods V178, V196 ff - - conventional Y178 - - endo-enol ethers 11198, I1201 - - endo-glycals V202 - -esters V216 - - exo-glycals V204 - - heterogeneous acid catalysis U196 - - I-hydroxy sugars U209 - - iodoetherification reaction I1201 - - isopropenyl glycosides U207 - - mild activation of 0-glycosyl N-ally1 carbamates with soft electrophiles V223 - - n e w V178 - - NIS-mediated I/204 - - orthoesters V223 - - oxazolines V223 - - phosphorus U229 - - sugar carbonates U22 1 - - sulfur derivatives Y229 - - use of ortho esters as intermediates V196 - - vinyl glycosides V206 -NOC-3 IU913 -NOC-4 IV913 - phenotype IV861 - product determining step U443 - rate-limiting step V443 - sites 111316,111851,111855, IU860 - strategy I/4 11 - - stepwise synthesis V411 - with a 2-thiophenyl-a-D-glucopyranosyl donor U381 - - stereochemistry of I1381 - with thioglycoside donors 1197 ff glycosylidene diazirines V155 ff - addition to aldehydes U170 - course of the glycosidation I/158 - effect of intermolecular hydrogen-bonding from the diazirine to the acceptor I/166 - glycosidation U158 - glycosidation of anomeric N-phthaloylated alosamine U166 - glycosidation of diols V164

- glycosidation of fluorinated alcohols

V163 - glycosidation of monovalent alcohols U163 - glycosidation of phenols Y162 - glycosidation of strongly acidic hydroxy compounds U162 - glycosidation of triols V164 - glycosidation of weakly acidic hydroxy compounds U163 -ketones V170 - precursors of glycosylidene carbenes V155 - stability V158 - synthesis V155 - synthesis of spirocyclopropanes 11168 glycosylphosphatidylinositol (GPI) IV130, IV135 f, 1U138, IU872 - anchored IV872 glycosyltransferase I1589 ff, V647 ff, U705, 11/20, IV29 f, IV264, IY266, IV314, 111382, 11/67], 111676 f, I11914 f, IV977, IV1036 - P-N-acetylglucosaminyltransferase IV IV9 15 - biological role 1/589 ff - families I1608 ff, IU267 - P-galactosides IV915 - Gal transferases IU382 - galactosyltransferase II/9 15 - GleA transferase 111382 - Golgi apparatus IU914 - inhibitors II/293 ff - - N-acetvlglucosaminvltransferases IV305 - - blood group A and B glycosyltransferase 111306 - - design II/294 - - donor analogs IV296 - - cll,2-fucosyltransferase IU300 - - al,3-fucosyltransferase 111300 - - al,3-galactosyltransferase W297 f - - - acceptor analogs IV298 - - PI ,4-galactosyltransferase I11296 f - - - acceptor mapping W297 - - natural I11293 - in solid-phase synthesis U705 - Kyl transferase 111382 - polypeptide glycosyltransferases IV67 1 - polysialyltransferases (PST) IY914 f - sialyltransferase (STX) IU915 a2,3 sialyltransferase 111676 -C2GnT IU676 - core 2Pl,6GlcNAc T I11676 ST3Gal I IU677 glypians Ill706 f -dally IV707 glycosylphosphatidyinositol IV706 - glypican-3 IIU07 - HS binding proteins b IV706 - ordered microdomains IV706 -rafts IU706 Simpson-Golabi-Behmel Syndrome IV707 I

~

-

~

-

Index

wingless W707 X-linked overgrowth IV707 GMI IV1016f GMI ganglioside - lysosomal digestive pathway II/952 G M Igangliosidosis - animal model lY948, W953 G Mganglioside ~ GM2 gangliosidoses IV462 -

-

GM2

- storage in HexA deficient mice II/952 GM2-activator W458 ff, IV462 GM3-dependent cell adhesion W778 GnT I-null mouse 111154 GnT-I1 111962 f GnT-I1 W963 f Golgi IV323, II/376, W378 f, IV383, IV385, W386 ff, II/389 ff, IV413, I1/962, IV964, W977 - apparatus IV146, II/274,11/855 - biosynthetic pathways II/925 - compartment 11/23I , W234 ff, 111925 - glycosyltransferases V597 - localization II/l 50 - membrane II/3 15 - sulfotransferases II/245 - - carbohydrates W245 - - tyrosine residues W245 - - tyrosine sulfation IV246 Gp 120 - binding W857 - H IV- 1 11/85I , 111852, II/855, II/860, II/86l, W862 - - V3 loop II/855,II/860 - - oligosaccharides 11/85I ff - - shedding W853 Gp 160, HIV- 1 IV852, W856 Gp 4 1, HIV- I II/852, II/855 GPlbcl 111246 GPI see also glycosylphosphatidylinositol II/417 ff, 111757 ff - acylated inositol II/420 - I-alkyl-2-acyl glycerol IV419 - anchored glycoproteins IU1033 - Anchors II/4 17 - biosynthesis W427, IU759 - - drug development IV427 - - - Plasmodium falciparum W427 - - regulation W427 - caveolae II/764, IU766 - ceramide II/4 1 9 - detergent insolubility W761 f, IV764 f - diacylglycerol IV419 - disease 111759 - dimyristoyl glycerol IV419 - endocytic pathways 111760 f -fyn IV762f - glycosphingolipis (GSLs) IV761

I 19

- - detergent insolubility IV761 - - membrane domains IV761 - inositol phosphoglycan (IPG) 111766 - insulin IU766 f - leishmania W758 - lyso-glycerolipid IV419 - mannosylation W422 - - Dol-P-Man W422 - - L- major IV422 - - PIG-B W422 - - substrate channeling W422 - - T. bmcei IV422 - membrane anchors IV1029 - membrane domains I1/761 ff - membrane release IV765 f - parasite coasts II/758 - paroxysmal nocturnal hemoglobinuria (PNH) IV757, IV7.59 - phospholipases IV765 ff - precursors IU420 - - glycolipid A IV420 - - synthesis W419 f - - - cell free systems IV420 - - - ER I1/419 - - - human IV419 - - - mutants I1/420 - - - T. brucei IV419 - - - yeast 11/419 - prion protein IU757, IU762 - second messenger 111766 - secretory pathway IV760 f - signaling IV764 ff - T. brucei IU758 - - African sleeping sickness IV758 - - trypanosomiasis W758 - - variant surface glycoprotein (VSG) W758 - T. cruzi I1/758 - trafficking W760 - transport 111760 - trypanosoma brucei W757 - trypanosoma cruzi W765 - yeast Ill758 ff - - cell wall W758 f - - Gas 1 p W759 - - Gcelp IV759 GQlb V313f - construction of oligosaccharide V314 - preparation of building block 1/3 13 - retrosynthetic analysis I/313 growth factor IU370, IV732 f, IV736,11/739 - bFGF IU370 - EGF 111370, IV732 f, -FGF-2 IV739 - PDGF-BB IV370 - receptors IU774, IV776 - TGFP II/734 - TGF-PI II/370

I2 0

Index

GSLs IU762 - clusters IU772 - detergent insolubility IU762, IU764 - -GSL interaction IU773 - membrane domains IV762 ff GTlb IU1016 ff, IV1022 guanosine diphosphate mannose IU131, I11133 ff, I11137 ff - and cell cycle IUI 39 f -transporter IU139 H-1 IU968, IU971 HA I11363 - concentrations IU363 - invertebrates IU363 - network IU363 - physiological function IU363 - synthesizing enzymes (HAS) IU365 Haemophilus influenzae IU437 2-halo-3-P-substituted sialic acid derivatives I1347 - sialy donors 11347 Hansch parameters I1874 HbsAg see hepatitis surface antigen HBV see hepatitis B virus heamagglutinin (Hag) IU839 ff HEF IJB39 - esterase domain IU844 Helicobacter pylori II/967 ff - antiadhesion therapy IU967 - antibiotics IU967 - atrophic gastritis IU967 - carbohydrate receptor analogs IU967 - duodenal ulcer I11967 - gastric cancer I11967 - gastritis IU967 - inflammnation IU967 - peptic ulcer IU967 - therapy I11967 - vaccines IU967 helminthic parasites I11879 helminths IU868 hemagglutinin IU23.5, IU486 - neuraminidase I11487 hemorrhagic fever IU832 HEMPAS (Hereditary Erythroblastic Multinuclearity with Positive Acidified Serum lysis test 111959, IU963 - HEMPAS disease IU89 heparan N sulfatase - in mucopolysaccharidoses W949 heparan sulfate 111395, IU701, I11969 ff - biosynthesis IU395 ff, IU701 f - - highly sulfated domains, HSDs IU702 - - macroscopic structure W702 - - unmodified domains, UMDs I11702

- - galactosyltransferases IU396 - - GlcA C5-epimerase IV398 ff - - GlcAlGlcNAc copolymerase IU397 - - GlcNAc (HexNAc) transferase 111397 - - GlcNAc N-deacetylase1N-sulfotransferase (NDST) IU398 ff - - glucuronyl transferase I IU397 - - linkage region IU396 ff - - regulation IU402 ff - - 2-0-sulfotransferase IU398 ff - - 3-0-sulfotransferase IU398 ff - - 6-0-sulfotransferase IU398 ff - - xylosyltransferase IV396 - functions, physiological IU396 - interactions IU395 ff - - antithrombin-binding pentasaccharide IU402 - - facilitated diffusion IU401 - - fibroblast growth factor I11403 - - platelet-derived growth factor I11403 - - specificity IV401 - lysosomal digestive pathway W947 ff - proteoglycans IV248, IV396, IU750 - - blood coagulation components IU248 - - cell-cell adhesion molecules IU248 - - chemokines IV248 - - cytokines IU248 - - extracellular IU248 - - growth factor IU248 - - lipid carrier molecules IU248 - - matrix proteins II/248 - - viral attachment receptors IU248 - structure IU400 ff - - anti-HS monoclonal antibodies IV400 - - domains IV400 ff - - fine structure 111401 heparin IU395 ff, IU701, IU750 - biosynthesis IU395 ff - - galactosyltransferases W396 - - GlcA CS-epimerase IU398 ff - - GlcA/GlcNAc copolymerase IV397 - - GlcNAc (HexNAc) transferase IV397 - - GlcNAc N-deacetylaselN-sulfotransferase (NDST) IU398 ff - - glucuronyl transferase I IU397 - - linkage region IU396 ff - - regulation IU402 ff - - 2-0-sulfotransferase IV398 ff - - 3-0-sulfotransferase IV398 ff - - 6-0-sulfotransferase I11398 ff - - xylosyltransferase IU396 - functions, physiological IU396 - /HS chains I11249 - - antithrombin IU249 - - FGFRl IU249 - - GlcNS 6-0-sulfate IU249 - - IduA-2-0 sulfate IU249 - - 6-0-sulfation of GlcN(Ac1S) IU249

Index

interactions 111395 ff - - antithrombin-binding pentasaccharide II/402 - - facilitated diffusion I1/401 - - fibroblast growth factor IY403 - - platelet-derived growth factor IY403 - - specificity W401 - pentasaccharide sequence U482 ff - - structure Y482 - - synthesis I/482 - proteoglycans IY396 - structure II/400 ff - - anti-HS monoclonal antibodies I11400 - - domains W400 ff - - fine structure W401 - synthesis I1482 - - antithrrombin-binding pentasaccharide sequence I/482 hepatitis B IV156 - virus 111833ff - envelope (g1yco)proteins IY833 f - - N-glycosylation I11834 - - 0-glycosylation IV834 - - L-protein II/834 - - M-protein II/834 - - S-protein I11834 - subviral particles 1V834 - virion formation II/835 - - glycosylation inhibitors II/835 hepatitis surface antigen II/833 chaperone II/835 hepatocellular carcinoma 111157 hepatoma II/ 157 heterologous (‘recombinant’) oligosaccharide production IB47 - concept U847 ff - E. coli US47 - methodology I1847 heterologous oligosaccharides U845 hetero-oligomer synthesis U418 HexA genes IY951 ff HexB genes II/951 ff hexosaminidases 111462, W469 P-hexosaminidases I11457 P-D-hexosaminidase see N-Acetyl---D-glucosaminidase HIV gpl20 W280 HIV-I 111851 - disease, pathogenesis IV862 - dissemination, spread I11862 - glycosylation variants W862 - molecular clones IV860 - infected lymphocytes II/853, IY854 - transmissibility 111853 HIV-2 IV8.53 HNK-1 II/387,1111018, IV1020, IU1022 f homogalacturonan II/788, IY795 f - cross-linked by calcium W795 -

I21

esterified In795 - localization 111795 - methyl esters IY795 - methyl esterified IV788 - rheological properties IY795 hormone non-specific P4-N-acetylgalactosaminyltransferase IU264 hormones II/370 - follicle stimulating hormone IY370 - biosynthesis of thyroid W477 ff host cell IY860, IY862 host surfaces I11968 f housekeeping gene IU151 HPAEC methods 111980 HPLC IU980 hspll0 IU905 htrB IV441 human IV419f -CD52 IY419 - paroxysmal nocturnal hemoglobinuria W420 -PIG-A IV420 Hunter disease see mucopolysaccharidoses, MSP I1 - animal model I11948 Hurler disease see mucopolysaccharidoses, MSP I - animal model IY948 hyaladherins IV687 hyaluronan IV685, IY744, IU745,11/747 - binding proteins IV687 ff - in vascular disease IY743 - receptors W687 f, 111694, 111696 - synthase II/691, W693 hyaluronate II/156, IU721 hyaluronic acid - digestive pathway in lysosomes IY947, IU950 hybrid N-glycans 111146 hybridoma-derived antibodies IY854 hydrogen bonding U871 hydrophobicity - analysis 11/25 -plot 11/25 hydroximolactones U157 - 0-Sulfonylation U157 6-hydroxyhexyl P-D-galactoside Y727 - enzymatic synthesis I/7272-D-hydroxyl fatty acids IU972 1-hydroxy sugars Y209 f - acidic activation Y210 - acidic activation with additional reagents 1/21 1 - dehydrative glycosylation 11212 -

I 03-GlcNAc-transferase 111320 I P6-GlcNAc-transferase I113 18, 111321 I antigen I11321 I,i antigens IU856 I22 IU722

I22

Index

I-cell disease - animal model IV948, IU954 iduronate-2-sulfatase - in mucopolysaccharidoses IU949 - in degradation of glycosaminoglycans IV948 ff a-L-iduronidase - in mucopolysaccharidoses IU949 - in degradation of glycosaminoglycans IY948 ff - a-lactalbumin IU266, W269 IgG glycosylation IU978 IgG sugars IU979 IL-6 IU446 1-imidazolylcarbonyl-glycosides U222 1-imidazolylthiocarbonyl-glycosidesY222 immobilized glycosyl acceptor V240 - polymer-supported oligosaccharide synthesis V240 immobilized glycosyl donors V242 - anomeric specificity U242 - for the synthesis of oligosaccharides I/242 - on polystyrene support U242 immune response IU968 - galectins IU640 immune system W678 - adaptive IU678 f - - MHC class I IV679 - - MHC class I1 IV679 - innate W678 immunogenic carbohydrates IU1043 - in insect cells II/1054, IV1056 - in plants IU1056, IU1059 immunogenicity and glycosylation of gp 120 IV856 immunoglobulins IV1036 immunoglobulin G IV977 f - superfamily IV579 - - cell recognition II/579 - - cell surface IU579 - - signal transduction IV579 immuno-suppression IU859 in vivo 111279 f inflammation 111862, IY967, II/972, IIll029 influenza IU839, IU973 - A and B viruses II/233 - C virus IU234, II/238 - - FIEF envelope glycoprotein IU238 - hemagglutinin IYlOOO -virus IU968 - - neuraminidase IV486 inhibitors of NAm IU847 inhibitors of N-linked oligosaccharide processing IU522 inhibitory potency IV553 ff innate immune system IU1030 inositol - acylation IV42 1 - - mammals W421

- - T. brucei IV421 - -yeast IV421 - glycosides V36 - - synthesis V36 - phosphoceramide IUI 30, IV137 - phosphoglycan (IPG) IV766 - - G P I IU766 - - insulin W766 - - second messenger II/766 - - signaling IU766 Insect cells - as a source for recombinant proteins IV1053 insects IU262 integrins - galectins IV639, IU640 cis interactions IU589 f - on cell surfaces IU590 - purified proteins IU590 - sialidase treatment IU590 interest cancer IV323 interleukin 1 (IL-1) IV446 intracellular proteoglycans II/703 intracellular sources of sugars -salvage IU5 - activation and interconversion of monosaccharides - - glycogen IU6 - - glucose IV7 - - glucuronic acid IU8 - - iduronic acid IU8 - - xylose IN3 - - mannose IU8 - - fucose II/9 - - galactose IUlO - - N-acetylglucosamine IVlO - - N-acetylgalactosamine 11/10 - - sialic acids IUl 1 intramolecular glycosidation reactions I/449 ff - carbon thethers U450 - definition U449 - silicon tethers U454 - tether does not participate in the reaction U459 - tethering to the leaving group I/459 - tether participation U450 invertebrate IV269 !.-inverting glucosidases I/725 iodoalkoxylation V372 IPG W767 - insulin 111767 - signaling IU767 isoantigen IU856 isoforms IU280 - overlapping substrate specificities IV283 isolation...1U516 - glycosidase inhibitor IU516 - radial chromatography IU5 16 - thin layer chromatography 11/516

-I;C-MS IU516 - Hand I3CNMR IU516 - high resolution mass spectrometry IU516 isothermal titration microcalorimetry U878 iterative glycal assembly strategy 1/62 ff jelly coat 111896 ff, IU906 juvenile chronic arthritis IU986 KDNase 111504 KDO U185 - anomeric 0-alkylation with primary triflates U184 - [3] glycosides Y199, U20l - - enol ether synthesis U198, U201 Kdo disaccharides V201 - enol ether synthesis U201 - glycosides I/199, U202, U205 - transferase IU441 keratan sulfate IU7 17 - digestive pathway in lysosomes IU947, IU950 keratin IU412, IU722 keratinocytes IU722 keratocan IU7 17 keratocytes, stromal IU720 keratosulfate IU407 2-keto-3-deoxy-nononic acid IU227 ketofuranosides I1201 - enol ether synthesis U201 KH-1 1/86 - synthesis via iterative assembly U86 kin recognition 111150 kinases II/370 - CAMP-dependent protein kinase IU370 - protein kinase C W370 kinetic analysis 11/151 Klebsiella IU439 knock-out IU7 18, IU970 Koenigs-Knorr method I/345 kojitriose U459 Krabbe disease IU466 - animal model 111948, IU953 KS IU717 ff -502 U69 - - synthesis via iterative assembly U69 KSPGs II/7 17 ff P-lactosamine linkage U272 - formation V272 a-D-lannosidase - lysosomal disorder IU948 P-D-lannosidase - lysosomal disorder IU948 a-lannosidosis - animal model IU948 P-lannosidosis

animal model IU948 lectin, C-type IU549 ff, IU597 ff - function IV598 - domain organization IV598 lectin, D. biflorus IU542 lectin, I-type IU579 L enzyme 1V3I 8 L. major IU418, I1/421 - gp63 IU418 LI IU720 lacdiNAc - analog Lewis' IU263, IU264 - synthase IV264 lactosamine II/408 lactosamine - saccharides U657 - - 1x3-fucosylation U657 - galectins IU639 lactosylceramide U183, II/464,11/969, IU972 - synthesis from acyl-protected lactose via anomeric 0-alkylation U183 laminin IU720 - galectins IU638 ff LAMPS - galectins IU639 ff laser photo chemically induced dynamic nuclear polarization (C1DNP)-NMR technique I1102.5 latent-active glycosylation U208 lateral pressure IU460 LBP IU446. IU1030 LDL IV745 LEC4 111159 LEC4A IU159 LECl4 I11161 LEC18 IU161 LECl CHO cell mutant IU152 lecloir pathway U663 - glycosyltransferases U664 - - sugar nucleotide substrates U664 lecticans IU1021 f lectins 111491, IU535 f, , IU539 ff, IU853, IU857, IV860 f, IU911 ff, IV999, IU1021 - anomeric preference IU539 - applications IU535 - apyrase IU540 - artificial neolectins II/912 - association constants IU539 - availability IV535 - carbohydrate crystal structures U994 ff - carbohydrate specificity IU536 - C o n A IU912 - crystal structure IU539 f, IU543 - endogenous lectin, called R 1 IU9 17 - galectin-1 IU916 f - galectin-3 IU916 f - galectin-4 IU916 - galectin-5 IU9 17 -

I24

Index

- galectin-6 IV915 - galectin-9 II/916 - homezygous mutant mice null for all three selectins IV917 - hydrophobic binding site 111543 f - hydrophobic sites IV540 - ligand interactions V562 -LNP IU540 - LPA IV912 - mice carrying a null mutation in the gene encoding galectin 1 or other galectins IY917 - monosaccharide specificities IV536 - multivalence II/539 - nuclear and cytoplasmic IV661 - - galectins IV66 1 - - CBP70 IV661 - pentraxias IV9 15 - PNA IV911 f - P-type IV563, IV569 - - cation-independent mannose 6-phosphate receptor (CI-MPR) IV563 ff - - cation-dependent mannose 6-phosphate receptor (CD-MPR) IV563 ff - - lysosomal enzyme recognition I1/569 f - - - cation dependence I1/569 - - - pH dependence IV569 - - - recognition of phosphodiesters IV.570 - - - recognition of diphosphorylated ohgosaccharides IV570 - quaternary structures IV541, II/543 - R C A I IV912 -SBA IV912 - selectins IV917 - E-selectin IV917 - P-selectins IV9 17 -SNA IU912 - specific carbohydrate ligands 11/911 - C-type lectins IV915 - P-type lectins IU915 - usual abbreviations IV911 - VVA II/911 -WGA IV912 leech sialidase L IV504 Legionella pneumophila II/437 Leishmania II/868, IV876 f - donovani IV875 - major IV417 - - gp63 IU417 - SP II/876 Leloir glycosyltransferases V846 letachromatic leukodystrophy - animal model II/947 f leucine zipper IV25 leukemia IV157 - cells IV318, IV323 leukocyte, extravasation IV862 leukotrienes IV446

leupeptin IV476, II/479,11/95I Lewis a IV968, IU971 Lewis b IV968, IV97 I Lewisc IV968 Lewis d I1/968 Lewis gene 11/97] Lewisx IV968 Lewisy IU968 LewisY V76, I/990, 111261 - conformations V990 - synthesis via iterative assembly 1/76 LeX lI/332,11/882, IV1030 - schistosomes IV88 1 ligand - coreceptors IV709 - - dimeric receptor state II/709 - - immobile complex IV709 - - cell-cell adhesion IV709 - - cell-ECM adhesion IV709 - - ligand activity IV709 - - microbial pathogenesis 111709 - - seven pass transmembrane receptors IV709 - - single pass transmembrane receptor tyrosine kinases IV709 - linkage I1905 - receptors I11708 - - clearance/intemalization receptors IV708 - - ligand localization IV709 linear B determinant IV321 linear polystyrene I/249 - polymer-supported synthesis U249 a-linked trisaccharide 11378 - synthesis with glycal epoxides as glycosyl donors U378 linkers 11250 ff - p-acylaminobenzyl V257 - p-alkoxybenzyl V258 - benzylphenol 1/251 - cleavable by photolysis U260 - crosslinked polystyrene resin V255 - diaryl(alky1)silyl 11258 - diethylsilyl 11258 - dioxyxylyl diether (DOX) Y252 - DOX I1251 - ethoxydimethylsilylbutylcarbonamide V257 - fluorene V251 - functionalized polystyrene cross-linked with divinylbenzene V255 - in two-phase (solid state) oligosaccharide synthesis U257 - 4-mercaptophenol I/257 - 4-mercaptophenol-acetamide V257 - mercaptopropanol I/258 - mercaptopropyl 11257 - mercaptothiopropanol U258 - nitrobenylphenolbenzoate V258 - nitrobenzyl V257

Index

- one-phase (solution) oligosaccharide synthesis 1/251 - p-oxybenzyl 1/258 - pentenyl I/257 - phenylaccetamide V25 1 - uhotochemicallv removable 11260 - pore glass (CPC) V255- dialkyl-or diary-silyl U259 - Rich’s linker I1257 - succinoyl V25 1, V257 - succinoyl diester V250 - succinoyl glycine I/257 - sulfonyl 11258 - thiglycoside linkers I/259 - thio ...I/251 - thioethanol U25 1 - two-phase systems 11255 lipase - lysosomal disorder 11/954 lipid II/963 - IVA II/440 - A W435 ff - linked oligosaccharide biosynthesis IV83 - remodeling W423 f - - ceramide W424 - - distearoyl glycerol II/423 - - human CD52-1 II/424 - - L. mexicana II/423 - - much-like proteins II/424 - - myristoyl CoA W423 - - porcine membrane dipeptidase IV424 - - sn- 1-alkyl group W424 - - T. brucei II/423 - - T. cruzi IV424 - - yeast II/423 - saccharide donor IV47 - sorting based on chain length IV815 - X II/440 lipoarabinomannan IV130, IU139 lipogranulomatosis IV465 lipopolysaccharide (LPS) 11/131, IU435 ff, 1111030 - Escherichia coli IU131 - Rhizobium 11/131 lipoprotein IV969 - oxidation IV746 Liver MBP IU607 LNP IV546 - apyrase IU546 - carbohydrate binding activity IV546 - root hairs IV546 locked anomeric configuration method 11189 f - epimerization at C-2 Y189 - isomerization of acetal V190 - mannosyl stannylene acetal V190 - rhamnosyl stannylene a c e d V190 - synthesis of methyl, allyl, benzyl glycosidases

I25

I/189 L-PHA IV156 LPS see lipopolysaccharide lpxA W439 lpxB IV440 lpxC IY439 IpxD IV440 lpxH IV440 lpxK IV441 lpxP W442 ff lpxY W442 ff LSC W315 L-selectin IV613, IV620 lucopolysaccharidoses - M S P I IV949 -MSPII IU949 - MSP IIIA IV949 - MSP IIIB lU949 - MSP IIIC 111949 - MSP IIID II/948 f - MSP IVA IU949 - MSP IVB IV949 -MSPVI Ill949 - MSP VII IV948 f lumican II/7 17 f lutropin 111262 Lymnaea stagnalis IV263 f lymphocyte see also T cell lymphotropic virus 111862 - homing IV249 - Gal6-0-sulfotransferate IV249 - GlcNAc-6-0-sulfotransferate IV249 - L-selectin ligands W249 1yosomal - enzymes II/564 - - synthesis IV564 - - acquisition of Man-6-P 1v564 - hydrolases IV473 ff - membrane transporters - - for cystine - - - lysosomal disorder IV946, IV954 - - for sialic acid - - - lysosomal disorder IV954 - sialidases W486 - storage diseases see aspartylglycosaminuria, cystinosis, fucosidosis, a-mannosidosis, plysosomal storage diseases IU953 f - in Australia IY953 f lysosomes IV456 ff, I1/564 - role in degradation of cytosolic oligosaccharides II/473 ff - role in degradation of glycoproteins W473 ff Mac-2-BP - galectins W639, IV640 macrocyclic glycosides 1/42

I26

Index

macrophages IV721 - infection IU860, IV861 - receptor, endocytosis IV852, IU861 macular corneal dystrophy I117 19 malaria parasites IV238 MALDI-TOF (matrix-assisted-laser desorption iorllzation -time of flight mass spectrometry) for monitoring glycans on recombinant proteins IV1050 MALDI-TOF-MS see matrix assisted laser desorption ionization-time of flight-mass spectrometry mammalian , - development IU153 - hepatic lectin IU549 ff mammals I11262 mammary carcinoma cells IU930 a-D-Man(l-2)-D-Man U727 - enzymatic synthesis U727 mannan IV134, IU136 ff, I11858 - binding protein IU861 mannobioses I11500 mannofuranose U179 - anomeric 0-alkylation with primary triflates V179 mannopyranose I1184 f - anomeric 0-alkylation with primary triflates V184 - complex formation Ul85 - coupling reaction V185 P-D-mannopyranosylamines U337 mannose binding protein (MBP) IU603 ff mannose - ligand binding I11603 - complement activation 111604 - C-linked IV130 - 0-linked 111136 ff - MBP-associated serine proteases (MASPs) IU604 - MBP deficiency IU605 - on recombinant proteins IU1052 -phosphate IV137f, I11140 - 6-phosphate receptor IV457 - recognizing proteins, and antibodies 111857, N861 mannosidase inhibitors 1V514, IU525 ff - deoxymannojirimycin (DMJ) IU5 14, IU526 - DIM (1,4-dideoxy- 1,4-imrno-D-mannitol IU527 - 1,4-dideoxy-1,4-imino-D-rnannitol IV5 15 - kifunensine IU514, IV527 - mannonolactam amidrazone 111527 - mannostatin I11528 - mannostatin A IU514 - swainsonine I11514 IU525 mannosidase inhibitors a-mannosidase U727, IV500, I11962 a-mannosidase I1 (MII) IV963

a-mannosidase IIx (MIIx) I11963 a-316-mannosidase I1 IV148 a-D-mannosidase - a (1 + 6) specific 111477, IY479 - cystolic W480 f - lysosomal I11475 ff, IV479, I11481 P-mannosidase IV500 P-D-mannosidase - lysosomal IV475 f, IU481 mannosidosis, salla disease IV88 - animal models - - N-acetyl-P-D-glucosaminidase(a-subunit) IV948,111951 ff - - N-acetyl-P-D-glucosaminidase(P-subunit) IV948, IV95 1 ff - - N-acetygalactosamine-4-sulfatase IV948 - - N-acetylglucosamine- 1-phosphotransferase 111948, IV954 - - N-acetylglucosamine-6-sulfataseIU948 - - acid phosphatase IU948 - - arylsulfatase A IV947 f - - cahepsin K IV948 ff - - cathepsin A (protective protein) 1V946, 111948, IV950 - - cathepsin D I11948 - - a-L-fucosidase IU948 - - a-D-galactosidase IU948 - - P-D-galactosidase IV948 - - P-D-galactocerebrosidase I11948 - - 0-D-glucuronidase IV948 - - P-D-glucocerebrosidase IV948 - - a-D-glucosidase IV948 - - glycosylasparaginase IU948 - - G Mactivator ~ IV948 - - a-L-iduronidase IU948 - - a-D-mannosidase I11948 - - P-D-mannosidase IU948 - - a-neuraminidase I11948 - - sphingolipid activator protein IV948 - - sphingomyelinase IV948 - pathology I11945 ff - storage disorders I11564 - - I-cell disease (mucolipidosis 11) IV564 - - pseudo-Hurler polydystrophy (mucolipidosis 111) IU564 f P-mannosides U13, U319 f, V329 ff - alkylation of I-0-metal complexes V332 - anomeric 0-alkylation V332 - de novo syntheses V334 - direct inversion V329 - enzymatic synthesis U337 - exomannosidases I1337 - from non-carbohydrate precursors V334 - mannosyl transferases Y337 - oxidation-reduction approach V329 - radical inversion of the anomeric chirality of a-D-mannopyranosides I1333

- reductive cleavage of cyclic orthoesters V334 - stannylene acetal method V332 - 2-ulosyl donor method 1/33] - glycosylation with mannosyl donors I1320 - stereoselective synthesis I13 19 - synthesis V13 a-mannosidosis 111477, II/479 P-mannosidosis IV478 mannosylation U320 ff - donor based methods I1327 - insoluble promoters I1320 - intramolecular V324 - silver oxide activation V320 ff - with sulfonates I1322 C-mannosylation see mannose, C-linked 0-mannosylation, see mannose, 0-linked mannosysltransferase 11141m IV129 - acceptor specificity IV132 - archaeal IV139 - cell cycle I11139 - donor specificity 11/131 - dolichol phosphate mannose-utilizing II/135 f - endoplasmic reticulum II/134 ff - eubacterial I11139 - families IV133 ff - golgi 111136 ff - structural features IU131 ff - yeast IU133 ff Marburg virus I11832 f - glycoprotein IV832 f - - function I11833 - - N-glycosylation IV832 - - 0-glycosylation 111832 - - sialylation I11833 Maroteaux-Lamy disease - animal model IV948 f mass spectra I1922 - fragmentation pathways V922 mass spectrometry I11972 MBGV see Marburg virus MBL 1111030 MCD I I11719 MCD I1 II/719 mechanism of sialic acid cleavage IV845 melanoma I11238 membrane protein - type 111 11/23 menstrual cycle IV723 M enzyme 111318 metabolic engineering V858 - definition IV1043 - in animal cells IV1044 - in insect cells 1111053 - in plants IV1056 - production of more complex oligosaccharides I1858 metachromatic leukodystrophy 111460, IU465

metastatic potential IV156 methyl a-thiogentiobioside V534 methyl 4,4”-dithiomaltotrioside I1547 methyl glucopyranosides V430 - rates of hydrolysis V430 N-(0-methy1)glycolylneuraminicacid IV234, 111239 MGATl IV151 MGAT2 IV153 MHC antigens - processing I11473 f Michael addition to unsaturated acceptors V549 microbial enzymes W36.5 - glycosyltransferase IU365 - Pasteurella multocida IV366 - recombinant enzyme 111365 - Streptococcus equisimilis IV365 - Streptococcus pyogenes IU365 microbial glycosidases IV497 microcal omega titration microcalorimeter U S 8 0 microglia IV720, IV860 microheterogeneity 1/59] Micromonospora viridifaciens 11/490 MI1 IU964 MIIx I11964 mimecan IV717 f mimicry I11968 mitsunobu - glycosylation 11214 - reaction U210, I1214 model of arthritis IV983 moenomycin-type analogs V12 - synthesis V12 molecular - chaperones IV997 - dynamics simulation IV970 - dynamics simulations 119.59 - modeling IV972 - molecular diversity IV923 - - generated by protein glycosylation I11923 monoclonal antibodies IV287 monocyte IV8.56, I11860 monoglcosylated oligosaccharides IU999 mononuclear cells IU858 monosaccharide polymerization Y227 monosaccharide transporters II/4 Monte-Carlo simulations U959 Morquio A disease - MSP IVA IV949 Morquio B disease (GMIgangliosidosis) - MSP IVB IV949 mouse - hepatitis virus I11235 - hepatocarcinomas IV930 MPS IIID - animal model IV948 MPS VII

I28

Index

- animal model IV948 MS analysis V924 f - protocols for U924 - sample loading for FAB-MS analysis I1924 - sample loading for MS-MS analysis on the Q-TOF I/925 - sample loading for LC-ES-MS on the Q-TOF I/925 - sample loading for LC-ES-MS-MS on the Q-TOF V925 - sample loading for MALDI-MS analysis VY25 - sample loading for NanoES-MS analysis on the Q-TOF U925 MsbA W441 msbB II/441 MUCl V275, JU284, IV41 I ; IU669, IV672, IU675 ff, 1V723 - as an antigen 1V679 - cell-cell interactions 111678 - different glycoforms W678 - effectson behavioral properties I11677 - expression IU677 - 0-glycosylation IV675 - immune responses IU677 -key building blocks U275 - synthesis V275 - tandem repeat IU675 MUC2 II/283, 111673 MUC2, 3 etc. IV669 f MUC4 11/672,II/675 MUCSAC II/673 MUC5B IU673 mucins 111237, IV313 f, IU316,11/322 f, IY410, 111970, IU1030 - colonic W237 - human nasal - genes IU313, IV670 - like glycoproteins IU3 13 - type 0-glycosylation Ill273 ff, IU669 ff mucus IV672 - barrier IV237 Mukaiyama promotor system V73 multiple - binding specificities IV972 - sulfatase deficiency II/465 multivalent binding IV588 - clustering IVS88 -high avidity IV588 - ligands U903 - very low affinity II/588 muramic acid I1566 mycoplasma pneumoniae IV856 myelin IV1013 ff, IU1021 - associated glycoprotein 111233, IV1017, IU1021 f myeloid enzyme: FucT IV IV205 - atherosclerosis 111206

- lung tumors W206 NAD(P)H cytochrome b5 reductase 11/231 NANA W485 nanomelia II/384 NAP IV969f nascent polypeptides IU46 ff naturally occuring alkaloids 111515 - indolizidines II/5 15 - nortropanes II/515 - piperidines IU5 15 - pyrrolidine alkaloids II/5 15 - pyrrolizidines II/5 15 NB-DNJ see N-butyldeoxynojirimycin NCAM II/1020 Neisseria meningitidis II/439 neoglycoproteins IU553 nerve growth cone guidance IU719 ff nervous system 1111013,11/1015,IV1019 Neu5Ac II/485 - de-N-acetylase IU230 neural cell adhesion molecule (NCAM) IV1020 neuraminic acid IU227 ff neuraminidases (Nam) 111485, IU839 - lysosomal IV476 f a-neuraminidase - in galactosialidosis IU946, IU948,1V950 - degradaton of Gh.12 by IV9S2 neurite IU720 ff - outgrowth IU720 neurocan IU377, W380 neuronal IU7 19 ff neuronal enzyme: FucT IX IV206 - neurons I11206 neutrophil- activating protein, NAP II/969 f - binding protein IV972 neutrophils IV972 Newcastle disease virus IV487 - sialidase IV503 NG2 IV377, IV381 Nicholas reaction V219 Niemann-Pick disease IV465 - animal model IV948, IV952 nod factors I/467 ff, IV546 - Beau synthesis I/471 - Hui synthesis V477 - metabolic pathway I1854 - precursors V854 - Nicolaou synthesis U468 - Ogawa snthesis U475 - structure V468 - syntheses I/467 nogalamycin V1110 nonhydroxyl fatty acids II/972 non-natural CMP-’Sia’ derivatives V638

Index

northern analysis IU287 NPD-sugars V666 - recycling system U666 NPG see pentenyl glycosides nuclear factor-KB (NF-KB) IV446 nucleoside diphosphatase IV21 nucleotide - a-diphosphate sugars V628 - activated donor sugars V627 - sugar - - CMP-sialic acid IU21 - - GDP-fucose IU22, IV29 - - synthesis IU21 - - UDP-galactose IV21 - - UDP-GlcNAc IV22 - - transporter 11/21 f, 11/24 - - - cell mutants 11/19 - - - cloning 11/22 f, II/25,1V28 - - - CMP-sialic acid IY19,11/26,11/29 - - - defects IV29 - - - deudrogram 11/24 - - - function 11/20 - - - mechanism IV20 - - - membrane topology IV19, IV25 f, IV30 - - - molecular defects 11/28 - - - structure IU25 - - - subcellular distribution 11/27 - - - UDP-galactose 11/19 - - - UDP-galactose transporter IY28 oligomannose N-glycans W145 oligosaccharides Y202, V239, U310,1/706,1/715, I/723, I/736 ff, V915,1/947 ff, - accessibility IV861 - amino-functionalized - backbone, type 1 and type 2 II/858 - biosynthesis V589 ff - - folding and quality control US93 - chains 11594 - - N-linked I/606 - - - invariable core I/606 - - - protein-specific processing U606 - - - site-specific processing U606 - clustering II/861 - complex type II/851, IU853, II/854, IU857 - conformation U988 - - packing features I1988 - conformational analysis in solution by NMR 11947 - conformational parameters U948 - 13C- 13C coupling-constants U9.53 - display IV860, IU861 - dynamics 11958 - - analysis in solution by NMR I/958 - electrophoresis IV992 - elucidention of conformation by NMR I/947

I29

enzymatic synthesis on insoluble supports Y707 ff - - on soluble supports V715 ff - - with endo-glycosidases (table) Y791 ff - - with em-glycosidases (table) I/736 ff - glycosidase-catalysed snthesis V723 - high mannose IU851, IU852, IU853, IU854, 111857, IU861, IV862 - in plant defense and cell signalling IV801 f - - chitin IV802 - - defense response IV802 - - developmental regulation IU802 - - elicitors IU802 - - endoglycanases W802 - - formation of IU802 - - glucan IV802 - - growth regulation II/802 - - inhibitor proteins W802 - - oligogalacturonides II/802 - - receptors IU802 - into glycolipids I/310 - ligand IU861, IV862 - lipid-linked IV134 ff - of gp 120 IU8S 1, IV852, IU853 - of gp160 IV856 - phenotype W861 - polymer-supported synthesis V239 - presentation 111861 - processing W147 - - N-linked II/5 I9 ff - - - endo a1,2-mannosidase IV.520 - - - glucosidase I IV.5 19 - - - glucosidase I1 IY519 - - - al,2-mannosidase IV5 19 - - - mannosidase IA IV520 - - - mannosidase IB IV520 - - - mannosidase I1 111.520 - signaling IV545 - solid-phase synthesis with glycosyltransferases I1706 - solution conformations U947 - structural analysis I/915 - synthesis I/407 ff, I/412 - - chemoselective glycosylations V407 - - efficiency U408 - - efficient chemical synthesis V407 - - iterative reactions I/409 - - orthogonal coupling concept V4 12 - - orthogonal strategy I/407 - - polymer-supported synthesis U407 - - strategic aspects V408 - - with endo-glycals V202 - transformation V3 10 oligosaccharides, N-linked 111854, IV856 - on gp 120 111855 - synthesis V609 - - major pathways U609 -

- -

I30

Index

C-oligosaccharides U495 - anionic pathways U496 ff - by applying a cycloaddition reaction V496 ff - by direct coupling I/496 ff - by the de novo synthesis U496 ff - radical pathways U496 ff - synthesis U4Y5 ff - synthetic approaches U496 ff C-oligosaccharides, synthesis of I/496 ff - cycloaddition and rearrangement I1527 - de Novo approach V5 18 - from smaller building blocks I/5 I8 - intermolecular anomeric radical addition U511, U513 - via C1-glycal carbanions U500 - via CS-alkynyl anions U496 - with anomeric samarium species US02 - with C-branched carbanions I/506 - with C6-phosphoranes U508 - Wittig olefination U508 S-oligosaccharides U531, U560 f - a-glucan-active enzymes U560 - P-glucan-active enzymes US60 - enzyme-substrate interactions U560 - synthesis US31 ff S-oligosaccharides, synthesis of U548 - chemoenzymatic methods V548 - solid-support synthesis U550 oligosaccharyl - dolichol 11/37, IU43 - - catabolism IV43 - transferase IV37, IU147 N-oligosaccharyltransferase IV45 oligosialyl chains IU239 oligosialylglycosides U353 - synthesis U353 olivomycin A U391 - synthesis V391 OMIM (online Mendalian Inheritance in Man) 111945 f oncogenesis 1111034 opaque, corneas IU7 19 ordered sequential Bi-Bi mechanism IU15 1 organ abscission IV794, IV801 - wall loosening IU801 organ transplantation IV469 orosomucoid see at-acid gl ycoprotein orthoester rearrangement U225 orthogonal - chain elongation of homo-oligosaccharides U414 - coupling U418 - glycosylation U410 ff - - current concepts U410 - - solid-phase oligosaccharide synthesis U414 - - strategy V414 - oligosaccharide synthesis U420

- - polymer-supported U420 - protecting groups M I 3 orthogonality I/4 13 - definition U413 orthomyxoviridae IU839 orthomyxoviruses IU486 OST complex I115 1 OSTl 11/55 ff OST2 11/57 ff OST3 11/55 ff OST4 IV56ff OST5 11/56 ff OST6 11/55 ff osteoadherin IU40Y, IU7 17 osteoclasts - role in pycnodysostosis 11/951 osteoglycin I117 17 osteoporosis - anti-cathepsin K drugs IU95 1 outgrowth, neurite II/720 overexpression IU156 oxonium ion U43 1 - torsional strain V431 PAMPs IU678 Panstrongylus IV873 Pantibody IU1057 f PAPS IU245 f, IU254 f, IU7 19 - synthetase polypeptide IV246 - translocase IU246 parainfluenza IU487 paramyxoviridae IU839 ff paramyxovimses II/486 parasites 111867 paromomycin U1117 paroxysmal nocturnal hemoglobinuria (PNH) II/757, IU75Y - GPI IU757, IU7.59 - PIG-A gene IU7.59 partially protected nucleophiles U187 - anomeric 0-alkylation U187 pathogen-associated molecular patterns 1u678 pea lectin IV536 pectin interactions IU794 f - covalently cross-linked IU795 pentasacchride core I1/145 n-pentenyl glycosides (NPG) V135 ff - as protecting groups U146 -chemistry of U137 - glycosylation I/135 - orthogonal protecting groups U148 - relative reactivities I/138 - solid-phase iterative couple-deprotect-couple strategy U146 n-pentenyl orthoesters U141, U144 - latent C2 esters V144

peptic ulcer II/967 peptide - antigens II/314 - N-gl ycosidase - - degradation of polymannose oligosaccharides inER IV480f - N-glycanase F IU506 pericellular matrices IU693 pericellular matrix II/690 f, IU694 perlecan IV710 ff - Arg-Gly-Asp motif 11/711 - chondrogenesis IU7 11 - Englebreth-Holm-Swarm murine tumor II/7 10 - Leu-Arg-Glu (LRE) cell adhesive motif I117 1 1 - rotary shadowing electron microscopy IU7 10 - SEA region 11/711 - sperm protein, enterokmase, agrin 11/711 - /syndecans (HSPGs) II/749 P-glycosides I15 I phagocytosis IU230, II/970, Ill972 - erythrocytes I11230 - lymphocytes II/230 - thrombocytes IV230 phase partitioning I11083 3 '-phosphoadenosine-5'-phosphosulfate II/7 1 9 phosphocan IV410, IU412 phosphokinase U638 phosphomannoisomerase (PMI) IV96 1 phosphomannornutase (PMM) I11960 f phosphomannose see mannose phosphate phosphoric acids 1/44 - glycosylation 1/44 photoaffinity labeling IU457, IU557 ff phthalic acid-tethered glycosidation 1/463 phylogenetic tree IV277 physiological role IU552 phytosphingosine IU972 pig, organ transplantation IU859 pigxenotransplants IV156 pituitary hormones IU264 plants - as a source for recombinant proteins IV1056 - cell expansion IU800 f - - cell elongation II/801 - - cellulose microfibri!s 11/801 - - environmental stimuli W800 - - hormones IU800 - - morphogenesis and differentiation 11/80 - - organogenesis II/801 - - rate of IV801 - - regulation IV800 - cell walls Y1066, 111786, IU791 - - angiosperms IU786 - - arabinoxylan IU786 - - biosynthesis IY800 - - - cell expansion IV800 - - - differentiation IU800

- - cellulose IU786 - - dicotyledons IV786 - - diversity IU786 - - glycoproteins IV796 f - - - arabinogalactan proteins 111796 - - - cross-linking IU796 - - - extensin W796 - - - GPI membrane anchors IU796 - - - hydroxyproline-rich glycoproteins IV796 - - - glycine-rich proteins IU796 - - - load-bearing structure W799 - - - proline-rich proteins IU796 - - - viscoelastic deformation 111799 - - - wall rigidification IU796 - - - wall stress relaxation 111799 - - - wall synthesis IU799 - - - water potential IU799 - - homogalacturonan IU786 - - monocotyledons IV786 - - rhamnogalacturonan I IV786 - - rhamnogalacturonan I1 111786 - - structural components IV786 - - type I IU786 - - type I1 I11786 - - xyloglucan IV786 - enzymes needed to improve recombinant protein production IV1048 - kingdom IV515 - - glycosidase inhibitors IV515 - primary cell walls II/798 - - function II/798 - - metabolism IV798 - - structural support IU798 - - turgor pressure II/798 - - cell expansion 111798 plant lectins, role II/543 f - cytokinins IV545 - hormone binding site II/546 - oligsaccharide signaling I11546 - rhizobium-legume symbiosis IU546 - plant defense IV544 plasma lipids IV744 plasmodium IU868 - genus II/868 P-loop motif IU254 PLP (proteolipid protein) 111774 pluramycins I/1111 PMI II/961 PMM IU960f PMMl IU960 PMM2 IU960 Pneumocystis carinii IU879 PNGase - degradation of polymannose oligosaccharides in ER W480 f PNH IY760 - PIG-A gene

I32

Index

polyacrylamide gel U707 - amino-functionalized I1707 - water-compatible v707 polylactosamine backbones, I-type I1653 ff - central branching I/654 - distal branching V653 - P4-galactosylation I1656 polyadenylation IU7 18 polycyclic glycosides V42 polyethylene glycol polyacrylamide (PEGA) V715 polyethyleneglyco1,monomethylether (MPEG) I1248 - polymer-supported synthesis V248 polyglycosylceramides 1U970 polyhydroxylated alkaloids IU5 13 - nitrogen, in the ring IYS13 - glycosidase, inhibitors of IUS13 - imino-sugars, alkaloids IU513 polyisoprenoid see polyprenol polylactosamines IV407, IY963 - backbones biosynthesis Y651 - - enzymatic in vitro synthesis V65 1 - galectins IV635, IU641 polylactosaminoglycans U603, IV265, IV.507 -chains Y605 - - specific modification U605 polylactosediaminoglycans IU265, IV268 polymannose oligosaccharides - degradative pathway IU473, 111479 ff polymer-supported - glycosylations 1148 - synthesis I1239 - - of oligosaccharides U24 1, V246 ff, I1260 - - - automation I/246 - - - cornbinatorial libraries V261 - - - examples of syntheses V260 - - - linkers U250 - - - one-phase V24 1, I1247 - - - polymer supports V246 - - - two-phase V241 poly-N-acetyllactosamine 111320, IU853 polyoma virus Ill159 polyoxin J V567 polypeptide - N-acetylgalactosaminyltransferasefamily U6 14 - bound oligosaccharide I1592 - a-GalNAc-transferases IU3 I5 - GalNAc-transferase IV273 - GalNAc-transferase IV273 ff - GalNAc-wansferase gene family IV275 - - GalNAc-TI IU276 f - - GalNAc-T2 IU277 - - GalNAc-T3 IU277 - - GalNAc-T4 IV277 - - GalNAc-T6 IU277 - - GalNAc-T8 II/277

- - kinetic properties IU278 polyprenols 11147 - reductase IV38 - chain length IV132 - pyrophosphate IU139 polysaccharide Ill785 - fractionation IV785 polysialic acid IU23, IY721, 1111020 polysialo ganglio-series gangliosides 113 11 - synthesis U3 11 polyvalent interaction Y1070 pompe disease - animal model IU948, IV954 posttranslational modification IU466 precipitin assays 1/877 predictive value of in vitro 0-glycosylation 1v288 pregnancy IV980 cis-prenyltransferase IU38 primary cell walls IU793, IU797, IU784 f, IU803 - biotechnology IU803 - Fourier transform infra-red (FTIR) spectroscopy IU797 - fractionation IV785 - gfycoprotein lU785 - hemicelluloses IV785 - heterogeneity IU797 - immunocytochemical methods IV797 - interacting networks IU793 - intercellular transport and storage 1y803 - isolation I11784 f - fruit ripening I11797 - organization IU793 - pectic polysaccharides IU785 - plasmodesrnata I11803 - pore size I11803 - purification IU784 - structural domains I1/797 - structural networks 111797 primary walls I11791 - apiogalacturonans IU791 - callose IU791 - coumaric IU791 - galactoglucomannans II/791 - P-D-glucans IU791 - ferulic IU791 - silicon IY791 - spezialized tissues IV791 - xylogalacturonans IU79 1 primates, New world, Old World IU857, IU859 prion protein IU757, 111762 -GPI IU757 processing of N-glycans IU146 progeroid syndrome IY386 promotor region 11/37] - transcription factors IU37 1 prosaposin I11458 proteases

Index

cathepsin A lysosomal disorder IV946,lV948,111950, IU954 - cathepsin K - - lysosomal disorder IU948 ff, IV954 proteasomes - degradation of deglycosylated proteins I11480 protecting groups 11147,11427,V430 ff - at the anomeric position U195 - - transformation into a good leaving group U195 - coupling efficiency V427 - effects on reactivity V427 - electronic and torsional effects I1430 - glycosylation stereoselectivity V427 - influence on donor reactivity V431 - influence on the acceptor I1441 - neighboring-group participation I/436 - n-pentenyl-based - reactivity control in stereoselectivity U437 - steric effects on glycosylation 11443 - stereoselectivity I1436 - strategy I1427 protective protein I11467 protein - C IV262 - carbohydrate - - binding US76 - - - evaluation Y876 protein-carbohydrate interaction YS63, US87 ff - - association in aqueous solution I1864 - - cluster glycoside effect 901 - - degradation IV1002 - - dipole-dipole interactions US64 - - dipole-induced dipole interactions 1/866 - - fundamental considerations 11863 - - gas phase non covalent interactions I1864 - - glycosylation and cancer IU923 - - N-glycosylation 1/591 f - - - initial steps 11592 - - 0-glycosylation V608 - - hydrogen bonding and n-- bonding V868 - - ‘hydrophobic’ interactions V872 - - machinery for GPI addition W426 - - - GAA 1 IV426 - - - GP18 I11426 - - 0-mannosyltransferase W136 - - misfolding 111852, IV855 - - role of multivalency 901 - - thermodynamic data (table) 11888 ff - - thermodynamics V887- conformation II/861 proteoglycans IV375 ff, IV379 ff, IV703, I11717 ff, IU729,1U731 ff, IU743 f - aggrecan IV73 1 ff, IV735, W738 - attachment sequence IV703 - biglycan lU734 ff - biosynthesis IV375 f, I11378 ff - core proteins 111375 f, IU378 ff, I11383 f, I11703 -

- -

I33

- - folding 111378, I11384 - - post-translational processing II/384 - - structure IU377, IV380 ff - - sugar nucleotides IU376 - - trafficking IU384 - - translation IV376, IV380, IV383 - - translocation IU376, IU383 - distribution IU375 - decorin I11734 ff - epiphycan 111734 f - families IV380 - function IV375 f - in vascular disease 111743 - keratan sulfate I117 17 ff - linkage region 111376,111378, IU381 f, I11385 ff - - biosynthesis 111385 ff - - DS I11387 - - galactosylation IV386 - - phosphorylation 111381. I11386 - - structure I11381 - - sulfation IV387 - - xybsylation 111378 - - xylosylation 111381, IU385 - part-time 111376, 11/381 - small leucine-rich repeat proteoglycan (SLRP) IV380 - structure I11375 f, I11379 - versican I11731 ff, 111735, IV739 proteolipid protein (PLP) I11775 proteoliposome IV21 f Proteus IV439 protioglycosylation I1369 protozoans I11868 P-selectin IU246, IV613 f, IV619 f, I111030 - glycoprotein ligand- 1 (PSGL- 1) U267 - - synthesis V267 Pseudomonas aureginosa I11439 PSGL- 1 IU246, IV6 16, IV670, IY677 psicosides V201 - enol ether synthesis U201 psychosine IU467 Pulmonary surfactant protein A (SP-A) IV608 Pulmonary surfactant protein D (SP-D) I11608 pycnodysostosis - animal model 111945, IV948 ff, 111954 2-pyridincarboxyl-glycosides I1222 pyrophosphatase V638 --pyrophosphate sugars V626 pyruvate I11229 quality control IV1002 of glycoproteins - - effect of castanospermine IU75 - - role of calreticulin and calnexin as lectin-like chaperones 11/69, IU75 ff - - role of endomannosidase IU75 ff -

I34

Index

- - role of glucosidases I and I1 IU75 ff - - role of monoglucosylated N-linked oligosaccharides 11/75 ff - role of UDP-Glyglycoprotein glucosyltransferase IV76 - schematic representation IU76 R-3-hydroxymyristate IV437 radical substitution 1/20] Raf kinase II/I 60 Rar W980 Ras IV772, IV777 RAS pathway W929 rat hepatocarcinomas II/930 rebeccamycin U69 - synthesis via iterative assembly 1/69 receptor -binding proteins IV840 - destroying activity IU839 - destroying enzymes 111235, IV237 ff recognition IV968 f recombinant - G n T I IV151 -GnTII IV1.53 - oligosaccharides 11845, I1852 - - purification I1852 - proteins see also Baculovirus, Insect cells, Plants, Chinese hamster ovary cells) IV1043, IV1044 reducing sugars Yl88 - anomeric 0-alkylation Vl88 regeneration systems V665 ff - application in carbohydrate synthesis V666 - carbohydrate-based V680 - CMP-NeuAc 1/671 - GDP-sugars U676 - UDP-galactose V666 - UDP-sugXs U669 regulation of HA synthesis IV370 regulatory elements IV285 remote activation V223 - concept U223 repertoire of GalNAc-transferases IV287 reptiles IU262 restonic lesions IV747 P-retaining glucosidases V725 retrovirusses 11/830,11/85I , II/855, 1V857,11/858, IV859 RG-I1 IV795 - 1:2 borate-diol ester W795 - cross-linked IU795 -dimer IV795 - self assembly IV795 RHAMM (Receptor for hyaluronan mediated motility) IV687 f, IU694, IU747 rhamnogalacturonan I IU786, II/788 f

- composition IU788 - ferulic acid IV789 a-L-rhamnosidase IV504 repeating disaccharide II/788 rhamnogalacturonan I1 IV786, IV789 f - backbone W790 - borate ester IU790 - composition II/789 - methyl esterified IV790 rheumatoid arthritis 111977, IV979 rheumatoid factors IU983 Rhizobium - legume symbiosis IV546 - leguminosarum IU437 - Rhizobium meliloti IY250 - - nod factors 111250, II/252 - - - lipooligosaccharide IV252 Rho 111772, W777 Rhodnius IV873 Rhodobacter capsulatus II/446 P(1-5)ribf V179 - stereoselective syntheses V179 ribfa I/179 - stereoselective syntheses II179 ribofuranose U178 - anomeric 0-alkylation with primary triflates I1178 ribophorins IV5 1 - I IU55ff - I1 11/54 ff Ricinus communis agglutinin IU980 RicR14 BHK cell mutant IV152 Rieske iron-sulfur protein I11232 RIP-motif IV490 Rnase B IU1001 rotaviruses IV238 Rous sarcoma virus IIll59

S. cerevisiae IV4 18 -Gaslp IU418 S. typhimurium sialidase IU504 saccharide-peptide hybrids (SPHs) U565, I/574 ff - biological activity V579 - conformation I/582 - definition V565 - directed synthesis V574 - solid-phase synthesis V578 Saccharomyces cerevisiae IV5 1 SA-LeX IU330 salidases IV485 salla disease IV474. II/954 Salmonella IV442 ff - typhimurium IV490 Sandhoff disease IU462, IV467, IV948, IV95 1 ff Sanfilippo A disease see MSP IIIA Sanfilippo B disease see MSP IIIB

index

Sanfilippo C disease see MSP IIIC Sanfilippo D disease see MSP IIID SAP-A 111466 SAP-B II/460, IV46.5 SAP-C IV465f SAP-D IV465 saposins IV459 SAP-precursor protein I11457 Sarcocystis spp IU879 SAT-3 IV339 saturation of IV132 Schindler disease IU954 Schistosoma IV262, IV868 - haematobium I11879 - japonicum IV879 - LDN glycans II/883 - LDNF glycans I11883 - mansoni I11263 f, II/879 sclera IU718 sea urchins IV239, IV896 f, 111899,11/904,11/9136 f secondary walls W798 - cellulose I11798 - lignin 1V798 secretory pathway IV378 ff, II/384 selectfluor V385 f - fluoroglycosylation of 3,4,6-tri-O-acetyl-D-glucal V386 selectins IV229, IV613 - E a n d P IU862 sensory nerves IU72 1 sepharose matrix V708 septic shock IU446 sequential N-glycopeptide synthesis I1299 sequon I11147 serglycin IU375, IU378 ff, I11381 f - and heparin IU704 - - anticoagulant W705 - - basophils IV704 - - low molecular weight heparin II/705 - - mast cells IV704 - - NK II/704 - - pentasaccharide sequence IV705 - - protein-free heparin IU705 - - secretory granules IV704 - - senne glycine repeats IU704 Serratia IV439 serum proteins IV960 SGGL II/1020, IV1022 f S-glycosides 1/49 shed effectors IU709 f - agonists IV709 - antagonists IV709 - HS oligosaccharides IU7 10 - tethered PG modulators I11709 shedding IV369 sialate - 4-0-acetylesterase IU230, W235

I35

- - mouse hepatitis virus IV237 ff - 9-0-acetylesterase 111230,111235,I11238 ff, I11839 - - influenza C virus II/237 ff - 0-acetylisomerase IV236 ff - 0-acetylmigrase 111235 - 7,8,9-0-acetylmigrase I11230 - 4-0-acetyltransferase II/230 - 7(9)-0-acetyltransferase IV236 ff - 7(9?)-0-acetyltransferase 111230 - 9-0-acetyltransferase 111230 - 9-0-lactyltransferase IV230 - lyasa 111229 - 8-0-methyltransferase 111230 sialic acid I/199, U566, IV227 ff, IV234 ff, 111238, 111485 f, II/583, IV721, IV839 ff, 111853, IU969 f, 111972, II/I 01 6 - 0-acetylated I1/234 - 0-acetyl migration II/235 - 0-acetylation, Golgi-membranes, bovine submandibular glands W236 - 4- and 9-0-acetylesterases I11234 - biological functions IU229 ff - biosynthesis 111227 f - diversity W227 ff - echinoderms IU228 - enol ether synthesis I1198 - evolution I1/229 - formation in liver IV1044 - from N-acetylmannosarnine 1111044 f, IU1047, I111050 - in cultured aortic cells I111044 - in human fibroblasts IUL044 - in insect cells 1054 f - in plants IU1059 - insects W229 ff - lactones II/239 -man IV228 - masking effect IV230 - 0-methyhted I11238 - microorganisms I11229 - modification W229 ff - occurence W227 ff - purification IV234 - recognition W843 - - sites I11231 ff - substituents I11233 - - biological significance I11233 - 0-sulfated I1/238 - synthetase V638 - 2,3-unsaturated IV228 sialidase IV228 f, IV235, I11238 ff, IVS03 - superfamily 111493 sialidation U345 ff - auxiliary group at C-3 Y347 - Koenigs-Knorr method V345 - mechanism I1356

I36

Index

- products V35 1 - - structures of V351 - reaction mechanism I/357 - special methods U359 ff - with phosphites I1356 - with phosphites of sialic acids U349 - with 2-thioglycosides V349 - with xanthates 11349, I1356 sialidosis - animal model IU948 - in Australian population IV954 sialoadhesins IV231, W233 ff sialoglycoconJugates 11345, IU229, IV232 - life-time IV232 sialoglycoproteins V345 sialomucin complex IV672, I11674 sialyl - donnors Y350 - glycosides V199 - 6’-sialyl-LacNAc V674 - - one-pot synthesis I/674 - Lewis antigen V296 -LewisX(sLeX)IU261, IV318, 11/321, 111323, IU614, II/977 - - bound to E-selectin VlOl I - - - conformation YlOlI - ohgosaccharide Y73 1 - - one-pot synthesis 11731 - I-6-sulfo Lewis’ II/229 - T antigen V278, II/854 - - synthesis U278 - Tn I11323 - Tn antigen U291, IV314, IV321 - transferase 1V213 ff, IU229, IV234, IU236 ff, IV322 - - classification and nomenclature IV216 - - family V610 - - a2,3-ST family IY216 - - a2.6-ST family II/2 17 - - a2,8-ST family IU2 I8 - - regulation and functionality IV219 2,3-sialyl T antigen Y293 a2,3-sialyltransferase IV304 a2,6-sialyltransferase IY302 - transition state analogs IY295 f - donor analogs IU302 a3-sialyltransferase U602, I1/322 f a6-sialyltransferase V602, 11/264,111268,IU322 f sialylation IV322 - lation and metastasis II/933 a-sialylation U272 a3-sialylation U656 siglecs IV231, IV233 ff, II/237 ff. I11580 f - CD33 II/581 - CD22 IM81 - MAG IV581 - sialoadhesin (Sn) IU581

signal - sequence for GPI addition IV426 - - hydrophobicity IV426 - - mutagenesis IV426 - - o spacer I1/426 - - w + 1 spacer W426 - - w + 2 spacer I11426 - transducers 111772 - transduction IV235, IV815 - - glycosphingolipids II/8 15 signalling W59I ff - B cell activation IY592 - CD22-deficient mice IU592 - ITIM-like motif II/592 - negative regulatory functions IV592 - SH2-containing II/591 - SHP- 1 tyrosine phosphatase IY592 - tyrosine residues IV591 silylated acceptors V216 1-0-silyl glycosides V215 - as glycosyl donor V215 simian immunodeficiency virus, SIV IV856 site of biosynthesis IV364 - eucaryotic IV364 - pericellular space 1V364 - streptococci 111364 SIV IV856 Sjogren’s syndrome IV986, IY990 skin 111734 ff, 1V739 Sly disease see MSP VII Small Leucine-Rich Proteoglycan (SLRP) IV410 SMC IV674f SMC IV677 smooth muscle cell IV746 snail I11263 f, II/264,1V266 f solid - oligosaccharides V45 - phase synthesis 1/45, V239, I/705 - S U P P O ~ ~ S1/255 - - crosslinked polystyrene resin I1255 - - functionalized polystyrene cross-linked with divinylbenzene V255 - - pore glass (CPG) V255 soybean agglutinin IV542 spatial arrangement IV556 species-specific 111897, IV899, IV904, IV906 f, IU905 a1 +2-specific fucosidases IV502 specificities IV318, IV323 spermatozoa I11239 sperm-egg - binding see also sperm-egg interaction IV904, IV907 - interaction I11895 sphingolipid activator protein I11458 sphingolipidoses W466 ff sphingolipids

index

- lysosomal digestive pathway II/951 ff sphingomyelin W465 sphingomyelinase IV465 sphingomyelin - lysosomal digestive pathway IVY52 sphingosine IV972 - derivatives and mimics V36 - - glycosylation V36 spirocyclopropanes, synthesis of V168 splicing, alternative IV7 18 Src family kinases I11772 Src kinase W160 c-Src 111772,111777 ST6Gal-I IV33Y stannanes U1 7 1 - formation using diazirines I/ 171 starfish IV896, IU8YY - Asterias rubens I11238 Stenotrophomonas maltophilia 11226 stomach pit cells IVY71 Streptococcus pneumoniae IV493 Streptomyces 1193 stress protein 111970 structural determination I1/516 - glycosidase inhibitor IU5 16 - radial chromatography IV5 16 - thin layer chromatography I115 16 - FC-MS I11516 - H and 13C NMR IU516 - high resolution mass spectrometry W516 structure-activity relationship IV518 - molecular modeling 1115 18 - molecular orbital calculations IV518 structure-based drug design 111487, 111969 STT3 11/58 STZ IV33Y subcellular Gbs trafficking I118 13 submaxillarv mucin IV674 substrate specificities IV151, 111205, IV279 f, 111280 subunit organization IV558 succinic - acid tether U461 - succinic acid-tethered U462 - - stereodifferentiation V462 sucrose synthase V642 sugar - acceptors U350 - combining site IV55Y - metabolism - - basic principles II/3 - nucleotide transporters 11/11 - nucleotides U625 - - synthesis I/625 - nucleotides V64 1 - - in situ generation U641 - nucleotides I1665

I37

- - biosynthetic pathways I1665 - nucleotides, synthesis Y629 ff - - CMP-activated sugars I/629 - - GDP-activated donors I1632 - printing 11/990 - specificities 11155I sugar nucleotides, synthesis of 11626, V635 ff - chemical synthesis I1626 - chemo-enzymatic synthesis I/635 - CMP-activated sugars I1637 - GDP-activated sugars I1639 - UDP-activated donors V626 - uridine diphosphate-activated donor sugars U635 sulfate polysaccharide I11897 sulfated - carbohydrates W247 - - monosaccharide modifications IV247 - 0-glycans I11323 sulfatide 111465, IU969, WY71 f, IV1013, IV1015 - lysosomal digestive pathway IV952 sulfatide (3’ sulfogalactosyl ceramide) IV8 13 - recognition I118 13 sulfation I11382 - linkage region II/382 - - structure II/382 ff N-sulfation I11408 sulfatransferase II/268 sulfoglucuronyl glycolipids (SGGL) II/1018 sulfonamidoglycosylation 1/68 N-sulfonylamines Vl71 -esters Y171 - formation using diazirines V17 1 sulfotransferase IV264, IV3 14,111324,1V390 f , IV4 I 3 - chondroitin 111390 f - KS IV71Y - umic acid II/391 6-sulfotransferase W3Y0 summary and future prospects IY528 - high degree of potency 111528 - new glycosidases I11528 - substrate specificity I11528 surface IgM I1/446 SVI W720 SV2 IU720 swainsonine IVY35 SWPl (suppressor of WBP) IV54 ff syncytium formation by HIV-1 II/857 syndecan 111378, IV749, IV376 f, IV38 I , IV705 f - cell scaffolding proteins IV706 - cytoplasmic domains II/706 - dominant negative inhibitors IV706 -ERM IV706 - GAG attachment I11706 -MAGUK IV706 - oligomerization domains IV706

I38

Index

- paracrine effectors IU706 - PDZ II/706 - phosphorylation II/706 - proteolytic cleavage IU706 - putative cell interaction IU706 - shedding IU706 - syntenin IU706 syndecan-1 II/382 synergistic inhibitors IV300 synthetic di- and tri-valent glycopeptides IV55.5 ff systemic lupus erythematosus IU990

T antigen (Gal-GalNAc) I/290, II/854 T. brucei IV418 ff - glycolipid A 11/420 - glycolipid C IV422 - inositol acyltransferase I1/422 - inositol deacylase IU422 - PARP II/4 18 ff - VSG 1U419,11/421 - VSG anchor IV418 T cell infections with HIV-I IU862 T cell lymphotropic virus IU861 T. cruzi II/419 T24 H-ras 1U159

T3-(3,3’,5-triiodo-L-thyronine) - formation via lysosomal degradation of thyroglobulin IU477 ff T4-(3,3’,5,5’,-tetraiodo-L-thyronine) - formation via lysosomal degradation of thyroglobulin IU477 ff tandem repeats II/670, 1U672 - sequences IU3 13 T-antigen (Gal-GalNAc) U276 - formation U276 targeting IU559 - to Golgi Apparatus IU150 TATA IV151 Tay Sachs - /Sandhoff disease IU95 1, IU952 - disease IV462, IU467 - - animal model IU948, IU95 1 ff - - mouse models IU951 ff tentaGel-N U261 - solid phase V267 - solution V267 tert-butyldimethylsilyl2-deoxyglycosides V394 tether I/449 -stable U449 - temporary U449 tetrasaccharides U972 - crystal structures U972 therapeutics IU860 therapy IV469, W967 thermodynamics of association US85 - binding site reorganization U885

- proton transfer I/885 - salt effects I1885 4-thiocellobiose U544 thiocyclodextrins V538 1,2-thiodisaccharides I/553 1,3-thiodisaccharides U55 1 1,4-thiodisaccharides 1641 thioethyl donors 1/74 - synthesis via iterative assembly 1/74 thiogalactosides I/435 - relative reactivities I/435 thioglycoses 1632 - selective S-deprotection V533 - synthesis U532 1-thioglycoses U532 2-, 3-, 4-, 5 - , or 6-thioglycoses 11532 thioglycosides U93 ff - applications I/I 10 - as acceptors Ul 10 - - thioglycosides as both donors and acceptors I/1 11 - - glycosyl phosphites V117 f - - for glycosylation V117 ff - - formation of glycosyl phosphonate V13 1 - - glycosylation mechanism V l 19 - - glycosylation of sialyl phosphites U122 - - glycosylation of C-2-acylated glycosyl phosphites U123 - - glycosylation with C-2-0-benzylated glycosyl phosphites U124 - - glycosylation with a-selectivity V124 - - glycosylation with P-selectivity V124 - - preparation U118 - - stereoselectivity in glycosylation U121 - - synthesis of CMP-NeuAc I/129 - - synthesis of GDP-fucose V129 - - transformation of glycosyl donor U131 - block syntheses V110 -donors U99 - - carbonium type VlOO - - direct activation U99 - - halonium type MOO - -heavy metal salt promotors V99 - - single-electron activation U106 - - sulfonium type UlO0 - donors with an anomeric sulfur V108 - from anomeric acetates U94 - from glycosyl halides U95 - glycosyl donor qualities U93 - intramolecular glycosidations V112 - orthogonal glycosylations U110 - protecting group manipulations 1/96 - solid phase synthesis Vl13 - synthesis V94 2-thioglycosides U349 - for sialyl glycoside syntheses V349 thio group migration U388

Index

1,I -thio linkages I1557 1,2-thio linkages 1/553 ff 1,3-thio linkages I1551 ff 1,4-thio linkages I1541 1,6-thio linkages U534 4-thiomaltose I1544 I-thio-P-D-mannopyranosidesI/336 thiooligosaccharides I1558 - conformation I1558 1,3-thiooIigosaccharides 11552 1,4-thiooligosaccharides U546 6-thiooligosaccharides I1538 6-thiosaccharides 11534 1-thiosucrose I/553 a,a-thiotrehalose I1557 threshold theory IU466 thrombomodulin I11376 ff, 11/381, W388 thyroglobulin - degradation in lysosomes 111474, W477 ff - degradation of carbohydrate chains 111478f - iodination IU477 f - proteolysis I11478 f thyroid hormones - biosynthesis 111474 - formation via lysosomal degradation of thyroglobulin I1/474, W477 ff tirosuine sulfation 111411 tissue-specific 11/718 tlc plates IV970 TLR2 I11446 TLR4 IU446 Tn antigen 1113 15 to cis to IV319 tobramycin Vl119 Toll-like receptor proteins IV446 topology IV457 - of GPI biosynthetic pathways IV424 f - - flippase IV425 - - L. major IU425 - - PIG-A I11424 - - PIG-B 111425 - - PIG-L IU424 - - Trypanosoma brucei 111425 Toulouse-Lautrec I11950 f Toxocara IV868 Toxoplasma gondii IV879 transcription - factor Spl IV151 - start sites IV161 transfer NOE experiments I11003 - study of carbohydrate-protein interactions U 1003 transfer of EtN-P IU423 - aminoethylphosphonate I11423 - PIG-F I11423 - T. brucei W423 - T. cruzi IU423

I39

transferrin 111960, IU962 f transgenic mouse W971 transglycosidation IU228 - intramolecular I11228 trans-Golgi I1597 - elongation I1597 - reaction V597 - termination reactions I1597 transition-state analogs I1/303 translocation 11142,111369 transmembrane domain IV150 transmissibility - of HIV-1 11/851 - of retroviruses 111851 transparency, corneal 111718 transplantation, organ I11859 trans-sialidases 111229, IV49 1 C-P,P-trehalose I1507 - synthesis 11507 triantennary structure 111553 ff Triatoma 111873 trichloroacetimidates VS f - formation I16 -reaction U6 - method V5, U7 - - basic principle I/5 - - glycosylation of carbohydrate acceptors 117 - - synthesis of oligosaccharides V7 Trichobilharzia ocellata W263 f triflates V541 - S~2-dispk1cement I1541 TRI-GP I11554 ff trimannan I1143 - synthesis from n-pentenyl orthoester trisaccharides I1972 - crystal structures I/972 tritylated acceptors I1224 Trypanosoma - brucei IV868, IU873 - - gambiense 111868, IV873 - - rhodesiense 111868,I11873 trypanosoma cruzi 111486, IV765, IV868, IU873 ff -GPI lU765 TSG-6 111747 tumor - associated changes IV1036 - cells I11932 - - adhesion W934 - - -endogenous lectins I11934 - - focal adhesions I11932 - - proliferation I11927 - necrosis factor-a I11446 tunicamycin IU40,11/408, II/I 000 p-turn structure 11148 ff type 1 - arabinogalactans IU791 - chains I11968

I40

Index

- collagen 111734, IV748 (Gal@1-3)GlcNAcP-R) chains I1607 - - synthesis U607 type I1 - arabinogalactans I11791 - collagen I11733 - hydroxyproline-containing proteins 111791 - (Gal(p1-4)GlcNAcp-R) chains I1607 - - synthesis I1607 - membrane protein IV315, IU320 type 111collagen 111748 type IX collagen IU377 type IX collagen 111381 tyrosine - phosphorylation IV156 - sulfation IU617 f tyrosylsulfotransferases IU246, IV252 f UDP-GIyg! ycoprotein glucosyltransferase - amino acid sequence IU122 - calcium requirement IU120 - distribution in nature IV120 - donor specificity I11120 - localization, subcellular IU120 - role in glycoprotein folding 111123 ff - reaction catalyzed IU119 - sensor of glycoprotein conformation IU121 f - sequence analysis IU122 f -size I11120

UDP-G1cNAc:N-acetylglucosamine-Iphosphotransferase in I-Cell disease I11954 UDP-GlcNAc-2 epimerase - in animal cells IV1044, W1046 UDP-glicose-4’-epimerase I/642 UDP-glucose: glycoprotein glucosyltransferase IU1003 UDP-N-acetylglucosamine IV439 UDP-N-acetylg1ucosaminyl:dolichylphospate N-acetylglucosaminyl phosphoryl transferase IU40 unprotected aldoses V187 - anomeric 0-alkylation with bromides V187 - anomeric 0-alkylation with cyclic sulfates V187 - anomeric 0-alkylation with primary triflates U1 87 -

virus shedding II/862 vaccines IU967 valency IV553 ff vascular - disease IV743 - endothelial cells IV749 - lessions W744

verotoxin (VT) mediated pathology IU811 versican (CSPGs) IU377, IU380, IU745 vertebrate synthases IV366 - HAS genes IV366 ff - HAS isoenzyme from mouse I11366 -human IV366 - structural features IU366 - DG42 protein from Xenopus laevis IU366 vibrio cholerae IU491 VIM-2 ganglioside U48.5 - total synthesis V485 ff VIP36 111545 viral - evolution I11235 - proliferation - - effect of glucosidase 1 and 1 1 inhibitors 11177 f - - role of endomannosidase pathway 1u78 - receptor destroying enzymes IU847 virion, virus particle IV859, IV862 virus IV821 f - classification IU822 - dissemination IU852 -entry IU860 - envelope IU851, I11862 - Epstein Barr IU856 - glycoproteins - - glycosylation IV853,11/859, IU862 - - macrophage-derived IU853 - inactivation 111851, IU859 - infection IV852, IU859, IV860, IV862 - infectivity 111851, IU855, IU860, IV862 - neutralization 111857, IV858 - pig, porcine IU859 - resistance IU859 - surface glycoproteins IU822 - transmissibility IU858 - transmission 111851, IU859 - tropism IU852, 111860, IV861 visceral Leishmaniasis IU87.5 vitelline coat see also vitelline envelope IU902 f vitelline envelope IY899, IU901 VMP-Neu5Ac IU229 von Willebrand factor IV246, IU279, I11673 WBPl (wheat germ agglutinin protein) IU54 ff wheat germ agglutinin IU536, IU543 Wolman disease IV954 worm development IV155

xanthan I11139 Xanthomonas manihotis IU499 xeno - antibodies W857, IV859 - antigen IU8.53 -graft IU859

Index

Xenopus see also frog - laevis 1U900 xenotransplantation IU859, II/I 030 X-ray - crystal structures I1/845 - crystalography IV1034 m-xylene-tethered glycosidation U464 xyloglucans 111787, W799 - cell expansion W799 - cellular differentiation II/799 - exoglycanases 111799 - fruit ripening IU799 - hydrolytic enzymes IIf799 - load-bearing structure W799 - poaceae II/787 - solanaceae IU787 - structural variation IU787 - transglycosylases W799

I41

P-xylosidase IU505 P-D-xylosides W386 ~l-2-xylosyltransferase II/148

yeast IU419, IU758 ff - GPI IU758, II/760 - cell wall I1/758 f - Gaslp 111759 - Gcelp IU7.59

zona pellucida (ZP) 111409,W902, II/896, IIf900, II/906